Single Cell Gene Expression in Bacillus subtilis: An Investigation into the Molecular Basis of Culture Heterogeneity by John David Chung B.S. Chemical Engineering University of California, Santa Barbara (1986) Submitted to the Department of Chemical Engineering in Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY in Chemical Engineering at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY May 1994 © Massachusetts Institute of Technology, 1994, All Rights Reserved lT-\ 4 X Signature of Author: Department of Chemical E ieering B by: a -S 1 -I Certified f Y y / - Jan A - / Gregory Stephanopoulos Professor of Chemical Engineering Thesis Supervisor ) I/, Accepted by: 25, 1994 1\' A . Robert E. Cohen Professor of Chemical Engineering Chairman, Committee for Graduate Students Science MASSACHUSETTS INSTITUTE OFTFCHNOLOGY JUN 06 1994 LIBRARIES Single Cell Gene Expression in Bacilw subtilie An Investigation into the Molecular Basis of Culture Heterogeneity by John David Chung Submitted to the Department of Chemical Engineering on January 25, 1994 in partial fulfillment for the Degree of Doctor of Philosophy in Chemical Engineering ABSTRACT An experimental and theoretical investigation was undertaken in order to better understand the molecular basis of physiologicalheterogeneity. Experimental studies were confined to differentiating cultures of Bacillus subtilis and involved the analysis of early developmental gene expression using single cell measurements and conventional genetic techniques. Quantitative studies of adaptive gene expression were performed to demonstrate the feasibility of the experimentally derived conclusions and to provide a framework in which to analyze adaptive cellular responses. It is concluded that: 1) threshold mechanisms control initiation into spore development via processes interacting with the phosphorelay (Burbulys et a. 1991),2) the intracellular concentration of phospho-SpoOAdetermines a transcriptional state of a B. subtilis cell and 3) that threshold mechanisms provide the cell with a stable means of initiating adaptive cellular responses. Experimental support for these conclusions was obtained from flow cytometry-based measurements of gene expression using lacZ fusions to loci known to be directly or indirectly activated by the phospho-SpoOAtranscription factor. Culture average measurements of phospho-SpoOA controlled gene expression were found to reflect the behavior of two cell populations. One population exhibited high expression while the other exhibited little or none. The size of the high expressing population was found to correlate with the final sporulation frequency. Mutations that reduce the intracellular level of phospho-SpoOA resulted in a redistribution of the relative sizes of these populations with fewer cells residing in the high expressing population and also a reduction in the sporulation frequency. These results indicate that the cell regulates the level of phospho-SpoOAvia processes interacting with the phosphorelay and that the level of phospho-SpoOAdefines a transcriptional state of the cell. Theoretical support for these conclusionswas obtained through the identification of a prevalent biological control structure that allows threshold mechanisms to exist. So long as the intracellular concentration of the transcriptional activator responsible for initiating adaptive gene expression is regulated by a positive feedback loop, a non-linear generation rate, of the cooperative type, and a concentration dependent removal rate, threshold mechanisms are possible. The fact that this control structure can be found in a variety of biological systems including those associated with phage development, nutritional adaptation and spore formation suggests that it represents a conserved mechanism used by cells to cope with changing environmental conditions. Thesis Supervisor: Dr. Gregory Stephanopoulos Title: Professor of Chemical Engineering 2 Acknowledgements One of the principal reasons behind coming to M.I.T. was to take advantage of the opportunities available at a premier research institution by expanding my horizons in a number of scientific disciplines with the organizing center being cell biology. The overall strength of the faculty at M.I.T. and the breadth of research being performed has not been disappointing. While such a purpose can lengthen ones tenure and is in many ways contrary to the focus needed to complete the doctoral program, such an opportunity arises only once in ones life. I leave M.I.T. having accomplished, in general, all that was sought. However, this would not have been possible without the support of a number of people. While they might all seem to contribute in different ways, I have learned that putting up with me can sometimes be an endeavor. I would like to first thank my advisor, Greg Stephanopoulos, for finding the right combination of support, guidance, freedom and of course pressure, needed to bring this work to completion. While there were undoubtedly tense moments, they can fortunately be looked upon as enriching experiences and perhaps well needed to persevere at a place like M.I.T. I am confident that this relationship has better prepared me for the real world. I am indebted to the assistance provided by the Cell Sorter Facility at M.I.T., particularly Stewart Conner, who not only was a great colleague, but also a supportive and encouraging friend particularly during the low point of my tenure. Much thanks also goes to Glen Paradis who picked up where Stewart left off and greatly assisted me during crunch time. Without the assistance of these two, this work would have never been completed. Their competence, support and friendship will not be forgotten. I would also like to thank Professor Alan Grossman for his advise and insight. If not for him, "my wheels might still be spinning." Working with him and Keith Ireton, previously one of his grad students, I have had a great opportunity to see top notch biologists in action. I would also like to thank my undergraduate UROP, Kirk Noda, for his assistance during those long experiments. He allowed me to realized how valuable the undergraduates can be to the research environment at M.I.T. Much thanks also goes to the other members of my thesis committee; Daniel Wang, Charlie Cooney, Arnold Demain, and Abraham Sonenshein. I can only hope that someday I will be able to exhibit the patience that was required of them during several of my "progress" reports. While it was often difficult to juggle between the disparate perspectives of the scientists and the engineers on my committee, valuable lessons were obtained through these interactions. In particular, all of them taught me the importance of presentation and the bottom line. Finally, I would like to thank the members of the 66-257Stephanopoulos lab, both past and present, for their support. Sung Park; thank you for your encouragement during a critical point in this work. Thanks also goes to Joe Vallino and Bob Kiss, for demonstrating that graduation was possible, and to Grace Colon, Catherine Shaw and Roy Kawamura for providing listening ears. 3 Table of Contents Abstra................................. ................... Aknowledgements ................................... 2 3 I Introduction 1.1 Present Interest in Species of Bacillus ....................... 1.2 Problems Associated with Use of Bacillus .................... 1.3Hypothesis............................... 1.4 Objective ............ ........................................... 8 9 10 1.6 Significance .......................................... 11 11 13 1.7 Thesis Organization .................................... 14 1.5 Approach ............................... 2 Background 2.1 Physiological Heterogeneity ............. .............................. 18 2.2 The Developmental Program of Spore Formation in B. subtilis .... 2.2.1 Signal Transduction via the Phosphorelay .............. 2.2.2 Transcriptional Activation via SpoOA-P ............... 2.2.3 Establishing Late Sporulation Gene Expression ......... 2.3 Mechanisms Leading to Physiological Diversity ................ 2.3.1 The Threshold Theory ............................ 2.3.2 The Continuum Theory ........................... 2.3.3 Common Ground ............................... 20 20 21 23 24 24 26 27 3 Development of Flow Cytometry-Based Techniques for Studying Gene Expression in Single Cells of a subtili 3.1 Introduction ................ 3.1.1 Flow Cytometry ....... 3.1.2 Bacterial Flow Cytometry 3.2 Problem Statement ........... 3.2.1 Signal Strength ....... 3.2.2 Bacterial Diversity ..... .......................... 36 ......................... .......................... .......................... 39 41 41 .......................... ........ ....... 43 .......................... 3.3 Approach .................. 3.4 Summary of Staining Conditions . .......................... 3.4.1 C,-FDG ............ .......................... 3.4.2 PKH26 ............. .......................... 3.5 Description of Molecular Probes . .......................... 3.5.1 f-Galactosidase Fluorogetaic Substrates ............... 3.5.2 The Second Fluorochrome : PKH26 .................. 3.6 The Internal PKH26 Negative Conitrol...................... ............................. 3.6.1 Tracking Background ............................. 4 37 44 45 46 47 48 48 50 51 51..4 51 ........ 3.6.2 Fluorescence Compensation ........................ 3.6.3 Controlled Mixtures ............................ 3.7 Staining Cells with 5-Dodecanoylaminofluorescein 53 55 di-p-D-galactopyranoside ................................ 55 3.7.1 Approach .................................... 3.7.2 Thermal Deactivation of p-galactosidase ............... 3.7.3 Constant Temperature Staining ..................... 3.7.4 The Two-Step Temperature-Time 55 Pulse Matrix ......... 3.7.5 Controlled Mixtures .............................. 56 57 58 59 3.8 Staining Cells with 5-Octanoylaminofluorescein di-f-D-galactopyranoside ................................ 3.8.1 Approach ..................................... 3.8.2 C8 -FDG is Optimal .............................. 3.8.3 Staining Time .................................. 3.8.4 Heterogeneous Expression in Stationary Phase Samples ... 3.8.5 Controlled Mixtures .............................. 61 61 61 63 63 65 3.8.6 Correlation Between Fluorescence and Average Enzyme Activity ...................................... 66 66 3.9 Discussion ........................................... 4 Characterization of Stationary Phase Cultures of R subtils 4.1 Introduction ................................... 4.2 Stationary Phase Protein Expression ................. 4.3 Single Cell Gene Expression: spoVG-lacZ ............. 4.4 spoVG-lacZ Cell Sorting .......................... 96 97 98 101 .... .tions . 4.5 Single Cell Gene Expression: spoIIA-lacZ ............. 103 4.6 Correlation Between Spores and the Expressing Cell Populiations .. 104 4.7 Discussion .................................... 106 5 Genetic Analysis of Threshold Phenomena 5.1Introduction ........ .... ............... ....... 119 ....... 120 ....... ....... ....... ................. ....... 123 123 123 124 5.4 kinA Null Mutations: Population Demographics ........ ...... 5.4.1 f-Galactosidase Expressing Populations ........ ...... 5.4.2 Spores ................................. ....... 5.5 kinA Null Mutations: Distribution of spo Gene Products .. ...... 5.5.1 Fluorescence Distributions .................. ...... 124 5.2 Stationary Phase Signal Transduction and Gene Expression 5.3 kinA Null Mutations: Culture Average Behavior ........ 5.3.1 spoVG-lacZ............................. 5.3.2 spoIIG-lacZand spoIIA-lacZ ................ 5.3.3spoIID-lacZ ..... ....... 5.5.2 spolID-JacZ Cell Sorting ................... 5.6 Observations on spoIIA-lacZ Gene Expression ......... 5.7 The Effects of spoOK Mutations .................... 5 ....... ....... ....... 124 126 127 127 128 129 129 5.8 Discussion ........................................... 131 5.8.1 Threshold Activation of Sporulation .................. 131 5.8.2 Multiple Thresholds .............................. 132 5.8.3 Implications of a Threshold Mechanism ............... 134 6 Quantitative Analysis of Threshold Phenomena 6.1 Introduction ..................... ..................... 6.2 Mathematical Conceptualization ........................... 161 163 6.3 Phage A 163 ........................................... 6.3.1 Background .................................... 163 6.3.2 Characteristics of Transcriptional Regulation ........... 6.3.3 The Rate Law for CI Generation .................... 165 166 6.3.4 The Single Cell Model ............................ 169 6.3.5 Steady State or Balanced Growth Solutions ............ 171 6.3.6 Initiating the Switch .............................. 6.3.7 The Dynamics Associated with the Transition ........... 6.3.8 Static versus Dynamic Thresholds ................... 174 175 177 6.4 The E. coli lac Operon .................................. 178 6.4.1 Background .................................... 178 6.4.2 Characteristics of the lac Operon ......... 6.4.3 Derivation of the Rate Law ........................ ........... 6.4.4 The Single Cell Model ............................ 179 180 183 6.4.5 The Steady States ............................... 6.4.6 The Hysteresis or Maintenance Phenomenon ........... 6.5 Sporulation by B. subtilis ................... ............. 6.5.1 The Requirements for Threshold Behavior ............. 6.5.2 The Threshold Requirements are Satisfied ............. 6.5.3 Threshold Model Describing Sporulation .............. 185 186 189 189 191 193 6.5.4 kinA Mutations ................................. 6.5.5 spoOK Mutations ................................ 6.6 Appendix: Fitting Experimental Observations ................. 6.6.1 Determination of A ................ .............. 196 197 199 199 6.6.2 Generation of the Hysteresis Curve .................. 199 7 Summary and Concusions 7.1 Summary ........................................... 7.1.1 Single Cell Measurements ......................... 222 222 7.1.2 Experimental Analysis of Gene Expression ............. 7.1.3 Quantitative Analysis of Gene Expression ............. 7.2 Conclusions .......................................... 8 Recommendations and Future Research Areas 8.1 Biological Studies ...................................... 8.1.1 Signals Inputs versus Signal Amplification 6 223 224 226 ............. 227 227 8.1.2 Adaptive Cellular Processes in Other Biological Systems ... 8.1.3 Biological Control Systems Theory ................... 8.2 Single Cell Measurements ................................ 8.2.1 Single Cell Analysis: Fluorescence Microscopy .......... 8.2.2 The Future: Strain Improvement via Flow Cytometry ..... 9 aterials and Methods 9.1 Strains, Maintenance and Cultivation Conditions ............... 228 228 229 229 230 231 ....................................231 9.1.1Strains .... 9.1.2 Routine Stratin Maintenance and Storage .............. 9.1.3 Cultivation 9.2 Analytical Technique s................................... 9.2.1 Specific f-G; alactosidase Activity (ONPG) ............. 9.2.2 Relative f-Gralactosidase Activity (MUG) .............. 231 .................................... 232 233 233 234 ....................................236 9.2.3 Spores .... .................................. 236 9.2.4 Cell Number rs ....................................237 9.3 Cell Staining ...... 9.3.1 PKH26 Staix iing ................................. 237 .................................... 238 9.3.2 C 12 -FDG .. .................................... 239 9.3.3 C8 -FDG .. ....................................240 9.4 Flow Cytometry ... 240 .......... 9.4.1 The Flow Cy(tometer ................... 241 9.4.2 Cell Sorting .................................... 242 9.4.3 Data Acquis ition and Analysis ...................... 9.5 Theoretical Computattions ................................ 10 References................... 243 248 7 Chapter 1: Introduction 1.1 Present Interest in Species of BaciEs Species of the genus Bacillus have a long history of industrial and academic use. For centuries, Bacillus subtilis (natto) has been used for the production of itohiki-natto, a fermented soybean foodstuff popular in Japan (Maruo and Yoshikawa 1989). Species of Bacillus have also been used for the production of a number of important products including amylases, proteases, nucleases, amino acids, nucleotides and antibiotics (Schaeffer 1969, Preist 1977, Maruo and Yoshikawa 1989). More recently, with the arrival of the biotechnology industry, the secretory apparatus of Bacillus has attracted much interest in these organisms as possible recombinant hosts. A number of heterologous proteins, both prokaryotic and eukaryotic, have been successfully expressed and secreted in species of Bacillus (Gray and Chang 1981, Palva et al. 1982, 1983, Honjo et al. 1985, 1986A, 1986B, 1987, Meyer and Fiechter 1985, Ulmanen et al 1985,Yamane et al. 1985,Nicaud et al. 1986,Schein et al. 1986, Nakayama et al. 1987, 1988, Chang 1987, Saunders et al. 1987). While Escherichia coli continues to be the most widely used recombinant host, advances in our understanding of Bacillus genetics and physiology should make this genus an attractive alternative to be considered in the future. In addition to its long time industrial use, the ability of species of Bacillus to initiate spore formation, a primitive form of cellular differentiation, has also generated a great deal of academic interest in this organism. The ease with which 8 species of Bacillus, particularly B. subtilis, can be genetically manipulated has provided scientists with an amenable system in which to study the complex processes regulating developmental gene expression. Over the past few decades, an enormous amount of research has been directed at identifying and understanding how environmental and physiological factors influence the process of spore formation. Such work has led to great insights into the molecular mechanisms responsible for initiating adaptive cellular changes in response to changing environmental conditions. Furthermore, in the quest for such knowledge, powerful genetic tools have been developed for studying this species, many of which might be applicable to other bacterial species (Errington 1986, Harwood and Cutting 1990, Sandman et al 1987, Yoshikawa and Sueoka 1963, Spizizen 1958, Zuber and Losick 1983). 1.2 Problems Associated With the Use of Bacillus Despite significant increases in our understanding of the molecular processes that control growth and protein expression processes in single cels of B. subtilis, explanations for the behavior often exhibited by culturesremains elusive. Cultures of Bacillus have been found to exhibit a rich variety of dynamic behavior. Long term oscillations,multiple steady states, genetic instabilities and uninterpretable transience have all been found to be associated with cultures of Bacillus (Humphrey et al. 1966, Heineken and O'Conner 1972, Fencl and Pazlarova 1982, Koizumi and Aiba 1989, this laboratory). In an effort to explain such behavior, a number of theoretical investigations have been undertaken (Humphrey et al. 1966,Heineken and O'Conner 9 1972, Koizumi and Aiba 1989), typically without much success. The inability of such studies to explain even qualitative observations suggests that cultures of Bacillus are much more complicated than often acknowledged. Clearly such behavior must reflect complex biological processes acting on the single cell and population levels. Past analysis have failed to account for the possibility that cultures of B. subtiis can be physiologicallyheterogeneous. Due to the ability of cells of B. subtilis to enter the developmental pathway of spore formation, the culture demographics might be anticipated to be more complex than those found in other bacterial systems (figure 1.1). Since only a fraction of the cell population eventually develops into spores, some eight to twelve hours after development is induced by starvation, late stationary phase cultures are observably heterogeneous. This observation introduces the possibility that the heterogeneity known to exist late in the culture is but a manifestation of heterogeneity that exists earlier during the stationary phase. Therefore, the proper analysis of gene expression will depend on if and when heterogeneity is established. L3 Hypothsis The central hypothesis behind this work is that culture heterogeneity and its effect on culture behavior are not sufficientlyunderstood. Since the observed culture behavior could "...represent the events occurring within the individual cells or some average of a heterogeneous population" (Novick and Weiner 1957),failure to account for culture heterogeneity in the analysis of data might act to obscure the real 10 behavior of cells within the culture. This might even lead to erroneous conclusions. A corollary of this hypothesis is that traditional approaches used to describe microbial culture dynamics may not be applicable. Hence, physiological heterogeneity could provide a plausible explanation for the major discrepancies that exist between experimental and theoretical studies. In contrast to homogeneous cultures, the behavior of cultures of Bacillus must now be addressed with respect to the different cell types that constitute the population. It appears that while much has been identified about the molecular mechanisms that control cell development and gene expression, the consequences of these results, and their ramifications on the demographic makeup of the culture, have not been fully appreciated. .4 Objective The principal objective of this work was to experimentally and theoretically investigate how early stationary phase events influenced the decision made by cells of B.subtilis to initiate sporulation. It was of particular interest to elucidate the degree to which active or threshold mechanisms influenced this decision. A natural consequence of this investigation would be a better appreciation of the effects, if any, that culture heterogeneity have on the observed culture average behavior. 1.5 Approach The investigative approach adopted is summarized in figure 1.2. Experimentally, the goal was to investigate how some of the earliest processes 11 involved in stationary phase signal transduction and early gene activation in B. subtilis influenced the cell's ability to acquire the terminally differentiated state. Since one of the earliest events associated with spore formation is the activation, via phosphorylation, of the developmental transcription factor encoded by spoOA (Burbulys et al. 1991, Grossman 1991), the pattern of SpoOA controlled gene expression was used to characterize early development. Gene expression was then studied in cells containing well characterized mutations in the signal transduction genes, kinA (Antoniewski et al. 1989, Perego et aL 1989, Burbulys et al 1991) and spoOK(Perego et al. 1991,Rudner et al 1991). Because kinA and spoOK mutations reduce the intracellular level of phospho-SpoOA, cells carrying such mutations provided the means of studying how reduced levels of phospho-SpoOAaffect early developmental gene expression and spore formation. To account for the possibility that not all cells expressed SpoOA controlled genes, a flow cytometry based technique suitable for studying gene expression on the single cell level was developed. This allowed data representing events occurring at the single cell level to be gathered as opposed to data that might reflect an average taken over a heterogeneous population. Theoretical studies focused on developing mathematical models aimed at incorporating system non-linearities that predicted the existence of multiple physiological states. All models were derived from first principles in a manner consistent with what is known about gene regulation. The consistency of the derived models, with reported observations, was then checked through simulations using 12 parameters reported in the literature. Since very little quantitative information on the sporulation system can be found in the literature, a detailed analysis of the molecular processes responsible for decision making was performed on two well characterized, qualitatively and quantitatively, genetic systems; cells of Eschenichia coli infected with phage A and the E. coli lac operon (Ackers et al. 1982, Ptashne 1987, Barkley and Bourgeois 1980). Under the appropriate conditions, heterogeneity of transcriptional states has been found in both systems (Bailone et al. 1979, Novick and Wiener 1957). The principles derived from this analysis were then used to develop a model aimed at describing the decision making processes involved in the initiation of spore development. The resulting model is shown to be consistent with the experimental results presented in this thesis and throughout the literature. 16 Significanoe Since many proteins of academic and industrial significance are produced by Bacillus during the early stationary phase, when culture heterogeneity might exist, information obtained from this study has fundamental and applied significance. Under heterogeneous culture conditions, the possibility that not all cells express the protein of interest exists. If it is desired to optimize the production of a particular protein, one would like to know this information. Not only would such information be valuable in its own right, but it would also lead to a more accurate characterization of the culture. Traditional definitions used to characterize culture performance, such as the specific productivity or the yield, could be replaced by more 13 precise terms reflecting the heterogeneous nature of the culture. Process manipulations that affect culture productivity could be assessed based on the effects they have on the constituent cell subpopulations. In particular, changes in culture productivity might result from a distribution of the culture demographics or from variations in the level of expression within individual cells. Information on the culture demographics might also provide a simple means of estimating the upper limits on culture productivity under a given set of cultivation conditions. Proper analysis of culture data will inevitably rely on the ability to address or account for culture heterogeneity. Pioneers in the area of molecular and cell biology realized the problems associated with heterogeneity and went to great length to either account for (Benzer 1953, Novick and Wiener 1957) or eliminate such factors (Herzenberg 1959). 1.7 Thesi Oranization In the followingchapters these ideas are further developed. Chapter 2 serves to introduce the concept of physiologicalheterogeneity. Also discussed is the present state of knowledge, which is evolvingat this very moment, surrounding the molecular processes known to be involved in regulating stationary phase gene expression in cultures of B. subti&s. In chapter 3, the development of the tools needed to study culture heterogeneity is described. The flow cytometry-based procedures used to study gene expression in single cells of B. subtilis, using lacZ fusions, are described. In chapters 4 and 5, these techniques are applied to study early stationary phase gene expression in differentiating cultures of B. subtilis. 14 In chapter 6, physiological heterogeneity is studied quantitatively. It will be shown that threshold mechanisms control adaptive gene expression in a number of biological systems. Chapter 7 summarizes the work performed and restates the principal conclusions. Chapter 8 suggests areas of future research. Chapter 9 is devoted to the materials and methods used throughout this work. Chapter 10 presents the literature cited throughout this thesis. Following each chapter, the tables and figures cited are presented. 15 Growth Phase Stationary Phase 8 -- 40 Homogeneous Homogeneous or Heterogeneous ? Homogeneous Heterogeneous a - (a 0- Growth Cycle uu -- 40 Development Time Figum1.1 Bacillus subtilis culture heterogeneity. During exponential growth, the culture may be characterized by a single cell type. Upon entrance into the stationary phase, some cells develop into spores (top right) and some do not (bottom). This introduces the possibility that early stationary phase cultures are comprised of more than one cell type. 16 Investigative Approach Biology of Bacillus Follow phospho-SpoOAcontrolled developmental gene expression. Perturb early gene expression with knA and spoOK mutations. Account for Heterogeneity Mathematical Modeling Identify mechanisms allowing multiple transcriptional states. I Apply principles to develop a sporulatlon Initiation model. I I I Figue 12 Approach taken to study the molecular basis of physiologicalheterogeneity. 17 Chapter 2: Background 21 Physiological Heterogeneity How genetically identical cells adopt different developmental fates is a question that biologists have focused on for quite some time (Horvitz and Herskowitz 1992). This question applies to relatively simple developmental systems like the lysislysogeny decision of phage lambda, as well as to development in complex multicellular organisms. In addition, even in organisms that do not undergo elaborate development, there can be physiologicalheterogeneity within a culture of genetically identical cells. With E. coli under some conditions, it is possible to maintain a subpopulation of cells that are induced for expression of the lac operon, while a distinct subpopulation remains uninduced (Novick and Weiner 1957). In principle, physiological diversity could be produced in two different ways (figure 2.1). In the simplest case, different cell types might be generated at the time of cell division. The sister cells that result from this asymmetric cell division would then be expected to behave quite differently. Each cell type might be expected to exhibit a pattern of gene expression as defined by its inherited constituents. The cellular diversity generated through this process is illustrated by the asymmetric cell divisionundergone by budding cells of Saccharomycescerevisiae. Upon cell division, two distinct cell types are generated. The mother cell, which is typically larger carries a scar on its surface whereas the surface of the daughter cell is entirely new. Additionally, the mother cell is capable of undergoing a mating type conversion 18 whereas this cannot normally occur in the daughter cell (Strathern and Herskowitz 1979). The asymmetric division exhibited by cells of the bacterium Caulobacter, also results in two distinct cell types (Newton and Ohta 1990). A motile swarmer cell and a non-motile stalked cell result from the division process. Both cell types behave quite differently and both have distinctly different patterns of developmental gene expression (Gober and Shapiro 1991). While the mechanisms that regulate the process of asymmetric cell division are likely to involve both spatial and temporal aspects acting within the predivisional mother cell, in the final analysis, culture heterogeneity results as a natural consequence of the reproductive process. Since culture heterogeneity is inherent to these systems, these cases will no longer be considered. In contrast to the heterogeneity that results from asymmetric division,cellular diversity can also arise when identical progeny adopt different fates as a consequence of some later event. This situation arises in sporulating cultures of B. subtilis where under appropriate conditions only a fraction of the cells develop endospores (Ireton and Grossman 1992). Since the sister cells that result from cell division during vegetative growth, appear identical, this system raises interesting questions regarding "When and Why ?" sister cells adopt different developmental pathways. The answers to such questions are difficult to obtain since they require detailed knowledge of the intracellular concentrations of key components within cells. However insights into such questions can be gained from a detailed account of the sequence of events that are set in motion once cells initiate the developmental 19 program. Because cells eventually develop into distinctly identifiable cell types, an accurate account of the events leading up to this point could in principle provide a means of localizing the point during development at which physiologicaldiversity is established. In the next section, the sequence of biochemical events that are set in motion, once cells receive the physiological and environmental signals required for spore formation, are described. 2.2 The Developmental Program of Spore Formation in R subtiis 2.2.1 Signal Transduction va the Phosphorelay When cells of B. subtilis encounter nutrient limitations, as characterized by entrance into the stationary phase, exponential growth ceases and the cells initiate a number of adaptive response mechanisms to cope with the changing environmental conditions. Depending on the exact nature of the cultivation conditions, the adaptive responses include degradative enzyme expression, antibiotic secretion, motility, competence development and spore formation (Ferrari et al. 1986, Dubnau 1991, Errington 1993, Strauch and Hoch 1993). However when faced with continued starvation, cells eventually initiate the developmental process of spore formation. One of the earliest events controlling the initiation of spore formation is the activation, va phosphorylation, of the developmental transcription factor encoded by spoOA This process involves, in part, the sequential transfer of inorganic phosphate by the kinA, spoOF, and spoOB gene products by what has been termed a multicomponent phosphorelay (Burbulys et al. 1991; figure 2.2). Upon entrance into the 20 stationary phase, KinA autophosphorylates in an ATP dependent reaction on a histidine residue and transfers the phosphate to an aspartate residue in SpoOF. SpoOF-P then serves as a phosphate donor by transferring this phosphate to SpoOB. Finally, SpoOB-Ptransfers this phosphate to the N-terminal domain of SpoOA. In addition to KinA, other proteins kinases have been identified (Burbulys et al. 1991, Rudner et al. 1991). SpoOA and SpoOF belong to a class of procaryotic regulatory proteins, the response regulators of the two-component regulatory systems (Albright et al. 1989, Ronson et al. 1989, Stock et al. 1989). The response regulators play a role in signal transduction in response to changes in the environment. These proteins are characterized by their conserved N-terminal domains and their activity appears to be modulated by their phosphorylated state. A histidine protein kinase, or sensor component, is the other component of the two-component regulatory systems. In response to certain signals, the sensor component autophosphorylates a histidine residue and transfers the phosphate to the cognate regulator. Histidine protein kinases are characterized by their conserved Cterminal domains. Two component regulatory systems have been found to control the adaptive responses of a number of processes including nitrogen regulation, phosphate regulation, motility and genetic competence. 2.22 Transcriptional Activation via SpoOA-P Upon activation, one of the earliest functions of SpoOA-P is to indirectly 21 activate transcription of a number of genes involved in the adaptive mechanisms available to the cell. SpoOA-P represses abrB, a gene which encodes for a repressor which is present during vegetative growth (Perego et al. 1988, Strauch et al 1990). AbrB in turn represses of a number of stationary genes involved in the adaptive responses available to the cell (figure 2.3A). AbrB has been found to prevent the transcription of genes involved in degradative enzyme expression such as aprE (Strauch et al. 1989), sporulation such as spoOE (Perego and Hoch 1987, 1991, Strauch et al. 1989), spoOH(Weir et al. 1991) and spoVG (Zuber and Losick 1987, Robertson et al. 1989), antibiotic synthesis such as tycA (actually a Bacillus brevis antibiotic synthesis gene, Robertson et al. 1989) and itself (Strauch et al. 1989). It is also known to play a role in competence development (Dubnau 1991). Hence the early action of SpoOA-P prepares the cell for possible entrance into a number of adaptive pathways. The eventual path chosen depends on the nature of the cultivation conditions. The removal of AbrB initiates a positive feedback loop that leads to more activated SpoOA (figure 2.3B). An important gene whose transcription is indirectly activated by SpoOA-Pis SpoOH(Yamashita et al. 1986, Dubnau et al 1987). spoOH encodes the oHsigma factor (Weir et al. 1991), which directs the transcription from loci that encode phosphorelay components including kinA, spoOF, and spoOA (Predich et al. 1992). Since the products of these loci lead to more SpoOA-P,the net result of this loop is the amplification of SpoOA-P. SpoOA-P also serves to activate transcription from the essential sporulation 22 loci spolIA (Trach et al. 1991), spollE (York et al. 1992), and spollG (Satola et al. 1991, 1992, Bird et al. 1993) (figure 24). The products of these loci are involved in establishing and maintaining differential gene expression in the mother cell and the forespore (Margolis et al. 1991, Losick and Stragier 1992). SpoOA-P directly activates transcription from these loci by binding to upstream regions of the respective promoters. Additionally, because transcription from these loci appear to be negatively controlled by the sinR gene product, SpoOA-P also appears to play a role in countering the effects of SinR (Smith et al. 1991, Mueller and Sonenshein 1992). While the mechanism by which this occurs has not been identified, recent findings suggests that it is indirect being mediated by the developmentally expressed sinlgene product (Bai et al. 1993). 2.2.3 Establishing Late Sporulation Gene Epression Following spollIA and spoIIG transcription, the sigma factors encoded by spolIAC, aF, and spoIIGB, au, are activated and the temporal program of gene expression is continued. Activation of ja involves the spollA and spoIIE gene products (Stragier ct al. 1988) and is believed to coincide with the morphological changes associated with the formation of the asymmetricseptum, the physical barrier that separates the mother cell from the developing forespore. Once this barrier is in place, gene expression in the mother cell relies in part on aE, and gene expression in the forespore relies in part on af. Support for this model follows from the fact that the spoIIGB gene product, oE, is needed to direct transcription from spolID, a 23 gene known to be expressed exclusively in the mother cell (Driks and Losick 1991). More detailed accounts of these and subsequent events can be found in a recent review by Errington (1993). 2.3 Mechanisms Leading to Physiological Diversity The sequence of events beginningwith signal recognition and transduction and ending at spore formation indicates that a sporulating cell must successfully pass through a number of developmental stages on its way to a spore. The question raised in section 2.1, namely " When and why identical cells adopt different developmental programs ?" can now be addressed in terms of the pattern of gene expression exhibited by sporulating cultures of B. subtlis. Specifically,"At what point along the sequence of events do sister cellsadopt different developmental fates ?" As discussed below, at least two explanations exist. 2.3.1 The Threshold Theory The threshold theory proposed by Aubert et al. (1961), hypothesized that the level of an extracellular carbon source is responsible for triggering the developmental changes associated with spore formation. This hypothesis was based on the observation that "mass sporulation" of Bacillus megaterium cells, in continuous culture, could be effected by a shift down in the feed rate to the reactor. Dawes and Mandelstam (1969, 1970) attempted to test this hypothesis,however their results were inconclusive (section 2.3.3). 24 There are two major implications of the threshold theory. The first follows straight from the mathematical concepts upon which threshold phenomena find their origins (Arnold 1986, Golubitsky and Schaeffer 1985). Namely that small changes in a certain system variable, often referred to as a bifurcation parameter, might lead to dramatically different system behavior. According to the Aubert hypothesis, the bifurcation parameter would be represented by the concentration of the extracellular carbon source and this parameter would determine whether or not a cell entered the developmenal pathway of sporulation. The second major implication of this theory is that physiological differences between sporulating and non-sporulating cell-typesmust be established at one of the earliest events in the sequence of events along the developmental pathway. Given that nutritional, intercellular and cell-cycle dependent signals are integrated, processed and, transmitted through the phosphorelay (Ireton et a. 1993, Ireton and Grossman 1992), this theory is consistent with the suggestion that continuous inputs into the phosphorelay generate bifurcated outputs. This is represented in figure 2.5, where the effect that the phosphorelay network might have on the distribution of a developmental specific molecule, within a cell population, is illustrated. Given a population of cells containing a smoothly distributed concentration of some threshold molecule, dramatically different cell types would result from the threshold mechanism associated with the phosphorelay. More recently, as genetic and molecular biologicalevidence has indicated that the spoOA gene product is of critical importance to both sporulation specific gene 25 expression and stationary phase signal recognition, the threshold hypothesis has been resurrected (Antoniewski et al. 1990). The effects that null mutations in kinA, a gene encoding for one of the earliest acting proteins in the sequence of events leading to sporulation specific gene expression, have on early gene expression has been investigated by a number of researchers (Perego et al. 1989,Antoniewski et al. 1990, Mueller and Sonenshein 1992). kinA null mutations have been found to have no or little effect on the expression of genes indirectly activated by SpoOA-P, via AbrB, including spoVG, aprE and itself (Perego et al. 1989,Mueller and Sonenshein 1992). However kinA null mutations have been shown to reduce spore formation and the expression of spollA, spollE and spoIIG to approximately 10% of wild type (Antoniewski et a. 1990, Mueller and Sonenshein 1992). In contrast to kinA, mutations in other components of the phosphorelay such as spoOA,spoOBand spoOF completely abolish spore formation and early sporulation gene expression. In contrast to the hypothesis set forth by Aubert, the activated SpoOA threshold hypothesis suggests that outputs from the phosphorelay are needed to elevate the concentration of a particular molecule to threshold levels. 2.3.2 The Contiuum Theory An alternative to the threshold theory is the "continuum theory." In its simplest form, such a theory might suggest that all cells initiate adaptive gene expression, as controlled by activated SpoOA, and that spore formation results only if cells successfully pass through a number of developmental checkpoints. This theory 26 implies that culture heterogeneity is a consequence of the cells failure to overcome the multiple barriers associated with the developmental program. Such an idea was proposed by Coote (1972) based on his studies of various oligosporogenous mutants (i.e. reduced sporulation mutants). Evidence for this hypothesis was provided by the finding that some oligosporogenous mutants resulted from leaky mutations in genes where null mutations otherwise resulted in asporogenous phenotypes. 2.3.3 Common Ground A closer examination of the finer points behind each theory reveals that both theories contain common elements. This follows from the fact that the ability of a cell, within a population, to obtain a threshold level of a particular molecule can be viewed as a stochastic event. Hence even if threshold mechanisms exist, random factors might still be involved in determining which cells initiate the developmental program. Conversely, once cells have initiated spore formation, the ability to successfully complete spore formation might be viewed in terms of the cells ability to overcome multiple thresholds. Support for this interpretation is provided by results of experiments aimed at testing the threshold hypothesis (Dawes and Mandelstam 1969, 1970). By following the incidence of spore formation in continuous cultures of B. subtilis, as a function of the feed rate to the culture, Dawes and Mandelstam hoped to distinguishbetween these hypothesis. They reasoned that if the threshold hypothesis were true, then as the feed rate to the reactor was continuouslychanged, a discontinuityin the incidence 27 of spore formation would be observed. Such a discontinuity was never observed. Instead, their results showed that the probability that a cell would form a spore varied smoothly, but inversely,with changes in the feed rate to the reactor. However, these results do not rule out the threshold hypothesis. To the contrary, such results would be expected if the threshold theory were true. Insight into this claim can be obtained by considering how changes in the feed rate to the reactor might affect the distributionof a particular molecule within the culture population. Given a threshold molecule, smooth changes in the reactor feed rate would be expected to alter the intracellular concentration of this molecule in a smooth and uniform manner in all cells. Furthermore, feed rate changes would be expected to change the average threshold molecule concentration in a smooth manner. Therefore, for a fixed threshold concentration, the fraction of the cells in the culture crossing this threshold would vary smoothly with the feed rate (figure 2.6). Given the complexity of the developmental process, it is likely that the actual situation involves elements of both theories. The fact that cell-cycle signals are involved in the initiation process (Mandelstam and Higgs 1974) clearly indicates that purely stochastic arguments represent an oversimplification. Likewise, even if the threshold theory were true, it is likely that multiple thresholds are involved. Evidence suggesting the existence of multiple thresholds can be found in early studies aimed at identifying the point at which cells of B. subtilis become committed to the sporulation (Mandelstam and Sterlini 1969). The results of this work suggest that no single point of commitment exist. Instead there appear to exist multiple checkpoints 28 along the developmental pathway, each of which appears to be associated with stable mRNA species (Mandelstam and Sterlini 1969). Additionally, as will be discussed later in this work (Chapters 5 and 6), the repressive effects exhibited by AbrB and SinR very much points to the existence of multiple thresholds. Given that the program of gene expression associated with spore formation represents a complex process involvingthe spatial and temporal coordination of multiple developmental sigma factors, a major challenge will be to reconcile what has been learned in the past with the molecular biological mechanisms that are presently being elucidated. 29 Mother Cells Sister Cells A A B A At Birth Saccharomyces cerevisiae * Caulobacter A After Birth Bacillus subtilis Escherichia coli FEge 2.1 Schematic illustration of the possible origins of population heterogeneity. Culture heterogeneity may result be a natural consequence of asymmetric cell division or a consequence of a later event (adopted from Horvitz and Herskowitz 1992). 30 KinA +ATP Phosphorelay KinA-P\ SpoOF SpoOB Stationary Phase Signals *. Ki-P / SpoOF-P S1oOB-P - V- - Activated . t SpoOA Figure 22 Schematic illustration of the phosphorelay signal transduction mechanism used by cells of B. subtilis to sense and transmit stationary phase signals (originallypresented by Burbulys and coworkers 1991). 31 SpoOA-P A AbrB aprE tycA spoOH spoVG spoOE degradative enzymes sporulation protein antibiotics -- Repression + Activation SpoOA B Signas ....... 4 SpoOA-P '1 I AbrB SpoOH Fgue 2.3 Schematic illustration of the initial events set in motion once SpoOA is activated via phosphorylation. The AbrB pathway is first activated (A) and this leads to more activated SpoOA via SpoOH (B). 32 SpoOA-P I t inH spollA/ spollG spollE + spollD Figure24 Schematic illustration of the stage II gene expression via activated SpoOA. 33 0 Threshod z Threshold Molecule INPUT i Z 3 z 50% 50% Phosphorelay Component Figure2. Population model explaining the threshold hypothesis of Aubert and coworkers (1961). It was hypothesized that the concentration of an extracellular metabolite acted as a threshold molecule. If it is assumed that the extracellular concentration effectively represents some intracellular metabolite, then the concentration of this intracellular species would be distributed throughout a cell population in a "normal" manner. Only those cells containing at least the critical concentration go on to form spores as illustrated by the bifurcated "output" population distribution. 34 0 C) D0 E z Threshold Molecule Figue 2.6 Model explaining the observations of Dawes and Mandelstam (1970). This conceptualization indicates that stochastic effects are not incompatible with threshold mechanisms since a reduced feed may simply shift the entire population so that more cells are able to attain the critical threshold. 35 Chapter 3: Development of Flow Cytometry-Based Techniques for Studying Gene Expression in Single Cells of R subtilis 3.1 Introduction In this chapter, the development of fluorescence-activated cell sorting (FACS) protocols suitable for detecting fi-galactosidase in single cells B. subtiis is described. These procedures are simple, easily implementable and take advantage of commercially available probes, 5-octanoylaminofluorescein di-f-D-galactopyranoside, C-FDG, 5-dodecanoylaminofluoresceindi-f-D-galactopyranoside, C12 -FDG and the membrane binding dye PKH26 along with standard flow cytometry instrumentation. The suitability of these protocols for physiological studies was assessed based on the ability of the procedures to accurately recover f-galactosidase expressing populations in controlled mixture experiments. Additionally a correlation found to exist between bulk culture assays and flow cytometry-based fluorescence measurements (with C8 - FDG) indicates that these procedures are suitable for quantitating the relative figalactosidase activity in a sample population. When the optimized staining conditions were used in conjunction with what is termed the "internal PKH26 negative control" (see section 3.6), a robust system capable of analyzing gene expression in single cells of B. subtilis results. The approach taken to develop these techniques is flexible and allows the procedures to be easily modified to meet the experimental demands of a given system. This was demonstrated by successfully applying these techniques to analyze cells obtained from two different growth media. The development of these protocols demonstrates that FACS-based technology can be applied successfully to 36 study single cell gene expression in bacteria. These developments should be viewed as steps towards the more widespread use of flow cytometry in microbiology. Since the completion of this work, an alternate technique for detecting B-galactosidase activity in single cells of the Gram-negative bacterium Myxococcus xanthus was published (Russo-Marie et al. 1993). While this alternate technique does not include a control for background fluorescence, it offers the advantage in that cell viability is maintained. 3.1.1 Flow Cytometry The advent of the flow cytometer represented a revolutionary advance in the area of single cell measurements. For the first time, the attributes of single cells could be measured for a population large enough to accurately reflect the culture demographics and to reliably detect rare events. In addition to rapid and robust analysis of large populations, flow cytometry also makes it possible to sort single cells or various subpopulations within the sample. The sorting process maintains cell viability, assuming the cells remain viable after the staining process, permitting the samples to be recultivated or subjected to further biochemical analysis. These capabilities make the flow cytometer arguably the most valuable instrument available today for studying the attributes of large populations of single cells. In theory, a flow cytometer can be used to quantitate any fluorescing constituent within or associated with a single cell. In practice, the molecule that one desires to detect (i.e. DNA, RNA or other cellular constituents) either does not 37 fluoresce or fluoresces very weakly effectively making it undetectable. Therefore fluorescence probes specific to the molecule of concern are commonly employed. Probes have been developed or identified for the detection of a number of cellular constituents including DNA, RNA, cellular protein, surface markers and various intracellular enzymes and molecules (Crissman et al. 1979). With the exception of intracellular enzymes and antigens, the procedures for identifying the above mentioned constituents are relatively straightforward. The complications with detecting intracellular enzymes and antigens result from the physical barriers presented by the cell wall and membrane. Many of the techniques that claim to have overcome the problems with intracellular molecule detection do not provide convincing data in support of these arguments (Srienc et al. 1983, Jaccoberger et al. 1986, Schepper et al. 1987). For an excellent categorization of the probes available for studying the attributes in single cells, the reader is referred to a number of references (Crissman et al. 1979, Waggoner 1990, Haugland 1992). Despite the numerous flow cytometry-basedtechniques and probes that have been developed for studying the attributes of single eukaryotic cells, the only intracellular enzyme that has been reliably detected in a heterogeneous population using flow cytometry is P-galactosidase (Srienc et al. 1986, Nolan et al 1988, Wittrup and Bailey 1988, Fiering et al. 1992). However, given the heavy reliance that modern day biology places on the use of lacZ gene fusions (Casadaban et al. 1983, Nielsen et al. 1983, Shapira et al. 1983, Zuber and Losick 1983), this "limitation" is by no means a major one. Presently lacZ fusions are used to study gene regulation and 38 expression in numerous bacterial and yeast species (Casadaban et al. 1983) as well as various mammalian cell lines (Shapira et al. 1983, Nolan et al. 1990). Hence the ability to detect P-galactosidase in single cells using flow cytometry provides a powerful tool that complements traditional biochemical techniques. 3.L2 Bacterial Flow Cytometry While flow cytometry has revolutionized many fields of science, particularly immunology, its use in bacterial microbiology is less common. In other words "... only a few dozen of several thousand papers in the flow cytometry literature deal with microbiological applications." (Shapiro 1988). Furthermore most of these applications have been aimed at identifying aggregate cellular constituents such as the total cellular protein or DNA (Bailey et al. 1977, Bailey et al 1978,Hutter and Eipel 1978, Boye et al. 1983,Robertson and Button 1989,Steen and Boye 1980,Steen et al 1982, Steen 1990). Other applications of flow technology in microbiology have been either aimed at detecting the cell itself or characterizing some physical feature of the cell such as size (Betz et al. 1984, Miller and Quarles 1990, Phillips and Martin 1982, Robertson and Button 1989). More recently, probes and protocols suitable for detecting rRNA in single bacterial cells have been developed (Delong et aL 1989, Amann et al. 1990, Wallner et al 1993). However such applications represent a small fraction of the techniques that have been developed for studying the attributes of single eukaryotic cells. Specifically,one of the most powerful features of the flow cytometer, the ability to 39 detect specific constituents such as proteins and enzymes within or associated with single cells, has eluded microbiological studies. The handful of attempts reported either do not provide convincing results in support of their claims (Schepper et aL 1987) or have sought to exploit peculiarities of a particular system. A number of workers have been able to correlate the presence of inclusion bodies, which presumably correlate with the intracellular content of a particular intracellular constituent, with light scatter measurements (Srienc et al. 1984, Wittrup et al. 1988). More recently, the detection of gramicidin S via FITC staining has been reported (Azuma et al. 1992). Again, this work was possible due to an apparent correlation found to exist between gramicidin S and green fluorescence measurements. In general these applications were possible only through the fortuitous existence of system specific correlations and do not represent approaches upon which rational strategies can be built. Furthermore, a closer examination of the results raise questions about the resolution of such measurements. The most convincingstudies where specific intracellular proteins have been detected in bacterial cells using flow cytometry involved the detection of consisting of hundreds -galactosidase in microcolonies of E. col, of cells (Nir et al. 1990A, 1990B). However, such measurements were only possible through the application of gel-microdroplet technology, which adds a complicating factor to the analysis (Weaver et al. 1984, Weaver 1987). The scarcity of applications of flow cytometry to studying gene expression in bacteria is due to a number of factors. While the technical barriers to overcome are 40 formidable (section 3.2) the implicit assumption that bacterial cultures are homogeneous, may represent an even greater barrier to the widespread application of flow cytometer-based analysis of bacterial cultures. While this homogeneity assumption is often justified, the existence of culture heterogeneity under commonly employed conditions suggests that this is not always the case (Benzer et al 1953, Novick and Weiner 1957, Tyler and Magasanik 1969, Maloney and Rotman 1973). Advances in eukaryotic applications have come about only after much time and effort was spent in designing and developing suitable probes and protocols. Until bacteriologists commit equivalent resources, there is no reason to believe that the problems specific for bacteria will be overcome. Certainly this will not occur unless the importance of culture heterogeneity is acknowledged by more microbiologists. 3.2 Problem Statement 3.z1 Sgnal Strength The small size of bacteria relative to eukaryotic cells presents a major technical challenge that must be overcome if FACS-based technology is to be used to study gene expression in single cells of bacteria (Steen 1990). Typically bacterial cells are two to three orders of magnitude smaller than eukaryotic cells, on a volume per cell basis. Since flow cytometers detect the total number of fluorescing molecules, as opposed to the concentration, signals generated by bacterial cells will be two to three orders of magnitude lower than their eukaryotic counterparts, for the same fluorochrome concentration. Additionally, the total amount of the cellular 41 constituent to be detected will be lower in bacterial cells. Whether it is desired to detect a bulk quantity such as the total cellular DNA or a specific enzyme such as pgalactosidase, the signal generating potential will be considerably less. Therefore probes and protocols developed for staining eukaryotic cells will not in general be applicable to bacterial systems. To date, very few of the procedures developed for staining eukaryotes have been successfullyapplied to bacteria. The weak signals generated by bacteria present formidable barriers since they must be detected above relatively high levels of background. originate from a number Background signals of sources including cellular autofluorescence (i.e. constituents within the cell that naturally fluoresce), electronic noise, and free fluorochrome that might be present in the staining mixture (Shapiro 1988, Wittrup and Bailey 1988). While these processes and problems are associated with the development of any staining procedures, they become of critical importance when weak signals are to be detected. Spurious fluorescence signals will act to reduce the already low signal to noise ratio inherent to bacterial systems. In many cases, detecting bacteria with forward light scatter, the most routine parameter acquired by the flowcytometer, may not be possible or requires that the detection system be fully optimized (Russo-Marie et al. 1993, Wittrup et al. 1988). The technical barriers to detecting fl-galactosidase in single bacterial cells, using a fluorogenic substrate, are illustrated in figure 3.1. The top object represents a cell that contains /-galactosidase. The goal is to identify this cell among cells that may not be expressing f-galactosidase (bottom object). The heterogeneity might 42 reflect a difference in the transcriptional state of promoter X, the promoter responsible for driving P-galactosidase expression. Alternatively, this heterogeneity might be genetic in origin. If the P-galactosidase expressing population is to be detected, it is first necessary for the non-fluorescent substrate to enter the cell where it may be cleaved by /-galactosidase. Upon cleavage, the fluorescent product must accumulate within the cell in order to generate a signal that is detectable above the background signals due to cellular autofluorescence and electronic noise. However, in the process of achieving this goal, spurious fluorescence signals typically develop. Two common sources are; 1) the leakage of fluorescing products from the - galactosidase positive cells into the extracellular medium where they may become associated with P-galactosidase negative cells and 2) fluorescing molecules that result from spontaneous substrate hydrolysis, either before (i.e. in the substrate stock solution) or during the staining process. Therefore the ability to detect /B- galactosidase positive cells relies on the ability to overcome these competing processes. If this is not achieved, the signal generated by a /-galactosidase positive cell will be indistinguishable from that generated by a /-galactosidase negative cell. Successful protocols that have been developed for detecting /-galactosidase in single eukaryotic cells rely on the ability to optimize these competing processes (Nolan et al. 1988, Wittrup and Bailey 1988). 3.2.2 Bacterial Diversity The detection problems inherent in the analysis of bacterial cells are further 43 complicated by the large diversity that exists within the bacterial kingdom and the plethora of cultivation conditions employed to study these organisms. Large variations in cell size exist between different genera. Also, the membrane and cell wall architectures are quite different in Gram-positive and Gram-negative bacteria. A further complication results from the ability of bacteria to alter their size, morphology and membrane properties depending on the cultivationconditions. Since these parameters greatly influence the ability to stain and detect cells,procedures that have been developed and optimized for a specific system will most likely require modifications if the cultivation conditions are altered. 3.3 Approach In order to study gene expression in single cells of B. subtilis, a FACS-based technique capable of detecting P-galactosidase in single cells was developed. This approach was motivated by the success that a number of researchers have had with detecting this enzyme in eukaryotes (Srienc et al. 1983, Nolan et al. 1988, Wittrup and Bailey 1988, Fiering et al. 1991). Additionally, a number of commercially available fluorogenic substrates specific for P-galactosidase could then be evaluated (Haugland 1992). Such a technique would then allow the expression of a number of genes to be studied due to the availabilityof numerous lacZ fusions and the relatively straightforward procedures by which new fusions can be created (Errington 1986). To address the problems outlined in the previous section, it was first desirable to assess the background problem under "typical" staining conditions. 44 This was accomplished by introducing a control population, consisting of lacZ- cells stained with the fluorochrome PKH26, into the staining mixture and independently monitoring the fluorescence accumulated by this population using two-color flow cytometry. This population is referred to throughout this chapter as the PKH26 negative control or the internal control. The internal control was used to identify the background levels associated with a number of staining conditions and to identify conditions that minimized spurious fluorescence. Additionally, use of the PKH26 negative control allowed the assay conditions to be rapidly modified when cells were grown under different cultivation conditions. 3.4 Summary of Staini Condition In this section, the assay conditions found to be suitable for detecting /galactosidase activity in single cells of B. subtilis are summarized. The rest of this chapter is devoted to the presentation and the analysis of the experiments that led to the identification of these conditions. All staining experiments performed with C12 -FDG employed SMY::pAF25 as the positive control. The strain contains a veg-lacZ fusion integrated into the amyE locus of the parental wild-type strain designated SMY. The veg promoter is constitutively active during vegetative growth. MB284 was employed as the positive control in experiments performed using C8-FDG. This strain contains a spo VG'-'lacZ fusion integrated into the SPfi site of the parental wild-type strain designated SMY. The spoVG promoter is developmentally regulated and is active during the early 45 stationary phase. The wild-type strain SMY, which does not contain a lacZ fusion, was used as the negative control in all experiments described in this chapter. Endogenous f-galactosidase activity was not detected in this strain. It was necessary to use different positive controls under different cultivation conditions since the promoters driving the expression of f-galactosidase in these two strains are subject to different regulatory controls (Zuber and Losick 1983,Fouet and Sonenshien 1990). The cultivation conditions and the strains used are described in detail in Chapter 9. 3.4,1 C-FDG Optimal conditions for staining cells of B. subtilis grown in nutrient broth with and without glucose (5g/l) are summarized in table 3.1. The staining solution consisted of a substrate concentration of 33/uM C8 -FDG or C12-FDG in phosphate buffered saline (PBS, pH 7.0) supplemented with 10mM Mg2 + (PBSM) and 0.4g/l sodium azide (PBSMA). A 33,uM fluorogenic substrate concentration was chosen since it was suitable for generating a detectable signal in P-galactosidase positive cells without the solubilityproblems encountered at higher concentrations (specificallyCl2FDG at a concentration of 66MM). The cell number concentration in the staining mixture did not exceed 1,000,000 cells/ml, excluding the PKH26 control population. This number concentration represented a compromise between the need to minimize the number of cells in the staining mixture, which reduces the amount of dye leaked into the medium, and the need to obtain a cell concentration that could be conveniently analyzed using flow cytometry. 46 Cells to be stained with C1 2 -FDG were subjected to a 500C temperature pulse for ten minutes then placed on ice. This pulse was found to permeabilize cells allowing the substrate to enter once the sample was placed on ice. Since the substrate was present during the temperature pulse, cell staining also occurred during this process. On ice, the cells were able to accumulate detectable levels of fluorescing molecules with minimal leakage into the extracellular medium. The samples were incubated overnight, eight to twelve hours, to allow a steady state fluorescence distribution to result, then analyzed by flow cytometry. A detailed protocol is presented in Chapter 9. Cells to be stained with C8-FDG were permeabilized with a 1% suspension of toluene, in PBSMA, and stained at room temperature. Cells present in a 1% toluene solution were vigorously vortexed for at least 10 seconds, then diluted 1/100 into a solution of 33iM C8 -FDG in PBSMA. This dilution was necessary in order to remove most of the toluene before the samples were analyzed by flow cytometry. The sample was incubated at room temperature for at least 30 minutes, to allow a steady state fluorescence distribution to result, then analyzed by flow cytometry. These conditions are described in greater detail in Chapter 9. 3.42 PKH26 Optimal conditions for staining cells of B. subtilis grown in nutrient broth, with and without glucose, with PKH26 are summarized in table 3.2. Cells were harvested approximately two hours after exponential growth ceased and added to a solution of 47 5MiMPKH26 in Diluent C (the solvent provided by the manufacturers of PKH26) and incubated at room temperature for five to seven minutes with occasional mixing. The staining reaction was stopped by diluting this suspension ten-fold with a solution of 20g/1 bovine serum albumin (BSA) in PBSM followed by a one minute incubation at room temperature. Since PKH26 binds to hydrophobic sites (Horan et al 1990), the step served to remove the residual PKH26 free in the mixture. The cells were recovered by centrifugation, washed three times with room temperature PBSM, resuspended in the original sample volume, then placed on ice. When fermentation samples were to be analyzed by flow cytometry, an aliquot of these cells was introduced into the sample when required. These conditions are described in greater detail in Chapter 9. 3. Description of Molecular Probes 3.5.1 -galactosidase Fluorogenic Substrates In order to overcome the detection problems associated with the small size of bacteria, it was necessary to employ a fluorogenic substrate whose fluorescing product exhibited: 1) a high extinction coefficient in order to efficiently absorb electromagnetic radiation (Shapiro 1988), 2) a high quantum efficiency in order to efficiently convert the absorbed light into fluorescent radiation (Shapiro 1988), and 3) the ability to stay associated with the cell following hydrolysis. While a number of probes satisfy the first two requirement, fluorogenic substrates that satisfy all of the above requirements are few. The third requirement is of extreme importance since 48 it represents the means by which the problems associated with the small size of bacteria can be overcome. Namely, a larger cell-associated fluorochrome concentration can compensate for the smaller cell volume. The Cn-FDG series of -galactosidase substrates appeared to satisfy these requirements. A comparison between the spectral properties of several fluorochromes upon whichmanyp-galactosidase substrates are derived, reveal several important features (table 3.3). Fluorescein-based P-galactosidase probes are among the most sensitive fluorochromes available today as estimated based on their ability to absorb electromagnetic radiation. A comparison between the various extinction coefficients presented in table 3.3 indicates that fluorescein-based substrates can be expected to be approximately twice as sensitive as resorufin-based -galactosidase substrates and about five times as great as f-methylumbelliferone-based substrates. Since fluorescein enjoys a relatively high quantum efficiency (0.98; Mathies and Stryer 1986), the sensitivity can be expected to be even greater in practice. Additionally, fluorescein-based substrates have the advantageous property in that they can be excited by light in the visible spectrum exhibiting an excitation maximum at 490nm. Also included in this table, for comparison purposes, are the spectral properties of one of the most sensitive class of fluorochromes available today; those derived from the phycobiliproteins (Mathies and Stryer 1986, Shapiro 1988). Unfortunately Pgalactosidase substrates employing this fluorochrome do not exist. The major attribute of the C,-FDG series of substrates results from the increased affinity that the fluorescing product exhibits for the cell, when compared 49 to fluorescein. Figure 3.2 serves to compare the chemical structures of fluorescein di-p-D-galactopyranoside (FDG; A) with that of C 12 -FDG (B). The difference between these two molecules is the covalent link between the 5 position of the fluorescein phthalate ring and the linear hydrocarbon. Members of the Cn-FDG family differ only by the length of the linear hydrocarbon. Upon hydrolysis by fl- galactosidase, the sugar residues shown on both of these molecules are released and a highly fluorescent fluorescein or fluorescein derivative is released. Since the product resulting from Cn-FDG hydrolysis is amphipathic, due to the hydrocarbon "tail", the fluorescing product can accumulate within the cell to a greater degree than that of fluorescein. While some leakage can be expected to occur, the increased affinity exhibited by the fluorescent product for the cell, suggests that it might be possible to identify staining conditions that would not exist had the hydrocarbon tail been absent. 3.5.2 The Second Fluorochrome: PKH26 The sole purpose of the second dye was to provide the means of directly assessing and tracking spurious fluorescence signal development. Therefore the following requirements were sought: 1) the emission spectrum of the dye must be different from that of fluorescein, 2) the dye must not interfere with the sample of interest and, 3) spurious fluorescence accumulated by cells stained with this dye must accurately reflect the spurious fluorescence accumulated by the sample cells. Based on the successful use of membrane binding dyes for cell tracking (Horan et al. 1990), 50 PKH26 was identified as a candidate probe. Figure 3.3 indicates that the first of these conditions is satisfied by PKH26. These spectra were obtained by exciting ethanol solutions containing fluorescein and PKH26 with 488nm light. These data indicate that both fluorochromes have well separated emission maxima (515nm for fluorescein and 667nm for PKH26). Based on the dichroic mirrors and fluorescence filters employed in this study (Chapter 8), the differences in these spectra were exploited. As will be shown in the next section, cells stained with PKH26 also satisfied the last two requirements. 3.6 The Internal PKH26 Negative Control 3.61 TrackingBacground To identify the level of spurious or background fluorescence accrued under a given set of staining conditions, two-color flow cytometry was employed. lacZ- cells, stained with PKH26 (i.e. PKH26 positive), were added to the assay mixture containing the fluorogenic substrate and the cells to be analyzed and the increase in green fluorescence of this population was monitored. This strategy is schematically illustrated in figure 3.4 where the various cell populations present in the mixture are represented. The top object represents a P-galactosidase positive cell (i.e. a cell that is expressing P-galactosidase). The object on the bottom right represents a galactosidase negative. This might reflect a lack of expression or a lacZ- genotype. Both of these populations would presumably be present in a sample to be stained. The object on the bottom left represents the lacZ- cells that have been stained with 51 PKH26. Should free C8 -fluorescein or C12-fluorescein interfere with the sample to be stained, the ability to independently monitor the PKH26 positive population using two-color flow cytometry, should allow this to be identified. Since any fluorescence associated with the PKH26 negative control represents spurious signals, this approach also provides a means of detecting fluorescing products free in the medium due to contaminated substrate solutions. This strategy was validated by performing an artificial staining experiment aimed at simulating the conditions that might be encountered when real culture samples are analyzed. Such an experiment is schematically represented in figure 3.5 where the three objects shown represent three distinct cell populations that were added to the test mixture. The bottom two objects represent lacZ- populations that were added to the mixture, one of which has been stained with PKH26 according to the conditions outlined in table 3.2. The lacZ- population stained with PKH26 represents what has been referred to as the background control. The second lacZpopulation would presumably represent the non-expressing cells in a real sample. The top object is meant to represent the B-galactosidase expressing population present in a real sample. For the purpose of identification, this population was stained with 10,uM PKH26. All three cell populations were combined and incubated in the presence of a 33MMClrFDG solution, at 500C, and the time course increase in fluorescence of the various populations was monitored using two color flow cytometry. Under these conditions, spurious fluorescence due to C12 -fluorescein was found to be due to both the spontaneous hydrolysisof C12 -FDG and leakage from the 52 Bf-galactosidase positive population. The means by which these population were experimentally tracked is illustrated in figure 3.6. This figure represents the two-color contour plot used both online and offline to track the three populations described in figure 3.5. The various populations are represented by the three "clusters"and correspond identicallyto the populations shown in this figure 3.5. The X-axis represents the fluorescence signal due to fluorescein. The movement of each population along this axis is our principal concern. The Y-axis represents the fluorescence signal due to PKH26. This signal represents the means by which each population is identified. These data where obtained from the initial time point associated with the experiment described. By monitoring the increase in green fluorescence (i.e. the fluorescence due to fluorescein) of each of the lacZ- populations, the accuracy with which the background control population tracks spurious fluorescence was assessed (figure 3.7). The parallel fluorescence profiles exhibited by both of the lacZ- populations (figure 3.7) indicates that both lacZ- populations experience the same level of non-specific staining. These data indicate that the background control strategy can be used to accurately monitor fluorescence increases due to non-specific staining processes. Similar results were obtained when three population mixtures were stained under a variety of conditions (see section 3.7.3). 3.6.2 Fuhaescence Compensation The finite C12 -fluorescein signal exhibited by the top population in figure 3.6 53 at time 0, reflects a technical feature associated with the use of multiple dyes in flow cytometry. This can be explained with reference to figure 3.3, where the emission spectra of the fluorescein and PKH26 are shown. Because the "shoulders" of each spectra overlap, PKH26 specific signals will generate an apparent fluorescein signal and vice-versa. The degree to which this "bleedover" occurs will depend on the signal strength and the optical filters employed. Greater PKH26 specific signals will generally lead to greater "bleedover," and thus a greater apparent fluorescein component. This is the reason why the top population, shown in figure 3.6, appears "green" while the middle population does not. While this "bleedover" may be aesthetically unpleasing, it does not affect changes in the green fluorescence signal and is therefore inconsequential to the conclusion drawn from the described kinetic experiment. In multi-color flowcytometry,the problems associated with spectra overlap are addressed using what is termed "compensation." A number of excellent references can be found on this topic (Loken et al. 1977, Alberti et al. 1987). Compensation involves electronically subtracting a percentage of the corrupting signal, in this case PKH26, from the corrupted signal, in this case fluorescein. This analysis assumes that all signals can be decomposed into elementary components as defined by the specific signals associated with each fluorochrome used (Loken et al. 1977). This analysis is mathematically identical to the projection methods employed in linear operator theory (Strang 1986). However in this case, the "basis" functions or vectors are represented by the signals specifically generated by each fluorochrome. 54 3.6.3 Controlled Mixtures PKH26 does not interfere with the ability to analyze real samples. The possibility that the PKH26 control could interfere with the sample of interest was addressed by determining how well the PKH26 control populations could be recovered after sample staining. The data presented in figure 3.8 indicate that known PKH26 positive populations can be recovered from controlled mixture experiments. These results were obtained from a number of controlled mixture experiments where known amounts of PKH26 positive cells were added to samples prior to staining with either C8 -FDG or C1 2 -FDG. It was therefore concluded that the PKH26 negative control does not interfere with the sample population of interest. 3.7 Staning Cells with 5-Dodecanylaminofluorescein di--D-galactopyranoside 3.7.1 Approach Since the membrane fluidity of cells has been found to be strongly temperature dependent (Overath et a. 1970), it was reasoned that temperature manipulations could be used to stain cells with C12 -FDG. This strategy was motivated by the structure of ClrFDG and its resemblance with the phospholipid molecules that constitute the cell membrane. Experiments were performed in order to identify suitable staining temperatures and time that lead to detectable specific signals with little background due to fluorochromes free in solution. Due to the possible thermal denaturation of -galactosidase (Cohn and Monod 1951), the permissible staining conditions were 55 constrained by the need to maintain enzymatic activity. The following sections describe the experiments that were performed in order to achieve these objectives. 3.7.2 Thermal Deactivation ofp-galactoidase Before the staining conditions could be identified, it was necessary to assess the effect that temperature variations had on the enzymatic activityof f-galactosidase. To identify the temperatures and times at which a sample could be incubated without dramatically inactivatingfB-galactosidase, temperature inactivation kinetic experiments were performed. Samples were incubated at different temperatures in a solution of PBSMA and the residual enzymatic activity, as determined through bulk culture assays employing the chromogenic substrate o-nitrophenyl-/f-D-galactoside (ONPG), was followed over time (figure 3.9A). The shape of these curves suggested that at each temperature, the inactivation of f-galactosidase obeyed a first order (in -galactosidase) kinetic process. Previous researchers have also reported that P-galactosidase inactivation obey first order kinetics (Cohn and Monod 1951). The fact that the residual activity curves appear as straight lines on a semi-log plot support this assumption (figure 3.9B). The slopes of these lines give the values of the first order rate constants (Levenspiel 1972). The rate constants determined from the plots presented in figure 3.9B were used to generate -galactosidase deactivation contours aimed at identifying the incubation times, at set incubation temperatures, over which a certain percentage of the original activity could be maintained. These results are presented in figure 3.10 56 where the area beneath each curve, and above the X-axis, represents the set of temperatures and time during which a sample may be incubated while maintaining the indicated percentage of its original activity. 3.7.3 Constant Temperature Staining Using the activity contours presented in figure 3.10 as a guide, a series of constant temperature staining experiments was performed. Two sets of samples consisting of f-galactosidase positive cells and -galactosidase negative cells were stained under constant temperature conditions in the presence of the internal PKH26 background control. Both sets of samples were incubated in a 33,1M C 12 -FDG solution, in PBSMA, for times determined by the 80% activity contour shown in figure 3.10. The time course increase in fluorescence of all populations was monitored and the average rate of fluorescence increase over this time frame determined (figure 3.11). These results indicate that signal generation due to B- galactosidase activity, leakage and spontaneous substrate hydrolysis are highly temperature dependent. At temperatures above 370C, increases in P-galactosidase specific signal generation are accompanied by increases in spurious signal generation, as illustrated by the top two curves. However below 37°C, very little fluorescence is generated-by any of the aforementioned processes. These results suggests that below the physiologicalgrowth temperature, 37°C, the membrane freezes and a transport problem is encountered. Hence it was concluded that constant temperature staining is not optimal. 57 3.7.4 The Two-Step Temperature-Time Pulse Matrix The rate data presented in figure 3.11 indicate that two staining regions exist over the range of temperatures studied. Above 370C, relatively high rates of specific and background staining occurred. Below 37°C, the rates of specific and spurious staining are both low. Therefore in order to take advantage of the high rate of specific staining, while minimizing background, a two-step staining process was devised. Cells to be stained were subjected to a brief high temperature pulse followed by incubation on ice. It was hypothesized that the high temperature pulse would act to load the cells with substrate and the low temperature would permit the substrate to be cleaved and retained. It was found that the high temperature pulse, in the presence of sodium azide, not only served to load the cells with substrate, but also served to permeabilize the cells. On ice, the substrate continued to enter the cell and the fluorescent product slowlyaccumulated to detectable levels. To test the feasibility of this scheme, 50/50 mixtures of P-galactosidase expressing and non-expressing cells were subjected to high temperature pulses for different periods of time. The temperature range was varied from 40°C to 52°C. Since the level of non-specific staining increased monotonically with temperatures, the goal was to identify the minimum pulse duration that led to good recovery of the 50/50 mixtures. Following the temperature pulse, the samples were incubated on ice and analyzed eight to twelve hours later. A ten minute temperature pulse at 50°Cwas found to be optimal. The results of the temperature-time pulse experiments are qualitatively summarized in table 3.4. 58 While a number of classes of staining have been qualitatively described, the results summarized in this table indicate that higher temperatures allow shorter incubation times to be employed. Representative green fluorescence distributions corresponding to the three qualitative classes described are presented in figure 3.12. Panel A represents the distributions exhibiting what has been termed "little staining." Panel B represents the distributions exhibiting two observable peaks. Panel C represents the distributions that exhibit what has been termed "excellentpeak resolution." Based on the distribution presented in panel C, the 50°C ten minute pulse was considered optimal. Finally it should be noted that the distributions resulting from these experiments were found to be stable for at least 24 hours. This is a desirable feature since often times the researcher will have limited access to a flow cytometer. 3.75 Cntroled Mixtures The results summarized in the temperature-time pulse matrix indicate that a ten minute 500C temperature pulse led to good staining with minimal background. Therefore a detailed series of controlled mixture experiments was performed to identify the resolution with which known positive populations could be recovered. Controlled mixtures comprised of 5% to 90% -galactosidase expressing cells were stained according to this strategy. Additionally, the PKH26 internal control was added to each sample. Representative C12-fluorescein fluorescence distribution (i.e. due to fluorescein) obtained from these experiments are presented in figure 3.13. 59 Panel A was obtained from a 100% lacZ- population. The fluorescence distribution obtained from this sample represents the amount of background that must be overcome in order to detect -galactosidase in a positive cell population. The distributions presented in panels B through F were obtained from samples containing known positive populations. The bimodal nature of these distributions indicate that the two cell populations are recoverable from each mixture. One population exhibits fluorescence signals equal to that of background (panel A). The second population is able to accumulate a fluorescence signal above background. The sizes of the various populations represented by the distributions presented in figure 3.13 correlate with the fraction of fi-galactosidase positives known to be present in the mixture. The relative proportion of the populations exhibiting high and low fluorescence was quantitated using two methods. The first method involved setting a fluorescence gate based on the green fluorescence exhibited by the internal control (figure 3.14). The second method involved the use of a non-linear regression package that could be used to fit two normal fluorescence distributions through the experimentally obtained composite distribution. A summary of these results is presented in figure 3.15. This figure indicates that the fraction of f-galactosidase positives added to the mixture can be accurately recovered using either method. These data suggests that this procedure can be successfullyapplied to detected figalactosidase populations in heterogeneous mixtures. Furthermore, the excellent recovery achieved indicates that loading of the substrate into the cells is relatively uniform under the employed conditions. 60 3.8 Stin Ces with 5-Octanminofluorescein di--D-galacpranosidc 3.&l Approach When the staining procedures developed in the previous sections were applied to stationary phase cultures of B. subtiis grown in nutrient media without glucose, very little staining resulted. Since bulk culture assays revealed that P-galactosidase was present in these cells, this lack of staining reflected a permeability barrier. In general it was found that stationary phase cells were considerably less permeable that growing cells. None of the Cu-FDG series of substrate were able to penetrate these cells using the two step staining procedure developed in the previous sections. To obtain information on the distribution of -galactosidase in these populations, the protocol had to be modified in order to overcome this permeability barrier. Three modifications were found to be necessary; 1) the method of cell permeabilization was altered, 2) C-FDG was used in place of C12 -FDG and 3) the staining was performed at room temperature. 3.&2 CFDG is Optimal To insure that the fluorogenic substrate had free access to intracellular fBgalactosidase, the cells to be analyzed were permeabilized by vortexing for at least ten seconds, at maximum speed, in a 1% toluene suspension in PBSMA Toluene was chosen over other organic solvents since toluene-permeabilized cells are routinely used in bulk biochemical assays (Miller 1972, Fisher et a. 1975). When cells were vortexed for longer periods of time, little if any difference was observed in the ability 61 of the cell to uptake the substrate. As shown in section 3.8.4, assays performed on sorted populations indicate that this permeabilization procedure allows C8-FDG to readily penetrate cells. The solution was then diluted 1/100 by the addition of a 33MuMsolution of C-FDG in PBSMA. The reason for this dilution was to facilitate the removal of toluene prior to analysis by flow cytometry. C-FDG was found to have the best penetration and retention properties of all substrates in the C.-FDG family,where n=2,4,8,12,16. Equal numbers of toluene permeabilized lacZ positive cellswere incubated with each of the Cn-FDG substrates, at room temperature, and the substrate penetration ability was assessed by determining the rate of fluorescence increase of each sample using a spectrofluorimeter. Figure 3.16A summarizes these results where the relative rates of fluorescence increase are presented for each substrate for comparative purposes. These results indicate that C2-, C4- and C-FDG are transported into the cells at approximately equal rates. Toluene permeabilized cells still managed to exclude C12and C16 -FDG. The ability of cells to retain the fluorescent product was assessed using the average single cell fluorescence obtained from each sample using flow cytometry. This is a useful criterion because flow cytometers measure the fluorochromes level within or associated with the cell and not the extracellular fluorochrome concentration. Equal numbers of toluene permeabilized cells, from the same batch of cells used in the previously described experiments, were incubated for thirty minutes at room temperature in a 33MMsolution of each substrate. The thirty 62 minute time point was chosen since this represented the last time point analyzed during the fluorimeter based experiments. Additionally, as will be shown in the next section, steady state fluorescence distribution had result by thirty minutes. The results of these experiments are summarized in figure 3.16B. Presented is the average single cell fluorescence obtained using each substrate. These data indicate that C8 -fluorescein is best retained by the cell. 3.&3 Staining Time In contrast to samples stained with C12 -FDG, samples stained with C8 -FDG more rapidly reached a stable time invariant fluorescence distribution. This is illustrated in figure 3.17 where the time course increase in the average fluorescence of a sample stained with C8-FDG is presented. After a rapid initial increase in the average single cell fluorescence, a steady state is reached. However prolonged incubation eventually led to the accumulation of considerable levels of free C- fluorescein as noted by the development of a faint green color in samples incubated overnight. Based on these data, samples were analyzed approximately thirty minutes after staining was initiated. Subsequent analysis of stained samples, one or two hours later, revealed that the thirty minute fluorescence distribution was stably maintained over a one to two hour time period. 3.84 Heterogeneous Expression in Stationary Phase Samples The fluorescence distribution obtained from a stationary phase sample of B. 63 subtilis, expressing spoVG-directed fl-galactosidase is heterogeneous (figure 3.18A). This data was obtained by staining stationary phase cells of MB284 according to the conditions outlined above (see table 3.1), in the presence of the internal background control. The results suggest that the culture sample is comprised of two population. One population expresses high levels of f-galactosidase and the second population expresses little or no fB-galactosidase. Based on the green fluorescence distribution of the internal control, the relative sizes of both population was determined (figure 3.18B). The observed fluorescence distribution reflect differences in single cell f- galactosidase content. To insure that the observed signals reflected differences in B- galactosidase and were not artifacts of the permeabilization procedure, cells from the high and low fluorescent populations, and the total population, were recovered via cell sorting and their respective enzymatic activities determined (figure 3.19). The relative enzymatic activities reported were obtained by assaying 50,000 sorted cells using the fluorogenic substrate 4-methylumbelliferyl -D-galactopyranoside (MUG). These values have been normalized such that the activity of the total population is unity. MUG was used since the increased sensitivityof this substrate, relative to the more commonly used chromogenic substrate ONPG, permitted fewer cells to be assayed in a shorter period of time (Chapter 9). The bulk culture activity,reflects the activity of the high expressing population. Using estimates of the sizes of the high and low expressing populations, the contributions of each population to the total sample activity was determined. 64 Whether the size estimates were determined using a fluorescence gate (67%), set by the green fluorescence distribution of the internal PKH26 negative control (figure 3.18B), or by curve fitting two distributions through the composite distribution (72%, see section 9.4.3), the percentage of the total sample activity residing in the high expressing population is estimated to be greater than 95% (section 4.4). 3.8.5 Contrled Mixtures Controlled mixture experiments were also performed with stationary phase cells expressing P-galactosidase. Cells expressing spoVG-directed P-galactosidase (MB284) were mixed with acZ' cells and the PKH26 internal control and stained according to the procedures described (table 3.1). Representative green fluorescence distribution obtained from these experiments are presented in figure 3.20. Similar to the controlled mixtures presented earlier (figure 3.13), panel A represents the distribution obtained from a 100% lacZ' population. The other panels correspond to various mixtures (see legend). These results indicate that the two populations can be recovered from the controlled mixtures. When the relative proportions of each of these populations is estimated based on the green fluorescence distribution of the internal PKH26 negative control, good recovery of the added populations results (figure 3.21). Because spoVG-lacZ expression was found to be heterogeneous, the results presented in figure 3.21 account for this heterogeneity. The low or nonexpressing population shown in figure 3.18 was considered to be part of the negative population. 65 3.86 Correlation Between Fluorescence and Average Enzyme Activity A correlation was found to exist between the steady state population average fluorescence obtained by flow cytometry, using C-FDG, and the specific /B- galactosidase activity determined from bulk culture assays. This is illustrated in figure 3.22 where the time course expression of spolIA-lacZ, as determined by bulk culture assays, is plotted along with the time course average relative fluorescence as determined using flow cytometry. The units of fluorescence intensity were normalized by the maximum value, exhibited at the time of peak bulk culture expression. Since the histograms shown throughout this chapter were obtained with logarithmic amplifiers, the fluorescence values reported were obtained after the logarithmic histograms were linearized (section 8.4.4). Similar results were obtained for other spo-lacZ fusions (chapter 5). These results are consistent with the theoretical work of Wittrup and Bailey (1990) where a relationship between enzymatic activity and single cell fluorescence was derived. Their model also suggests that the steady state fluorescence distributions obtained (figure 3.17) result from a balance between the rate of substrate hydrolysis,or product formation, and the rate of product loss. 3.9 Discusion The principal conclusions derived from the results presented in this chapter are: 1) that low temperature staining, whether on ice or at room temperature, alleviates many of the background problems associated with the use of the C-FDG family of fluorogenic substrates, 2) the PKH26 background control concept provides 66 a means of assessing the background problems that may be encountered during protocol development, 3) the PKH26 background control provides the means of controlling for spurious fluorescence when "real" samples are to be analyzed, 4) the adaptable nature of bacterial cells make it unlikelythat a single staining protocol will be optimal under all conditions and 5) that the average steady state fluorescence was found to correlate with enzyme activities determined via bulk culture assays. A number of workers have found that low temperature staining can be used to minimize background fluorescence, particularly when larger eukaryotic cells are to be analyzed (Nolan et al. 1988, Wittrup and Bailey 1988). Since the completion of this work, Russo-Marie and coworkers (1993) were able to develop a FDG based technique for staining the Gram-negative bacterium Myxococcus xanthus. Again, this technique relied on low temperature staining. The commonly provided explanation behind the success of low temperatures staining is that the cell membrane freezes, or becomes less fluid, and this serves to prevent the fluorescing product from leaving the cell. This explanation also appears to apply to cells of B. subtilis stained with Cr2- FDG. When the Cl2 -fluoresceingenerated fluorescencewas determined before and after cells were washed with cold and room temperature PBSM, cells washed in cold PBSM retained more of their original fluorescence. Given that low temperature staining is essential to minimizing background fluorescence due to free fluorochromes, it is likely that any of a number of permeabilization procedures may be used. In this work, temperature and toluene were found to be sufficient. Yet a number of permeabilization methods have been 67 successfully used to analyze the attributes of single cells. Isoamyl alcohol (Maloney and Rotman 1973), triton X-100 (Wittrup and Bailey 1988), and toluene (Rotman 1961) have been used to permeabilize E coil and Saccharomyces cerevisiae. Clearly the choice will depend on the goals and specifics of the system of interest. The best feature of the protocols developed in this work is the use of the PKH26 negative control. Given that low signal to noise ratios can be expected when bacteria are to be analyzed using flow cytometry, due to their small size, this control allows the results to be embraced with a certain degree of confidence. However the ability of the PKH26 positive cells to accurately track background fluorescence relies on the use of cells that are similar in size to the cells to be analyzed. This reflects the nature of the tracking process. Since spurious signals result from the association of free fluorochrome with the surface of the cells care must be taken in order to minimize size variations that exist between the control cells and the sample cells. This was accomplished by using stationary phase control cells to track stationary phase sample cells. It would not be advisable to use exponentially growing cells of B. subtiis, which exhibit filamentous behavior, to track the accumulation of spurious fluorescence signals by stationary phase cells, which exist as single cells. The principal objective of this work was to develop procedures that would permit flow cytometry to be used to assay the activity of various promoters within a population of B. subtilis using commercially available equipment. Given that many of the promoters of interest were active during different stages of a batch culture, it was necessary to overcome the problems associated with weak signals and high 68 background while maintaining a flexible means of adapting the procedure based of differences in the cell physiology. The great diversity that exists between bacterial genera, in terms of cell size, morphology and membrane properties, and the numerous cultivation conditions and media employed, will most likely require some modification of the staining procedures described. However given the relative scarcity of such information, the conclusions obtained from these studies should greatly facilitate future developments. 69 Table 3.1 Optimized C.-FDG Staining Conditions Conditions Substrate concentration C 2-FDG (M) C-FDG 33 33 Sample concentration (cell/ml) 106 106 PKH26 lacZ- concentration (cell/ml) l 106 106 500C (10 min) 1% toluene (10 sec) 0 25 Permeabilization procedurez 3 Staining temperature (C) Staining time 8 hours 30 minutes 'These cells represent the internal PKH26 negative control population that was added to all samples analyzed (see text). 2The heat treatment was performed in the presence of 0.4gn/ sodium azide. 3 Following the 1% toluene treatment, the cells were diluted 1/100 into the staining solution. This results in a 0.01 % toluene concentration during sample staining. 70 Table 3.2 PKH26 Staining Conditions Substrate concentration (M) Reaction volume (l Diluent C)' 5 200 Cell concentration (cell/ml) 107 Staining temperature (C) 25 Staining time (minute) 2 5 Reaction quencher (g/l BSA solution) 3 20 1 Diluent C is the hydrophobic buffer supplied by the manufacturers of PKH26. 2 During sample staining, the reaction mixture was gently mixed. 3 The reaction mixture was diluted 1/10 in a solution of 20g/l BSA in PBSM. 71 Table 3.3 Spectral properties of fluorochromes commonly used in fi-Galactosidase fluorogenic substrates Fluorescent Excitation' Emission' Extinction' Quantum2 Product Maximum (nm) Maximum (nm) Coefficient (cm-lM-') Yield fluorescein 490 514 90,000 0.99 resorufin 571 595 55,000 f-methylumbelliferone 360 450 15,000 B-phycoerythrin 3 540 575 2,410,000 0.98 Obtained from Handbook of fluorescentprobes and researcherchemicals, Haugland, R., Molecular Probes, Inc., 1992, pp 77,81. 2 Mathies, 3 R.A. and Stryer, L (1986). Belongs to the phycobiliprotein class of fluorochromes. P-galactosidase fluorogenic substrate. 72 Not available as part of a Table 3.4 Staining of 50/50 acZ+/lacZ Mixtures (Temperature-Time Pulse Loading Matrix)1 Loading Temperature (C) 40 43 45 48 50 52 22 - - - - - - 5 - +- + + + 10 - +- + + ++ ++ 20 - +- -- ND ND ND -- ND ND ND ND Loading Time (min) 40 (-) (-) (+-) (+) Little staining High non-specific staining Two peaks identified Good staining (++) Excellent peak separation (ND) Not determined Following the heat treatment, the mixtures were incubated on ice overnight then analyzed by flow cytometry (see text). 73 Positive Promoter X lacZ _ I C)NA Beta-Gal Specific Hydrolysis Loading Cn-DG< Cn-FDG Non-specific "< Product_< Fluorescent Hvdrolvsis . . _. Negative Figre 3.1 Problems associated with staining cells with a fluorogenic -galactosidase substrate. In addition to the specific signals generated by substrate hydrolysis,spurious fluorescence can result from free fluorochrome present in solution. 74 AIA OH Mn H( H OH OH fn. U'. He < 'O H tfOHO OH 0 lpgurs 3.2 Comparison between the molecular structures of FDG (top) and C 2 -FDG (bottom). The C12 tail provides the fluorescing product with a higher affinity for the cell, thus increasing the specific signal associated with P-galactosidase positive cells. 75 A - 1.0 0 0 1.O a:ID o 0.5 0 500 540 580 Wavelength 620 (nm) Fguare 3.3 Emission spectra of fluorescein and PKH26 in ethanol. Excitation was performed at 488 nm. The differences in these spectra permit both fluorescence signals to be independently monitored. 76 Positive C8-FDG or C12-FDG Fluorescent Product Non-specific Hvdrnlvsin Non-Specific Staining PKH26 Negative Control Negative Fige 3.4 Schematic illustration of the application of the PKH26 internal negative control. Cells stained with PKH26 can be used to independently monitor the spurious increase in green fluorescence due to fluorescein related products free in the staining mixture. 77 Positive Cn-FDG 0-< Spontaneous Fluorescing Product Fgure 3. Schematic illustration of the experiment used to validate the use of the PKH26 negative control. Cell populations, both f-galactosidase positive and negative, were staining with different concentrations of PKH26 in order to independently monitor the increase in green fluorescence of each population. 78 &A s-r I 141 ··· · r······l·l··· · · c) 0) 0 r· ' · · ·· · · · · · · · · · r II · r r · · · · r r·rirYhC···r· · · · · · · r··i · · r·i··· · · · · · · · · · · · · · · · · · · · · · · ····· · · · ·r·· · · · ·r 1II LI· ·· b· · · I1C·ra··-------· · · rrr·r··y-·r-----I LI L1 · rC ry· · · · · · IY·Y··YLIYI··· r·r·· ·YYII· __ · · · · · · · ·· f63r··r L · · · · ···I· r·i· r· · L ·· · · · · ·· ,· · · r I· _r·· r· i · u· rr· · · · I · · · · · · · · · · · · · · · · · mrYIIYP··llr ·· r··leeLYI· r· ri""' ·r··r ·· ·· ·· ·r·r· · · IICAI-LC · · ··L···I·I······ · · · · · · · ·· -C · · · ···· · ru···· ·r -- · · ·· Fluorescence (fluorescein) Fxge 3.6 Two-color contour plot used representing the populations described in figure 3.5. The fluorescence increase of the two lacZ- populations, which contain low PKH26 fluorescence (i.e. bottom two populations), was monitored. Since the mixture contained fBgalactosidase positive cells, the fluorescence increase of the bottom two populations resulted from both dye leakage and spontaneous substrate hydrolysis. 79 4 C I, 00 0I. U 3 L. 0 2 0 i0 0 Cr. aD 1 (R 0 0 1200 2400 3600 Time (sec) Figure 3.7 Time course fluorescence increase of the lacZ- populations shown in figure 3.6. 80 50 la 0 L r 40 0 at 0 (4 . 30 4 O 0 a. a- (pJ 4-. C e0 IX. 20 10 L. 0 CL 0 0 10 20 30 40 50 Percent PKH26 Positives Added Figur 38 Recovery of PKH26 positive cells from mixtures containing PKH26 positive and negative populations. The abilityto recover the knownPKH26 positivepopulations indicates that the dye is stably maintained by cells. 81 100 75 ,. 25 O 0 20 40 60 80 100 120 20 40 60 80 100 120 100 4. 10 <10 0 Time (min) FIge 3.9 Time course decreases in l-galactosidase activity following incubation at various temperatures (A). The fact that the decay curves exhibit straight lines when plotted on a semi-log scale (B) indicates that the decay process is first order in //-galactosidase. 82 300 C E 0 - 2000%X E - 80% .0 100 90% C 0 I 45 _ I i . i 50 . · 55 Incubation Temperature (C) Figue 3.10 Deactivation contours generated from the decay kinetics presented in figure 3.9B. The area bounded by the X-axis and each line delineates the temperature and time domain where at activity can be maintained at least above the value indicated (i.e. the percentage of the original activity indicated by the various lines). 83 80 60 C E U IC S U a, 40 0L o m: 20 0 0 10 20 30 40 Incubation Temperature 50 60 (C) Figum 3.11 Average rates of fluorescence accumulation obtained from a series of constant temperature experiments performed with C 12 -FDG. -galactosidase positive and negative populations were individuallyincubated in the presence of the PKH26 control population and the average rate of fluorescence increase for all four populations determined. 84 0so A 40 20 0 0 C,) 4-' C- 200 400 600 600 1000 200 4000 0 0 800 1000 400 600 800 1000 60 B a) LLJ 40 0 20 E z 0 DU 0 C 40 20 0 0 200 Channel Number Figuir 3.12 Green fluorescence distributions resulting from 50/50 fi-galactosidase positive/negative mixtures stained according to the two-step staining procedure (see text). Panel A illustrates the case where little staining resulted. Panel B illustrates the case where two populations are somewhat detectable. Panel C illustrates the case where the two populations can easily be identified. 85 75 A 60 45 i 0 15 .. 0 200 400 600 600 _ 4- 1000 0 . . "1 a) LU 0 200 400 600 200 400 600 200 400 600 S00 1000 D 60 45 a) -o E zZL 15 0 0 200 400 600 500 1000 0 o00 1000 75 F 60 30 15 0 0 200 400 600 00 1000 0 Channel Number Figure 3.13 Representative 600 1000 Channel Number green fluorescence distributions obtained from a series of controlled mixtures stained with C 2-FDG under optimized conditions in the presence of the PKH26 negative control. The -galactosidase positive population added to each sample is 0% (A; i.e. 100% lac,), 5% (B), 25% (C), 50% (D), 75% (E) and 100% (F). percentages do not include the PKH26 background control population. 86 These CA cMj 4 I-r ··· ·· · · · r·r · r C. .. · ·· · ·· · ·)·. ICII r· =i; 32 .. c= · r·rr C) · I······ · o 0 · · · · · II· · ·.. I ·· .··. · · · · · *· 16 · ·.· ·······..· · · · · ·· I, .· II r·· · I· .. · · .·.·. · · · · · · · ·· · ·.... 11 r r · ,· · · · r·· · · r····, · · ,··r· r······ · r · · · · · · · · · · · · · · · · · r · I · I · · · I· ······ · · · · · r···· ·. · · · r·r· ·. · 111·.. III····I·· · r···i·I r····· r· · ·. · · · r · · · · · · · · · r · · r I· ·· , · r . · ···· · · · · · · · · · · · I. ..· · · · · r r · I· ·· · · · I · · · · · ·. · · ·· · .r· · · rr .ri· · · II r I r · · · · r· r··rii r · · ri · · · · · · ·. · · · · .····,·· · · . · · · · · · · I 16 r r r fluorescein) Fluorescence (fluorescein) Figure 3.14 The two color contour plot representing the 50/50 mixture shown in figure 3.13D. The top population represents the PKH26 negative control. The bottom two populations resulted from the 50/50 controlled mixture. 87 1.0 L- O O 0.8 00 a: C 0.6 0 0 I. 0.4 0 0 0.2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 Positive Fraction Added Figure 3.15 Quantitative summary of the populations recovered from controlled mixture experiments. The distributions presented in figure 3.13 were analyzed either by the use of fluorescence cutoffs, set by the green fluorescence distribution of the PKH26 negative control or by curve fitting (see chapter 8). Both methods yield approximately equilvalent results. 88 1.2 I - A 1.0 - U ab nlA L- I 0.6- O . 0.4- .I , 0.2An IIV_. I I I I C2 C4 C8 C12 C16 I I I C2 C4 C8 V A . %q I.Z - B VI1.0I o 0.80.6- 0.4 ._ - Ot 0.2- nn !l IJ L -- M ~--r-I C12 I C16 FDG Hydrocarbon Tail Figure 3.16 Summary of C.-FDG screening experiments used to identify C-FDG as the optimal substrate for staining sporulating cultures of B. subtiis. Panel A represents the rate C.-FDG hydrolysis obtained using a spectrofluorimeter. Panel B represents the average steady state fluorescence obtained using flow cytometry. 89 AA UV C 60 I I, I 0 IIL 40 .__ c: 20 0 0 5 10 15 20 25 Time (min) Fgwe 3.17 Time course increase in the average fluorescence, due to Cs-FDG hydrolysis, obtained using flow cytometry. 90 80 A 67X Above C Background 60 LU 0 40 .0 E z 20 0 0 200 400 600 800 1000 Channel Number .. W.V B 67% Above Background 4S " .. i· · ,11 · ...... e' . . : ·: ., %. · ·~~~~ .:I,. · · oil··· · o ".. '- .... ... .... .. ...... ... e . .· · .... . ,w ....... x · ^YI~ ............ @........ ..... ........... ..... .. ~ .- ·-*· i-,- .. : ii '. is *.---:; · · - @ ^ - *s__-@el @@@@@@ ·· · e@ @ .,.r ··· · _ .... :11 _....... ii...... ..... @-e·-@@ @ ... 4b @@*_ e @ r··· @ o YT@@ @ * @@ @--w @ @ _·· __ __ I. ___ ,U I l, Fluorescence (fluorescein) gue 3.18 C-fluorescein generated fluorescence distribution obtained from cells containing a spoVG-IacZ fusion (A) in the presence of the PKH26 negative control. The PKH26 negative control population is shown in the corresponding two color contour plot (B). 91 75 A a44C 0 w Bottom 26% :W 50 L Top 54 I; sorted sorted '6- 0 - e 25 - E z 0 l - 200 0 400 - - 800 600 1000 Channel Number 1.5 B 1.0 - 2:a3 0 0.5 - 'U -i 0.0 -~~~ - - Total High Expressing Low Expressing Cell Populations Fgure 3.19 Summary of the cell sorting results obtained from the sample presented in figure 3.18. 92 100 U 4- C 0 w A 60 4- 0 0 40 -0 E z 20 0 0 200 400 600 600 1000 0 ' 200 400 200 400 600 800 1000 an~~~~~~~~' . D C 0 w 0- 40 30 0 20 E 10 0 z 0 0 200 400 600 00 1000 0 Channel Number Figure 3.20 Representative 600 600 1000 Channel Number green fluorescence distributions obtained from a series of controlled mixtures stained with Cs-FDG under optimized conditions in the presence of the PKH26 negative control. Thef-galactosidase positive populations added to each sample are 0% (A; i.e. 100% lacZ, 21% (B), 49% (C) and 67% (D). These percentages do not include the PKH26 negative control population. 93 75 l, 0 60 0 45 0 0a, 0a 30 C 0 L. 15 0 0 15 30 45 60 75 Percent Positives Added Figure 321 Quantitative summary of the populations recovered from the controlled mixture experiments. These results were obtained from the distributions presented in figure 3.20 using a fluorescence cutoff determined by the green fluorescence distribution of the PKH26 negative control (3.18B). 94 30 · P 1.0 N 4.' :W a,M C) 0 20 Co .- 1.0 < o-n a. co a £0 o (n 0 o 10 0.5 a l 0£0 'U 0 a) 4, I, nn 0 0 2 4 6 8 10 Time (hr) FIgurm3.22 Time course correlation between the P-galactosidase specific activityobtained from bulk culture assays and the culture average fluorescence obtained by flow cytometry. ,8-Galactosidase activity was driven from the spolIA locus (KI938; chapter 9). 95 Chapter 4: Caraterization of Stationary Phase Populations of R subtiis 4.1 Introduction The hypothesis advanced in this work is that cell population heterogeneity plays a significant role in determining culture behavior. If this is indeed the case, it should be of particular concern as the culture enters the stationary phase. The reasons for this assertion is that upon entering the stationary phase, cells initiate a number of adaptive response mechanisms in order to adjust to the changing environmental condition of nutrient limitation. Included in these responses is the induction of a number of industrially important degradative enzymes, as well as those genes needed for spore formation. In this chapter the techniques developed and described in Chapter 3 are used to characterize the populations associated with stationary phase cultures of B. subtilis. Experimental evidence is presented that demonstrates that culture heterogeneity is established very early during the stationary phase. Additionally, though unobservable by bulk average measurements, this heterogeneity plays a significant role in explaining culture behavior. These conclusions follow from the finding that expression of some of the earliest sporulation or stationary phase marker genes is restricted to a subpopulation of cells. The simplest interpretation of these results is that two distinct cells types are generated. Furthermore, once these differences are established, each cell type adopts a different developmental program. Evidence in support of these statements was obtained by analyzing single cell gene expression in cultures of B. 96 subtilis using spo-lacZ fusions, the fluorogenic substrate C8-FDG, and flow cytometry. The results of these experiments indicate that the expression of early sporulation genes is cell-type specific. These finding suggests that induction into the developmental pathway of spore formation is and all-or-none phenomenon. Such behavior is consistent with the existence of threshold-like molecular mechanisms that control the cell's decision to initiate the developmental process. 4.2 Stationary Phase Protein Expression Great advances in our understanding of the molecular interactions that control stationary phase protein expression have been made over the past few years (Burbulys et a. 1991, Grossman 1991). While much still needs to be clarified regarding the sensing mechanisms used by the cell (i.e very upstream elements), it is now known that the environmental and physiological signals needed to activate stationary phase gene expression are transmitted via the transfer of inorganic phosphate through a multicomponent phosphorelay (Burbulys et al. 1991). When the appropriate conditions are satisfied, which include nutritional, cell cycle and intercellular factors, the principal role of the phosphorelay appears to be the activation, vWaphosphorylation, of the important transcription factor encoded by spoOA Once activated, SpoOA-P plays a major role in coordinating the cell's transition from exponential growth to the stationary phase. One of the earliest functions of SpoOA-P, is to indirectly activate the transcription of spoVG and spoOH,among other genes, by repressing abrB (Perego 97 et al. 1988, Strauch et al. 1990). AbrB is a constitutively expressed DNA binding protein that represses spoVG and spoOH(Perego et al. 1988, Strauch et al. 1990), among other genes. Although the biological role of the spo VG product has not been determined, its pattern of expression has made it an excellent marker for monitoring the initial events associated with the developmental process. In contrast, spoOH encodes for the sigma factor, ao , which is absolutely required for spore formation (Weir et al. 1984, Yamashita et al. 1986, Dubnau et al 1987, Weir et al. 1991). Following the activation of spoVG transcription, SpoOA-P directly activate transcription from the spolIA locus (Trach et al. 1991). Transcription from this locus is directed by the RNA polymerase complex containing aH. Since both phospho- SpoOA and a are required for spollA transcription, the pattern of spolIA transcription can be used to mark developmental events occuring after the initiation of spoVG transcription. The spolIA locus encodes, among other proteins, the sporulation specific sigma factor, uF, which is needed to establish differential gene expression in the mother cell and the forespore (Losick and Stragier 1992). This temporal sequence is illustrated in figure 4.1. In the following sections, the expression of spoVG-lacZ and spoIIA-lacZ are monitored in detail. 4.3 Single Cen Ge Exprsion apVG-lcZ Due to the central role that SpoOA plays in regulating the transition from exponential growth to the stationary phase, and its absolute requirement for spore formation, I sought to identify how the products of genes under SpoOAcontrol were 98 distributed throughout the cell population. Such information was considered important since the transcription of genes under SpoOAcontrol represent some of the earliest stationary phase phenomena. Therefore such knowledge could help to localize the stage, in the sequence of developmental events, where physiological differences between sporulating and non-sporulating cells might be established. This was accomplished using spo-lacZ fusions in conjunction with the cell staining techniques developed and described in the previous chapter. spoVG-lacZ expression is induced early during the stationary phase. Figure 4.2 presents the growth and P-galactosidase specific activity profiles obtained from a batch culture of B. subtilis(MB284; spoVG'-'lacZ) grown in nutrient broth (Schaeffer 1965). Cell growth was monitored using turbidity measurements at 660nm. The Pgalactosidase specific activity reflects the amount of o-nitrophenyl -D- galactopyranoside (ONPG) hydrolyzed by the "average" cell in the culture (i.e. it reflects a culture average measurement; see Chapter 9 for an explicit definition). galactosidase expression was directed from the spoVG promoter. In order to eliminate ambiguities associated with the fermentation time, it is customary to define a time scale relative to the point at which the culture departs from exponential growth. This convention was used in all experiments. The time course profile of spoVG-lacZ expression, presented in figure 4.2, indicates that 1) spoVG is expressed at low levels during exponential growth, 2) spoVG-IacZ expression increases rapidly soon after the culture enters the stationary phase and 3) spoVG-lacZ expression is transient, as noted by the decline in specific activity curve. Such results are consistent 99 with those obtained by other researchers (Zuber and Losick 1983, Mueller and Sonenshein 1992). Variations in bulk culture specific activity measurements reflect variations in the population demographics (figure 4.3). In addition to bulk culture activity measurements, samples at the indicated intervals were stained with C8 -FDG according to the protocol described in the previous chapter. The purpose behind this measurement was to determine whether the specific activity, as measured using bulk average assays, reflected the behavior of all cells or of a subpopulation of cells. The parallel profiles exhibited by the curves representing the size of the expressing population and the bulk culture specific activity indicate that bulk culture variations in the specific activity reflect variations in the size of the expressing population. Percent positive refers to the percentage of cellsthat were expressingf-galactosidase as judged by green fluorescence above background. Since the total cell numbers change little, if any, during the first several hours of the stationary phase in nutrient broth media, the percentage of the population staining positive reflects the absolute size of the positive population. Bulk culture measurements of spoVG-lacZ expression represent averages taken over a very heterogeneous population. The green fluorescence distributions, originating from C8 -fluorescein, obtained from samples withdrawn at different times during the stationary phase are presented in figure 4.4. Panel A represents the fluorescence distribution obtained from a sample harvested at time zero or To. This distribution indicates that heterogeneity is manifested very early during the stationary 100 phase. Also shown are the distributions obtained from samples harvested two and four hours into the stationary phase as denoted by T2 (B) and T4(C). Panel D shows the background fluorescence distribution associated with the T2 samples (i.e. the PKH26 negative control). The bimodal nature of these distributions indicate that the culture is comprised of two populations. One populations expresses relatively large amounts of f-galactosidase and the other expresses little if any. The distributions shown in figure 4.4 indicate that /-galactosidase is-unstable in stationary phase cells of B. subtilis grown in nutrient broth. Initially, between To and T2, the size of the expressing population increases. This is reflected by the increase in the area under the peaks at high fluorescence. Since the total cell numbers remained essentially constant from T 2 to T 4, gains in the size of the non- expressing population must have occurred at the expense of the expressing population. The degradation of f-galactosidase in sporulating cultures has also been found by other workers (Errington and Mandelstam 1986). Yet, the technique employed in these experiments provides a non-intrusive means of acquiring this information. Previous studies have relied on the use of protein synthesis inhibitors such as chloramphenicol. 4.4 spoVGO-acZce Sorting The fluorescence distributions obtained via flow cytometry reflect true differences in the level of intracellular P-galactosidase activity. The experiment described in the previous section was repeated in order to identify the level of P- 101 galactosidase present in the low expressing population. The T2 sample, which corresponded to the peak in bulk culture -galactosidase expression was stained, analyzed and sorted by flow cytometry. Similar to the results described in the previous section, a bimodal distribution resulted (figure 4.5). The high expressing population was estimated to comprise 67% of the total population, using a fluorescence cutoff set by the green fluorescence distribution of the PKH26 negative control (figure 3.18). Cells from both populations were sorted, according to the fluorescence gates shown, and the recovered populations were assayed for Bgalactosidase using the fluorogenic substrate 4-methylumbelliferyl -D-galactoside (MUG; see Chapter 9). Due to the well separated emission spectra exhibited by MUG and fluorescein (the emission maxima exhibited by these compounds being 450nm and 514nm respectively), use of this assay was possible without interference due to fluorescein. The greater sensitivity afforded by this assay (Harwood and Cutting 1990, Fiering et al. 1991) permitted as few as 5000 cells to be analyzed. Sorting analysis revealed that bulk culture specific activity measurements, obtained using the chromogenic substrate o-nitrophenyl-f-D-galactoside (ONPG) reflected the B-galactosidasecontent in the high expressing population. MUG assays revealed that the high expressing population expressed greater than twenty fold more f-galactosidase than the low expressing population on a per cell basis (inset to figure 4.5). The relative activities reported were obtained from the rate of fluorescence increase of each population as determined using a spectrofluorimeter, due to MUG hydrolysis, on a per cell basis (see Chapter 9 for more details). Each measurement 102 was obtained by assaying 50,000 cells. Normalized activities are reported relative to the specific activity of the total cell population, which was arbitrarily assigned a value of one. Based on estimates of the relative sizes of each population, 67% percent high expressing versus 33% low expressing, the contributions of each population to the total sample activity was determined (figure 4.6). These results allowed the conclusion that greater than 95% of the total culture activity resided in the high expressing population. Inherent in this analysis is the assumption that the specific activity of the top 54% of the sample population (i.e. the population that was actually sorted) reflected the specific activity of the top 67% of the population, as estimated using a fluorescence cutoff. 45 Sinl CellGeneE ixpress spoli-lacZ The results presented in the previous sections indicate that physiologically distinct cell populations appear very early during the stationary phase. If the heterogeneity observed in cultures expressing spoVG was sporulation specific, then heterogeneous expression of spoIIA-lacZ would be expected. This follows from the fact that transcription from this locus is known to occur after spoVG transcription. To investigate this possibility, experiments similar to those described in the previous section were performed on cells containing a spoIIA-lacZ fusions (MB296, spoIAlacZ::pPP81; see Chapter 9). Cultures expressing spoIIA-lacZ were found to be heterogeneous being comprised of high and low expressing populations. The time course profiles of the 103 bulk average specific activity (ONPG assays) and the proportion of the population expressing f-galactosidase, as directed from the spoIIA locus, are presented in figure 4.7. The observation that the size of the expressing population, at the point of maximal bulk culture expression of spolA-lacZ, is similar to the size of the expressing population obtained from cultures expressing spoVG-lacZ suggests the cells that earlier expressed high levels of spoVG-lacZ correspond to the population that later expressed high levels of spoIIA-lacZ. Together, these results suggest that the observed heterogeneity is a sporulation specific phenomenon. As in section 4.3, it is possible to conclude: 1) variations in bulk culture activity reflect variations in the size of the high expressing population, 2) bulk culture activity measurements principally reflect the behavior of a cell subpopulation and 3) spoIIA-acZ is transiently expressed. It is interesting to note that in contrast to spo VG-lacZ expression (figure 4.3), the decay time associated with the size of the population expressing spoIIA-lacZ is longer. This indicates that spoIIA-lacZ expression continues over a longer period of time, relative to spoVG-lacZ expression. 4.6 Correlation Between Spores and the Expressing Cell Populations Since the spoVG and spolA gene products are needed for spore formation, the previous results suggests that the cells expressing high levels of f-galactosidase, as directed from these loci, are the same as those that proceed to form spores. Therefore, a series of experiments aimed at investigating this hypothesis was 104 performed. If this hypothesis were true, a correlation between the size of the populations expressing various spo-lacZ fusions and the spore population would be expected. A correlation exists between the number of cells expressing spoVG-lacZ and spollA-lacZ and the number of spores. Figure 4.8 presents the time course profiles of the number of cells expressingfi-galactosidase as directed from the spoVG(A) and spoILA (B) loci and the total number of spores. Initially, the size of both positive populations attains a maximum, then eventually decreases. This was to be expected since these loci were found to be transiently expressed (sections 4.3 and 4.5). However, the fact that the decline in the number of cells expressing fi-galactosidase mirrors the later appearance of spores, strongly suggeststhat the cells that developed into spores earlier expressed these sporulation genes. It is interesting to note that when the sporulation frequency is compared to the fractionof the population that earlier expressed spoIIA-lacZ and spoVG-lacZ a correlation was also found (figure 4.9). A correlation between the fraction of the population developing into spores and the fraction of the population that earlier expressed a variety of spo-lacZ fusions was also found. Using different genetic backgrounds and different spo-lacZ fusions (figure 4.10), single time point comparisons were made between the sporulation frequency (at T12 for SMY strains or T18 for JH642 strains) and the proportion of the culture that earlier expressed f-galactosidase. Each datum shown in figure 4.10 was obtained from an independent experiment performed using one of four genetic 105 backgrounds (SMY, JH642, JH642 kinA::Tn917 and JH642 spoOKAC) and one of four spo-lacZ fusions. Plotted against the sporulation frequency is the percentage of the culture that earlier expressed P-galactosidasefrom the indicated spo-lacZ fusion. The time point at which the various samples were analyzed using flow cytometry coincided with the point of maximal bulk culture expression. Depending on the spolacZ fusion, this varied between T 2 and T4. To obtain cell lines that exhibited low sporulation frequencies, oligosporogenic mutants (i.e. cell lines exhibiting reduced sporulation compared to wild type) were employed. This included both kinA and spoOKmutants. Despite the use of a variety of strains and spo-lacZ fusions, and the fact that single time points, separated by well over ten hours, were compared, this correlation appears quite good. 4.7 Dicussion The immediate conclusions that can be drawn from these studies are: 1) that bulk measurements of gene expression, among other variables, in cultures of B. subtiis reflect averages taken over heterogeneous populations, 2) that heterogeneity results from the inability of certain cells in the population to initiate early developmental gene expression 3) that the subpopulation of cells that express various spo-lacZ fusions appear to be responsible for the appearance of spores and 4) that incomplete sporulation results from the inability of cells to express some of the earliest genes associated with the developmental process. The implications of these results, in terms of the processes involved in initiating development are further 106 discussed in Chapter 5. These conclusions suggest that entrance into the sporulation process is an all or none process. This is precisely what would be expected if threshold-like mechanisms control the initiation process. Furthermore, the decision made by the cell to initiate development occurs very early during the stationary phase. Such a conclusion is in direct contrast with the view that all cells express early sporulation genes and incomplete sporulation results from abortive processes along the developmental pathway. As alluded to in Chapter 2, the actual situation probably involves elements of both views. It will be shown in the next chapter that the ability of the cell to induce expression of one of the earliest observed sporulation genes, spoVG, is necessary but not sufficient for the cell to develop a spore. Such a finding is in agreement with the hypothesis that multiple threshold barriers must be overcome by the cell in order to successfullycomplete endospore formation. This chapter is concluded with a speculative observation regarding the inability of all cells in the culture to initiate spore formation. A explanation as to why high levels of P-galactosidase, as directed from different spo loci, are induced in only a fraction of the culture could involve cell-cycledependent factors (Mandelstam and Higgs 1974, Dunn et al. 1978, Ireton and Grossman 1992). Mandelstam and Higgs (1974) found that there exists an optimal time within the chromosomal replication cycle during which cells can be induced to sporulate. About fifteen minutes into the cycle, induction into the sporulation pathway was found to be optimal. Since cells in the culture would be expected to be at different stages of the chromosomal 107 replication cycle, uniform induction might not be expected. Theoretical analysis performed by Dunn and coworkers (1978), based on the concept of an optimal induction time, allowed the prediction that a maximum of 66% of the population could be induced to sporulate at any one time. To arrive at this number, Dunn and coworkers (1978) assumed: 1) that each cell contained multiple replication forks and hence chromosomes, according to the Cooper-Helmstetter model of chromosomal replication (Cooper and Helmstetter 1968), 2) that cells whose chromosomes had been replicated past the critical point were unable to sporulate, unless a new round of replication could be initiated and 3) that the chromosomal replication and the cell doubling times are 55 and 40 minutes respectively (for growth on hydrolyzed casein at 37C). Applying the age distribution theorem of Sueoka and Yoshikawa (1965), the 66% number was calculated. It is interesting to point out that when P-galactosidase, as driven from both the spoVG and spoIlA loci, was detected in single cells of the highly proficient sporulating cell line SMY, the maximum fraction of the population staining positive ranged from 6070% (figures 4.3, 4.5, 4.7, 4.9). 108 SpoOA Stationary Phase --------- (Transcription Factor) - Signals Nutritional Cell Cycle SpoOA-P Intercelular abrB -I- - spollA spoOH l I spoVG| + )I.- Activation Repression Figure 41 Schematic illustration of a subset of the events set in motion followingSpoOA activation. As illustrated, both spoVG and spolE4 are under SpoOAcontrol. 109 10 0l 200 o 0 CA E To C 0 (0 ff (0 a: 1 0uC, 4-1 n CD C 'j) 0 100 o_. 0.1 0 0 t 4-' 0 0.01 -4 ,< I I I I I -2 0 2 4 6 0 _ 8 Time (hr) F'gure 4.2 Time course profiles of growth (;OD60) and -galactosidase expression (0) as directed from the spoVG locus. 110 100 I L) 75 a CD 0 0 U) CD 100 o a_ O 0o" 25 O -2 O0 _. I I I I 0 2 4 6 0_,.,8 Time (hr) Figure 43 Time course profiles of the culture average gene expression (spoVG-lacZ) and the percentage of the population staining positive for L-galactosidase. 111 40 60 B A 38% Positive 73% Positive 40 0 20 4- 20 0 0 60 200 400 600 0 800 1000 0 200 400 1000 D 4% Positive 0) w 0 800 40 C 47% Positive C 600 40 20 L o .0 E 20 z 0 0 200 400 Charnel 600 800 0 1000 0 ~Naber 200 400 600 800 1000 Charnel Nunber Figue 4.4 Green fluorescence distributions obtained from culture samples harvested at To (A), T2 (B) and T4 (C) (figure 4.3). Panel D represents the green fluorescence distribution of the PKH26 negative control present in the T2 sample. 112 100 Specific Activity 0 C 75 4..m 1 Total Sort 1.4 High sort 0.07 Low sort I.- 26 Low Sort 50 0 L. -54X High Sort t.-o E z 25 0 0 200 400 600 800 1000 Channel Number Figure4S Green fluorescence distribution obtained at T2 from a culture expressing spoVGlacZ Also shown are the fluorescence gates used during cell sorting and the relative specific activity of each sorted cell population. 113 100- 7 101' U 75 E (0 50so 0 25 - 0 -l L I- - - - ____1 Total High Expressing I r - t Low Expressing Figure 4.6 Summary of the relative contributions of the sorted cell populations to the total culture average specific activity. 114 100 30 C) I 75 0 Z0 .,m 0 0 2o 0 50 0 To Q. 10e _,. 10 25 . v 0 -2 0 2 4 6 8 Time (hr) Figure 4.7 Time course profiles of the culture average gene expression (spoIIA-lacZ) and the percentage of the population staining positive for P-galactosidase. 115 600 co I w 400 0 E 0 E 200 0 0 5 10 15 20 0 5 10 15 20 600 to I w 400 O L. a E 200 z 0 Time .(hr) Figum 4.8 Time course profiles of the P-galactosidase positive () and the spore (O) populations obtained from cultures expressing spoVG-lacZ (A) and spoIlA-lacZ (B). 116 100 0 Lo 75 C) Cn la 0 ct O) 50 0 .a, 0 cl 14 A_ 25 )Q ._ 0 0 5 10 15 20 0 5 10 15 20 100 0 L o 75 la C, 0 0 enr c 50 0. 25 0 N_ 0 Time (hr) Fgure 4.9 Time course profiles of percent f-galactosidase positives (0) and spores (O) (spoVG-lacZ A and spoIA-lacZ; B). 117 ad 9 IUU 10 0 o0. ) .1 1 .1 .1 1 10 100 % Positive Figure 410 Plot of the percentage of the population expressing -galactosidase, from a number of spo-lacZ fusions, versus the sporulation frequency. The proportion of the population expressing -galactosidasewas evaluated between T2 and T4 near the peak bulk culture expression. The sporulation frequency was determined at least twelve hours after exponential growth ceased. 118 Chapter 5: Genetic Analysis of Threshold Phenomena 5.1 Introduction In Chapter 4, the techniques developed in Chapter 3 were used to characterize the cell populations present in stationary phase cultures of B. subtilis. One of the major conclusions reached is that culture heterogeneity, associated with the expression of early sporulation marker genes under SpoOA control, is established soon after the culture enters the stationary phase. These results imply that processes involved in activating SpoOA, or processes acting further upstream, are responsible for the observed heterogeneity. In this chapter, single cell measurements of gene expression are combined with genetic perturbations to explore how processes involved in generating phosphoSpoOA affect culture heterogeneity. The approach employed involves identifying how null mutations or deletions in stationary phase signal transduction genes, known to interact with the phosphorelay, alter the distribution of P-galactosidase as expressed from genes known to be under SpoOA control. The signal transduction genes involved are kinA and spoOK Since these mutations reduce the level of activated SpoOA in cells, the observed effects will allow us to infer how changes in the level of activated SpoOA affect gene expression. The characterization of the kinA (previously spoIIJ) gene product has recently led to a revival of the threshold hypothesis (Antoniewski et a 1990). kinA null mutations and spoOKC deletions cause partial defects in sporulation and reduced expression of early sporulation genes (Antoniewski et al. 1990, Perego et al. 1991, Rudner et al. 1991, Mueller and Sonenshein 1992). The reduced expression of these genes could be due to a decrease in expression in all cells and/or a decrease in the number of cells expressing these genes. The results presented in this chapter indicate that mutations in kinA and spoOK reduce the size of the cell subpopulations expressing certain early sporulation genes. Furthermore, the resulting subpopulations express at wild type (spolIG) or near wild type (spoIIA) levels, as determined using flow cytometry. These results support the hypothesis that SpoOA-P, or some component of the phosphorelay, must reach a critical threshold in order to activate early sporulation gene expression. Since the size of the populations expressing early sporulation genes correlates with the proportion of cells developing into spores (figure 4.10 and this chapter), this threshold appears to represent one of the major hurdles that cells must overcome for successful spore development. 5.2 Stationay Phase Signal Tansduction and Gene Expression When certain cell-density,nutritional and cell-cycle conditions are met, cells of the Gram-positive bacterium B. subtilis can differentiate into heat-resistant endospores. One of the key events controlling the initiation of sporulation is the activation, via phosphorylation, of the developmental transcription factor encoded by spoOA Phosphorylation of SpoOA involves the sequential transfer of phosphate through a phosphorelay comprised of the products of kinA, spoOF and spoOB(top of figure 5.1; Burbulys et al. 1991, Grossman 1991). KinA autophosphorylates and 120 donates phosphate to SpoOF. The phosphate is then transferred to SpoOB, and finally to the SpoOAtranscription factor. In addition to kinA, there appears to exist at least one other protein kinase, KinB, that serves as a source of phosphate for SpoOF (Burbulys et al. 1991). KinA is the best characterized (Antoniewski et al. 1990, Perego et al. 1989). The phosphorelay's involvement in the sporulation process appears to be the generation and possibly amplification of phospho-SpoOA. In contrast to other components of the phosphorelay, sporulation is absolutely dependent upon the SpoOAgene product. Double mutations that bypass the sporulation defects exhibited by kinA, spoOB and spoOF mutation have been identified (Shoji et al. 1988, Kawamura and Saito 1983). Such mutations have been shown to be alleles of spoOA that lead to alterations in the N-terminus of SpoOA. Additionally,mutations in spoOB and spoOF can be partially overcome through kinA over expression (Perego et al. 1989). These results support the notion that the kinA, spoOB and spoOF gene products act upstream of SpoOAand play a major role in regulating its intracellular concentration. Recent evidence suggests that multiple environmental and physiological signals are integrated through the phosphorelay network (Burbulys et aL 1991, Grossman 1991, Ireton et al. 1993, Bird et aL 1993). Nutritional, cell density and cell cycle signals have been found to be integrated through the activation of the SpoOA transcription factor (Ireton et al. 1993, Olmeda et al 1990). Through the isolation of spoOA mutations, resulting in various deletions or alterations in the N-terminal 121 region of SpoOA, transcriptional activation of spollA and spollG can bypass the normally required nutritional and cell density signals. Additionally, an altered function mutation in spoOA, rvtAll, can bypass the dependence of early sporulation gene expression on cell-cycle or DNA replication signals (Sharrock et al. 1984). The signal integrating function of the phosphorelay is further supported by the finding that the oligopeptide permease transport system encoded by the spoOK operon also provides signal inputs into the phosphorelay (Rudner et al 1991,Perego et al 1992). While the specific role that this operon plays has yet to be identified, mutations in spoOK have been found to result in partial sporulation defects. Additionally, over-expression of kinA can partially overcome mutations in spoOK (Rudner et al. 1991). The model presented in figure 5.1 summarizes these observation. Once activated, SpoOA serves to coordinate the cell's entrance into the sporulation pathway by setting in motion a series of developmental events that eventually result in the cell's "commitment" to the developmental pathway. The genes activated by SpoOA-Phave been discussed in great detail in Chapter 2 and are summarized in figure 5.1. The bottom part of this figure illustrates that upon SpoOA activation, a sequence of well regulated and temporally separated events are set in motion. The temporal sequence includes: 1) abrB repression, 2) spoVG transcription, 3) spolIG and spolIA transcription, 3) spolID transcription and 4) eventual spore formation. In the followingsections, single cell measurements of gene expression are used to track cells along this developmental sequence. 122 5.3 kinA Null Mutations: Culture Average Behavior 5.3.1 spaVG-lacZ Recent results indicate that null mutations in the kinA loci lead to an oligosporogenic,or reduced sporulation, phenotypes (Antoniewski et at 1990,Mueller and Sonenshein 1992). Such mutations have also been shown to reduce expression of certain early sporulation genes (Antoniewski et aL 1990, Mueller and Sonenshein 1992). The results presented in figure 5.2 illustrate that the expression of spoVGlacZ is unaffected by the presence of the kinA null mutation. These results, which are consistent with those obtained by Mueller and Sonenshein (1992), indicate that the reduced level of phospho-SpoOA resulting from kinA null mutations is still sufficient to repress abrB. 5.3.2 pllGacZ and poIA-lacZ In contrast to the pattern of spoVG-lacZ expression, transcription from the early sporulation genes spoIIA and spollG, as determined through lacZ fusions, is attenuated approximately one order of magnitude by kinA null mutations (figure 5.3 and 5.4). These operons encode, among other proteins, sporulation sigma factors that are needed to establish cell type specific gene expression in the mother cell and the forespore (Losick and Stragier 1992). Since transcription of these genes is directly activated by phosphorylated SpoOA, these results suggests that the level of phospho-SpoOA in kinA mutants cannot initiate transcription from these genes at wild-type levels. 123 5.3.3 spaxlDacZ Transcription from the spollD locus is known to require the gene products of the spollA and spoIIG operons (Rong et al. 1986, Clarke and Mandelstam 1987). The sporulation sigma factor ae, encoded by the second gene in the spoIIG operon (spollGB), is known to direct transcription from this locus (Clarke et al. 1986,Clarke and Mandelstam 1987, Errington 1993). The post transcriptional processing required for oB activation has been shown to involve the spollIA gene products (Errington 1993). Based on the requirements for spolID transcription, the pattern of spoIIDlacZ gene expression exhibited by wild-type and kinA cells lines would be expected to be similar to patterns exhibited by spoIIA-lacZ and spoIIG-lacZ expression. The time course profiles of B-galactosidase as directed from a spoIID-lacZ transcriptional fusion are presented in figure 5.5. These results, indicate that the adverse affects that kinA null mutations have on spolIA-lacZ and spoIIG-lacZexpression are propagated through to spoIID-lacZ expression. 5.4 InA Null Mutations: Population Demographics 5.4.1 p-alactosidase Expressing Populations The results presented in the previous section indicate that early sporulation gene expression, as determined from bulk culture assays, is reduced by kinA null mutations and this affect is propagated into later development. The question that arose was whether the observed reduction in gene expression was due to a small 124 population of cells expressing at wild type levels or whether gene expression was reduced in all cells. If the expression of SpoOA target genes, in kinA null mutants, was the result of a small cell population expressing at wild-type levels, a threshold level of activated SpoOA, or some component of the phosphorelay, would be implicated as a critical determinant in initiating spore formation. Alternatively, if kinA null mutations reduced SpoOA controlled gene expression in all cells, the appearance of spores might simply reflect the ability of a small population of cells to "escape" the effects of reduced levels of necessary sporulation gene products. Mutations in kinA cause a decrease in the proportion of cells that express high levels of early sporulation genes. Figures 5.6-8 present the time course profiles of the fraction of the culture expressing -galactosidase as directed from the different sporulation loci. The results were obtained by staining cell samples, at the indicated intervals, with C3-FDG according to the procedure outlined in Chapter 3. Since the total cell numbers change little, if any, during the stationary phase in cultures grown in nutrient broth, these percentages reflect the number of positive cells. These results indicate that standard measurements of gene expression actually mask the behavior of a small population of cells expressing high levels of P-galactosidase. Fluorescence distributions, reflecting C 8-fluorescein, obtained from samples harvested at the point of peak wild type expression are presented in figures 5-9-12. kinA fluorescence distributions were obtained from samples harvested at the same time point (as wild type). The fluorescence cutoff used to estimate the relative sizes of the two populations was set by the green fluorescence distribution of the PKH26 125 background control (figure 5.13). A comparison between the fluorescence distributions obtained from wild type and kinA cultures, containing spoIIG-lacZ, spoIlA-lacZ and spolID-lacZ reveals that kinA null mutations act to reduce the size of the expressing population. The fluorescence distributions obtained from cells containing spoVG-IacZ were unaffected, as could be anticipated from the specific igalactosidase activity profiles. 5.4.2 Spores The appearance of spores in wild-type and kinA null mutants was monitored during a time course experiment in order to determine whether the pattern of appearance was qualitatively similar to those exhibited early during development by the early sporulation marker genes. The time course of appearance of spores obtained from wild type and kinA::Tn917 cell lines are presented in figure 5.14. These results confirm the observation that kinA null mutations reduce spore formation approximately ten fold. However, in contrast to reports that kinA mutations lead to delayed sporulation on solid media (Perego et al 1989), these data indicate that no delay in spore development occurs in liquid cultures. Prior to the first datum, at 4.75 hours, no spores were detected. Furthermore, the similar population profiles exhibited by spores and positive populations (figure 5.6-8) provides further evidence in support of the conclusion that the spores evolve from the high expression populations. 126 5.5 inA Null Mutations: Distributions of spo Gene Products 5.5.1 Fluoreence Distributions The specific fluorescence of the samples analyzed by flow cytometry, obtained by averaging the C 8 -fluorescein fluorescence distributions (see Chapter 8), correlates with the bulk culture specific activity (figure 5.15). These data were obtained from the flow cytometer based analysis of the spollG-lacZ (A) and spoIID-lacZ (B) expression. Also reproduced for comparative purposes are the time course profiles of fl-galactosidase activity as determined through standard methods (figures 5.3 & 5.5). Similar results were presented in chapter 3 (section 3.8.6) for cells expressing spoIIA-lacZ. The positive population present in kinA cultures behaves similarlyto wild type. Table 5.1 contains specific fluorescence data, obtained from a finer analysis of the bulk culture data (figure 5.15) via the histograms presented earlier (figures 5.9-12). Two methods of analysis were employed in the analysis of the fluorescence distributions. In the first method, the specific fluorescence was obtained by averaging over all cells (i.e. the approach used to generate figure 5.15). In the second method, the average was taken solely over the high expressing subpopulation. A comparison between the specific fluorescence obtained from each wild type-kinApair reveals that differences apparent at the total population level essentially disappear (or are diminished for spoIIA-lacZ) when the average is taken over the high expressing subpopulation. These results indicate that the high expressing populations in kinA cultures behave similarly to the high expressing population in wild type cultures. 127 Furthermore, cells that are capable of activating early sporulation gene expression appear to be responsible for the residual sporulation (table 5.1; spores versus positives). 5.5.2 spaPD-acZ Cel Sortin Enzymatic assays performed on the high expressing subpopulations of both wild type and kinA cells indicated that both populations behave similarly. Wild type and kinA cells containing a spollD-lacZ fusion were harvested two hours into the stationary phase (cultivated in DSM), stained with C8 -FDG, analyzed by flow cytometry and sorted according to green fluorescence. The resulting green fluorescence distributions for wild type and kinA cell populations are presented in figure 5.16. In contrast to the distributions presented in the earlier figures, the wild type and kinA distributions are comprised of 100,000and 500,000events respectively. By increasing the number of events taken from the kinA sample, the bimodal distribution is more apparent (inset to figure 5.16B). Also shown are the two parameter contour plots used to define the sort window (the rectangles shown in figures 5.17A & B). The X-axis represents the green fluorescence signal and the Y- axis represents the right angle light scatter measurement. That both high fluorescent populations behave similarly is shown in table 5.2 where in addition to the data presented in table 5.1, the relative specific activityof the total and the high expressing populations are shown. These results indicate that the level of spoIID-lacZ expression in both high expressing populations is essentially identical and thus provide 128 direct support for the contention that the wild type and kinA high expressing populations behave similarly. 5.6 Observations on spaA Gene Expression The effect of the kinA mutation was to reduce the fraction of cells expressing spoIIA-lacZ (table 5.1). However, for spoIA, the average fluorescence per cell in the expressing population was found to be lower in kinA mutants relative to wild type (table 5.1). This difference could reflect differences in the binding affinity of SpoOAP for the spolIA regulatory region. This difference might also reflect the fact that transcription of spollA is directly controlled by aH (Wu et al. 1989, Wu et al. 1991), the spoOH gene product, while transcription of spollG is controlled by aA (Kenney et al. 1989). Yet, since the size of the expressing population generally reflects the proportion of cells eventually developing into spores, it is concluded that the level of spoIL4 gene products in cells containing kinA null mutations is still sufficient to carry out its in vivo roles in spore formation. 5.7 The Effects of spOKAC Mutations The spoOK operon has also been found to play a role in the signal transduction processes leading to the activation of SpoOA(Rudner et al. 1991, Perego et al. 1991; figure 5.1). Similar to kinA null mutations, mutations in this operon lead to an oligosporogenic phenotype. This operon encodes five proteins that exhibit significanthomologywith the oligopeptide permease (opp) operon of the Salmonella 129 typhimurium (Perego et al. 1991,Rudner et al. 1991) and has been postulated to play a role in sensing and importing extracellular signaling peptides (Grossman and Losick 1988) needed for efficient spore formation. While the exact role that this operon plays in transducing early stationary phase signals remains unknown, its involvement in these processes has been deduced by the finding that overexpression of the kinA gene product can partially overcome spoOK derived sporulation defects. Previous experiments indicate that deletions in the spoOKC, the third gene in the operon, lead to an oligosporogenic phenotype (Rudner et al. 1991). Deletions in spoOKC also cause a decrease in bulk culture measurements of P-galactosidase as directed from the spoIIG locus (figure 5.18). Given its postulated role in sensing and transmitting stationary phase signals, it was of interest to identify if the pattern of single cell gene expression was similar to that resulting from kinA mutations. Null mutations in spoOKCcause a decrease in the fraction of cells that express high levels of spoIIG-lacZZ. The time course experiments presented in figure 5.18 suggests that little or no p-galactosidase is expressed in a spoOK mutant. However, analysis of single cell gene expression indicates that the bulk culture profiles conceal a small population of cells expressing high levels of spoIIG-lacZ (figure 5.19; table 5.3). These distributions were obtained from the Tz75 samples shown in figure 5.18. As with the kinA mutations, a comparison between the size of the population expressing high levels of spoIIG-lacZ and the proportion of cells that developed into spores (table 5.3) reveals a correlation. These results provide additional support for the conclusions drawn in the previous section and further reinforce the postulated 130 role that the spoOK operon plays early during sporulation. 5.8 Discussion 5.&1 Threshold Activation of Sporulation The principal conclusion obtained from the results presented in this chapter is that threshold mechanisms control entrance into the developmental pathway of spore formation. Based on these results, the results presented in Chapter 4 and, what is known about early gene expression in B. subtilis, it is concluded that the threshold is associated with the phosphorelay. This conclusion is illustrated in figure 5.20 where the effect that a kinA mutation has on the concentration distribution of a phosphorelay component or the level of signal input into the phosphorelay is schematically illustrated. Similar to the distributions presented throughout this work, the abscissa represents the amount of the "threshold" molecule and the ordinate represents the number of cells. The effect that kinA, or spoOK,mutations have on the level of this threshold molecule is schematically represented reduction throughout the cell population. by a uniform Therefore, for a fixed threshold concentration as set by the cell, such mutations act to reduce the number of cells above the critical threshold so that fewer cells are capable of entering the developmental pathway. Cells that are able to overcome this threshold barrier initiate stage II gene expression and eventually develop into spores. While it can be inferred with a great deal of certainty that the level of phospho-SpoOA is reduced, on the average, in cells containing the kinA or spoOK 131 mutations, the results presented in this chapter are not able to discriminate how the level of phospho-SpoOA is distributed throughout the population. If the level of signal inputs into the phosphorelay must reach a threshold, then one might anticipate that the distribution of phospho-SpoOAwithin the population would be high or low. In contrast, if the level of some phosphorelay component must need to reach a threshold level, then it is likely that the level of phospho-SpoOA may need to reach a threshold. This statement followsfrom the assumption that the components of the phosphorelay; KinA, SpoOF, SpoOB and SpoOA, and their phosphorylated counterparts are close to equilibrium so that changes in the level of each component leads to equilibrium driven changes in all other components. In Chapter 6, a threshold model is developed in order to provide an unambiguous framework in which to analyze these possibilities. This model will also be shown to be consistent with the results presented in this thesis. 5.82 Multiple Thresholds If one accepts the operational definition of a threshold, namely; the mechanism by which small or continuous changes in a parameter or variable are dramatically amplified so as to lead to very different outcomes, then it is likely that multiple thresholds exist. This follows from the molecular mechanisms known to regulate early stationary gene expression in B. subtilis. The first developmental switch appears to be associated with the AbrB pathway (figures 2.3A & 5.1) which regulates spoVG expression. AbrB exists as a 132 hexamer in vivo and it is the hexameric form that is responsible for repressing transcription from a number of stationary phase genes (Strauch et a. 1989). Additionally,AbrB has been found to repress transcription, from certain loci, through a cooperative binding process (Strauch et al. 1989B). Therefore small decreases in the in vivo level of AbrB, resulting from both degradation and decreased transcription via repression by phospho-SpoOA, can be expected to have an amplified effect. These observations, along with the pattern of spoVG expression in kinA mutants suggests that only a small amount of phospho-SpoOA is required to overcome this threshold. The second threshold appears to involve the repression of stage II genes via SinR and possibly GsiA (Mueller and Sonenshein 1992, Bai et al. 1993). Mutations in sinR or gsiA have been found to bypass the partial sporulation and gene expression defects exhibited by kinA mutants (Louie et al. 1992, Mandic-Mulec et aL 1992, Mueller and Sonenshein 1992). Also, mutations in sinR result in increased sporulation and stage II gene expression while over-expression has the opposite effect (Mandic-Mulec et al. 1992). It also appears that SinR represses spoOA expression (AD. Grossman, personal communication). While very little is known about the mechanisms by which phospho-SpoOA inactivates GsiA, it is known that SinR is inactivated by the developmentallyexpressed sinI gene product. The functional role that SinR plays in repressing certain target genes appears to be inactivated by protein-protein interactions with the sinIgene product (Bai et al 1993). Since SinR appears to perform its role as a tetrameric complex, the mechanism by which cells 133 relieve this repression might very well mimic the philosophy employed by cells to inactivate the lac repressor, which also functions as a tetramer (Barkley and Bourgeois 1980). Finally, because sinI expression relies on intact spoOA and spoOH genes (Bai et al. 1993), it is likely that phospho-SpoOA is involved in countering the effects of SinR. Based on the results presented in this work, this threshold appears to require higher levels of phospho-SpoOA. In Chapter 6, a model is developed in order to demonstrate that these threshold arguments are consistent with what is known about early stationary phase gene regulation in cultures of B. subtilis. This model is used to argue that the AbrB and the SinR checkpoints are associated with the need of the cell to accumulate a threshold signal input into the phosphorelay and to subsequently amplify this signal via the phosphorelay. In the context of this model, the spoOK gene products are hypothesized to be associated with the signal input threshold. The kinA gene product is hypothesized to be associated with the need for signal amplification. 5.83 Implications of a Threshold Mechanism The existence of threshold mechanisms in biological systems has several important implications: 1) systems controlled by threshold mechanisms can be expected to be sensitive to initial conditions and to slight variations in cultivation conditions and 2) stochastic and deterministic processes will appear to govern such systems. The first implication is a natural consequence of the definition of a threshold. Namely that small changes in the concentration of a particular molecule 134 can lead to dramatics changes in the behavior of the system. Because there is no way of knowing a pnon which "small" changes are important, system variability must be accepted as the rule, not the exception. It is therefore not surprising that complex biological systems, particularly those involving differentiating systems such as cultures of B. subtilis, often exhibit day-to-day variability. Despite the deterministic nature of the molecular mechanisms that lead to threshold behavior, the ability of cells to initiate the developmental process of spore formation also involves stochastic elements. The existence of random factors that control the initiation into sporulation was demonstrated by the early work of Schaeffer (1965) and Dawes and Mandelstam (1970). They found that there exists a finite probability that a cell would initiate spore formation under a given set of cultivation conditions. More recently, the results reported by a number of researchers (Weinrauch et al. 1990, Rudner et al. 1991) permit a similar conclusion. These researchers determined the sporulation probabilities (i.e. frequencies) in various signal transduction mutants involvingsingle mutations (in kinA, spoOK and come and double mutations (kinA spoOKand kinA comP). All of these genes are believed to be involved in regulating the initiation process. These researchers noted that the sporulation probabilities observed in the double mutants were more severe than in any of the single mutants. However, further analysis of their data indicates that the sporulation probabilities observed in the double mutants simply reflected the multiplicative probability of the individual mutants. This is illustrated in figure 5.21 where the calculated probabilities, as determined from the individual mutants, are 135 plotted versus the experimentally determined double mutant sporulation probability. These data indicate that stochastic elements are involved in determining which cells in the population accumulate enough activated SpoOA to initiate spore formation. 136 Table 5.1 Sporulation and P-Galactosidase Population Data Fluorescence % % all positive spores' positive 2 cells3 cells4 26 30 22 64 strain fusion kinA KI1179 K11398 spoVG-JacZ + spoVG-lacZ - 0.5 31 23 69 SIK122 AG1350 spoIIG-lacZ spoIIG-lacZ + - 32 2 23 4 11 2 71 62 KI1183 KI1397 spoIID-lacZ spoIID-lacZ + - 22 1 25 5 16 3 85 64 KI938 spoIIA-lacZ + 21 33 41 109 KI1001 spoIIA-lacZ - 4 2 5 57 l Percent spores were determined between 12 and 18 hours after the end of exponential growth as described in chapter 9. Percent positive refers to the percentage of cells that were expressing galactosidase as judged by green fluorescence above background. 2 /B- 3 Average relative fluorescence (all cells) is the average linear fluorescence above background per cell. In all experiments > 10,000 events were counted. Average relative fluorescence (positive cells) is the average fluorescence above background per cell in the positive or expressing subpopulation. 4 137 Table 5.2 Sporulation and spoIID-lacZ Expression Data genotype wild type kinA % spores' % positive2 Specific Activity3 all positive cells cells Fluorescence 4 all positive cells cells 17 19 0.21 1 0.25 1 0.5 2.6 0.06 0.93 0.09 1.2 1 Percent spores were determined 18 hours after the end of exponential growth as described in chapter 9. 2 Percent positive refers to the percentage of cells that were expressing - galactosidase as judged by green fluorescence above background. 3 Specific activity was determined by assays 44,000 cells from each population (i.e. total population sorts and positive population sorts) using the MUG assay. The wild type positive population has been normalized to unity for comparison purposes. 4 Same as in Table 5.1 except that the data has been normalized with the wild type positive population set to unity. 138 Table 5.3 Sporulation and spollG-lacZ Expression Data in Oligosporogenous Mutants strain genotype SIK122 wild type AG1350 1 Percent % spores' % positive2 Fluorescence all positive 3 cells cells4 31 20 16 120 kinA::Tn917 3 6 4 140 spoOKAC 1 4 2 90 spores were determined 18 hours after the end of exponential growth as described in chapter 9. 2 Percent positive refers to the percentage of cells that were expressing/f- galactosidase as judged by green fluorescence above background. 3 Average relative fluorescence (all cells) is the average linear fluorescence above background per cell. In all experiments 2 10,000events were counted. 4 Average relative fluorescence (positive cells) is the average fluorescence above background per cell in the positive or expressing subpopulation. 139 Stationary Phase Signals KnA Signal Transduction (Phosphorelay) * SpoOK OF wKinA-P OB .Kin-P f OF-P Kt Kin + Activation sp oIIG OB-P OA A-P1 1 spollA soabrB spollA spoVG I- Repression Transcriptional Activation spollD spore Figure5.1 Summary of stationary phase signal transduction via the phosphorelay and transcriptional activation va phospho-SpoOA. Upon entrance into the stationary phase, the the spoOKgene products provide input signals to the phosphorelay (KinA, SpoOF, SpoOB and SpoOA) 140 150 4-0 O O 100 ._ C/) CD co m 50 ar O 0 I n -1 0 1 2 3 4 Time (hr) Figure 5.2 Time course profiles of culture average P-galactosidase expression as directed from the spoVG locus in wild type () 141 and kinA null mutants (O). 60 4._ 4:0 O C, . 40 0 0O 20 la 0 0 e4. I Ca 0 -1 I 0 1 2 3 4 Time (hr) Figre 5.3 Time course profiles of culture average B-galactosidase expression as directed from the spollIG locus in wild type () 142 and kinA null mutants (O). 100 .l 4. 0 75 o cY 0a, 0 50 -o 0 0 0t0o 25 I cm 0 -1 0 1 2 3 4 Time. (hr) Figure 5.4 Time course profiles of culture average Bf-galactosidase expression as directed from the spollA locus in wild type () 143 and kinA null mutants (). 40 a. O O 30 0 5. co 20 to C, 10 I 0 0 1 2 3 4 5 Time (hr) Fiure 5.5 Time course profiles of culture average P-galactosidase expression as directed from the spolID locus in wild type (0) and kinA null mutants (O). 144 30 0 20 ,_ H 13. eQ 10 0 -1 0 1 2 3 4 Time (hr) Figure 5.6 Time course profiles of the proportion of the population expressing galactosidase as directed from the spoIIG locus in wild type () mutants (0). 145 and kinA null A 4U 30 0O CL 10 0 m 1 0 1 2 3 4 Time (hr) Figure 5.7 Time course profiles of the proportion of the population expressing igalactosidase as directed from the spoILA locus in wild type () mutants (O). 146 and kinA null 30 020 . 10 0 0 1 2 3 4 5 Time (hr) Figure 5.8 Time course profiles of the proportion of the population expressing fBgalactosidase as directed from the spoIID locus in wild type () and kinA null mutants (). 147 feI n ou (D40 LU 0 -0 E z 20 0 0 200 400 600 800 1000 0 200 400 600 800 1000 r,,^ ou ,) c- a) 40 LL 0 4o 0) E 20 0 Channel Number Fisure 5.9 Green fluorescence distributions obtained from cells expressing spoVGlacZ in wild type (A) and kinA null mutants (B). The samples analyzed were harvested at T 2 (figure 5.2). 148 r, TOU C') C- () 40 LUJ 0 L) () E 20 0 0 200 400 600 800 1000 0 200 400 600 800 1000 ou n _ U) Cn (i 40 LU 0 ,,) E 20 0 Channel Number Figure 5.10 Green fluorescence distributions obtained from cells expressing spoIIGlacZ in wild type (A) and kinA null mutants (B). harvested at T2S (figure 5.3). 149 The samples analyzed were OU (I C- a) 40 LU a) E 20 0 0 200 400 600 800 1000 0 200 400 600 800 1000 f:' ou ,n a) I1 40 L a) - 20 0 Channel Number Figure5.11 Green fluorescence distributions obtained from cells expressing spollAlacZ in wild type (A) and kinA null mutants (B). The samples analyzed were harvested at T2 (figure 5.4). 150 r' rX ou LI 0 a) .0 E 20 0 0 200 400 600 800 1000 0 200 400 600 800 1000 f"n ou U) ) 40 LU 0 () E z 20 TO Channel Number Figure 5.12 Green fluorescence distributions obtained from cells expressing spolDlacZ in wild type (A) and kinA null mutants (B). The samples analyzed were harvested at T3 (figure 5.5). 151 60 a, 40 < 0.1% Positive Hij N- o0 E 20 z 0 0 200 400 600 800 1000 Channel Number Figure 5.13 Green fluorescence distributions obtained from the PKH26 internal negative control present in the wild type spoIIG-lacZ sample (figure 5.10A). 152 25 20 u) 15 QI en 10 5 0 0 5 10 Time 15 20 (hr) Figure 5.14 Time course profiles of the sporulation frequency in wild type (0) and kinA null mutants (). 153 -d - 13 -n 0(D C 60 (D E3 o c, 40 5c C, (y) - Q 0 0 D (D 0 0 1 2 3 -- 4 4U -I ,U (D O '0 30 (I 15 0 D (D 10 20 ,cI C, C-) ° 10 0 (D 0 0- 0 1 2 3 4 5 Time (hr) Figure 5.15 Correlation between the culture average specific activity, obtained from bulk culture assays, and the culture average fluorescence, obtained from flow cytometry. Panel A represents spoIIG-lacZ expression (figure 5.3) and panel B represents spoIID-lacZ expression (figure 5.5). 154 400 A C 0 300 0 200 E 100 z 19% Positive 0 200 400 B 2000 )O w 0 L- e .0 1000 E z 0 0 200 400 600 800 1000 Channel Number Figure 5.16 Fluorescence cutoffs used to sort and quantitate wild type (A) and kinA (B; inset) cell populations expressing high levels of spoIID-lacZ 155 230EC06 Real-time 20 dot disp CHUNC SSC ·. r vs FITC ru.rum.c .·o i~~~~~:'·.. ·. . .'· ....a .:·....·· ·. ·.. -.:..,t· '. ·- . ..'.. · ._"-.'' · _-,-" ;,'._·*' t;.; ~... .... .41~~~~~~~~~~~~~~~.· I ', *·,·_ * *. :: _P::1__-. - .:.. _,...>::=== ;= ~.'l~;, *:.. , ,..,". ___ , ~"..__._:; .,.~r-S:_ .....'.dl.·,, -,rP,.*P1.~e Wmmm m".":":,. 6 .* r s , · *· ,. * .-. ·.. ' ' . .:-. ',·. ''.-' ·: · · .. =::..r... r .. =.·.: . ~-"~ .': ,-.. ... _.~-.:..w,~-...--.:~:,~ : ;&r.' o. ..r· -.. '-.--._ :.·.u z' _ z ~ W,,O e[,td. :.,,. ,l.S'-.! ::,' .-. .::~-.~.~.--:.. * ·· ' *;-{,' ':-'.'I-,~' ..:.··..*·..'.':".".. -..:,....-e.,~ -.{:.~,,; . ". '8·.~':.:...' -;: -'··..... :.,. · · ':''.: _ :..·-.~',;,7 -. '* ' ' .-* .. .' ·. · . · - Real-time 20 dot disp 230EC07 CHUNG l . - ... :..' -: .. .. *:.. .'-;....... .- : *: ' ·. .....- ,..".--.. ,..: - - vs FTC SSC · :. :. . .. ·- . .* *..:. .·, ., i..: . *.. ;;... i i. · ·''.: . · '': 1 - Figue 5.17 Two parameter, right angle light scatter (Y-axis) and green fluorescence (X-axis), dot plots displaying the windows used to sort wild type (top) and kinA (bottom) cell populations expressing high levels of spoIID-lacZ 156 U 150 0 100 0 0 44 .' (U) f, la .z 50 (9C 1J ea 0 -1 0 1 2 3 4 Time (hr) Figure 5.18 Time course profiles of spoIIG-lacZ expression in various isogenic cell lines. 157 4M IW 100 A B C, a) 75 20% Positive 75 6% Positive w I4- O 50 50 25 25 I- c~ E Z 0 0 200 400 600 800 100 0 1000 0 400 600 800 1000 60 C C u) 200 D 75 4% Positive 40 1.6% Positive 4- 50 D E Z 20 25 0 0 0 200 400 600 800 1000 Channel Number 0 200 400 600 800 1000 Channel Number Figure 5.19 Green fluorescence distributions obtained from cells expressing spolIGJacZ in wild type (A), kinA (B) and spoOKAC mutants. The samples analyzed were harvested at T27 5 (figure 5.16). Also presented is the green fluorescence distribution of the PKH26 negative control present in the wild type sample. 158 IThreshold Initiation Mechanism I Wild Type Onto Spores V- 0 kinA \ I .0 E I1 z , ' I ID Ir I I II I I I II Threshold Phosphorelay Component or Signal Inputs Figure 5.20 Proposed threshold initiation mechanism used by cells of B. subtilis to control early developmental gene expression. 159 10-2 0 V 10- 3 -0 0 0 10-4 C · - kinA spoOK v - kinA spoOK141 Q U 10-s * - kinA comP L_ * - kinA spoOK L * - kinA spoOK141 10 / I,,,,,,1 110-6 Ig 10-5 I wlW1 I I I IW~g1 I . 10 - 4 I I I 10-3 10-2 Sporulation Frequency (Experimental) Figure 5.21 Comparison between the sporulation frequencies exhibited by various double mutants. The ordinate represents the sporulation frequencies calculated from the single mutant sporulation frequencies assuming the mutations acted independently. The abscissa represents the experimentally determined double mutant sporulation frequencies (the spoOK data was obtained from Rudner et. al. 1991 and the comP data was obtained from Weinrauch et. al. 1990). 160 Chapter 6: Quantitative Analysis of Threshold Phenomena 6.1 Introduction In the previous chapters the concept of a biologicalthreshold was introduced in order to explain how genetically identical cells, in the same environment, could adopt dramatically different developmental fates. In this sense, a threshold is meant to describe how small changes in a system variable or parameter can elicit "catastrophic" changes in the system behavior (Arnold 1986). Experimental results were also presented, suggestingthat threshold-like mechanisms control entrance into the sporulation pathway. The purpose of this chapter is to show that threshold mechanisms can be predicted theoretically in a manner that is fully consistent with what is known about gene regulation in a number of biological systems. The generality of threshold mechanisms is argued for by showing they control adaptive gene expression in two paradigm biological systems, phage A and the lac operon. Since both systems are well characterized, both qualitatively and quantitatively, a theoretical analysis of how the biological processes that regulate these systems is possible. The principal goal of this analysis is to identify the biological control structures responsible for the existence of threshold behavior. By restricting the analysis to the experimentally determined parameter space found in the literature, the identification process can proceed without the ambiguities that would be associated with less characterized biological systems. Quantitative analysis will be used to show that threshold-like mechanisms are 161 initiated when the actively maintained balanced growth state of the cell is disrupted, by environmental and physiological forces, beyond the cells ability to compensate. Single cell material balances on key transcriptional activators are used to study the relationships that exist between the various rate processes that determine the intracellular concentration of these transcriptional activators, and hence the transcriptional state of the cell Consistent with what is known about microbial growth, it is shown that the cell stably maintains a homeostasis or balanced growth state, by balancing these rate processes. However, when encountered with extreme and prolonged disturbances, the capacity of the cell to maintain its original state is exceeded and adaptation becomes the only course of action available to the cell. Adaptation is shown to coincide with the crossing of a threshold and this ultimately results in the cells acquisition of a different transcriptional state. Beneath the obvious differences that exist between the lac, the A and the B. subtilis sporulation systems, a common control "structure" is shown to regulate adaptive gene expression in all cases. This structure and its implications were derived from quantitative studies on both the ac and the A systems. So long as the intracellular concentration of the transcriptional activators responsible for initiating adaptive gene expression is regulated by a positive feedback loop, a non-linear generation rate, of the cooperative type, and a concentration dependent removal rate, threshold potential exists. This finding has allowed a simple conceptual framework to be put forth that is capable of succinctlyorganizing the plethora of information available on early stationary phase gene expression in B. subtilis. This includes, and 162 is not limited to, the data presented in the previous chapters. The pervasiveness with which this control "structure" can be found in nature suggests that it represents a conserved biological mechanism used by microbes to efficiently cope with changing environmental conditions. 6.2 Mathematical Conceptualization Figure 6.1 illustrates how the biological systems analyzed are conceptualized both physically and mathematically. It is assumed that physiologically distinct cell types represent cells with distinct transcriptional states. The focus is placed on identifying and understanding the processes and mechanisms that control the transition between states. Material balances on single cells are then used to translate what is physically known about these systems into a mathematical model. The physical states are assumed to be representable by stable mathematical states identified by the governing equations. Insight into the physical origins of threshold phenomena are obtained by identifying and analyzing the mathematical conditions required for the existence of threshold behavior. The consistency of these results is then checked against experimental observations. 63 PhageA 6.3.1 Bacgound Under the appropriate conditions, lysogens of E. coli may opt to initiate lytic development (figure 6.2, Jacob and Monod 1961, Bailone et al. 1979, Ptashne 1987). 163 Once induced, a distinct pattern of gene expression is initiated. After some time, this primitive developmental process culminates in the maturation of phage particles and the subsequent lysis of the host. However it has also been found that depending on the induction conditions, not all phages initiate this developmental program (Bailone et al. 1979). Such a situation results in a heterogeneous culture comprised of both lysogens and phages undergoing lytic development. Since dramatically different phage types coexist under nearly identical environmental conditions, this suggests that mechanisms leading to threshold-like behavior are involved. It is now known that the decision made by lysogens is controlled by the intracellular concentration of the CI transcription factor, or JArepressor (Ptashne 1987). Repressor is a DNA binding protein that is actively expressed in lysogens. Repressor functions as a dimer and its principal role is to maintain lysogeny by repressing cro, a lytic specific gene. Figure 6.3 illustrates how CI dimer represses cro. cI and cro are adjacent on the Achromosome and transcription from these loci occur in opposite directions via the PRMand PR promoters, respectively. Segments of both promoters comprise, in part, the OR operator and it is the state of this operator that determines which promoter is utilized. In the lysogenic state, repressor binds to OR near cro and this effectively prevents cro from being transcribed. At the same time, this binding configuration allows cI to be transcribed from PRm. Conversely, during lytic development the state of OR is such that cro is transcribed and cI is repressed by Cro (figure 6.3B). Hence the transcriptional states of OR directly determine the fate of A. 164 In the following sections, the processes that control the transition from lysogeny to lysis are analyzed quantitatively. The goal is to quantitatively identify the key processes responsible for this switch in order to better appreciated what might be involved in less characterized systems. In so doing, precise arguments will be used to define commonly used terms associated with adaptive gene expression which include "threshold," "trigger" and "switch." 6.3.2 CQhracteristicsof Transcriptional Regulation The lysogenic transcriptional state of OR is stably maintained by the intracellular concentration of repressor. Figure 6.4 reveals that OR is comprised of three adjacent sites OR , 1 OR2 and OR3 upon which CI dimer may bind. The relative binding affinities of CI dimer for these sites are such that OR1 is first filled, followed by OR2 then OR3 . At low repressor concentrations, transcription of cro is blocked due to the binding of dimer to OR1 (figure 6.3A and 6.4A). At higher repressor concentrations, OR2 is filled. By binding to OR2 , dimer serves to activate transcription from c. At the same time, the transcriptional block of cro is further reinforced. Finally as the repressor concentration is further increased, OR3 is filled and this results in a shut down of cI transcription. This autoregulatory function (i.e. self repression at high dimer concentrations) prevents the excessive build up of repressor within the cell. 165 6.3.3 The Rate Law for CI Generation The regulatory interactions described above indicate that the level of CI dimer is crucial to maintaining the state of lysogeny while preventing lytic development. Therefore the model developed in this and the next few sections is specifically aimed at summarizing the processes that determine the intracellular concentration of CI. In this section, a rate law describing CI generation is derived. In the next section, this rate law is incorporated into a single cell material balance on CI. Finally the model is studied in order to identify the features responsible for threshold behavior. To greatly simplify the algebraic manipulations involved, it is assumed that CI dimer binds to OR only in the manner described (figure 6.4). It is also assumed that transcription is initiated from PRmonly after both OR1 and OR2 are occupied. These assumptions are reasonable since thermodynamic considerations indicate that these binding states account for greater than 92% of the occupied configurations (Ackers et al. 1982). Finally, the rate of CI generation is assumed to be directly proportional to the rate of transcription of cI. In other words, transcription, or the initiation of transcription, is assumed to represent the rate limiting step in the series of steps leading to the generation of CI. Under these assumptions, the rate of generation of CI is given by: RGEN=k [o (CI) 2(CI) 2] (1) [O(CI)2(CI) 2 J represents the concentration of the doubly occupied operator. [(CI)2] represents the intracellular concentration of CI dimer. k represents the lumped first 166 order rate constant relating the rate of transcription to the rate of generation of CI. An alternate interpretation of equation (1) is that the rate of generation of CI is proportional to the probability that both OR1 and OR2 are filled. Equation (1) is made useful by replacing [O(CI)(CI) 2 ] with variables and parameters that may be determined experimentally. This is accomplished by assuming that the interactions between the operator sites (i.e. OR1 etc.) and CI dimer are close to equilibrium. The reactions and their corresponding equilibria relationships are given by: o] +[ CI)o()]= (CI) [o(CI)2] + (cI) 2 ] =o (CI) [o(CI) 2 (CI) 2] +[ (CI) 2 ] =[o(CI) (2) 2] 2 (CI) 2] 2(cI)2 (cI)2]. (3) (4) with K 1= [o (cI) C[o [ (CI) ]2 (5) 2] [o((CI)2 (c)) 2 2K2 [o((CI)][ (CI2) [o(CI)2 (CI) 2 (CI)] Here and C ~~~~~~~~(6) 2 represent the intracellular concentration of the(7) Here [O(CI)21and [O(CI)2(CI)CI)z ] represent the intracellular concentration of the 167 singly and triply occupied operator respectively. The final relationship needed, the OR site balance, is given by: [OT]=O0] +[O(CI)2] + O(CI)2(CI)2] +[O(CI)2 (CI)2(CI)2 ] * (8) [Or] and [0] represent the total and the unbound operator concentrations respectively. When equations (2-8) are solved for the doubly occupied operator complex, and the result substituted into equation (1), one obtains: klKl[OT ] [ (CI)2] RGEN 2 1+Kl[ (CI) 2 +K1K2[(CI) 2) 3 +K1 K2 K3 (CI) 2 (9 ) ] This equation relates the rate of generation of CI to: 1) the intracellular concentration of CI dimer, 2) the equilibrium binding parameters and 3) the kinetic parameter kl. With the exception of kl, estimates of all parameters can be found in the literature. A compilation of these parameter values, along with the source, is presented in table 6.1. Using these values, it is possible to study how this equation varies as a function of [CI] (figure 6.5). To reflect the fact that k, is unknown, curves for different values of kl are presented. The relationship used to convert monomer concentration, [CI], to that of dimer concentration, [(CI)2 ], is presented in the next section. Several important features can be derived from figure 6.5. The first involves the qualitative form of these curves. At low repressor concentrations, the generation rate depends non-linearly on [(CI)2 , accelerating with increasing repressor. 168 This "activation" region reflects the fact that the doubly occupied complex promotes transcription (figure 6.3B). At very large [CI], the generation rate decreases with increasing [CI]. Physically this corresponds to the autoregulation, or self-repression, exhibited by CI dimer on c transcription (figure 6.3C). The second important feature is that the activation and repression regions only depend on the equilibrium constants associated with the binding of CI dimer to the different OR sites. Variations in the value of kl only affect the amplitude of the generation curves. 6.3.4 The Single Cell Model To examine the in vivo ramifications of equation (9), molar balances on the intracellular levels of CI monomer and dimer are derived. The latter is needed because equation (9) is a function of the intracellular dimer concentration. Assuming the cell is a well mixed variable volume reactor and that repressor does not cross the system boundaries, the general molar balances on monomer and dimer are given by: d( [CI Vc) ( RGMI-RDERj) dt d( [ RoENand RDEGR represent (CI)Vc] dt (R = (RaN R, -RDWal) V V (10 ) (1) VC the molar rates of generation and removal, respectively, on a per volume basis. Vc represents the cell volume. Contributions to the various summation terms present in equations (10) and (11) are assumed to be limited to 169 equation (9) for the generation of CI and the forward and reverse reactions associated with the dimerization reaction: 2 [CI] = (CI) 2] (12) The equilibrium constant associated with this dimerization reaction is written in terms of the forward and reverse reactions rates as: Kakf ((C)] kr [CI] (13) ( 3) Substituting the contributions from the dimerization reaction into the generation and degradation summations in equations (10) and (11) yields: dt ] ) =RG+k, [ (CI) 2] -kf [CI] 2 -k2 [CI] dw( [ (CI)]) =_ k [ (CI) 2] +knt[CI] 2-k3 [ (CI) 2] dt 2 2 (14) (s ) where the forward and reverse kinetic constants, k and k, are based on moles of monomer. The last term, on the right hand side of both equations, represents the dilution effect due to the expanding cell volume. Under balanced growth conditions, they are given by (i=2,3): 170 k dlnV _ n2 dt d (16) where d represents the doubling time of the cell in minutes. In the derivation of equations (14) and (15), mechanisms that lead to the specific degradation of monomer and dimer have been ignored. When these processes cannot be ignored, first order kinetic mechanisms can be assumed and the values of k2 and k3 are altered accordingly. This approach is demonstrated in section 6.3.6 where mechanisms leading to the specific degradation of CI monomer are incorporated into the material balances. 63.5 Steady State or Balanced Growth Solutions The steady state solutions are obtained by setting the time derivatives in equations (14) and (15) to zero. When this is performed, equation (15) yields: [(CI) 2] = [CI] 2 KA. (17) with KKa1 A 1 +1 k, (18) KA can be considered to be the apparent equilibrium constant. Since KA is smaller than K,, the [(Cl)2 ] level under balanced growth will be smaller than expected in the absence of kinetic effects (i.e. under equilibrium conditions). The difference reflects the dilution of dimer as the cell volume increases. In the analysis to follow, 171 KA will assume the reported value of K,. This assumption is justified since the dissociation of dimer occurs rapidly (Pirrotta et al. 1970, Chadwick et al. 1970). Therefore the specific rate constants associated with dimer dissociation, k, and cell growth, k2 or k3, are such that k, > > k3 . Combining equations (17) and (18), with the steady state form of equation (14), one obtains: kKlK O ] 2 k3KA[CI] klK2 1 +K, A[ cI ] +KK2 T 2 C ] (19) [C] 4 CI (9) KK2K3 K [ CI ] Equations (17) and (19) along with the associated parameters (table 6.1) determine the intracellular concentrations of CI monomer and dimer under steady state or balanced growth conditions. The simplest interpretation of this equation is that at steady state the rate of repressor generation (right hand side) must be balanced by the rate of repressor removal (left hand side) in order for the cell to maintain a balanced growth environment. The right hand side of this equation yields curves that are identical to those appearing in figure 6.5. The left hand side of this equation represents the rate of repressor removal due to the expanding cell volume. The linear and quadratic terms, on the left hand side, result from the dilution of monomer and dimer respectively. Solutions of this equation can be determined graphically in a manner similar to that employed in nonisothermal chemical reaction engineering (Fogler 1992). This is demonstrated in figure 6.6 where the repressor removal rate is superimposed on the repressor generation curves presented in figure 6.5. The removal curve was 172 determined assuming repressor loss is solely due to cell growth (a 30 minute cell doubling time was used). The points of intersection between the generation and removal curves identify the real solutions of equation (19). Figure 6.6 (A and B) indicates that three real solutions are possible. Also indicated in figure 6.6A is the range of intracellular repressor concentrations reported in the literature (Pirrotta et al. 1970, Ackers et al. 1982). The lower and upper boundaries that delimit this region, as indicated by the dashed vertical lines, correspond to total repressor concentrations of 1X10-7Mand 2X10-7Mrespectively. This corresponds to 100-200 -' molecules of repressor per cell, assuming a cell volume of approximately 1X10 5 1. Since intersections between the generation and the removal curves represent solutions to equation (19), intersections in the range of reported repressor concentrations identify "realistic" solutions. Realistic solutions are possible when the specific generation rate constant, kl, ranges between 4 min-' and 8 min-' (see legend to figure 6.6). In subsequent sections, a k, of 4 min - is shown to lead to system behavior that is consistent with other experimental observations. The physical significance of solutions to equation (19) can be explained with reference to figure 6.3. The lysogenicstate corresponds to the stable steady state at high concentrations of CI (figure 6.6A). When CI is present in sufficient quantities, the phage stably maintains lysogenyby blocking transcription of cro. The lytic state corresponds to the trivial steady state where no repressor is present (figure 6.6B). With repressor absent or present in small quantities, cro is expressed and the transcriptional switch is thrown. Between these physical states there exists an 173 unstable steady state that is not experimentally realizable. The physical significance of this unattainable state is discussed in section 6.3.7. 6.3.6 Initiating the Switch In the laboratory, lytic development is induced by subjecting cells to stimuli that activate the proteolytic action of RecA (Bailone et al. 1979, Ptashne 1987). RecA is a protein that is normallyinvolved in catalyzingrecombination between DNA molecules (Ptashne 1987). When DNA is damaged, RecA exhibits proteolytic activity and acts to degrade the monomeric form of A repressor. It has little or no effect on the dimer. Assuming the degradation of CI monomer obeys a first order kinetic process, the action of RecA may be described mathematically by increasing the value of k 2. Figure 6.7 illustrates what happens to the steady state solution "structure" associated with equation (9) as the value of k 2 is increased. Small increases in k 2 lead to a small leftward shift in the removal curve. Because the generation and removal curves still intersect, at high repressor concentrations, the phage is able to maintain the state of lysogenydespite such perturbations in k2. However as the severity of the perturbation is increased, as reflected by a larger value of k2, the capacity of the phage to maintain the lysogenic state is exceeded. When this occurs, the only transcriptional state available to the phage is associated with the lytic state. This imbalance, between the generation and removal rates, directs the phage towards lytic development as defined by a zero repressor concentration state. 174 k2 represents a threshold or bifurcation parameter. As demonstrated in figure 6.7, as the value of k2 is varied slightly around the neighborhood of seven times the growth rate, dramatic changes in the system behavior are effected. Since k2 reflects the degradative action of RecA, this indicates that the concentration or the specific activity of RecA must reach a threshold level in order to initiate lytic development. 6.3.7 The Dynamics Associated with the Transition In this section, the dynamics associated with the transition to lytic development, once the threshold barrier has been overcome, are investigated. The goal of this section is to demonstrate that initiation into the developmental pathway requires that several conditions are met. In addition to achieving a threshold concentration of RecA or equivalently the value of k2, this threshold level must be maintained for a certain period of time. Unless both conditions are met, the phage will maintain its state of lysogeny. Figure 6.8A illustrates the time course profile of CI, monomer and dimer, following a step increase in k2 above the critical threshold. These curves were generated using equation (14) and the steady state form of (15). The steady state form of equation (15) was used in order to eliminate the terms involving kf and k, from equation (14). Since the dynamicsof dimer dissociationare rapid, the dynamics of equation (15) can also be expected to be rapid. As the concentration of monomer decreases, due to the action of RecA, the concentration of dimer can be expected to adjust rapidly according to the steady state form of equation (15). 175 The profiles shown in figure 6.8A represent the dynamics associated with the transition from lysogeny to lytic development. Following the perturbation, the rate of monomer removal exceeds its rate of generation. Therefore the intracellular level of repressor is driven to zero. The time duration of this transition is of the order of 30 minutes. A similar time constant, determined experimentally, has been reported by Bailone and coworkers (1979). Because the time associated with this transition depends on the value of k2, the value chosen for k2 represents a fit. Yet this "fit" seems reasonable since it corresponds to a k2 slightly higher than the threshold value (see legend to figure). Since lytic development is not initiated by all phages, a value of k2 close to the threshold appears reasonable. Followinga transient perturbation in k2, the eventual fate of the phage, or cell, is decided by the repressor concentration at the unstable steady state. Figure 6.9A presents the time course response curves, of [CI], following two different pulse perturbations in k2 The duration of these pulses differ by only one minute (figure 6.9B). These pulses can be viewed as simple approximations to the action of RecA, which has been found to be transient in nature (Bailone et aL.1979). As indicated in figure 6.9A, the fate of the phage depends very much on the length of time the perturbation persists. When the cell is subjected to a 30 minute pulse, the phage maintains its state of lysogeny. This is possible because [CI] is driven to a value just above the unstable steady state concentration, as indicated by the dashed horizontal line. Therefore following the perturbation, the phage remains within the attractive region of the lysogenic state. In contrast, a 31 minute perturbation is sufficient to 176 commit the phage to lytic development (dotted curve). In this case, [CI] is driven to a value just below the unstable steady state monomer concentration so that once the original phase portrait is reestablished, the phage resides in the attractive region of the lytic state. These results are consistent with the finding that lysogenymay or may not be maintained when lysogens are transiently induced by thymine starvation (Bailone et a 1979). The repressor concentration at the unstable steady state indicates that approximately 85% of the monomer, prior to the perturbation, must be degraded in order to commit phages to the lytic pathway. This value is in good agreement with the 90% value reported by Bailone and coworkers (1979). 6.3.8 Static versus Dynamic Thresholds The monomer concentration at the unstable steady state represents a dynamic threshold. Since the fate of the phage, and hence the cell, depends on whether its intracellular monomer concentration falls above or below the unstable steady state concentration, slight changes in this value can lead to dramatic changes in the system behavior. The term dynamic is used to explicitly indicate that time dependent factors are involved in determining the state adopted by the phage. This was graphically illustrated in figure 6.9 and it will resurface in section 6.5.3. In contrast, the term static threshold can be used to described the barrier that must be overcome in order to alter the number of steady states available and does not involve time. This was illustrated in figure 6.7 were the activity of RecA, via the paramater k2, was required to reach a critical level before lytic development was possible. 177 Both types of thresholds have been alluded to in the literature (Novick and Wiener 1959, Lewis et a). 1977). The arguments presented in this section allow the subtleties associated with both thresholds to be appreciated. 6.4 Tme R c ac Operon 6.4bA1Bacground It has long been known that when the E coli lac operon is induced under suboptimal conditions, it is possible to generate and maintain physiologically distinct cells types (Benzer 1953, Novick and Weiner 1957, 1959, Cohen and Horibata 1959). This heterogeneity is associated with the transcriptional state of the cells. One population is induced for expression of the lac operon while the other remains uninduced. The explanation for this phenomenon, provided by Novick and Weiner (1957) is presented in figure 6.10. They hypothesized that in order for cells to become induced for expression, the amount of lac permease, encoded by lacY, must reach a threshold level within a cell. Upon attaining this threshold concentration, inducer could be imported into the cell in order to inactivate lac repressor. This results in the expression of more permease and eventually leads to full induction. Cells that did not achieve this threshold level remain uninduced. The autocatalytic activation mechanism described in figure 6.10 cannot explain the existence of two stable populations. By the arguments presented above, the uninduced state would be expected to be unstable. Any permease molecules associated with the cell would rapidly import inducer and the positive feedback loop 178 would be initiated. Since the time scale associated with cell growth is much larger than the time scale associated with inducer transport or permease expression, the amount of permease per cell would be expected to rapidly increase despite the fact that it is diluted upon each division cycle. Novick and Weiner (1959) realized this stability problem and accounted for it using hypothetical arguments. In this section, these issues are studied quantitatively using models derived from single cell material balances. The stability problem will be addressed and the conditions required for the existence of multiple transcriptional states identified. Additionally, the "maintenance" phenomenon (Novick and Wiener 1957, section 6.4.6) or the dependence that inducibility has on initial conditions will be explained. All analysiswill be performed in the concentration and parameter ranges reported in the literature. It will be demonstrated that the information needed to explain these phenomena is already known. 64.2 Chaeristics of the lac Operon When cells of E. coli are cultivated in non-glucose containing media, the trans acting factor encoded by acI negatively regulates transcription from the lac operon (see Reznikoff 1992 for recent review). lacI encodes a repressor that blocks transcription by specifically binding, as a tetrameric complex, to the lac operator. Since lacI is expressed constituitively, little or no transcription occurs under noninducing conditions. However, when an appropriate inducer is added, molecules of inducer bind to repressor in such a way as to reduce the affinity of repressor for the 179 operator (figure 6.11B). Once this occurs, the operon is freely transcribed (figure 6.11C). An important detail involved in inducing the operon arises from the fact that two inducer molecules, at least, must bind to the tetrameric repressor complex (Boezi and Cowie 1961, Yagil and Yagil 1971, Barkley and Bourgeois 1980 for rev.). Standard biology and biochemical engineering texts often omit this detail. This omission is of little consequence when the operon is fully induced. In either case, the rate of transcription can be expected to saturate with respect to increasing inducer concentrations. However, under suboptimal induction conditions, failure to account for this detail will lead to inexplicable system behavior. In other words, the dynamical system will be dramatically altered (Othmer 1976, Tyson and Othmer 1978). The ramifications of this detail form the core of the analysis that follows. The approach used to study the heterogeneity associated with suboptimal induction of the lac operon, is identical to that employed in the previous sections. A rate law describing the generation of permease is first derived. This rate law is then incorporated into a single cell material balance on the lacY gene product. The resulting model is then used to explain the hysteresis phenomenon noted by Novick and Weiner (1957) and others (Cohn and Horibata 1959). 6.43 Derivation of the Rate Law The rate of expression of lac permease is assumed to obey the following equation: This equation assumes that the rate of generation is proportional to the concentration 180 RGEklJ10] (20) of free operator, [0]. It also assumes that transcription is the limiting step in the series of steps from transcription of the operon to translocation of the permease to the cell membrane. The first order kinetic constant k [min-'] relates the concentration of free operator to the rate of permease generation. The units associated with this equation are [M-min-1]. As before, [0] is replaced with variables and parameters that can be determined experimentally. The approach taken is essentially identical to that of Yagil and Yagil (1971). The following reactions are considered to lie close to equilibrium: 2 [I] +[R]=[ 2 RR] (R] +O] =[RO]. (21) (22) Here [I], [R], [I2 R] and [RO] represent the concentration of free inducer, free repressor, inducer-repressor complex and repressor-operator complex respectively. The associated equilibrium constants are given by: K CI2R] 1= (R][I]2 K =[RO Theappropriate site balances are given[R [oby: The appropriate site balances are given by: 181 (23) (24) [OT,] =[o] (25) +[RO] (26) [Rr ] =[R] +[I 2 R ]. [RT] and [OT] represent the concentrations of the total repressor and the total operator respectively. Equation (26) neglects [RO] relative to [R]. This assumption greatly reduces the algebraic manipulations required and is justified on the basis that the concentration of repressor is much greater than that of the operator. Solving equations (21-26) for [0], followed by substitution into equation (20), one obtains: RGENk IOT ] 1+K1[I] 1+K1 [I] 2 2 +K2[Rr] (27) This equation describes the rate of generation of permease as a function of [I]. Figure 6.12 illustrates the shape of this function for different values of k. All parameters associated with this equation, except the kinetic parameter kl, can be found in the literature (table 6.2). The shape of this curve is in good agreement with the data obtained from Herzenberg (1959). This includes both the S-shape and the concentration of inducer needed for saturation. The dominant physics associated with equation (27) are contained in the [I]2 term present in the numerator. This term originates from reaction (21) where the fact that two inducer molecules are needed to inactivate the repressor is explicitly accounted for. Due to this term, the rate of permease generation varies 182 approximately as [I]2 at low inducer concentrations and an S-shaped curve results. Had only one inducer molecule been sufficient to inactivate repressor, a hyperbolic rate expression would have resulted. Both rate expressions exhibit saturation kinetics, however the latter would have depended linearly on [I] at low inducer concentrations. Therefore considerably different behavior would be expected to result under suboptimal induction conditions. 6A4 The Single Cel Model In this section, the rate law given by equation (27) is incorporated into single cell material balances derived on the species shown in figure 6.10. Equation (26) represents the repressor balance. The molar balances on permease and inducer are given by: d( [ V) d t -RGVC dt =(RACTIVr+RPASSIVE) VC, (28) ' (29) [YJ and [I] represent the intracellular concentrations of permease and inducer respectively. In both equations, degradation processes have been ignored. RGBN is given by equation (27). RACTVB [M-min-'] represents the rate of active transport of inducer via permease. RPASSVi represents the transport rate of inducer viapermease independent processes. 183 The rate of active transport of inducer via permease can be represented by (Rickenberg et al. 1956, Kepes 1960): R"z a I R~ACTzrVz Y1 I[Iz(30) P (30 + where a [min-'] represents the turnover number of lac permease and i [EsM] represents the permease saturation constant. The turnover number represents the number of molecules of inducer that can be transported into the cell per unit time per molecule of permease under saturating conditions. [IAx] represents the extracellular concentration of inducer [AM]. RpAsswvis assumed to be given by: RPASSZV=8( [EX] - [] (31 ) where 6 represents the volumetric mass transfer coefficient [min"]. The form of this rate expression was chosen to reflect; 1) the concentration dependent nature of this mode of transport (Herzenberg 1959) and 2) the equilibration of the intra- and the extracellular inducer concentration under balanced growth conditions in lacy strains (Kepes 1960, Yagil and Yagil 1971). When equations (30) and (31) are substituted into equation (29), one obtains: -[] Ra (2Y[ dt,Nk d[Y] _R _k2 [ Y] d(I dt dt pzl+ + [IM] 8[ ( 184 I])-J k[ I]- I ) -k2 I] (32) (32) (33) where k 2 1 dVc_ ln2 v dt d (34) k2 represents the specific growth rate of the cell. If necessary, degradative processes can be easily incorporated into these equations by altering the value of k2. 6.4.5 Te Steady States Substituting the steady state form of equations (33) and (27) into the steady state form of equation (32), one obtains: 1+K1 [i]2 X, [I] -X 2 -a 1+K [] 2+K2 [RT] (35) where _ kl k,_[OT] 3 aa k2 Da* (36) (38) = P X2- + Iz=x] p (37) In the derivation of (35), the dilution of inducer due to cell growth has been neglected relative to the loss inducer by export out of the cell or that > >k2 (table 185 6.2). Equations (35-37) along with the associated parameters (table 6.2) determine the intracellular concentration of inducer under balanced growth conditions. Because [Y] is linearly related to [I], via the steady state form of equation (33), equation (35) also determines [Y]. The parameter A was introduced in order to group the unknown parameters given by a, d and kl. With the exception of A, everything else associated with this equation is known. The approach used to determine A is presented in section 6.6.1. Since the right and left hand sides of equation (35) scale linearly with the generation and removal rates of [Y], respectively, they will be referred to as the such. This greatly simplifies the comparison to experimental data (i.e. one of the principal goals behind this analysis). The generation rate is easily interpretable using equation (27). The removal rate represents the loss of permease due to cell growth. Its disguised form results when [Y] is eliminated from equation (32) using the steady state form of equation (33). This manipulation reduces the dimensionality of the system and facilitates further analysis. Another advantage of this form of equation (32) stems from the fact that this equation controllable variable, to [I]. relates [Idx], an experimentally When solutions of equation (35) are sought (next section), the existence of multiple solutions will be shown to depend on [IEJ. 6.4.6 Te Hysteresis or Maintenance Phenomenon The significantconclusion derived from the "maintenance"phenomenon is that the transcriptional state of the lac operon is not a unique function of the 186 environment, as represented by [IBx]. Therefore, two physiologically distinct cell types can coexist in the same environment. This phenomenon is described in figure 6.13, where the transcriptional state of the lac operon is shown to depend on both [IEX] and the previous historyof the cell. If the environment encountered by an uninduced cell is adjusted in a manner consistent with path I, (A) -> (B) -> (C), then induction occurs only after the concentration of inducer reaches a "high"value. Intermediate values of [Iwdj will not induce an uninduced cell. In contrast, if induced cells are placed in an environment characterized by intermediate [IEX],they remain induced (Path II). Taken together, paths I and II form a classic hysteresis loop. An explanation for the maintenance phenomenon can now be provided by figure 6.14 where the right and left hand sides of equation (35) are plotted as functions of the intracellular inducer concentration with [IEX]as a parameter. The parameters in table 6.2 were used. Since the generation rate is independent of [IEX], a single curve results. These figures indicate that the number of real solutions of equation (35) depends on [IEx]. Path I in figure 6.13 can now be explained by the series of lines presented in figure 6.14A At low [IEA],only the uninduced state is available to the cell as noted by the single point of intersection between the removal and generation curves. This corresponds to a low rate of permease generation and a low [I]. As [IBx] is continuously increased, three intersections between the generation and removal curves eventually result. Yet due to the stability of the uninduced state, the cell will retain its uninduced state. However as [IEx]is further increased, the removal curve eventually becomes tangent to P1. Any further increase 187 in [IEx]will result in the dissolution of the uninduced state. Because the generation rate exceeds the removal rate, beyond P1, the cell becomes "attracted" to the induced state. Consistent with the previous use of the term, [IEx]at P1 defines an induction threshold. Path II, in figure 6.13, can be explained by the series of lines shown in figure 6.14B. Starting from the induced state, when [Iex] is reduced to the point where the removal rate becomes tangent to the generation rate at P2, further decreases in [Iex]result in the disappearance of the induced state. Since the induced state cannot be maintained at this lower [IEx],P2 defines a second threshold with will be termed the "maintenance threshold". As in the case with RecA, both thresholds are static in nature since they qualitatively rearrange the phase portrait. This hysteresis loop is more clearly visualized when the loci of intersections, as shown in figures 6.14A and 6.14B, are plotted against a continuously varying [IEX] (figure 6.15). This curve was generated by determining the solutions [I] of equation (35) as [IEx] was varied from 10 to 1000MuMof TMG (see section 6.9.2 for details). This curve is partitioned into three branches by P1 and P2. Each branch defines a loci in the [IBx]-[I]plane that corresponds to a qualitatively different physical solution (Iooss and Joseph 1990). The branch beginning at low [IEx]and continuing to P1 corresponds to path I in figure 6.13 and represents the uninduced cellular state. The uninduced state is maintained until the induction threshold is reached (i.e. P1). The branch beginning at large [IE] and continuing to P2 corresponds to path II of figure 6.13 and represents the induced cellular state. This state is maintained so long as the value of [IE, remains above the value identified by P2. The branch between P1 and 188 P2 identifies the locus of unstable states. As in the case with phage A, this locus identifies the set of dynamic thresholds. Near this locus, small variations in [I], or equivalently [Y], lead to the adoption of dramatically different cellular states, depending on which side of this curve branch the perturbation places the cell. This represents the threshold that was alluded to by Novick and Weiner (1957,1959). This curve is consistent with several experimental observations. The maintenance or hysteresis phenomenon is explainable simply by tracing the path sketched out by the arrows. This corresponds identically to the closed path shown in figure 6.13. Also the induction threshold corresponds to an [IEx] of approximately 100,uM TMG. This is consistent with experimental data since Novick and Weiner (1957) fully induced cultures using an [IEx] of 250MuM.This curve also describes the ability of cells to concentrate inducer. Rickenberg and coworkers (1956) and Kepes (1960) found that cells in the induced state could concentrate inducer to levels approximately 100 times that of the extracellular medium. This concentration factor is described over a large range of [IEX]. The agreement between the maintenance threshold predicted by this curve and that reported by Novick and Wiener is exact since it represents a "fit" (section 6.6.1). 6.5 Sporlation by R subtil 6.5.1 T Requirements for Threshold Behavior Despite the apparent differences between the lac and the A systems, a common regulatory scheme was found to control the initiation of adaptive gene 189 expression in both systems. So long as the concentration of the transcriptional activator responsible for induction is regulated by: 1) a positive feedback loop, 2) a non-linear generation rate, of the cooperative type and 3) a concentration dependent removal rate, threshold behavior is possible. These features were found to be inherent to the molecular mechanisms known to regulate adaptive gene expression in both systems. Positive feedback was found to be necessary for an autocatalytic activation mechanism to exist. This allowed the rate of generation of the transcriptional activator to be expressed as a function of its intracellular concentration. In the A system, positive feedback results from the direct ability of CI dimer to activate its own transcription. The positive feedback associated lac operon transcription is less direct. According to figure 6.10, permease must first concentrate inducer. Only after inducer is present at "high"intracellular concentrations will repressor be inactivated and permease transcription initiated. The importance of positive feedback becomes quite apparent when the feedback loop is "opened". In lacY strains, physiological heterogeneity, associated with the transcriptional state of the lac operon, is no longer possible (Herzenberg 1959). Non-linear generation rate laws, of the cooperative type, were found to be necessary in order to provide the system with the mathematical ability to exhibit multiple stable steady states. Such rate laws were found to be a natural consequence of the need for two molecules of "inducer" to activate transcription, either directly (phage A) or indirectly (the lac operon). This class of rate laws led to a point of 190 inflection at low inducer concentrations when the generation rate was plotted as a function of intracellular inducer. When the removal rate curve was superimposed on the generation rate curve, three intersections were possible. Each stable intersection was found to correspond to a different transcriptional state. Given a purely hyperbolic generation rate law, of the Michaelis-Menten type, this would not have been possible (Othmer and Tyson 1978). 6.5.2 The Threshold Requirements are Satisfied The identification of the mathematical "structures" required for threshold behavior provide the necessary framework in which to discuss the initiation of spore development. Based on what is known about the initiation process, all three conditions appear to be met. The positive feedback loop associated with the B. subtilis sporulation system is illustrated in figure 6.16. A detailed account of the interactions implicit in this figure can be found in the manuscript by Strauch and coworkers (1993). For comparative purposes, the positive feedback loop associated with the lac system is also shown. In both systems, information on environmental changes, which originate from "starvation" associated phenomena (sporulation system) or from the concentration of a well defined inducer (lac system), is relayed to the transcriptional apparatus of the cell. While the transduction mechanisms associated with both systems might appear very different, functionally they are identical since both result in the initial accumulation of intracellular transcriptional activator or inducer. In the ac system, this information is relayed through the direct 191 transportation of the inducer into the cell. In the sporulation system, this information is relayed through the direct transfer of phosphate via the phosphorelay. Following this initial boost of inducer, if the action of various negative regulatory elements can be countered, an inducer amplification mechanism is set in motion via the positive feedback loops. In the lac system, the lacI gene product must be countered. In the sporulation system, AbrB and SinR must be rendered ineffective. Upon repressor inactivation, more "transporter" proteins are expressed and eventually the fully induced state is achieved. In the sporulation system, the transporting proteins amplified are the phosphorelay gene products encoded by the kinA, spoOF and spoOA Full induction eventually results in a high level of phospho-SpoOA and the subsequent expression of stage II genes (i.e. spollG, spolIE and spollA are transcribed). An important question is "Where does the non-linear generation rate, of the cooperative type, originate ?" While a definitive answer is unavailable, a possible explanation could involve the sinR gene product. sinR encodes for a developmental switch protein that has been found to play a major role in determining which of a number of adaptive responses the cell chooses (Bai et al. 1993). SinR is a repressor of early sporulation gene expression and recent evidence suggests that it represses stationary phase induction of spoOA (A.D. Grossman, personal communication). SinR functions as a tetramer in vivo and its inactivation seems to be mediated through the binding of multiple molecules of SinI (Bai et al. 1993). Since sinIl expression relies on the spoOA and spoOH gene products (Bai et al. 1993), if two 192 molecules (more than two will also work) of SinI are required for SinR deactivation, a nonlinear phospho-SpoOAgeneration rate, of the cooperative type, would result. Also represented in figure 6.16 are the molecular processes that lead to concentration dependent inducer removal rates. In the lac system, inducer removal occurs by growth and inducer export (Kepes 1960). Under suboptimal induction conditions, the inducer removal rate was found to represent a critical barrier to full induction. In the sporulation system, the removal rate will depend on which phase of the culture the cell resides. During the exponential phase, inducer removal might be expected to occur via cell growth. During the stationary phase, recent evidence suggests that phospho-SpoOA removal occurs by dephosphorylation via the spoOE gene product (Perego and Hoch 1991). spoOE encodes for a phosphatase that acts to degrade phospho-SpoOA and its expression has been found to be under SpoOA control. As will be shown in section 6.5.4, under suboptimal induction conditions, the rate of removal of phospho-SpoOAcan be expected to represent a critical barrier to full induction. 6.5.3 Threshold Model Descnbing Sporulaftion The three components required for threshold behavior can be summarized by the equation: d[I] d[ [I ] ) -R =RGmn( ( [ I]) where [I] represents the intracellular concentration of phospho-SpoOA. 193 (39) RGEN and RREM represent the generation and removal rates of [I] and the [I] argument is shown to denote the dependence of these rates on phospho-SpoOA. For simplicity, the dependence of the generation and removal rates on other factors is not addressed explicitly. This equation states that when an imbalance between the generation and removal rates exists, the intracellular concentration of inducer will change. Based on the arguments presented in the previous section, the functional dependence of RGEN and RREMon [I] might be expected to be qualitatively similar to the curves drawn in figure 6.17. The S-shaped generation curves results from the assumption that two molecules of SinI are required to inactivate SinR, an assumed repressor of spoOA The linear removal curves follow from the assumption that SpoOA-P is removed according to a first order kinetic process. The intersections at low and high phosphoSpoOA concentrations would reflect the growth and sporulating states respectively. Under balanced growth conditions, phospho-SpoOA removal curves similar to those shown in figure 6.17A might be expected. The slope of these lines represent the specific growth rate of the cell with faster growth rates leading to larger slopes. As before, these lines represent the removal of phospho-SpoOAdue to the expanding cell volume. Without loss of generality, the slope of these lines could also include the action of any unknown phosphatases. Since sporulation occurs even under growth conditions (Schaeffer et al. 1965, Dawes and Mandelstam 1969,1970, Dawes and Thornley 1970), this possibility is accounted for by the intersection at high phosphoSpoOA concentrations. This model predicts that increased sporulation would be expected under 194 slower growth conditions. This follows from the observation that the phospho-SpoOA concentration at the unstable steady state increases with increasing growth rates (i.e. P1<P2<P3). Recall that this concentration represents the dynamic concentration threshold (see section 6.3.8) that separates the growth and sporulation states. Experimental support for this prediction can be found in the literature (Schaeffer et al. 1965, Dawes and Mandelstam 1969,1970, Dawes and Thornley 1970). The ability to initiate stage II gene expression depends on the ability of the cell to accumulate phospho-SpoOA. When nutrients become limiting and cells enter the stationary phase, the slope of the removal curve decreases (figure 6.17B; curve A) because the cell either stops dividing or divides more slowly. Under these circumstances the rate of phospho-SpoOAgeneration will exceed its rate of removal and phospho-SpoOA can be expected to accumulate within the cell according to equation (39). Yet this rate imbalance does not continue indefinitelysince the spoOE gene product also accumulates during the early stationary phase. The accumulation of SpoOEwithin the cell can be expected to lead to a time dependent increase in the rate of phospho-SpoOA removal, as reflected by the series of lines with increasing slopes (figure 6.17B). Therefore unless the cell can accumulate a level of phospho- SpoOA above a certain threshold (P) before the rate of phospho-SpoOA removal balances its rate of generation (curve B), it will be locked in the uninduced state. Whatever phospho-SpoOAis present eventually decreases since the rate of removal dominates. In contrast, those cells with levels of phospho-SpoOA above this critical concentration continue to accumulate phospho-SpoOA and eventually initiate stage 195 II gene expression. 6.5.4 A Mutations kinA mutations act to decrease the rate of generation of phospho-SpoOA. Because the kinA gene product is involved in the positive feedback loop needed for phospho-SpoOA generation (i.e. kinA is a phosphorelay component), the phosphoSpoOA generation curve in kinA mutants will be attenuated. This might be reflected by the generation curves shown in figure 6.18A. This conclusion was arrived at by analogy with the ac system. In the ac system, an attenuated feedback loop could be manifested by a lower permease turnover number, a, or a lower rate specific permease generation constant, kl. Recall that both parameters directly effect the amplitude of the generation rate curves shown in figure 6.14 via the parameter A. The dynamic phospho-SpoOA threshold concentration is greater in kinA mutants. As discussed previously, upon entrance into the stationary phase a series of lines with increasing slopes would result due to the accumulation of SpoOE (figure 6.18B). The removal curves would be unaltered since the AbrB pathway is not affected by kinA null mutations (Perego et al 1989, Mueller and Sonenshein 1992, this work). Therefore for a fixed removal rate (figure 6.18A) the critical threshold will be greater in the kinA mutant relative to wild type as illustrated by points P1 and P2. The increased threshold makes it more difficult for cells to attain the induced state so that fewer cells in the population would be expected to initiate stage II gene expression. The probability that a cell will overcome this increased threshold barrier 196 is further reduced by the lower phospho-SpoOA generation rate in kinA mutants. According to equation (38), the rate of accumulation of phospho-SpoOA varies directly with the generation rate. These predictions are in full accordance with the data presented in chapter 5. This analysis suggests that even if a kinA mutant cell is able to accumulate phospho-SpoOA above the boundary concentration, P2, it will always contain less phospho-SpoOA relative to wild type even in the maximally induced state as defined by P3 and P4. This might explain why the fl-galactosidase expression in kinA cells containing a spoIA-lacZ fusion was found to be lower than in wild type (table 5.1, figure 5.11). These arguments are all consistent with the effects that SpoOE variations have on sporulation. spoOE loss of function mutations have been found to restore sporulation in kinA mutants to wild type levels (Perego and Hoch 1991). Since the phospho-SpoOA removal curves would not increase rapidly, if at all, during the stationary phase in spoOE mutants, phospho-SpoOA could accumulate unchecked. In contrast, overexpression of spoOEwould lead to a more rapid increase in the rate of removal of phospho-SpoOA. This would lead to decreased sporulation since it would reduce the amount of time available for the cell to accumulate phosphoSpoOA. This prediction is in accordance with experimental observations (Perego and Hoch 1991). 6.5.5. spoOKMutations Similar to the effects of the kinA mutant, this model predicts that the rate of 197 phospho-SpoOA generation will be lower in a spoOK mutant. In contrast to kinA mutations, the mechanism by which spoOK mutations reduce the rate of phosphoSpoOAgeneration is predicted to be different. Since the spoOK gene products are presumably involved in mediating signal inputs into the phosphorelay (figure 6.16), independent of the positive feedback circuitry, their contribution to the inducer generation rate should not depend on the level of phospho-SpoOA. This is supported by the finding that the spoOK operon is expressed constitutively (A.D. Grossman, personnel communications). Under this assumption, figure 6.18B would reflect the effect that spoOKmutations have on the inducer generation rate. In contrast to kinA, the spoOK mutation decreases the entire generation curve in a phospho-SpoOA independent manner. This figure suggests that when growth ceases and wild type cells began to accumulate phospho-SpoOA, few spoOK cells even begin to accumulate phospho-SpoOA,since the phospho-SpoOAinducer removal rate remains greater than its generation rate, for phospho-SpoOA concentrations below P. This analysis suggests that processes acting on at least two levels control entrance into development via the phosphorelay. Figure 6.18B suggests that a signal input threhold, mediated in part by the spoOK gene products, must be overcome if the cell is initiate development. Figure 6.18A suggests that even if the signal barrier is overcome successfully, initiation will also depend on the cell's ability to amplify or accumulate phospho-SpoOA above a critical threshold. Support for this decoupling was presented in Chapter 5 (figure 5.18) where the data obtained by several research groups (Rudner et al. 1991, Weinrauch et al. 1990) was further analyzed. 198 It was shown that the probability that a spoOK kinA double mutant will successfullyinitiate spore formation, assumed to be reflected by the sporulation frequency, is given by the product of the single mutant initiation probabilities. 6.6 Appendi: Fitting Experimental Observations 6.6.1 Determination of A The parameters gathered from the literature (table 6.2) determined equation (35) up to the adjustable parameter A. "A" is a lumped parameter comprised of both reaction and transport terms (equation 36). By noting that an extracellular inducer concentration of approximately 25M TMG: 1) sustains the induced state in induced cells, 2) fails to induce uninduced cells and 3) defines the minimum value below which cells cannot maintain the induced state (Novick and Weiner 1957), A can be determined according to figure 6.17. The removal line shown represents the right hand side of (35) for fixed f (table 6.2) and a [Ix] of 25MuM.A is determined by varying its value until the left hand side of equation (35) becomes tangent to the removal line. This results in an approximate value of A of 240. 6.6.2 Generation of the Hysteresis Curve The approach used in to generate figure 6.15 involves the application of elementary differential geometry. Equation (35) can be viewed as defining a curve in the [Iy]<-[I] plane. This curve is traced out as follows. differentiated with respect to [I] to yield: 199 Equation (23) is d[I EX] g 1 +2K1[ I ]g 3g 4+g 2g 4 d[I] (A+AK [ 1 I ) +g 2 g2 g 3 g4 - p (40) - where gl = -2AK 1 g 2 =K2[] EX]i III (41) 2 (42) +K[I] g3= [I] -IEx] I [ IEX] g4 =1 +P (43) (44) Combining equations (40-44) with the condition: [IEx]=0.1 @ [I] =0.1, figure 6.15 can be generated. 200 Table 6.1 Parameters used in Phage A Simulations Parameter Value' Source2 [OT] 1 X 10-9 M Johnson et. a 1981 KA 5 X 107 M-l Ackers et. a. 1982 AG, -11.7 Kcal/mol Ackers et. a. 1982 AG 2 -12.1 Kcal/mol Ackers et. al 1982 AG 3 -10.1 Kcal/mol Ackers et. al. 1982 K1 1.77 X 108 M-l AG,=-RTlnK, K2 3.40 X 108 M-' AG 2 =-RTInK 2 K3 1.32 X 108 M-' AG 3 =-RTlnK 3 k 2,k3 0.023 min' ki=ln2/rd ' The AG 2 value used includes the cooperative binding free energy. The AG3 value does not include the cooperative binding free energy, which results from interactions with CI dimer bound to the OR2 operator, since cooperation does not occur when both OR1 and OR2 are occupied (Ptashne 1987). 2 Physiological temperature was employed (i.e. 370C). 201 Table 6.2 Parameters used in lac Operon Simulations Parameter Value' Source [O]1 1 X 10 9 M Johnson et. al. 1981 [RT]K2 1 X 10S Bourgeois and Monod 1970 K1 0.012 M- 2 Barkley and Bourgeois 1980 6 0.82 min -' Kepes 1960 ~ 4500 IM k2 0.023 min- ' k2= ln2/rd A 240 Fit to data (see appendix) Kepes 1960 ' The value of K, shown is the square of the reported value. The reported value was determined based on the binding of one molecule of methyl-/-D-galactoside (TMG) to repressor. 202 Physiological Heterogeneity PhysicalTransitions Material Balances II A I I I I I I I I I I I I I I I V ! Single Cell Models I I Mathematical State Simulated Transitions ----------------------------------- 1 Mathematical State 2 Multiple Stable Steady States Fgure 6.1 Conceptualization of physiological heterogeneity from a biological and a mathematical perspective. Distinct cell types are assumed to reflect cells with different transcriptional states. The processes that determine how the cells acquire different states are studied using mathematical models that exhibit multiple steady state behavior. 203 Lysogeny . Lysogeny I : OR I I T Mathematical State 1 -v Mathematical State 2 -= Figure 62 The decision faced by bacteriophage A when integrated into cells of E. col. Under suboptimal induction conditions, some phages will initiate lytic development while others maintain a state of lysogeny. The divergent pathways undertaken by phages suggests that a developmental "threshold"has been achieved by a fraction of the population. The goal is to develop a mathematical model that can explain this threshold-like behavior. 204 OR Operator A I~~~~~~~~~ I I !i L PRM PR -- - - - - - - - - - - - - - - - - _- _- -_ - cl cro ye NN Transcription Blocked Polymerase Lysogeny B - PRM cl ----------- -----------cro Transcription Blocked RNA Polymerase , . Lysis Figure 6.3 The genetic switch controlling initiation into lytic development. The switch, associated with the OR operator, can be in the lysogenic or the lytic positions. In the lysogenicposition (A) cI is transcribed and cro, a lytic specific gene is repressed. In the lytic position (B), cro is transcribed and cI is repressed. 205 OR Operator cro block 0R21dime ORI IOR3 0 dimer 0 c- Activated State r- cro block c) C) 0 C Ct Transcript cl block Repressed State cro block ta-~~~ o Figure 6.4 Maintaining the lysogenicstate of the ORoperator. Three binding sites comprise OR and CI dimer binds dominantly in the order shown (adopted from Ptashne 1987). Positive and negative feedback loops maintain a stable lysogenicstate. 206 o0 8 C E 6 0 4 .o 0 c 2 0oL aL_ a: 0 0 2 4 Monomer Concentration 6 8 [M] X 10 8 Figre 6.5 Parametric, in kj, study of equation (9). All parameters are reported in table 6.1. The curves shown are for kl=2, 4, 6, 8 and 10 min-'. 207 r% a) 0 6 E 4 +-' 2 'I- cE 0 0 2 4 6 8 A U.4 0) Q0 ' 0.3 X _ E 0.2 '0" 0.0 0.0 0.2 0.4 Monomer Concentration 0.6 [Ml X 108 Figure6.6 Multiple steady state solutions to equation (19). The intersections between the removal and generation curves (A), identify the solutions of equation (19). The generation curves shown are for kl=2, 4, 6, 8 and 10 min-'. The removal curve is set by the value of the specific growth rate. For each generation curve, three solutions exist. Panel B represents a blow up of the low concentration region where the two solutions at low repressor levels become obvious. This plot allows a fit of kt to the reported concentration range (A). 208 f ^ U.0 O 0.6 0.4 (O a: 0.2 nn Uv. 0 1 2 Monomer Concentration 3 4 [M] X 108 Figure 6.7 Activation of the biological threshold. As the value of k 2 is slowly increased, a point is reached where the phage (or cell) can no longer adjust to the increased "stress." This results in the dissolution of the balanced growth state, that characterizes lysogeny, and induction into lytic development. The degradation of monomer, by RecA, is represented by an increase in k2, above the value set by the specific growth rate of the cell. The removal curves shown are for k 2= 0.023 (removal solely due to cell growth), 0.092, 0.162 and 0.254 min-'. k2 =0.162 min-' approximates the position of the threshold. 209 6 a, X 1o x c 4 4- o u) o 2 0aC: 0 co 0 -20 0 20 40 60 -20 0 20 40 60 f% n,' U..U C 1u. a) irX c 0.20 o 'o aa) 0.10 0 l- Q O) 0.00 Time (min) Figure 6.8 Transient profiles of monomer and dimer (A) following a step increase (B) in k 2 from 0.023 (the growth rate effect) to a value above the critical threshold (i.e. 0.184 min -1). 210 4 0 Go x 3 0 co t_ Cu L. c a) 2 o 0 c o L. (D E 1 0 c 0 oi 0 -2 0 20 40 60 80 S e f%^ U.3U - B At=31 min At=30 nin M 0.20- k2=0.184 min z o co L 8 0.100 0.00 k2=0.023 min- ' 0.00 -20 k2=0.023 rrin'l I 20 0 20 40 60 Time (min) Figure6.9 Transient response of monomer (A) followingpulse perturbations in k2 (B). The width or time of the two pulses differ by one minute. When the phage (or cell) is subjected to a thirty minute perturbation, it is able to reestablish lysogeny. However, lyticdevelopment is initiated when the duration of the perturbation is increased by one minute. 211 LacY (Permease) T Inducer Import -- Inducer Repressor Inactivation Repression Lacl (Repressor) Fgure 610 Autocatalytic activation mechanism used to describe culture heterogeneity. Inducer is transported into the cell by the lacy gene product. Inside the cell, inducer indirectly activates lacYexpression by inactivating lac repressor (encoded by lacd). Cells that have attained a threshold level of permease activate this loop and eventually become fully induced. 212 Promoter A) Uninduced Operator lad lacZ lacY M~ Repressor Promoter lad I Operator WlacZ I lacY B) Induction Promoter C) Fully Induced lad Operator VII I lacZ lacy URNS RA/ I I I Transcript O Figue 6.11 Negative regulation of the lac operon. In the uninduced state (A), repressor blocks transcription. The affinity of the allosteric repressor complex for the operator is reduced upon binding to two inducer molecules (B and C). 213 6 c) co 4 0 aQ) 4+J c- o 2 C_ cO 0 0 2000 4000 6000 Inducer Concentration 8000 10000 [PaM] Figure 6.12 Parametric, in kl, study of equation (27). All parameters employed are reported in table 6.2. The curves shown are for k 1=30, 40, 50 and 60 min'. 214 F 0 0 n F w un Path I II (dotted) -_. V A Path B C (solid) 0 Uninduced + Induced Increasing Extracellular Inducer gurfte6.13 Hysteresis or maintenance phenomenon associated with induction of the ac operon. When inducer is added to uninduced cells (Path I) induction occurs only at "large" inducer concentrations. In contrast, when inducer is partially removed from induced cells, the induced state is maintained at inducer concentrations that cannot induce uninduced cells (Path II). 215 8 6 a) (V a: Cn C, 4 a) o0 (D C: 2 E 0 -2 0 100 200 300 400 -1r% 11 )UU a) 200 o E 100 0 0 1X10 4 2X10 4 3X10 4 Inducer Concentration [M] Figure 614 Explanationfor the hysteresis phenomenon in terms of threshold phenomena. Equation (35) is plotted with different removal rates, as determined by [IEx](see table 6.2 for parameter values). Panel A is a blow-up of the low concentration region. P1 (A) represents the induction threshold. P2 represents the maintenance or extinction threshold (see text). 216 4 '5 Il- 2 104 0 L. 0 - 0 c- 105 C In 1 101 Extracellular Inducer (M Fgue 10 3 102 TMG) 6.15 The hysteresis curve generated by equation (35). As [Ia] is increased, the number of intersections between the generation and removal curves (figure 6.14) varies from one to three to one (excluding the points of tangency). These intersections appear as points on this curve for each [IEx]. P1 and P2 represent the induction and the extinction thresholds respectively. When [IEx] is increased beyond P1, full induction occurs. When [IEX] is decreased below P2, cells lose the ability to maintain the induced. For [IEx]between P1 and P2, the cell can be in either state depending on its past history. 217 Sporulation Signals SpoOE, Growth via SpoOK I Phosphate Transport Phosphorelay l Components Inducer Removal , Phospho SpoOA Inadivation (direct and indirect Indirect Repression of Transporters Extracellular Signlals AbrB, SinR (Repressors) Export, Growth via lac Induce3r Inducer Removal LacY 1 Transport I(Permease) Intracellular Inducer II I nactivation Repression of Transporter Lacl (Repressor) Figure 6.16 Positive feedback loops associated with the amplification of phospho-SpoOA (top) and lac inducer (bottom). 218 a) r0 E 1) c- c (U U) (o P1P2 P3 (Ur Fa 0 E c co C 0 c) (9 P Phospho-SpoOA Concentration Fgue 6.17 Threshold model describing initiation into sporulation. Panel A represents how the cell might maintain a homeostatic balanced growth state by balancing the rates of phospho-SpoOA generation and removal. Panel B represents how the time dependent increase in SpoOE eventually balances the unchecked accumulation of phospho-SpoOA during the early stationary phase. 219 a) 0 CC O CE (o -O (c P1 P2 P3 P4 a) cr 4-A 0 E a) CC c 0 c a) (9 P Phospho-SpoOA Concentration Figure 6.18 Model describingthe effects that kinA (A) and spoOK(B) mutations might have on the rate of phospho-SpoOAgeneration and hence the ability of the cell to accumulate phospho-SpoOA. 220 300 0 4-, 40 200 CC 'I- 0r0 40 E 100 c 0 0 0.5X1 0 4 104 Inducer Concentration [pM] Fegu 6.19 Generation rate curve fit to the data of Novick and Wiener (1957). The point of tangency represents the static or "maintenance threshold." If the slope of the removal rate curves is slightly increased (i.e. for slightly lower [IEX], see text), the cells are no longer capable of maintaining the induced transcriptional state. 221 Chapter 7: Summary and Conclusions 7.1 Summary An investigation aimed at better understanding the origins of physiological diversity associated with sporulating cultures of B. subtilis was undertaken. The objective was to better understand how the heterogeneity known to exist when cells acquire the terminal differentiated state was influenced by events occurring much earlier in the stationary phase. Single cell measurements of gene expression were combined with genetic analysis in order to identify how events occurring at the level of stationary phase signal transduction affected early developmental gene expression as determined through the use of a number of temporally separated developmental markers. Theoretical studies, made possible through the development of mathematical models, were used to demonstrate that the experimentally derived conclusions were not only physically possible but consistent with the molecular processes known to regulate early sporulation. 7.L1 Single Cell Measurements Gene expression was studied on the single cell level, through the development of a flow cytometry-based technique capable of estimating the relative sizes of B. subtiis populations expressing -galactosidase. Phospho-SpoOA controlled gene expression was then studied, on the single cell level, through the use of chromosomally integrated spo-lacZ fusions. The technique involved taking samples 222 from stationary phase cultures, staining with a fluorogenic P-galactosidase substrate and analyzing green fluorescence, due to C8 -fluorescein, by flow cytometry. The staining conditions were optimized in order to minimize non-specific fluorescence while permitting the generation expressing cell populations. of detectable signals in known -galactosidase The ability to apply this technique experimentally was assessed based on the ability to recover known cell subpopulations in controlled mixture experiments and to correlate sample fluorescence with culture average measurements of f-galactosidase activity. In all cases when fermentation samples were analyzed, an internal control was added to the experimental cells in order to control for background fluorescence and leakage of the fluorescent product out of cells containing -galactosidase. The internal control was an aliquot of cells that did not contain a lacZ fusion, and that had been stained with PKH26 to allow the control cells to be distinguished from the experimental cells using two color flow cytometry. This insured that the level of fluorescence background was not problematic and permitted a quantitative estimate of the size of the cell population exhibiting fluorescence above background. 7.1L2 perImental Analis of Gene Expresion The processes associated with early development were studied at the level of signaltransduction and transcriptional activation viathe phospho-SpoOAtranscription factor. Single cell measurements were used to estimate the distribution of spo gene products, using lacZ reporter fusions, within a population. The spo markers analyzed 223 spanned early development ranging from the indirect activation of one of the earliest genes expressed upon entrance into the stationary phase, spoVG, to the expression of a gene known to be exclusively expressed in the mother cell, spollD. Culture average measurements of phospho-SpoOAcontrolled gene expression were found to reflect the behavior of two cell populations. One population exhibited high expression with the other exhibiting little or none. The size of the high expressing population was found to correlate with the final sporulation frequency. Well defined genetic mutations, in genes involved in signal transduction, were used to perturb the level of intracellular phospho-SpoOA and the effects monitored using single cell measurements. It was found that mutations that reduce the level of phospho-SpoOA resulted in a redistribution of the relative sizes of these populations with fewer cells residing in the high expressing population. Again, the size of the high expressing population was found to correlate with the final sporulation frequency. 7.L3 Quantitative Analysis of Gene Expression Quantitative studies were used to demonstrate the feasibility of the experimentally derived conclusions and to provide a conceptual framework in which to study the relationship between the biochemical rate processes that control cell behavior. Feasibility was demonstrated by the ability to extract mathematical models, from the accepted state of genetic regulation, capable of explaining how distinct cell types can coexist in the same experimental environment. Initiation of adaptive cellular processes was analyzed in terms of the rate processes that determined the 224 intracellular level of transcriptional activator. The resulting model was shown to be consistent with experimental results presented in both, this thesis and the literature. The conceptual framework provided by the model allowed a hypothesis to be put forth regarding how various processes interact with the phosphorelay. In the simplest analysis, two states are assigned to the phosphorelay and both act to control entrance into sporulation. The default state, present during vegetative growth, regulates entrance into development at the level of signal inputs into the phosphorelay. This state is characterized by basal levels of certain phosphorelay components including kinA, spoOF and spoOA and constitutive expression of the spoOK operon. The induced or the "activated" state of the phosphorelay is attained once cells overcome the signal threshold barrier. This state is characterized by elevated expression of certain phosphorelay components, as made possible by activation of the AbrB pathway and SinR deactivation. To fully induce stage II gene expression, the activated state of the phosphorelay state must succeed in amplifying phospho-SpoOA to a critical level, by accelerating its rate of generation, before competing processes begin to dominate. 225 7.2 Conclution 1) Threshold mechanisms control entrance into the sporulation pathway and these threshold are associated with processes that interact with the phosphorelay. 2) The phosphorelay represents a critical bottleneck to early developmental gene expression and represents one of the principal means by which cells regulate the concentration of phospho-SpoOA. 3) The intracellular level of phospho-SpoOAdetermines a transcriptional state of a B. subtilis cell. 4) Culture heterogeneity, associated with incomplete spore formation, results in part by the inability of cells to expressed some of the earliest genes known to be involved in the sporulation process. 5) So long as the concentration of the transcriptional activator responsible for initiating gene expression is control by a positive feedback loop, a non-linear generation rate of the cooperative type, and a concentration dependent degradation rate, threshold potential exists. 6) Threshold potential is inherent to the molecular mechanisms known to regulate adaptive gene expression in a number of genetic systems including; 1) induction of the lac operon, 2) induction of lytic development by phage 1 and induction into the sporulation process. 7) Threshold mechanisms provide the cell with a stable means of maintaining homeostasis while allowing it to initiate adaptive responses when extreme perturbations exceed its ability to compensate. 226 Chapter & Recommendations and Future Research Areas 81 Biological Studies 81.1 Signal Inputs versus Signal Amplification The experimental and theoretical results obtained from the analysis of the signal transduction processes needed to generate phospho-SpoOA, suggests that the kinA and spoOK gene products act at different levels. Both are involved with phospho-SpoOAgeneration however it is hypothesized that the spoOKgene products are involved transmitting the initial signals into the phosphorelay. In contrast, it is suggested that the kinA gene product in principally involved in amplifying the level of signal inputs into the phosphorelay in order to amplify the rate of phospho-SpoOA generation. This conceptual framework might provide the vehicle in which to analyze the finer regulatory features associated with stationary phase signal transduction. A number of experiments are suggested in order to explore this hypothesis. It would be interesting to look at spoVG expression in a spoOK kinA double mutant. Since spoVG expression is normal in a kinA mutant, the multiple threshold hypothesis would predict that the double mutant would behave similar to a single spoOKmutant. Because kinA is expressed at low levels during exponential growth, it would also be interesting to investigate how early gene expression is altered, if at all, by mutations in the kinA vegetative promoter in wild type and spoOKmutants. Such experiments could help to sort out whether the initial signals responsible for activating the AbrB 227 pathway, presumably originating from spoOK mediated processes, flow into the phosphorelay via kinA. 8&2 Adaptive Cellular Processes in Other Biological Systems The approach employed in this thesis to study heterogeneity should be applicable to the analysis of other biological systems. Analysis of competence development in B. subtilis, mating in S. cerevisiae, chemotaxy and a host of other adaptive response mechanisms exhibited by cells are in theory amenable to similar analysis. The conclusions derived from this study could possibly provide a conceptual framework in which to study signaling processes and adaptive cellular mechanisms in higher organisms. Such systems are not amenable to the advanced genetic techniques available to bacterial systems. This includes a number of industrially relevant biological systems such as stem cell differentiation, neural signal transduction and Tcell development. Since single cell measurement technologies are quite advanced in these systems, the convoluting effects that might be introduced by physiological heterogeneity can easily be taken into account. &3 Biologica Control Systems Theory The theoretical studies performed were aimed at identifying important biological control structures capable of imparting decision making abilities to cells. Strict emphasis was placed on interpreting these control structures in terms of known regulatory mechanisms. The generality with which positive feedback and cooperative 228 generation, can be found in biological systems suggests that this simple approach to analyzing biological systems might prove to be useful in the future. Such an approach is in direct contrast to the methods currently aimed at describing biochemical networks in their entirety. For example, it might be interesting to analyze how the cell regulated its intracellular level of PEP, by regulating the flow of glucose and related products via the PTS system and glycolysis in terms of the control structures analyzed in chapter 6. The PTS system exhibits striking similarities with the phosphorelay as first observed by Burbulys and coworkers (1991). Furthermore, PEP serves to positively regulate the flow of reactants into the network and hence indirectly regulates its own intracellular concentration. The approach taken to analyze such systems (chapter 6) might be fruitful since emphasis is only placed on the key control structures as opposed to all processes involved. &2 Single Cell Measurements 82.1 Single Cell Analsis: Fluorescence Microscopy The technique used to obtained single cell measurements of gene expression involved the sole use of flow cytometry. However it might be useful to compare the results obtained from this technique with fluorescence microscopic observations. The advantage afford by FACS is that samples may be analyzed less subjectively and simultaneous multicolor analysis can be used to control for background. This allows assay conditions to be readily assess and subsequently optimized. In conjunction with 229 FACS optimized assay conditions, fluorescence microscopic observations would provide reliable visual observations which could be used to calibrate and compare the sensitivities inherent to each technique. However the drawbacks with fluorescence microscopy is that in order to practically count a large number of cells (i.e. several hundred), the concentration of cells must be relatively high compared to the concentrations amenable via flow cytometry. Preliminary work along these lines indicated that interference could be a problem. &2.2 The Future: Strain Improvement via Flow Cytometry Strain improvement and selection via flow cytometry represents a extremely challenging area of industrial microbiology, which is very much in its infancy, with the potential for great payback. While much effort is currently being directed towards pathway modification in order to increase culture productivity, the excellent screening features provided by FACS should not be forgotten. Presently the limiting steps lie in the areas of instrument sensitivityand probe development. Yet this is somewhat understandable since the development of FACS technology has been principally guided by eukaryotic biology where such issues are less of a problem. Eukaryotic biologists have taken advantage of this feature and strain selection in the form of clone isolation and hyperproducer isolation is routinely employed. Until equivalent resources are committed to the development of microbiologyspecific instrumentation and protocol development, bacterial flow cytometry will remain in its infancy. 230 Chapter 9: Materials and Methods 9.1 Strains, Maintenance and Cultivation Conditions 9.L1 Strains The B. subtilis strains used throughout this work are listed in table 9.1. All strains derived from JH642 contain the trpC2 and phel mutations (Perego et al. 1988). SMY was obtained from the laboratory of A.L Sonenshein (Mueller and Sonenshein 1992). The spoVG-lacZ translational fusion, present in both the JH642 and SMY backgrounds, was carried in the specialized transducing phage SPf and has been described (Zuber and Losick 1983). The spollG-lacZ transcriptional fusion, present in the JH642 background, was carried in the specialized transducing phage SPf and has been described (Kenney et al. 1989). The spoIIA-lacZ transcriptional fusion, present in both the JH642 and the SMY backgrounds, was carried on the integrative plasmid pPP81 and has been described (Piggot and Curtiss 1987). The spoIID-lacZ (Stragier et al. 1988) transcriptional fusion was located in the nonessential amyE locus and has been described. The veg-lacZ (Fouet and Sonenshein 1990) transcriptional fusion was carried on the integrative plasmid pAF25 and has been described. The kinA::Tn917 null mutation (Sandman et a. 1987) and the spoOK mutation (Rudner et al. 1991) have been described. 9.1.2 Routine Strain Maintenance and Storage LB medium (10g/l Tryptone, 5g/l Yeast Extract, 10g/l NaC1, pH 7.0, Davis et 231 al. 1986) supplemented with 5g/l glucose was used for routine strain growth and maintenance. All strains were stored at -800 C in a 17% glycerol solution. The cell stocks were grown in Difco sporulation media (8g/1 Difco nutrient broth, lg/lKCI, 0.25gAlMgSO 4-7H 2 0, mM Ca(NO 3 )4, 0.01mM MnCl2, 0.001mM FeSO 4; Harwood and Cutting 1990) or LB (5gl glucose) and just prior to the onset of the stationary phase lml of culture was added to 0.5ml 50% glycerol and the mixture was stored at -800C. 9.L3 Cultivation Prior to each experiment, -80°Ccell stocks were reconstituted using LB plates supplemented with 5g/l glucose and the appropriate antibiotic concentrations (Harwood and Cutting 1990). Following approximately twelve hours of growth at 37oC, single exponentially growing colonies were used to inoculate 25ml of prewarmed Difco Sporulation Media (DSM), with and without 5gl1glucose, in 250ml baffled flask without antibiotic selection. vigorous agitation. The flasks were cultivated at 37°C with Genetic stability was verified by replica plating late stationary phase cultures onto the appropriate antibiotic plates. Single colony inoculations insured that at least five doublings had occurred before the cells entered the stationary phase. SMY::pAF25 (Fouet and Sonenshein 1990) was used as the -galactosidase positive control in controlled mixture staining experiments involving C12-FDG. Cells were grown in DSM supplemented with 5g/l glucose to an OD"o of 3.5-4.5, recovered 232 by centrifugation, washed two times with room temperature PBSM (50mM phosphate; 100mM NaCI, 10mM MgC12; pH 7.0), resuspended in the original volume, then placed on ice for further use. The cells used to generate the controlled mixture distributions shown in figure 3.13 contained an average specific activity of 240 Units as defined below. The cell number density, corresponded to approximately 1.5x109 cells/ml as determined using a Coulter Counter. This corresponded to approximately 3xlO8 cells/mlper OD660.SMY, which does not contain a acZ fusion was used as the negative control in all staining experiments performed using the SMY background and was cultivated under similar conditions. MB284 (spoVG-lacZ in SMY; Mueller and Sonenshein 1990) was used as the B-galactosidase positive control in controlled mixture staining experiments involving C8 -FDG (sections 3.84-3.8.5). Cells were grown at 370C in 25 ml (DSM) without glucose and harvested two hours (i.e. at T) into the stationary phase. This corresponded to an OD60 of approximately 2.0 and a cell density of approximately lx109 cells/mi. The culture average specific f-galactosidase activity of samples used in controlled mixtures (section 3.8.5) was approximately 112 Units as defined below. 9.2 Analytical Techniques 9.21 Specific 4alactosidase Activity (ONPG) Cells were recovered by centrifugation, washed twice with phosphate buffered saline supplemented with magnesium (PBSM; 50mM phosphate; 100 mM NaC1, 10 mM MgCI2; pH 7.0) and either stored at -20°C for later analysis or on ice for 233 immediate analysis. Bf-Galactosidase was assayed in whole cells using the substrate o-nitrophenyl-f-D-galactoside (ONPG) according to the procedure of Miller (1972). Cell suspensions were combined with Z-buffer (60mM Na2 HPO4 , 40mM NaH2 PO 4, 10mM KCI, lmM MgSO4, 0.27 % v/v P-mercaptoethanol, pH 7.0) such that the final volume was 1 ml. 10ul toluene was added (1% final concentration) and the solution vortexed vigorouslyfor at least ten seconds. The tubes were then gently shaken at 37oCfor 30-40 minutes to allow the toluene to evaporate. The cell suspensions were then incubated at 30°C for ten minutes and the reaction was started by the addition of 0.2ml of prewarmed ONPG (4g1 ONPG in an aqueous solution of 10.5g/1K2HPO4, 4.5 KH2 PO 4, g/l (NH4 )2 SO4 , 0.5g/l sodium citrate-2H 2 0). The reaction was allowed to proceed until a faint yellow color appeared (at least ten minutes elapsed in all samples) and the reaction stopped by the addition of 0.5ml 1M Na2CO3. The optical density was recorded at 420nm and 550nm. Light scatter due to cell debris was corrected using the formula given by OD42-1.75OD55s (Miller 1972). One unit of activity is defined as AA420per minute per ml culture per OD660X 1000 at 30°C. Spectrophotometric spectrophotometer measurements were made using a Konitron, Uvikon Model 810 using a 1 cm lightpath. 9.2.2 Relative.Galactosidase Activities (MUG) Relative enzymatic activities of sorted populations were determined using the fluorogenic substrate 4-methylumbelliferyl-f-D-galactopyranoside(MUG) using a modification of the procedure described by Fiering and coworkers (1991). 5,000 to 234 50,000 sorted cells were added to Z-buffer such that the final volume was 48 0&1. fB- mercaptoethanol was added so that the final concentration was 2.7 parts to 1000 (Miller 1972). Toluene was added to a concentration of 1% and the mixture was incubated at 37°C for 10 minutes in to equlibrate the samples. 12 0 1lof a 3mM MUG solution was added to the equilibrated mixture and the reaction time recorded. The 3mM solution was obtained by diluting one part of a 30mM MUG stock solution (stored at -200 C, in N,N-Dimethylformamide) with nine parts Z-buffer without P- mercaptoethanol. Samples were incubated at 37°C anywhere from ten minutes to five hours with periodic monitoring, in order to insure a detectable signal had developed, and the reaction was stopped by adding 300julof MUG stop solution (15mM EDTA, 300mM glycine,pH 11.2). A blank consisting of an equal number of lacZ' cells, incubated for an identical time period, was used to record the fluorescence background. Under the cultivation conditions employed, background due to endogenous P-galactosidase was not detected. The fluorescence intensity for each sample was determined using a Perkins Elmer Model LS-5spectrofluorimeter. The excitation and emission monochrometers were set to 3nm centered about 355nm and 460nm respectively. An arbitrary enzymatic activity unit was defined, as AFluorescence per minute per cell, in order to allow comparisons between samples analyzed under identical conditions (i.e. at the same time). Since only the relative activities of various subpopulations within a culture sample were desired, absolute measurements were not needed (section 3.8.4). The linearity of this assay with respect to time and cell numbers, or equivalently 235 enzyme activity, is shown to figure 9.1. 9.23 Spores Sporulation frequencies were determined by serially diluting culture samples in T'-Base (2g/l (NH4) 2 SO4, 14g/l K2HPO4, 6g/l KH2 PO4 , lgfl Na 3 -citrate-2H 2 0) followed by heating the samples for 15 minutes at 80C. 100zl aliquots were plated onto LB plates supplemented with 5g/l glucose. After a 12-15hour of incubation, at 37°C, colony counts were performed. Colony counts were repeated after 24 hours to account for delayed spore germination. At least 100 colonies were counted in order to determine the sporulation frequency. The frequency of sporulation represents the ratio of the number of colonies obtain with and without the heat treatment. 9.2.4 Cell Numbers Cells numbers were determined using a Coulter ZBI Coulter Counter equipped with a 30Mum orifice (Coulter Electronics, Hialeah, FL). The electrode electrolyte was a 4% NaCI solution. Isoton II (Curtin Matheson, Boston MA) was used as the sample diluent. 1/Ampere was set to 0.354 and 1/Amplification was set to 1/4. 25ml of Isoton II was dispensed into the sample Acuvettes (Curtin Matheson) and typically from 5 to 20/zl of cells were added to the solution. This dilution resulted in a cell concentration from 1x105 to 3x10s cells/ml so that counts between 10,000-30,000 were obtained per 100l of dilute cell solution. All cell counts were 236 performed in triplicate and the arithmetic means employed. Both, the Isoton II and the 4% NaCI electrolyte solution were filtered twice using a 0.2tm filters prior to use. 9.3 Cel Staining 9.3.1 PKH26 Staing lacZ' cells to be stained with the membrane binding dye PKH26 (Zynaxis, Malvern PA) were harvested according to the procedures described above. cells, corresponding to approximately 2x106 cells, were pelleted 2 00 /l of in a 1.5 ml microcentrifuge tube, resuspended with 10/ul PBSM then rapidly but mildly mixed with 1001Alof Diluent C (Zynaxis). This solution was immediately added to a room temperature solution containing 51AMPKH26 in 100/zl Diluent C (i.e. 2/1 of the 1mM PKH26 stock was added to 1001 l Diluent C). The solution was gently but thoroughly mixed then incubated at room temperature for five to seven minutes. The staining reaction was stopped by diluting the solution 1/10 with room PBSM containing 20g/1l bovine serum albumin (BSA). The mixture was incubated for one minute then the cells were recovered by centrifugation. The cells were washed three times with room temperature PBSM, resuspended in 200/A1PBSM then placed on ice. It was found to be absolutely essential that all solutions used be filtered. 0.2/Mmlow protein binding filters were used (Part No. 4192, Gelman Sciences, Ann Arbor MI) in this work. To insure that the level of PKH26 specific fluorescence was greater than that of background (detected by the PKH26 optical system; figure 9.2), it was sometimes necessary to increase the dye concentration to 7.5aM. Alternatively, the number of 237 cells to be stained could be reduced. The precaution was taken because the final PKH26 fluorescence was found to depend on the amount of total surface area in the sample to be stained relative to the dye concentration. 9.3.2 c-FDG Cells taken from late exponential phase cultures containing glucose were stained with the fluorogenic -galactosidase substrate C 12 -FDG. 1x106 cells were added to lml of PBSM containing 33MM C12-FDG, 0.4 g/l sodium azide and x106 lacZ' cells, stained with PKH26. All mixture constituents were contained in Falcon 2054 polystyrene tubes (Becton Dickinson), since these tubes allowed analysis via flow cytometry without the need to transfer the sample. The solution was incubated at 50°C for 10 minutes then immediately placed on ice. Samples were analyzed by flow cytometry 8-12 hours later. The hydrophobic nature of the C12-FDG placed upper limits on the dye concentration in PBSM. To insure that the substrate was dissolved, the solution was heated for 5 minutes at 45°C, then filtered through a 0.2/M low protein binding filter (Gelman Sciences). Filtering was found to remove the fluorescent particles that tended to accumulate in the substrate stock solution. Low protein binding filters, along with heat treatment were necessary to prevent the substrate from being removed from solution upon filtering. In contrast to the SMY strain, SMY::pAF25 eventually sporulated in DSM (5gAlglucose). Presumably this was due to an exhaustion of glucose, facilitated by the 238 expression of veg-lacZ Under such conditions, the C 12 -FDG could not be sufficiently loaded into the cell under the described conditions. This problem was overcome using the C8-FDG protocol described in the next section. In fact, this observation led to the identification of CS-FDG as the optimal substrate for stationary phase cells (section 3.8.2). 9.33 CFDG Stationary phase samples to be stained with the fluorogenic -galactosidase substrate Cs-FDG were harvested as described above. 4 50 5 0/p1of cells were added to 1Ml of PBSMA containing 1% w/v toluene and 1x106 lacZ cells stained with PKH26. The mixture was vortexed vigorously for at least ten seconds then diluted 1/100 with a solution of PBSM containing 33/AM C-FDG. The samples were incubated for 30 minutes at room temperature to allow a steady state fluorescence distribution to be reached, then analyzed by flow cytometry. As shown in figure 3.17, a steady state fluorescence distribution was obtained after a few minutes at room temperature. CS-FDGwas purchased from Molecular Probes (Eugene OR) in the form of a desiccated solid. To reconstitute the substrate, 627;/l of a solution containing lml H20 and 250/u1dimethyl sulfoxide(DMSO) and added to 5mg C8-FDG. The mixture was vortexed then stored at -20°C. This stock solution was then used to makeup a 33/aM solution for cell staining. Prior to use, the staining solution was filtered using a 0.2,um low protein binding filter. In contrast to C12-FDG, the heat treatment was 239 not found to be necessary. 9.4 Flow Cytometr 9.4.1 The Flow Cytometer The flow cytometer used in this work is a FACStar Plus (Becton Dickinson, Mountain View CA). A simplified schematic of this instrument is presented in figure 9.2. Cells were hydrodynamically focused in order to pass cells single file through argon laser outputing 100mW of 488nm light. For routine analysis, the sheath fluid consisted of Isoton II (Curtin Matheson). Upon illumination, the amount of light scattered and given off by fluorescing molecules associated with the cells was monitored. Forward light scatter was detected using standard photodiodes. In order to detect cells using forward scatter, the forward scatter neutral density filter (a standard instrument feature) was removed. Fluorescent light was collected and passed through a 560nm short pass dichroic mirror. The short wavelength light generated by fluorescein (i.e. the light less than 560nm which is passed by the dichroic mirror) was then passed through a 530 band pass filter (30nm bandwidth at half transmission; 525 BP 3 band pass optical filter; Omega Optical, Brattleboro, VT). The longer wavelength light generated by PKH26, the component reflected by the 560nm short pass dichroic mirror, was then passed through a 575nm band pass filter (26nm bandwidth at half transmission; 575 BP 3 band pass optical filter; Omega Optical). The fluorescing light generated by C.-fluorescein was gathered using a photomultiplier tube operating between 700-800 volts. The resulting signal was 240 discretized using a four log decade amplifier. The fluorescing light generated by PKH26 was gathered using a photomultiplier tube operating between 600-700volts. This resulting signal was discretized using a five log decade amplifier. The photomultiplier outputs transferred to a MicroVax 3000 real-time work station using a 12-bit ADC. 9.4.2 CeSorting Sorting experiments were performed using a 70/Amjet in air nozzle and cells were sorted into cold Z-buffer (Miller 1972) without P-mercaptoethanol. In sorting experiments, Z-buffer, without fi-mercaptoethanol, was also used as the sheath fluid. The drop drive was operated at 23,000 Hz and the sheath pressure was set to 10psi. Due to the noise inherent in the forward scatter (FSC) signal with the drop drive operating, it was necessary to trigger sorting on the side scatter (SSC) signal. The fluorescence gate used to sort the high expressing population was set by the green fluorescence distribution of the PKH26 background control contained in a sample identical to the sample to be sorted. The actual sample sorted did not contain an internal control in order to avoid possible contamination of the sorted populations by the artificiallyintroduced population. The samples to be sorted were stained according to the procedures described above (C-FDG) then maintained at 4°C, using a refrigerated water bath, during the duration of the sort. To obtain a certain number of sorted cells, the base assay volume (i.e. 0.5 ml) was scaled linearly in all components in order to achieve the desired cell numbers. 241 The cell density of the sorted population was set by the capabilities of the flow cytometer. Based on the number of events sorted and the volume of the sorted sample, the cell number density of the sorted sampled was determined to be approximately 3.3x10 cells/ml. This calculation was performed in order to determine the volume of sorted sample to be used when relative enzyme activities were to be determined according to the MUG procedure (section 9.2.2). 9.43 Data Acquisition and Analysis The flow cytometer was interfaced to MicroVAX 3000 operating under the VMS operating system. ACQ8 was used to acquire data in real-time (Becton Dickinson). Off-line one and two color data analysis was performed using DISP4, DISP2D (Becton Dickinson, San Jose CA) respectively. Parameter conversions were performed using CALC4 (Dr. Robert Murphy, Carnegie Mellon). Non-linear regression used for fluorescence distribution analysis was performed using the MODFIT program operating under the Windows environment (Verity Software House, Topsham ME). This program utilizes the Levenberg-Marquart algorithm (Press et a. 1990) in order to minimize the chi-squared quantity. An example histogram is shown in figure 9.3 (this distributions corresponds identically to the distribution shown in figure 3.16B). MicroVax 3000 based Flow Cytometry Standard (FCS) files were downloaded to an IBM compatible (Northgate 486) using the KERMIT file transfer protocol. The use of logarithmic amplifiers necessitated the conversion of the 242 fluorescence data to a linear format. This was performed using the CALC4 program (Robert Murphy, Carnegie Mellon). The relative linear fluorescence data presented throughout this work were obtained by linearizing both the experimental cell population and the internal control population, prior to substrate addition and after the sample obtained a steady state fluorescence distribution. The reported fluorescence values were obtained using the formula: RFL=ARFL(Sampl e) -ARLF(PKH26Control ) (1) where ARFL(sample) represents the difference in average fluorescence between the steady state sample distribution and the unstained sample, obtained via CALC4, and ARLF(PKH26 Control) represents the difference in average fluorescence between the initial and final PKH26 control distributions, obtained via CALC4. 9.5 Theoretical Computations Model simulations were performed using a fourth order Runge-Kutta algorithm with automatic step size adjustment (SIMNON; Engineering Software Concepts, Palo Alto CA). 243 Table 9.1 Bacillus subtilis Strains strain genotype reference JH642 SMY SMY::pAF25 MB284 MB296 KI1179 KI1398 trpC2 phel Prototrophy amyEl:(veg-lacZ) cat SPf::(spoVG-IacZ) cat spoIIA+::pPP81(spoIIA-lacZ) cat SPp::(spoVG-lacZ) cat SPBf::(spoVG-lacZ)kinA::Tn917cat SIK122 AG1350 SPpB::(spollG-lacZ) cat SP::(spoIIG-lacZ)kinA::Tn917::neo KI1183 KI1397 KI938 KI100o amyE:(spoIID-lacZ) cat amyE:(spollD-IacZ)kiA::Tn917 cat spoIIA+::pPP81(spollA-lac cat spoIIA+::pPP81(spoIIA-lacZ)kiA::Tn917::spc spolIG-lacZspoOKAC 244 Perego et aL 1988 Mueller and Sonenshein 1992 Fouet and Sonenshein 1990 Mueller and Sonenshein 1992 Mueller and Sonenshein 1992 Zuber and Losick 1983 Zuber and Losick 1983 cat Kenney et aL 1989 Kenney et aL 1989 Stragier et aL 1988 Stragier et aL 1988 Piggot and Curtis 1987 Piggot and Curtis 1987 Rudner et aL 1991 1;U 3 100 vs c o 60 0C 0 o 40 0 IL 20 n 0 2 4 6 8 10 Time (hr) . ,~ 1ZU I 100 c 0 0o 0 C 60 0 0 3 40 IL 20 0 0.0 30000 60000 Number of Cells Fgure 9.1 Linearity of the MUG assay with respect to time (A) and total cell number (B). 245 a, a, E C 0 .......P. :te 00 .0 7C 0rz @1 0 0 0 EDi 00 0A t00 -0 Fqure 92 Schematic illustration of the flow cytometer used in this work. 246 L 300-- 200-- 100-- 0- ,i,,1i,,,1I,,,i,,,,i,,l1l1l,i,,,1I,,,,i,,i,111111l 30 60 90 120 150 180 210 240 270 300 Fgure 9.3 Sample output obtained from the MODFIT curve fitting program. The X-axis scales with the green fluorescence signal and the Y-axis scales with the number of events. The population exhibiting high fluorescence was determined to be 72% of the total population. 247 Chapter 10: References Alberti, S., Parks, D.R., Herzenberg, LA. (1987). A single laser method for subtraction of cell autofluorescence in flow cytometry, Cytometry, 8, 114-119. Ackers, G.K., Johnson, AD., Shea, M.A. (1982). Quantitative model for the regulation by A phage repressor, Proc. Natl. Acad. Sci. USA, 79, 1129-1133. Amann, R.I., Kromholz, L, Stahl, D.A (1990). Fluorescent-oligonucleotide probing of whole cells for determinative, phylogenetic, and environmental studies in microbiology, J. Bacteriol., 172, 762-770. Antoniewski, C., Savelli, B., Stragier, P. (1990). The spolIJgene, which regulates early developmental steps in Bacillus subtilis, belongs to a class of environmentally responsive genes, J. Bacteriol., 172, 86-93. Arnold, V.I. (1986). Catastrophe Theory, Springer-Verlag, New York, NY. Aubert, J.P., Millet, J., Castoriadis-May, C. (1961). Relation entre le taux de croissance et la sporulation chez Bacillusmegaterium, Compt Rend., 253, 1731-1733. Azuma, T., Harrison, G.I., Demain, AL. (1992). Isolation of a gramicidin S hyperproducing strain of Bacillusbrevis by use of a fluorescence activated cell sorting system, Appl. Microbiol. Biotech., 38, 173-178. Bai, U., Mandic-Mulec, I., Smith, I. (1993). SinI modulates the activity of SinR, a developmental switch protein of Bacillus subtilis, by protein-protein interaction, Genes & Dev., 7, 139-148 Bai, U., Lewandoski, M., Dubnau, E., Smith, I. (1990). Temporal regulation of the Bacillus subtilis early sporulation gene spoOF,J. Bactenol., 172, 5432-5439. Bailey, J.E., Fazel-Madjlessi, J., McQuitty, D.N., Lee, LY., Allred, J.C., Oro, J.A (1977). Characterization of Bacterial Growth by Means of Flow Microfluorimetry, Science, 198, 1175-1176. Bailey, J.E., Fazel-Madjlessi, J., McQuitty, D.N., Lee, LY., Oro, J.A (1977). Measurement of structured microbial population dynamics by flow microfluorimetry, AIChE Journal, 24, 570-577. Bailone, A, Levine, A, Devoret, R. (1979). Inactivation of prophage A repressor in vivo, J. Mol. Biol., 131, 553-572. 248 Barkley, M.D., Bourgeois, S. (1980). Repressor recognition of operator and effectors, In: The Operon, Miller, J.H. & Reznikoff, W.S. (eds), CSH Laboratory, USA, pp. 177-220. Benzer, S. (1953). Induced Synthesis of enzymes in bacteria analyzed at the cellular level, Biochim. Biophys. Acta, 11, 383-395. Betz, J.W., Aretz, W., Hartel, W. (1984). Use of flow cytometry in industrial microbiologyfor strain improvement programs, Cytometry, 5, 145-150. Bird, T.H., Grimsley, J.K., Hoch, J.A., Spiegelman, G.B. (1993). Phosphorylation of SpoOA activates its stimulation of in vivo transcription from the Bacillus subtilis spoIIG operon, Mol. Microbiol., 9, 741-749. Boezi, J.A., Cowie, D.B. (1961). Kinetic studies of P-galactosidase induction, Biophys. Journal, 1, 639-647. Bourgeois, S., Monod, J. (1970). In vitro studies of the lac operon regulatory system, In: Control Processes in Multicellular Organisms, Wolstenholme, G.E.W., Knight, J. (eds.), J & A Churchill, London, pp. 3-27. Bouvier, J., Stragier, P., Bonamy, C., Szulmajster, J. (1984). Nucleotide sequence of the spoOB gene of Bacillussubtilis and regulation of its expression, Proc. Natl. Acad. Sci. USA, 81, 7012-7016. Boye, E., Steen, H.B., Skarstad, K. (1983). Flow cytometry of bacteria: a promising tool in experimental clinical microbiology, J. Gen. Micro., 129, 973-980. Burbulys, D., Trach, KA., Hoch, J.A (1991). Initiation of sporulation in B. subtilis is controlled by a multicomponent phosphorelay, Cell, 64, 545-552. Carter, H.L, Moran, C.P. (1986). New RNA polymerase a factor under spoOcontrol in Bacillus subtilis, Proc. Natl. Acad. Sci. USA, 83, 9438-9442. Chadwick, P., Pirrotta, V., Steingerg, R., Hpokins, N., Ptashne, M. (1970). The A and 434 phage repressors, CSH Sypm. Quant. Biol., 35, 283-294. Chang, S. (1987). Engineering for protein secretion in gram-positive bacteria, Meth. in Enzymol., 153, 507-516. Clarke, S., Lopez-Diaz, I., Mandelstam, J. (1986). Use of lacZ gene fusions to determine the dependence pattern of the sporulation gene spolID in spo mutants of Bacillus subtilis, J. Gen. Microbiol., 132, 2987-2994. 249 Clarke, S., Mandelstam, J. (1987). Regulation of stage II of sporulation in Bacillus subtilis, J. Gen. Microbiol., 133, 2371-2380. Cohn, M., Horibata, K. (1959). Analysis of the differentiation and of the heterogeneity within a population of Escherichia coli undergoing induced fBgalactosidase synthesis, J. Bacteriol., 78, 613-623. Cohn, M., Monod, J. (1951). Purification et properties de la P-galactosidase (lactase) d'Escherichia coll, Biochim. Biophys. Acta, 7, 153-174. Collins, J.F., Mason, D.B., Perkins, W.J. (1964). A microphotometric method for the estimation of penicillinase in single bacteria, J. Gen. Microbiol., 34, 353-362. Cooper, S., Helmstetter, C.E. (1968). Chromosomal replication and the division cycle in Escherichia coli B/r, J. Mol. Biol., 31, 519-540. Coote, J.G. (1972). Sporulation in Bacillus subtilis. Characterization of oligosporogenous mutants and comparison of their phenotypes with those of asporogenous mutants, J. Gen. Microbiol., 71, 1-15. Crissman, H.A., Orlicky, D.J., Kissane, R.J. (1979). Fluorescent DNA probes for flow cytometry, J. Histochem. Cytochem., 27, 1652-1654. Dawes, I.W., Mandelstam, J. (1969). Biochemistry of sporulation of Bacillus subtilis 168: continuous culture studies, In: Continuous cultivation of Micro-organisms. Proccedings of the 4th Symposium, Prague, Prague:Academia, Publishing House of Czechoslovak Academy of Sciences, pp. 157-162. Dawes, I.W., Mandelstam, J. (1970). Sporulation of Bacillus subtilis in continuous culture, J. Bacteriol., 103, 529-535. Dawes, I.W., Thornley, J.H.M. (1970). Sporulation in Bacillus subtilis. Theoretical and experimental studies in continuous culture systems, J. Gen. Microbiol., 62, 49-66. Davis, L.G., Dibner, M.D., Battey, J.F. (1986) Basic Methods in Molecular Biology, Elsevier Sci. Publish., New York, NY. Delong, E.F., Wickman, G.S., Pace, N.R. (1989). Phylogenetic stains: Ribosomal RNA-based probes for the identification of single cells, Science, 243, 1360-1363. Driks, A., Losick, R. (1991). Comparmentalized expression of a gene under the control of sporulation transcription factor oBin Bacillus subtilis,Proc. Natl Acad. Sci. ,88, 9934-9938. 250 Dubnau, E.J., Cabane, K., Smith, I. (1987). Regulation of spoOH, n early sporulation gene in Bacilli, J. Bacteniol., 169, 1182-1191. Dunn, G., Jeffs, P., Mann, N.M., Torgensen, D.M., Young, M. (1978). The relationship between DNA replication and the induction of sporulation in Bacillus subtlis J. Gen. Microbiol., 108, 189-195. Elion, E.A., Brill, J.A., Fink, G.R. (1991) FUS3 represses CLN1 and CLN2 and in concert with KSS1 promotes signal transduction, Proc. Natl. Acad. Sci. USA, 88, 9392-9396. Errington, J. (1993). Bacillus subtilis sporulation: Regulation of gene expression and control of morphogenesis, Microbiol. Rev., 57, 1-33. Errington, J. (1986). A general method for fusion of the Escherichia coli lacZ gene to chromosomal genes in Bacillus subtilis, J. Gen. Microbiol., 132 2953-2966. Errington, J., Mandelstam, J. (1986). Use of a lacZ gene fusion to determine the dependence pattern of the sporulation operon spollIA in spo mutants of Bacillus subtilis, J. Gen. Microbiol., 132 2967-2976. Fencl, Z., Pazlarova, J. (1982). Production of extracellular enzymes in continuous culture, Folia Microbiol., 27, 340-349. Ferrari, F.A., Trach, K., LeCoq, D., Spence, J., Ferrari, E., Hoch, J.A. (1985). Characterization of the spoOA locus and its deduced product, Proc. Natl. Acad. Sci. USA, 82, 2647-2651. Fiering, S.N., Roederer, M., Nolan, G.P., Micklem, D.R., Parks, D.R., Herzenberg, LA. (1991). Improved FACS-Gal: Flow cytometric analysis and sorting of viable eukaryotic cells expressing reporter gene constructs, Cytometry, 1Z 291-301. Fisher, S., Rothstein, D., Sonenshein, AL (1975). Ribonucleic acid synthesis in permeabilized mutant and wild-typecells of Bacillussubtilis,In: Spores VI, Gerhardt, P., Costilow, R.N., Sadoff, H.L (eds), ASM, Washingtpn D.C., pp. 226-230. Fouet, A., Sonenshein, AL. (1990). A target for carbon source-dependent negative regulation of the citB promoter in Bacillus subtilis, J. Bacteriol., 172, 835-844. Gaur, N.K., Oppenheim, J., Smith, I. (1991). The Bacillussubtilis sin gene, a regulator of alternate developmental processes, codes for a DNA-binding protein, J. Bacteriol., 173, 678-686. 251 Gaur, N.K., Cabane, K., Smith, I. (1988). Structure and expression of the Bacillus subtilis sin operon, J. Bacteriol., 170, 1046-1053. Gaur, N.K., Dubnau, E., Smith, I. (1986). Characterization of a cloned Bacillus subtilis gene that inhibits sporulation in multiple copies, J. Bacteriol., 168, 860-869. Gray, O., Chang, S. (1981). Molecular cloning and expression of Bacillus lichenifornnis P-lactamase gene in Escherichia coli and Bacillus subtilis, J. Bacteriol., 145, 422-428. Gholomhoseinian, A., Piggot, P.J. (1989). Timing of spolI gene expression relative to septum formation during sporulation of Bacillussubtihs, J. Bacteriol., 171, 5747-5749. Gober, J.W., Shapiro, L (1991). Temporal and spatial regulation of developmentally expressed genes in Caulobacter, Bioessays, 13, 277-283. Golubitsky, M.G., Schaeffer, D.G. (1985). Singularities and Groups in Bifurcation Theory, Springer-Verlag, New York, NY. Grossman, AD. (1991). Integration of developmental signals and the initiation of sporulation in B. subtilis, Cell, 65, 5-8. Grossman, AD., Losick, R. (1988). Extracellular control of spore formation in Bacillus subtilis, Proc. Natl. Acad. Sci. USA, 85, 4369-4373. Guzman, P., Westpheling, J., Youngman, J. (1988). Characterization of the promoter region of the Bacillus subtilis spollE operon, J. Bacteriol., 170, 1598-1609. Harwood, C.R., Cutting, S.M. (1990). Molecular biologicalmethods for Bacillus, John Wiley and Sons, Chichester, England. Haugland, R.P. (1992). Handbook of fluorescentprobes and research chemicals, Molecular Probes Inc., Eugene, OR. Heineken, F.G., O'Connor, R.J. (1972). Continuous culture studies on the biosynthesis of alkaline protease and a-amylase by Bacillus subtilis NRRL-B3411, J. Gen. Microbiol., 73, 35-44. Herzenberg, LA. (1959). Studies on the induction of -galactosidase in a cryptic strain of Escherichia coli, Biochim. Biophys. Acta, 31, 525-538. Hoch, J.A., Trach, K, Kawamura, F., Saito, H. (1985). Identification of the transcritpional supressor sof-1 as an alteration in the SpoOA protein, J. Bacteriol., 161, 552-555. 252 Hofmann, J., Sernetz, M. (1983). A kinetic study of the enzymatic hydrolysis of fluoresceindiacetate and fluorescein-di-/-D-galactopyranoside, Anal. Biochem., 131, 180-186. Honjo, M., Akaoka, A., Nakayama, A., Furutani, Y. (1986). Secretion of human growth hormone in Bacillus subtilis using prepropeptide coding region of Bacillus amyloliquefaciens neutral protease gene, J. Biotech., 4, 63-71. Honjo, M., Akaoko, A., Nakayama, A., Shimada, H., Furutani, Y. (1985). Construction of the secretion vector containing the prepro-structure coding region of Bacillus amyloliquefaciens neutral protease gene and secretion of Bacillussubtilis aamylase and human interferon-beta in Bacillus subtilis, J. Biotech., 3, 73-84. Honjo, M., Nakayama, N., Iio, A., Kawamura, K., Sawakura, A., Furutani, Y. (1987). Construction of a highlyefficient host-vector system for the secretion of heterologous protein in Bacillus subtilis, J. Biotech., 6, 191-204. Honjo, M., Akaoka, A., Nakayama, N., Shimada, H., Kawamura, K., Sawakura, A., Furutani, Y. (1986). Construction of secretion vector and secretion of hIFN-, in Bacillus moleculargenetics and biotechnologyapplications,Ganesan, A.T., Hoch, J.A. (eds), Acad. Press, Inc., Orlando, FL, pp. 89-100. Horan, P.K., Wheeless, LL (1977). Quantitative single cell analysis and sorting, Science, 198, 149-157. Horan, P.K., Melnicoff, M.J., Jensen, B.D., Slezak, S.E. (1990). Fluorescent cell labeling for in vivo and in vitro cell tracking, Meth. Cell Biol., 33: Chapter 42, 469490. Horvitz, H.R., Herskowitz, I. (1992). Mechanisms of asymmetric cell division: Two Bs or not two Bs, that is the question, Cell, 68, 237-255. Humphrey, A.E., Kitai, A., Cooney, C.L (1966). Dynamics of bacterial spore cultivation in batch and continuous culture, J. Ferm. Technol., 44, 283-300. Iooss, G., Joseph, D.D. (1990). Elementary Stabilityand Bifurcation Theory, SpringerVerlag, New York. Ireton, K., Rudner, D.Z., Siranosian, K.J., Grossman, A.D. (1993). Integration of multiple developmental signals in Bacillus subtilis through the SpoOAtranscription factor, Genes & Dev., 7, 283-294. 253 Ireton, KI, Grossman, A.D. (1992). Coupling between gene expression and DNA synthesis early during development in Bacillus subtilis, Proc. Natl. Acad. Sci. USA, 89, 8808-8812. Jacob, F., Monod, J. (1961). Genetic regulatory mechanisms in the synthesis of proteins, J. Mol. Biol., 3, 318-356. Jacobberger, J.W., Fogelman, D., Lehman, J.M. (1986). Analysis of intracellular antigens by flow cytometry, Cytometry, 7, 356-364. Johnson, AD., Meyer, B.J., Ptashne, M. (1979). Interactions between DNA-bound repressors govern regulation by the A phage repressor, Proc. Natl. Acad Sci., 76, 5061-5065. Johnson, AD., Poteete, AR., Lauer, G., Sauer, R.T., Ackers, G.K., Ptashne, M. (1981). A repressor and cro-components of an efficient molecular switch, Nature, 294, 217-223. Kallio, P.T., Fagelson, J.E., Hoch, J.A., Strauch, M.A. (1991). The transition state regulator Hpr of Bacillus subtilis is a DNA-binding protein, J. Biol. Chem., 266, 13411-13417. Kawamura, F., Saito, H. (1983). Isolation and mapping of a new suppressor mutation of an early sporulation gene spoOF mutation in Bacillus subtlis, Mol. Gen. Genet., 192, 330-334. Kell, D.B., Ryder, H.M., Kaprelyants, AS., Westerhoff, H.V. (1991). Quantifying heterogeneity: flow cytometry of bacterial cultures, Antonie van Leeuwenhoek, 60, 145-158. Kenney, T.J., Kirchman, P.A., Moran, C.P. (1988). Gene encoding ua is transcribed from a a^-like promoter in Bacillus subtilis, J. Bacteiol., 170, 3058-3064. Kenney, T.J., York, K., Youngman, P., Moran, C.P. (1989). Genetic evidence that RNA polymerase associated with aA factor uses a sporulation-specific promoter in Bacillus subtili& Kepes, A. (1960). Etudes cinetiques sur la galactoside-permease D'Escherichia coli, Biochim. Biophys. Acta, 40, 70-84. Koblan, KIS., Ackers, G.K. (1992). Site-specific enthalpic transcription at bacteriophage A OR, Biochemistry, 31, 57-65. 254 regulation of DNA Koizumi, J., Aiba, S. (1989). Oscillatory behavior of population density in continuous culture of genetically-engineeredBacillusstearothermophilus, Biotechnol. Bioeng., 34, 750-754. Levenspiel, 0. (1972). Chemical Reaction Engineering, John Wiley & Sons, New York, NY. Lewandoski, M., Dubnau, E., Smith, I. (1986). Transcriptional regulation of the spoOF gene of Bacillus subtilis, J. Bacteriol., 168, 870-877. Lewis, J., Slack, M.W., Wolpert, L (1977). Thresholds in development, J. Theor. Biol., 65, 579-590. Loken, M.R., Parks, D.R., Herzenberg, LA. (1977). Two-color immunofluorescence using a fluorescence-activated cell sorter, J. Histochem. Cytochem., 25, 899-907. Losick, R., Stragier, P. (1992). Crisscross regulation of cell-type-specific gene expression during development in B. subtilis, Nature, 355, 601-604. Louie, P., Lee, A., Stansmore, K., Grant, R., Ginther, C., Leighton, T. (1992). Roles of rpoD, spolIF, spolIJ, spolIN, and sin in Regulation of Bacillus subtilis Stage II sporulation-specific transcription, J. Bacteiol., 174, 3570-3576. Maloney, P.C., Rotman, B. (1973). Distribution of suboptimally induced Bf-Dgalactosidase in Escherchia coli, 73, 77-91. Mandelstam, J., Higgs, S.A. (1974). Induction of sporulation during synchronized chromosome replication in Bacillus subtilis, J. Bacteiol., 120, 38-42. Mandelstam, J., Sterlini, J.M. (1969). "Commitment" to sporulation in Bacillus subtilis, In: Spores IV, Campbell, LL (ed), ASM, Bethesda, MD, pp. 180-185. Mandic-Mulec, I., Gaur, N., Bai, U., Smith, I. (1992). Sin, a stage-specific repressor of cellular differentiation, J. Bactenril., 174, 3561-3569. Margolis, P., Driks, A., Losick, R. (1991). Establishment of cell type by compartmentalized activation of a transcription factor, Science, 254, 562-565. Marsh, L, Neiman, A.M., Herskowitz, I. (1991). Signal transduction during pheromone response in yeast, Annu. Rev. Cell Biol., 7, 699-728. Maruo, B., Yoshikawa, H. (1989) in Bacillus subtilis:Molecular biologyand industrial applications, Maruo, B., Yoshikawa, H. (eds), Kodansha LTD., Tokyo, Japan. 255 Mathies, R.A., Stryer, L (1986). Single-molecule fluorescence detection: A feasibility study using phycoerythrin, In: Applications of fluorescence in the Biomedical Sciences, Taylor, D.L, Abbott Lab. (eds), Alan R. Liss, Inc., New York, pp. 129-140. Meyer, H.-P., Fiechter, A. (1985). Production of cloned human leukocyte interferon by Bacillus subtilis: Optimal production is connected with restrained growth, Appl. Environ. Microbiol., 50, 503-507. Miller, J.H. (1972). Experiments in molecular genetics, Cold Spring Harbor, New York, NY. Miller, J.S., Quarles, J.M. (1990). Flow cytometric identification of microorganisms by dual staining with FITC and PI, Cytomety, 11, 667-675. Moran, C.P. (1991). expression of stage II genes during sporulation in Bacillus subtilis, Res. Microbiol., 142, 841-845. Mueller, J.P., Sonenshein, A.L (1992). Role of the Bacillus subtilis gsiA gene in regulation of early sporulation gene expression, J. Bacteriol., 174, 4374-4383. Nakayama, A., Kawamura, KI, Akaoka, A., Mita, I., Honjo, M., Furutani, Y. (1987). Extracellular production of human growth hormone by a head portion of the Bacillus amyloliquefaciens neutral protease in Bacillus subtilis, J. Biotech., 5, 171-179. Nakayama, A., Ando, K., Kawamura, K, Mita, I., Hori, M., Honjo, M., Furutani, Y. (1988). Efficient secretion of the authentic mature human growth hormone by Bacillus subtils, J. Biotech., 8, 123-134. Newton, A., Ohta, N. (1990). Regulation of the cell division cycle and differentiation in bacteria, Annu. Rev. Microbiol., 44, 689-719. Nicaud, J.-M., Mackman, N., Holland, I.B. (1986). Secretion status of secretion of foreign proteins by microorganisms, J. Biotech., 3, 255-270. Nir, R., Lamed, R., Gueta, L, Sahar, E. (1990). Single-cell entrapment and microcolony development within unifrom microspheres amenable to flow cytometry, Appl. Environ. Micro., 56, 2870-2875. Nir, R., Yisraeli, Y., Lamed, R., Sahar, E. (1990). Flow cytometry sorting of viable bacteria and yeast according to -galactosidase activity, Appl. Environ. Micro., 56, 3861-3866. 256 Nolan, G. P., Fiering, S., Nicolas, J.F., Herzenberg, LA. (1988). Fluorescence- activated cell analysis and sorting of viable mammalian cells based on -Dgalactosidase activity after transduction of Escherchia coli lacZ, Proc. Natl. Acad Sci. USA, 85, 2603-2607. Novick, A., Weiner, M. (1957). Enzyme induction as an all-or-none phenomenon, Proc. Natl. Acad. Sci. USA, 43, 553-566. Novick, A., Weiner, M. (1959). The kinetics of f-galactosidase induction, In: A Symposium on Molecular Biology, Zirkle, R.E. (ed), Univ. Chicago Press, Chicago, pp. 78-90. Olmeda, G., Ninfa, E.G., Stock, J., Youngman, P. (1990). Novel mutations that alter the regulation of sporulation in Bacillus subtlis. Evidence that phosphorylation of regulatory protein SpoOA controls the initiation of sporulation, J. Mol. Biol., 215, 359372. Othmer, H.G. (1976). The qualitative dynamics of a class of biochemical control circuits, J. Math. Biol., 3, 53-78. Overath, P, Schairer, H.U., Stoffel, W. (1970). Correlation of in vivo and in vitro phase transitions of membrane lipids in Escherichia coli, Proc. Natl. Acad. Sci. USA, 67, 606-612. Palva, I., Lehtovaara, P., Sibakov, M., Kaariainen, L, Sibakov, M., Cantell, K., Schein, C.H., Kashiwagi, K., Weissmann, C. (1983). Secretion of interferon by Bacillus subtilis Gene, 22, 229-235. Palva, I., Sarvas, M., Lehtovaara, P., Sibakov, M., Kaariainen, L (1982). Secretion of Escherichia coli P-lactamase from Bacillus subtillisby the aid of a-amylase signal sequence, Proc. Natl. Acad Sci. USA, 79, 5582-5586. Peng, S., Sommerfelt, M.A, Berta, G., Berry, AK, Kirk, K.L, Hunter, E., Sorscher, E.J. (1993). Rapid purification of recombinant baculovirus using fluorescenceactivated cell sorting, BioTechniques, 14, 274-277. Perego, M., Higgins, C.F., Pearce, S.R., Gallagher, M.P., Hoch, J.A. (1991). The oligopeptide transport system of Bacillus subtilis plays a role in the initiation of sporulation, Mol. Microbiol., 5, 173-185. Perego, M., Hoch, J.A (1987). Isolation and sequence of the spoOE gene: its role in initiation of sporulation in Bacillus subtilis, Mol. Microbiol., 1, 125-132. 257 Perego, M., Hoch, J.A. (1988). Sequence analysis and regulation of the hpr locus, a regulatory gene for protease production and sporulation in Bacillus subtihs J. Bacteiol., 17 2560-2567. Perego, M., Hoch, J.A. (1991). Negative regulation of Bacillussubtilis sporulation by the spoOE gene product, J. Bacteriol., 173, 2514-2520. Perego, M., Wu, J.-J., Spiegelman, G.B., Hoch, J.A. (1991). Mutational dissociation of the positive and negative regulatory properties of the SpoOA sporulation transcription factor of Bacillus subtilis, Gene, 100, 207-212. Perego, M., Cole, S.P., Burbulys, D., Trach, K., Hoch, J.A. (1989). Characterization of the gene for a protein kinase which phosphorylates the sporulation-regulatory proteins SpoOA and SpoOF of Bacillus subtilis, J. Bacteriol., 171, 6187-6196. Perego, M., Spiegelman, G.B., Hoch, J.A. (1988). Structure of the gene for the transition state regulator, abrB. regulator synthesis is controlled by the spoOA sporulation gene in Bacillus subtilis, Mol. Microbiol., 2(6), 689-699. Phillips, A.P., Martin, K.L (1982). Immunofluorescence analysis of Bacillus spores and vegetative cells by flow cytometry, Cytometry, 4, 123-131. Piggot, P.J., Curtis, C.A.M. (1987). Analysis of the regulation of gene expression during Bacillus subtilis sporulation by manipulation of the copy number of spo-lacZ fusions, J. Bacteriol., 169, 1260-1266. Pirrotta, V., Chadwick, P., Ptashne, M. (1970). repressors, Nature, 227, 41-44. Active from of two coliphage Predich, M., Nair, G., Smith, I. (1992). Bacillus subtilis early sporulation genes kinA, spoOF, and spoOA are transcribed by the RNA polymerase containing aH, J. Bactenol., 174, 2771-2778. Preist, F.G. (1977). Extracellular enzyme synthesis in the genus Bacillus, Bactenol. Rev., 41, 711-753. Press, W.H., Flannery, B.P., Teukolsky, S.A., Vetterling, W.T. (1990). Numerical Recipes: The Art of Scientific Computing, Cambridge Univ. Press, New York, NY, pp.523-528. Ptashne, M. (1987). A genetic switch: Gene control and phage lambda, Cell Press (Cambridge, MA) and Blackwell Sci. Public. (Palo Alto, CA). 258 Rickenberg, H.V., Cohen, G.N., Buttin, G., Monod, J. (1956). La galactoside- permease D'Escherichia coli Annal. Inst. Pasteur, 91, 829-857. Robertson, B.R., Button, D.K. (1989). Characterizing aquatic bacteria according to population, cell size, and apparent DNA content by flow cytometry, Cytomety, 10, 70-76. Robertson, J.B., Gocht, M., Marahiel, M.A, Zuber, P. (1989). AbrB, a regulator of gene expression in Bacillus, interacts with the transcritpion initiation regions of a sporulation gene and an antibiotic biosynthesis gene, Proc. Natl. Acad. Sci. USA, 86, 8457-8461. Rong, S., Rosenkrantz, M.S., Sonenshein, AL (1986). Transcriptional control of the Bacillus subtilis spolID gene, J. Bacteriol., 165, 771-779. Rotman, B. (1961). Measurement of activity of single molecules of P-D-galactosidase, Proc. Natl. Acad. Sci. USA, 47, 1981-1991. Rudner, D.Z., LeDeaux, J.R., Ireton, K., Grossman, AD. (1991). The spoOK locus of Bacillus subtilis is homologous to the oligopeptide permease locus and is required for sporulation and competence, J. BacterioL, 173, 1388-1398. Russo-Marie, F., Roederer, M., Sager, B., Herzenberg, LA, Kaiser, D. (1993). B- Galactosidase activity in single differentiating bacterial cells, Proc. Natl. Acad. Sci. USA, 90, 8194-8198 Sandman, K., Losick, R., Youngman, P. (1987). Genetic analysis of Bacillus subtilis mutations generated by Tn917-mediated insertional mutagenesis, Genetics, 117,603617. Satola, S.W., Baldus, J.M., Moran, C.P. (1992). Binding of SpoOA stimulates spolIG promoter activity in Bacillus subtilis, J. Bacteriol., 174, 1448-1453. Satola, S.W., Kirchman, P.A, Moran, C.P. (1991). SpoOA binds to a promoter used by aA RNA polymerase during sporulation in Bacillus subtilis, Proc. Natl. Acad. Sci USA, 88, 4533-4537. Saunders, C.W., Schmidt, B.J., Mallonee, R.L, Guyer, M.S. (1987). Secretion of human serum albumin from Bacillus subtilis, J. Bacteriol., 169, 2917-2955. Schaeffer, P. (1969). Sporulation and the production of antibiotic, exoenzymes, and exotoxins, Bacteiol. Rev., 33, 48-71. 259 Schaeffer, P., Millet, J., Aubert, J.-P. (1965). Catabolic repression of bacterial sporulation, Proc. Natl. Acad Sci. USA, 54, 704-711. Schein, C.H., Kashiwagi, K., Fujisawa, A., Weissman, C. (1986). Secretion of mature IFN-a2 and accumulation of uncleaved precursors by Bacillus subtilis transformed with a hybrid a-amylase signal sequence-IFN-a2 gene, Bio/lech., 4, 719-725. Schepper, T., Hitzman, B., Rinas, U., Schugerl, K. (1987). Flow cytometry of Eschenichia coli for process monitoring, J. Biotech., 5, 139-148. Schmidt, R., Margolis, P., Duncan, L, Coppolecchia, R., Moran, C.P., Losick, R. (1990). Control of developmental transcription factor oa by sporulation regulatory proteins SpoIIAA and SpoIIAB in Bacillus subtilis, Proc. Natl. Acad. Sci. USA, 87, 9221-9225. Shapiro, H.M. (1988). Practical Flow Cytometry, Alan R. Liss, New York, NY. Shoji, K, Hiratsuka, S., Kawamura, F., Kobayashi, Y. (1988). New suppressor mutation surOBof spoOBand spoOFmutations in Bacillus subtilis,J. Gen. Microbiol., 134, 3249-3257. Smith, I., Mandic-Mulec, I., Gaur, N. (1991). The role of negative control in sporulation, Res. Microbiol., 142, 831-839. Spizizen, J. (1958). Transformation of biochemically deficient strains of Bacillus subtilis by deoxyribonucleate, Proc. Natl. Acad. Sci. USA, 44, 1072-1078. Srienc, F., Bailey, J.E. (1984). Characterization of intracellular accumulation of ployB-hydroxybutyrate (PHB) in individual cells of Alcaligenes eutrophus H16 by flow cytometry, Biotechnol. Bioeng., 26, 982-987. Srienc, F., Campbell, J.L, Bailey, J.E. (1983). Detection of bacterial -galactosidase activity in individual Saccharomyces cerevisiae cells by flow cytometry, Biotech. Lett, 5, 43-48. Steen, H.B. (1990). Flow cytometric studies of microorganisms, In: Flow Cytomety and Sorting, Melamed, M.R., Lindmo, T., Mendelson, M.L (eds.), Wiley-Liss Inc., New York, NY, 605-622. Steen, H.B., Boye, E. (1980). Bacterial growth studied by flow cytometry, Cytometry, 1, 32-36. 260 Steen, H.B., Boye, E., Skarstad, K., Bloom, B., Godal, T., Mustafa, S. (1982). Applications of flow cytometry on bacteria: cell cycle kinetics, drug effects, and quantitation of antibody binding, Cytometry, Z 249-257. Stragier, P., Bonamy, C., Karmazyn-Campelli, C. (1988). Processing of the sporulation sigma factor in Bacillus subtils: How morphological structure could control gene expression, Cell, 52 697-704. Strang, G. (1986). Introduction to applied mathematics, Wellesley-Cambridge Press, Wellesley, MA. Strauch, M.A., Wu, J.-J., Jonas, R.H., Hoch, J.A (1993). A positive feedback loop controls transcription of the spoOF gene, a component of the sporulation phosphorelay in Bacillus subtilis, Mol. Microbiol., 7, 967-974. Strauch, M.A., Hoch, J.A. (1993). Transition-state regulators: sentinels of Bacillus subtilis post-exponential gene expression, Mol. Microbiol., 7, 337-342. Strauch, M.A., Trach, KA., Hoch, J.A. (1992). SpoOA activates and represses its own synthesis by binding at its dual promoters, Biochimie, 74, 619-626. Strauch, M.A, Webb, V., Spiegelman, G., Hoch, J.A. (1990). The SpoOA protein of Bacillus subtilis is a repressor of the abrB gene, Proc. Natl. Acad. Sci. USA, 87, 18011805. Strauch, M.A, Perego, M., Burbulys, D., Hoch, J.A. (1989). The transiton state transcription regulator AbrB of Bacillus subtilis is autoregulated during vegetative growth, Mol. Microbiol., 3, 1203-1209. Strauch, M.A., Spiegelman, G., Perego, M., Johnson, W.C., Burbulys, D., Hoch, J.A (1989). The transiton state regulator abrB of Bacillus subtilis is a DNA binding protein, EMBO, 8, 1615-1621. Sueoka, N., Yoshikawa, H. (1965). The chromosome of Bacillussubtilis I. Theory of marker frequency analysis, Genetics, S2 747-757. Trach, K., Strauch, M., Wu, J.-J., Dhillon, N., Jonas, R., Hanstein, C., Kallio, P., Perego, M., Bird, T., Spiegelman, G., Fogher, C., Hoch, J.A. (1991). Control of the initiation of sporulation in Bacillus subtilis by a phosphorelay, Res. Microbiol., 142, 815-823. Tyler, B., Magasanik, B. (1969). Molecular basis of transient repression of galactosidase in Escherichia coli, J. Bacteniol., 97, 550-556. 261 f- Tyson, J.J., Othmer, H.G. (1978). The dynamics of feedback control circuits in biochemical pathways, Prog. Theoret. Biol., 5, 1-62. Ulmanen, I., Lundstrom, KI, Lehtovaara, P., Sarvas, M., Ruohonen, M., Palva, I. (1985). Transcription and translation of foreign genes in Bacillus subtilis by the aid of a secretion vector, J. Bacteiol., 162, 176-182. Waggoner, A.S. (1990). Fluorecscent probes for cytometry, In: Flow Cytometry and Sorting, Melamed, M.R., Lindmo, T., Mendelson, M.L (eds.), Wiley-Liss Inc., New York, NY, 209-225. Wallner, G., Amann, R., Beisker, W. (1993). Optimizing fluorescent in situ hybridization with tRNA oligonucleotide probes for flow cytometric identification of microorganisms, Cytometry, 14, 136-143. Weaver, J.C., Seissler, P.E., Threefoot, S.A, Lorenz, J.W., Huie, T., Rodriguez, R. (1984), Ann. NYAcad Sci., 434, 363-372. Weaver, J.C. (1987). Gel microdroplets for microbial measurement and screening: Basic principals, Biotech. Bioeng. Symp., 17, 185-195. Weaver, J.C., Williams, G.B., Klibanov, A, Demain, AL. (1988). Gel microdroplets: Rapid detection and enumeration of individual microorganisms by their metabolic activity, Bio/Technol., 6, 1084-1089. Weinrauch, Y., Penchev, R., Dubnau, E., Smith, I., Dubnau, D. (1990). A Bacillus subtilis regulatory gene product for genetic competence and sporulation resembles sensor protein members of the two-component signal-transductionsystems, Genes & Dev., 4, 860-872. Weir, J., Dubnau, E., Ramakrishna, N., Smith, I. (1984). Bacillus subtilis spoOHgene, J. Bacteiol., 157, 405-412. Weir, J., Predich, M., Dubnau, E., Nair, G., Smith, I. (1991). Regulation of spoOH, a gene coding for the Bacillus subtilis oa factor, J. Bactenol., 173, 521-529. Wittrup, KID., Bailey, J.E. (1988). A single-cell assay of P-galactosidase activity in Saccharomyces cerivisiae, Cytometry, 9, 394-404. Wittrup, KID., Bailey, J.E. (1990). Mathematical modeling of a single-cell enzyme assay, Biotechnol. Bioeng., 35, 525-532. 262 Wittrup, K.D., Mann, M.B., Fenton, D.M., Tsai, LB., Bailey, J.E. (1988). Single-cell light scatter as a probe of refractile body formation in recombinant Escherichia coli, Bio/Technol., 6, 423-426. Wu, J.-J., Piggot, P.J., Tatti, KM., Moran, C.P. (1991). Transcription of the Bacillus subtilis spollA locus, Gene, 101, 113-116. Wu, J.-J., Howard, M.G., Piggot, P.J. (1989). Regulation of transcription of the Bacillus subtils spollIA locus, J. Bacteriol., 171, 692-698. Yagil, G., Yagil, E. (1971). On the relationship between effector concentration and the rate of induced enzyme synthesis, Biophys. Journal, 11, 11-27. Yamane, K, Shiroza, T., Furusato, T., Nakamura, K, Nakazawa, K, Yanagi, K, Yamasaki, M., Tamura, G. (1985). Bacillus subtilis secretion vectors for proteins and oligopeptides constructed from B. subtilis a-amylase genes, in Molecular biology of microbial differentiation, Hoch, J. and Setlow. P. (eds.), ASM, Washington, D.C. Yamashita, S., Yoshikawa, H., Kawamura, F., Takahashi, H., Yamamoto, T., Kobayashi, Y., Saito, H. (1986). The effect of spoO mutations on the expression of spoOA- and spoOF-lacZfusions, Mol. Gen. Genet., 205, 28-33. Yariv, J., Kalb, A.J., Yariv, M. (1972). Labelling of lac-permease, FEBS Letters, 27, 27-29. Yashphe, J., Halvorson, H.O. (1976). Bi-D-Galactosidase activity in single yeast cells during cell cycle of Saccharomyces lactis, Science, 191, 1283-1284. York, K, Kenney, T.J., Satola, S., Moran, C.P., Poth, H., Youngman, P. (1992). SpoOA controls the aA-dependent activation of Bacillus subtilis sporulation-specific transcription unit spollIIE,J. Bacteriol., 174, 2648-2658. Yoshikawa, H., Sueoka, N. (1963). Sequential replication of Bacillus subtilis chromosome. I. Comparison of marker frequencies in exponential and stationary growth phases, Proc. Natl. Acad. Sci. USA, 49, 559-566. Zuber, P., Losick, R. (1987). Role of AbrB in SpoOA- and SpoOB-dependent utilization of a sporulation promoter in Bacillus subtlis, J. Bacteriol., 169,2223-2230. Zuber, P., Losick, R. (1983). Use of a lacZ fusion to study the role of the spoOgenes of Bacillus subtilis in developmental regulation, Cell, 35, 275-283. 263