Development of Glucose Valves for Metabolic
Engineering Applications in E. coli
by
Kevin Val-Murvin Solomon
MS, Chemical Engineering Practice
Massachusetts Institute of Technology, Cambridge, MA, USA, 2008
ARCNES
B Eng BioSci, Chemical Engineering & Bioengineering
McMaster University, Hamilton, ON, Canada, 2006
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
September 2012
@ 2012 Massachusetts Institute of Technology. All rights reserved
Signature of Author
Department of Chemical Engineering
August 6, 2012
,1
1
Certified by
U,
Kristala Jones Prather
Theodore T. Miller Career Development Associate Professor of Chemical Engineering
Thesis Supervisor
Accepted by
Patrick S. Doyle
Professor of Chemical Engineering
Chairman of the Committee for Graduate Students
2
Development of Glucose Valves for Metabolic
Engineering Applications in E. coli
by
Kevin Val-Murvin Solomon
Submitted to the Department of Chemical Engineering on August 6, 2012 in Partial Fulfillment of
the Requirements for the Degree of Doctor of Philosophy in Chemical Engineering
ABSTRACT
Microbial production platforms have extraordinary potential to help meet the energy, material
and health needs of tomorrow. Through the reactions of cellular metabolism, microbes are an
attractive option for the sustainable production of commodity and enantiopure specialty
chemicals. One outstanding challenge, however, is the engineering of these systems for
economic viability. As a solution to this issue, a metabolite valve is proposed: a biochemical
device which may be used to dynamically redirect metabolite flux away from endogenous
processes into production pathways.
In this work, we develop and characterize a glucose valve. First, a novel E. coli strain was
engineered to allow the redirection of glucose flux from central metabolism via glucokinase.
Using a promoter library, glucokinase expression was varied with an attendant change in specific
growth rate and carbon flux. A model pathway was then constructed to utilize the redirected
carbon demonstrating that the efficiency of such pathways may be controlled through
glucokinase expression. Next, inducible antisense RNA and inverting genetic circuits were
developed to dynamically control glucokinase expression. With dynamic control, carbon was
redirected from endogenous processes only once sufficient cellular resources had accumulated,
further improving performance. In this manner, yields and titers of the model pathway were
increased with a concomitant decrease in acetate waste. Finally, elements of this system were
modeled to gain mechanistic insight and to establish a control envelope of viable expression and
production regimes.
This thesis represents one of the first reports of glucose redirection in E. coli and is an example
of the ongoing development of a relatively new paradigm in metabolic engineering: dynamic
flux control. With a glucose valve, the needs of cellular health and demands of heterologous
production may be balanced, enabling the development of efficient processes with glucose as a
sole carbon source for both cell growth and biochemical production. This ability to use a single
carbon source simplifies process design, lowering capital and operating costs. Furthermore, a
glucose valve has potential applications in the optimization of existing processes where carbon
is underutilized and wasted as fermentation products such as acetate.
Thesis Supervisor: Kristala L. J. Prather
Title: Theodore T. Miller Career Development Associate Professor
3
ACKNOWLEDGEMENTS
As I complete the marathon that is a doctoral degree, I can't help but reflect on the time I've
spent here. The journey was extraordinary, and while painful at times, extremely educational,
formative and gratifying. However, I don't think I would have been able to make this journey
without the support and influence of those around me.
First and foremost, I would like to thank my advisor, Professor Kristala Prather. You have been
an excellent mentor and advisor and I am forever in your debt. I am extremely grateful for your
patience, support and optimism which have allowed me to persevere over the years. I am also
thankful for the opportunities that you have afforded me to grow beyond the bench including
the mentoring of students and participating in leadership positions within SynBERC. I would also
like to thank my thesis committee members: Professors Narendra Maheshri, Drew Endy and
George Stephanopoulos. Your thoughtful and critical feedback and comments throughout my
thesis has aided greatly in its development and my growth as a professional scientist. I would
also like to thank the Practice School Program. I thoroughly enjoyed the work experiences and
the opportunity it presented to explore potential career paths. I would also like to acknowledge
the NSF, SynBERC and NSERC for the financial support which has made this work possible.
I am grateful to all the undergraduate students I have mentored. Not only have they allowed
me to practice and hone my skills as an educator, they have made productive contributions to
the development of this thesis. Specifically I'd like to acknowledge their contributions over the
years: Aziza Glass who helped with initial development of the qRT-PCR assay; Matt Luchette
who performed the reporter gene assay of Chapter 3; U'Kevia Bell who helped identify a
problem with the initial Gnt- mutants that were tested; Zach Waxman who helped to rule out
potential strategies to construct the MG1655 mutants of Chapter 3; Brian Ma who developed
the ATP stoichiometric model presented in Chapter 4 and is a coauthor of one of my
manuscripts; and finally, Tarielle Sanders who validated the qRT-PCR assay to distinguish mRNA
and asRNA transcripts, generated the qPCR data presented within this dissertation and is a
coauthor of two of my manuscripts.
My friends and colleagues within the Prather Lab, MIT and beyond have been also instrumental
with their support. This list is far from exhaustive but I'd like to acknowledge a few. To
labmates Diana Ritz (nee Bower), Micah Sheppard, Eric Shiue, David Nielsen and Himanshu
Dhamankar, I'm extremely appreciative of your support and positive attitude. You have made it
a joy to work in Prather Lab with your sound technical advice, friendship and our communal
commiseration. To Fen, Ashley, David and Renee, thank you for reminding me that fitness can
be fun. Not only have you started me off on a path to a newer and better me, you have been
pivotal in managing my stress these last few months. To my friends back home, Desiree
Walters, Mirza Ahmic and Shane Phillippe, thank you for grounding me and reminding me of life
outside MIT. To my classmates, thank you for being there as we navigated the trials and
tribulations of MIT grad school together.
Last but not least, I'd like to acknowledge my parents: Val and Winsome Solomon. You have
been a constant source of strength and support. For almost a quarter century you have fostered
and endured my career as a student. I would not be here today without the many sacrifices that
you have made and for that, I am extremely grateful.
KEVIN SOLOMON
4
TABLE OF CONTENTS
Abstract
Acknowledgements
List of Figures
List of Tables
List of Abbreviations
1.
INTRODUCTION
1.1.
Motivation
1.2.
Potential strategies to pathway optimization
1.2.1. Downregulation of related pathways
1.2.2. Dynamic Expression Profiles
1.2.3.
Emerging paradigms: Synthetic Biology
1.3.
Scope of thesis
1.4.
Thesis Organization
2.
13
14
15
15
18
19
21
22
TUNING PRIMARY METABOLISM FOR HETEROLOGOUS PATHWAY
PRODUCTIVITY
24
2.1.
Introduction
2.2.
Results and Discussion
2.2.1. Generation of a tunable glucose flux platform
2.2.2. Range of in vivo expression of glk
2.2.3. Glk expression controls gluconate productivity
2.3.
Summary & Concluding Remarks
2.4.
Methods
2.4.1. Strains and plasmids
2.4.2. Culture conditions
2.4.3. Quantification of mRNA levels
2.4.4. Enzyme activity assays
2.4.5. Metabolite analysis
3.
3
4
7
9
10
25
27
27
28
32
35
36
36
40
40
41
42
A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL
CARBON METABOLISM
43
3.1.
Introduction
3.2.
Results
3.2.1. Inducible antisense RNA as a dynamic glucose valve actuator
3.2.2. Effect of asRNA expression on product yields
3.2.3. Inverting gene circuit as a dynamic glucose valve activator
3.2.4. Effect of inverter expression on product yields
3.3.
Discussion
3.4.
Materials and Methods
3.4.1. Strains and plasmids
3.4.2. Inverter construction
3.4.3. Culture conditions
3.4.4. Enzyme activity levels
3.4.5. Reporter gene assay
3.4.6. Quantification of mRNA levels
3.4.7. Metabolite analysis
44
46
46
52
55
60
63
65
65
68
69
70
70
70
71
5
4.
GLUCOSE VALVE PERFORMANCE POTENTIAL AND IMPACT ON
PHYSIOLOGY
4.1.
Introduction
4.2.
Effects of Glk on physiology: ATP Regeneration and Cell Health
4.3.
Effects of Physiology on Valve Performance
4.3.1. Sensitivity of gluconate productivity to system parameters
4.3.2. Significance of growth mediated buffering as a regulatory mechanism
4.4.
Perspectives
4.5.
Materials and Methods
4.5.1. Strains and plasmids
4.5.2. Culture conditions
4.5.3. Enzyme activity levels
4.5.4. Metabolite analysis
73
74
83
83
89
95
97
97
97
98
98
5.
CONCLUSIONS AND FUTURE DIRECTIONS
99
5.1.
5.2.
5.3.
Thesis Summary
Future Directions
Concluding Remarks
100
101
103
6.
REFERENCES
APPENDICES
Al I Generating a recA expression system for chromosomal manipulation
A1.1
Introduction
A1.2
Materials & methods
A1.3
Results & Discussion
A1.4
Works Cited
A21 Extending the Glk expression family
A2.1
Introduction
A2.2
Materials and Methods
A2.3
Results and Discussion
A31 MATLAB Code
A3.1 Gdh-Glk competition model
A3.2
Effect of Growth Mediated Buffering
6
72
104
121
122
122
122
123
125
126
126
126
126
130
130
133
LIST OF FIGURES
Figure 1-1| The zero sum challenges with traditional pathway optimization strategies. ...................
15
Figure 1-21 A hypothetical example of a complex regulatory circuit utilizing multiple modules or parts.
..............................................................................................................................................
20
Figure 1-31 Thesis organization as the engineering design cycle ........................................................
23
Figure 2-11 Glycolytic uptake and utilization in wildtype E. coli and the modified host ......................
26
Figure 2-21 Growth rate recovery of AptsHlcrr gal#P mutants...........................................................
28
Figure 2-31 Glucokinase expression family construction...................................................................
29
Figure 2-41 Expression family activity data........................................................................................
30
Figure 2-51 Gluconate molar yield on glucose at 72 h as a function of Glk activity and IPTG induction .. 34
Figure 3-1| Schematic of the inducible antisense constructs. ...........................................................
47
Figure 3-21 Reporter gene assay of pBAD30 and pKVS45 in KTSO22..................................................48
Figure 3-31 Antisense glucose valve ................................................................................................
50
Figure 3-41 Effects of medium composition on antisense performance ...........................................
52
Figure 3-51 Potential impact of valve on productivity at 72 h ............................................................
54
Figure 3-61 Effect of asRNA on gluconate yield .................................................................................
55
Figure 3-71 Genetic inverter as a glucose valve .................................................................................
56
Figure 3-81 Initial characterization of a genetic inverter ...................................................................
57
Figure 3-91 Genetic inverter performance........................................................................................
58
Figure 3-101 Dynamic performance of inverter as a glucose valve ....................................................
60
Figure 3-11| Effect of inverter on gluconate yield .............................................................................
62
Figure 3-12| Gdh activity as a function of Inverter state...................................................................63
Figure 4-1 Overview of aerobic respiration......................................................................................76
Figure 4-21 Protein, RNA and DNA content as a function of specific growth rate ...............................
79
Figure 4-31 ATP consumption rates as a function of specific growth rate and Glk in vitro total activity in
minimal medium supplemented with glucose. ...................................................................
80
7
Figure 4-41 Com peting reactions for glucose.............................................................................84
Figure 4-5I Gluconate yield as a function of GIk and Gdh activity, and glucose transport rate...........86
Figure 4-61 Impact of Glk activity on Gdh expression .......................................................................
87
Figure 4-71 Gluconate yield as a function of Glk and Gdh expression A) Experimentally measured
B)M odel predictions....................................................................
..........
--... --------.................
88
Figure 4-81 Predicted Glk activity as a function of time when mRNA levels are halved at t = 0.......91
Figure 4-9| Buffering capacity of cells perturbed from the initial steady state (Relative gik mRNA = 1) as
a function of the final m RNA level ......................................................................................
Figure 4-101 Prototypical characterization datasheets .....................................................................
94
96
Figure A - 1| Photograph of exposed and unexposed plates after 20 h............................................124
Figure A - 21 Performance of selected BIOFAB promoters vs the Anderson library...........................129
8
LIST OF TABLES
Table 2-1| Glucokinase expression family promoter sequences ........................................................
29
Table 2-21 Titers of the major fermentation products .......................................................................
33
Table 2-31 Main strains and plasmids used ......................................................................................
37
Table 2-41 Oligonucleotides used ........................................................................................................
38
Table 3-1 Titers of the major fermentation products as a function of Glk expression. .....................
53
Table 3-21 Main strains and plasmids used ......................................................................................
66
Table 3-31 Oligonucleotides used ........................................................................................................
67
Table 4-1| Approximate energetic costs of transcription, translation and replication ........................
80
Table 4-21 Glk-Gdh competition model parameter definitions and values used .................................
85
Table 4-31 Growth mediated buffering model parameter definitions and values used......................90
Table A - I| Oligonucleotides used....................................................................................................
123
Table A - 21 Oligonucleotides used....................................................................................................
127
Table A - 31 BIOFAB promoters tested ..............................................................................................
128
9
LIST OF ABBREVIATIONS
5KG
5-ketogluconate, 5-ketogluconic acid
AadA
aminoglycoside resistance protein
AckA
acetate kinase A
AmpR
ampicillin resistance
asRNA
antisense ribonucleic acid
aTc
anhydrotetracycline
ATP
adenosine 5'triphosphate
BioCAD
computer aided design tools for biological systems
Bla
betalactamase
bps
base pairs
Cat
chloramphenicol acetyltrasnferase
CDS
coding sequence
CmR
chloramphenicol resistance
CTP
cytidine 5'triphosphate
DAD
diode array detector
endA
endonuclease I
G6PDH
glucose-6-phosphate dehydrogenase
GalP
galactose permease
GaIR
galactose repressor, a ga/P regulatory protein
GalS
galactose isorepressor, a ga/P regulatory protein
Gdh
glucose dehydrogenase
GFP
green fluorescent protein
10
Glk
glucokinase
Glu
glucose
Gnt
gluconate, gluconic acid
GntK
gluconate kinase
GTP
guanosine 5'triphosphate
HPLC
high performance/pressure liquid chromatography
Idi
isopentyl diphosphate isomerase
IdnK
idonate kinase
IPTG
isopropyl -D-1-thiogalactopyranoside
KanR
kanamycin resistance
kcat,
k2
turnover rate
Km
Michaelis-Menten constant
Lad
lac repressor
LB
Luria-Bertani rich medium
LVA
protein degradation tag with terminal Leucine Valine Alanine sequence
M9
M9 minimal medium
MgIABC
-methyl galactoside transport system
mRNA
messenger RNA
NAD*
nicotinamide adenine dinucleotide
NADH
reduced nicotinamide adenine dinucleotide
NADP4
nicotinamide adenine dinucleotide phosphate
NADPH
reduced nicotinamide adenine phosphate dinucleotide
neo
kanamycin resistance gene
OD, OD600
optical density
11
ORF
open reading frame
PCR
polymerase chain reaction
PEP
phosphoenolpyruvate
Pps
phosphoenolpyruvate synthase
Pta
phosphotransacetylase
PTS
phosphoenolpyruvate sugar phosphotransferase system
qPCR
quantitative polymerase chain reaction
qRT-PCR
quantitative reverse transcription polymerase chain reaction
RBS
ribosome binding site
RecA
homologous recombination protein
RID
refractive index detector
RNA
ribonucleic acid
RT
reverse transcription
SD
standard deviation
SEM
standard error of the mean
sGFP
superfolder green fluorescent protein
SpecR
spectinomycin resistance
TetR
tet repressor
Tris-HCI
tris(hydroxymethyl)aminomethane pH adjusted with hydrochloric acid
TTP
thymidine 5' triphosphate
UTP
uridine 5'triphosphate
UTR
untranslated region
12
CHAPTER 1
Introduction
ABSTRACT
With increasing price volatility and a growing appreciation for more sustainable processes,
microbial chemical production has been tapped as a promising renewable alternative for the
generation of diverse, stereospecific compounds.
Nonetheless, many attempts to generate
them are not yet economically viable. Due to the zero sum nature of microbial resources,
traditional strategies of pathway optimization are attaining minimal returns. This result is in
part a consequence of the gross changes in host physiology resulting from such efforts and
underscores the need for more precise and subtle forms of gene modulation. In this chapter,
alternative strategies and emerging paradigms to address this problem are described. Potential
solutions from the emerging field of synthetic biology are also highlighted.
Using this
framework, a solution to the problem of glucose flux manipulation is proposed.
This chapter contains material adapted from:
Solomon, KV, Prather, KU. (2011) The zero-sum game of pathway optimization: Emerging
paradigms for tuning gene expression, Biotechnol. J. 6, 1064-1070.
13
1.1 Motivation
1.1.Motivation
Microbial production systems display a remarkable flexibility in the diversity and
enantioselectivity of the compounds that they can generate. These compounds have historically
been natural products such as ethanol, amino acids, acetone and antibiotics. However, with the
introduction of ever more sophisticated tools, a range of natural and unnatural products have
been made in engineered hosts including compounds such as hydroxyacids (1-3), isoprenoids (4,
5), polyketides (6, 7), and biopolymers (8, 9).
While several of these processes have been
successfully commercialized (10-12), many remain economically infeasible and are the subject of
intense optimization efforts.
In optimizing microbial pathways, the objectives are to maximize product flux, yield and
selectivity. Traditionally, this problem has been approached by an analysis of the metabolic
pathway; branch points that lower product yield and selectivity (gene inactivation) are removed
and the flux of intermediates through the pathway (gene overexpression) is increased.
The
power of such methods has improved tremendously with the advent of computational tools
such as Flux Balance Analysis (FBA) (13, 14) and bilevel optimization (15-17) to identify flux
bottlenecks. They are still, however, fundamentally constrained by the interconnectedness and
finite nature of microbial resources (Figure 1-1).
supplementation,
Gene inactivations may necessitate media
impair cellular function and are sometimes infeasible for non-linear
production pathways. Overexpression of pathway genes, on the other hand,
comes at the
expense of endogenous ones due to consumption of common precursors and titration of cellular
machinery such as polymerases and ribosomes and may lead to growth inhibition, reduced
expression and even cell death (18-20). In certain hosts the heat shock response is stimulated
by protein overexpression (21, 22) further limiting the degree of overexpression possible.
Moreover, successfully overexpressing or knocking out genes does not guarantee improved
14
CHAPTER 1| INTRODUCTION
productivity. Decoupling the native regulation of flux within the pathway in these ways may
lead to the accumulation of intermediates that can inhibit pathway enzymes (3, 23) or are
bacteriostatic (1, 24, 25). These challenges are not insurmountable, but they do underscore the
need for more tools in pathway optimization.
A
9
C?9
4
4
Figure 1-1| The zero sum challenges with traditional pathway optimization strategies.
In the original pathway (A), only one media supplement (yellow circle) is needed to generate product
(blue circle) and essential metabolites (aqua circle) in the cell (pink). However, gene inactivation (B)
necessitates additional supplementation to generate the essential metabolite while overexpression (C)
increases the pool of desired intermediate at the expense of expression and flux through the other
enzymatic steps potentially limiting growth. The sum of these effects on the host's health controls the
degree of success on overall pathway production. Metabolite flux is proportional to the line thickness
while metabolite pool size is represented by the circle area. Dashed lines indicate an absence of
metabolite flux/pools when compared to wildtype.
1.2.Potential strategies to pathway optimization
1.2.1.
Downregulation of related pathways
Modulation of gene expression, such as downregulation of undesired branch points, has
been identified as a fruitful avenue for increased pathway productivity (16, 17). In contrast to
gene inactivation, downregulation offers the ability to redirect metabolite flux into production
pathways while maintaining sufficient flux for endogenous processes.
Moreover, in cases of
drastic differences in catalytic efficiency of competing enzymes, it may prove more efficient than
15
1.2 Potential strategies to pathway optimization
overexpression of pathway enzymes. Downregulation may be implemented in many different
ways.
One promising method, amenable to implementation in a wide variety of hosts and
pathways, is the use of antisense RNA (asRNA) mediated inhibition of translation (26-30).
One such example of asRNA use in pathway optimization is found in the engineering of
Clostridium acetobutylicum. Predating the rise of petrochemical sources, C. acetobutylicum was
an industrially relevant source of solvents such as acetone and butantol (31) which it naturally
ferments as part of its lifecycle (32, 33). Recent volatility in the price of chemical feedstocks and
increasing concern regarding the sustainability of traditional chemical synthetic routes have led
to renewed interest in the species with a focus on controlling the distribution of products (27,
29, 34, 35).
The Papoutsakis group used an asRNA approach to downregulate the CoA
transferase which catalyzes the formation of acetone (ctfAl) to shift these strains to a primarily
alcohologenic mode of production (ethanol and butanol), obtaining the highest ethanol titers
reported at the time in C. acetobutylicum (29, 34). Similar success has been reported for the
engineering of glutamate synthesis from Corynebacterium glutamicum.
C. glutamicum is a
natural overproducer of amino acids and an industrial source of several of these including
glutamate (36) which is produced from the transamination of a-ketoglutarate, a citric acid cycle
intermediate. Utilizing an asRNA approach, Kim and coworkers (28) increased the cell specific
productivity of glutamate by inhibiting activity of 2-oxoglutarate dehydrogenase thereby
allowing sufficient flux of at-ketoglutarate through the citric acid cycle for energy production
while diverting additional precursors to increase glutamate synthesis. Finally, asRNA technology
has been utilized in the synthesis of cobalamin (Vitamin B12) in Bacillus megaterium to improve
titers and yields by 20% (30).
The use of downregulation extends beyond the realm of small molecule synthesis and
has similar applications in recombinant protein production where acetate has been established
16
CHAPTER 1 I INTRODUCTION
to have an inhibitory effect on specific protein expression and bacterial growth (37-40).
Controlling acetate production by inactivation of phosphotransacetylase (pta) or acetate kinase
(ackA) genes in E. coli, which shunt excess acetyl-CoA to acetate, has a deleterious effect on the
cellular redox state (40), carbon flux (41), and ultimately growth (41). Diverse solutions such as
process-based schemes (37) and metabolic engineering of the host to shunt the excess acetylCoA to acetoin (39) have been developed to address the issue. Nonetheless, these solutions are
not scalable to all methods of culture and inhibit ATP synthesis by acetate secretion. Thus, Kim
and Cha (42) chose an antisense based scheme to minimize detrimental physiological effects.
Through minor antisense inhibition of ackA and pta, Kim and Cha were able to reduce acetate
formation by more than 20% while simultaneously observing a 60% improvement in the
production of green fluorescent protein with negligible impact on cellular growth.
In addition to these examples of asRNA inhibition, tools utilizing the effect of codon bias
on translation efficiency in C. glutamicum (43, 44), repressible promoters in S. cerevisiae (45-47)
and titrating inducible promoters in E. coli (48) have been used with great success to
downregulate competing pathways and increase product yields and/or titers. Moreover, the
last decade has seen intense efforts to regulate genes at the transcriptional and post
translational levels culminating in several novel methods such as regulated suppression of
amber mutations (49), inducible protein degradation (50), engineered allostery (51) and
ear><RecNum>111(52, 53). Despite the fact that many of these emerging technologies have yet
to mature and attain widespread adoption, particularly in an industrial context, the growing
interest in asRNA points to its relative ease of implementation.
While unexplored in these
studies, another advantage of downregulation is the possibility of dynamic control of gene
expression.
17
1.2 Potential strategies to pathway optimization
1.2.2. Dynamic Expression Profiles
When maximizing product titers and yields for industrial scale fermentation, carbon flux
is shifted from the normal balance of metabolic intermediates and shunted into the desired
product. This shift is frequently at odds with the goals of the cell, i.e. maintaining metabolite
flux levels and maximizing biomass. Thus, genetic alterations that alter metabolite flux will incur
a redistribution of metabolites to compensate for the change with some inhibition of growth.
Gadkar et al. (54) studied this issue in silico as it applied to glycerol and ethanol production. In
their work, they pursued a bilevel optimization strategy analogous to that of OptKnock (15)
where product titers are maximized subject to growth maximization and other physical
constraints to determine gene candidates for upregulation or deletion.
However, unlike
OptKnock, they also optimized the timing of these genetic changes. For glycerol production,
simulations of a biphasic approach to gene expression resulted in a 30% improvement in titers
over a static strategy. Similarly, ethanol titers were improved by 40% over a static strategy and
90% over wildtype behavior. These cases and more were further studied by Anesiadis and
coworkers (55) with the simulated behavior of genetic elements from synthetic biology, as
opposed to instantaneous switching in expression, and came to a similar conclusion: dynamic
control of gene expression may be implemented to increase pathway productivity.
One of the first experimental demonstrations of this paradigm was elegantly performed
in 2000. In trying to produce lycopene in E. coli, Farmer and Liao (56) sought to overexpress 2
key rate limiting enzymes: phosphoenolpyruvate synthase (Pps), which controls the pool of a
glycolytic intermediate
needed for lycopene biosynthesis, and isopentenyl diphosphate
isomerase (Idi), which pulls glycolytic intermediates into the lycopene biosynthetic pathway.
However, overexpressing them statically from a tac promoter hindered growth, yields and titers.
Thus, they engineered a gene circuit/metabolite control system in which expression of pps and
18
CHAPTER 1| INTRODUCTION
idi was directly tied to the availability of acetyl phosphate, a proxy for glycolytic flux and cellular
health. Using this approach, they were able to overexpress these enzymes to higher levels than
that seen using a static approach while maintaining cellular viability and ultimately improve
titers by 50%, productivity three-fold and carbon yields by more than an order of magnitude.
Similar control systems have also been developed to drive protein expression through the use of
quorum sensing in E. coli (57, 58). Such systems allow for coordinated delayed induction across
multiple cellular populations in addition to transmitting the metabolic load state of the host (59)
thereby mitigating potential challenges associated with protein overexpression. Moreover, they
are modular and readily amenable to integration in complex circuits (57) where Boolean logic
and sensor functions can be implemented for tight pathway regulation in combination with
other strategies for cumulative effects. A hypothetical example of this is presented in Figure 1-2
where sensing and logic (AND) operations are used to drive expression of pathway genes and
product only when high cell densities and carbon flux are achieved.
1.2.3. Emerging paradigms: Synthetic Biology
With an eye towards the creation of sophisticated gene circuits and networks for both
pathway regulation and biosynthesis, the emerging discipline of synthetic biology has
established a paradigm of developing reusable modules or "parts" and "devices" to control gene
expression (60-62). Towards this end, libraries of sensors, control elements, promoters (63-65),
and ribosome binding sites (RBS) (66) among others have been developed.
Many of these
libraries are curated within the Registry of Standard Biological Parts (http://partsregistry.org)
and are freely available to the community.
Through these libraries of parts, network
components may be individually selected, tuned and regulated to achieve the necessary
phenotype.
19
1.2 Potential strategies to pathway optimization
AHL
LuxR
GinAp2
gC
Pathway
4..-
Central
Metabolism
A
genes
AND
gate
Product
Figure 1-21 A hypothetical example of a complex regulatory circuit utilizing multiple modules or parts.
The LuxR/Luxl quorum sensing system (luxl not shown), mediated by N-acyl homoserine lactone (AHL), is
used to drive the expression of gInAp2, an acetyl phosphate (ACP) sensor (67). Sufficient carbon flux
through central metabolism will lead to accumulation of ACP. The presence of GInAp2 and ACP serve as
inputs to an AND gate (binding of ACP to GlnAp2) whose output is expression of pathway genes and,
ultimately, synthesis of product (triangles). Product is only produced when biomass and carbon flux is
high, i.e. from a healthy culture.
The rise of part libraries has also been accompanied by the development of computer
aided design (BioCAD) tools to facilitate the design of ever more complex circuits (68-72).
However, they are dependent on the availability of datasheets (73, 74) or other experimental
characterization to describe them which are typically context dependent and not readily
generalizable to all scenarios. Moreover, the current lack of generic insulators for these parts
results in feedback from downstream parts, or retroactivity (75), which can further perturb
performance from expectation. Nonetheless, there has been some success with the engineering
of systems from these libraries using both theoretical and experimental approaches.
For
example, using an equilibrium statistical thermodynamic model, Salis and coworkers (76) were
able to evaluate the effects of the 5' UTR on translation culminating in the design of novel RBSs
able to achieve expression levels spanning 5 orders of magnitude.
20
Their software tool,
CHAPTER 11 INTRODUCTION
RBSCalculator (https://salis.psu.edu/software/), also allows for relative expression tuning of a
given sequence. Empirical and combinatorial approaches to the tuning of gene expression from
library components have also proven successful in optimizing yields of lycopene and mevalonate
production pathways (63, 77).
Finally, a combination of both theoretical and experimental
characterization has been used to design and develop tuned systems with little post hoc
adjustment (78).
1.3.Scope of thesis
As described previously, one of the ongoing challenges of metabolic engineering is that
of flux optimization.
In particular, when traditional paradigms such as pathway gene
overexpression or competing gene knockouts are infeasible, there are a dearth of tools
available. To that end, this thesis proposes a new device for this problem: a metabolite valve.
With a metabolite valve, engineers can dynamically tune the flux through the competing
pathway in such a way as to optimize pathway productivity and satisfy endogenous needs. Such
a device would be invaluable in contexts where glucose itself is taken from glycolysis and
redirected towards heterologous production.
A glucose valve would enable the design of
glucose demanding heterologous pathways in systems where glucose is utilized as the sole
carbon source, potentially reducing operating costs and simplifying process design. Moreover, a
glucose valve would also allow for the optimization of existing pathways by reducing carbon
waste to fermentation byproducts such as acetate.
Valves may potentially take many forms. Examples of suitable control elements range
from self-splicing inteins (79) and riboregulators (53) to repressible promoters (45) and
antisense RNA (27).
In this thesis, emphasis will be placed on elements developed with
synthetic biological applications in mind.
The advantage of such an approach includes well
documented and curated parts with a rich and vibrant community of researchers for open
21
1.4 Thesis Organization
discourse. With these tools, this thesis aims to develop and explore the potential of a glucose
valve. Specifically, the work presented in this dissertation will:
e
Develop a host with the potential for glucose redirection
*
Identify and validate a suitable control node for glucose redirection
e
Develop and evaluate several methods of dynamic control of glucose flux
*
Analyze and characterize the performance of such systems
1.4.Thesis Organization
This dissertation is organized around a single iteration of the engineering design process
(Figure 1-3).
In Chapter 1, a survey of the literature has been presented that has defined the
problem of flux manipulation and detailed several promising strategies to its solution.
Chapter 2, this problem statement is narrowed to that of glycolytic flux redirection.
In
Here, a
ptsH/crr galp4 E. coli mutant is developed with glucokinase forming a viable control node for
glucose flux. System constraints such as the bounds of feasible glucokinase expression, its
effects on cellular physiology and potential for improved pathway productivity are explored.
Dynamic valve designs, such as antisense RNA and an inverting gene circuit, are proposed in
Chapter 3. Their ability to control glycolysis, dynamic behavior, and impact on the productivity
of a model heterologous pathway are also evaluated. In Chapter 4, the behavior of this system
is analyzed. Using theoretical models and experimental observations, insight is gained into the
inner workings of such a system and the potential of a glucose valve in the current context is
bounded. The findings of Chapters 2-4 are summarized in Chapter 5 with potential refinements
and future directions indicated.
22
CHAPTER 1| INTRODUCTION
Define
r
the
ChuPter I
I
*
Figure 1-31 Thesis organization as the engineering design cycle
23
CHAPTER 2
Tuning primary metabolism for heterologous
pathway productivity
ABSTRACT
Tuning expression of competing endogenous pathways has been identified as an effective
strategy in the optimization of heterologous production pathways. However, intervention at the
first step of glycolysis, where no alternate routes of carbon utilization exist, remains unexplored.
In this chapter we have engineered a viable E. coli host that decouples glucose transport and
phosphorylation enabling independent control of glucose flux to a heterologous pathway of
interest through glucokinase (g/k) expression.
Using community sourced and curated
promoters, g/k expression was varied over a 3-fold range while maintaining cellular viability.
The effects of gik expression on the productivity of a model glucose-consuming pathway were
also studied. Through control of glycolytic flux we were able to explore a number of cellular
phenotypes and vary the yield of our model pathway by up to 2-fold in a controllable manner.
This chapter contains material adapted from:
Solomon, KV, Moon, TS, Ma, B, Sanders, TM, Prather, KJ. (2012) Tuning primary metabolism
for heterologous pathway productivity, ACS Synth. Biol., doi: 10.1021/sb300055e.
24
CHAPTER 2 | TUNING PRIMARY METABOLISM FOR HETEROLOGOUS PATHWAY PRODUCTIVITY
2.1.Introduction
Metabolic engineering and microbial cell factories are powerful tools for the generation
of commodity (31, 80) and specialty (5, 46, 56) chemicals in a renewable manner. While new
advances have diversified the product portfolio from natural products such as amino acids and
acetone to unnatural products such as biopolymers (8, 9) and biofuels (81-84), there still
remains the challenge of making these pathways competitive with traditional chemical
synthesis. One solution to this problem is to maximize product flux, yield and selectivity by
removing competing enzymes at key pathway branchpoints and/or overexpressing pathway
enzymes. However, in cases where the competing enzyme is essential or its catalytic efficiency
(kcat/Km) is significantly greater than that of the production pathway, these methods may
necessitate more costly medium supplementation or ultimately prove ineffective. To address
these pitfalls, we examined a third strategy: tuning expression of the competing pathway.
Tuning endogenous gene expression has been previously identified as an effective
strategy to increase pathway productivity (16, 17) as it allows the balancing of endogenous
cellular needs with that of pathway efficiency. Through dynamic and static implementations,
such as codon substitution (43, 44), inducible/repressible promoters (45, 47), and antisensemediated gene silencing (28, 30), researchers have explored this strategy with varying degrees
of success. More recently, groups have begun to examine control of the nodes of central carbon
metabolism (85-88).
However, there still exists a gap with regards to early intervention in
central metabolism where alternate routes of carbon utilization do not exist. More importantly,
the ability to introduce heterologous pathways that compete directly with central carbon
metabolism at these nodes remains unexplored. We thus set out to design a system where
glucose flux could be redirected from glycolysis and into a heterologous production pathway.
25
2.1 Introduction
Wildtype E coli
Modified Host
----
-----------
Glucoe4"
'TAP
Glucose
-
Glucose
AOP
6
Gik
Glucose
H
P
___
--
P
-
~ Glucose+ H
L ---
--------
16
G6P0
Periasml
Cytoplasm
Periplasmi
Cytoplasm
Figure 2-1| Glycolytic uptake and utilization in wildtype E.coil and the modified host
In wildtype E.coili, glucose is primarily transported and phosphorylated through the PTS system (ptsHlcrr)
for consumption in endogenous processes. In the modified host, constructed here, glucose transport is
decoupled from phosphorylation allowing the diversion of glucose into heterologous production
pathways. Mg|ABC - 0-methyl galactoside transport system encoded by mg/BAC; GaIP - Galctose
permease encoded by ga/P; PTS - glucose and mannose specific components of the phosphoenolpyruvate
phosphotransferase system encoded by ptsHlcrr; Gik - glucokinase encoded by g/k; PEP phosphoenolpyruvate.
The first step of glucose metabolism is transport into the cell and phosphorylation for
entry into glycolysis (Figure 2-1). In wildtype E. coili, this is accomplished predominantly by the
phosphoenolpyruvate (PEP):carbohydrate phosphotransferase system (PTS) (89-91) where
glucose is translocated across the cellular membrane and simultaneously phosphorylated to
glucose-6-phosphate (G6P) with the consumption of one PEP. G6P is subsequently oxidized in
glycolysis to provide ATP and other valuable metabolite precursors. This phosphorylation step,
however, reduces the amount of free glucose available as a substrate for heterologous
production pathways. Furthermore, the coupling of glucose transport and phosphorylation in
PTS makes it unsuitable as a modulation target for the redirection of free glucose. There are,
however, two other known pathways of glucose uptake.
Alternative glucose transporters
include the low affinity galactose:H+ symporter GalP and the ATP-dependent MgIABC system,
which are able to internalize glucose in an unphosphorylated state (92) while the ATPdependent glucokinase (Glk) is able to phosphorylate glucose for glycolysis (93). Several studies
26
CHAPTER 2 TUNING PRIMARY METABOLISM FOR HETEROLOGOUS PATHWAY PRODUCTIVITY
have further demonstrated that wildtype-like growth rates may be recovered in minimal
medium supplemented with glucose in the absence of PTS when GaIP is overexpressed (94-96).
In such a host, glucose transport is decoupled from phosphorylation, allowing for intracellular
redirection of glycolytic flux (Figure 2-1).
In this study, we hypothesized that the downregulation of glucokinase would allow for
the redirection of free glucose into a competing heterologous pathway (Figure 2-1). Using the
well characterized Anderson promoter library (http://partsregistry.org, Parts J23100-J23119),
we created a family of mutants with varying expression of glucokinase and examined the limits
of cell viability. We then evaluated the effect of expression level on the productivity of a model
pathway, the one step oxidation of glucose to gluconic acid, a top value added commodity
chemical from biomass used in a variety of industrial processes (97, 98).
In this manner, we
demonstrated that control of primary metabolism over a 3-fold range is indeed possible with
the potential to control heterologous pathway productivity.
2.2.Results and Discussion
2.2.1.
Generation of a tunable glucose flux platform
To enable control of intracellular glucose flux, it was necessary to generate a PTS- Glu*
phenotype. We achieved this in E. coli DH10B by first deleting the ptsH/crr operon encoding
components of the glucose/mannose specific permeases and constituents of the cascade
involved in transferring a phosphoryl group from PEP to glucose (26). As expected, the deletion
resulted in a Glu~ phenotype. To restore growth on glucose, galP was upregulated with the
introduction of the strong constitutive lac/ promoter (Part J56015 from the Registry of Standard
Biological Parts, http://partsregistry.org) to replace the native ga/P promoter. Native regulation
of ga/P was also removed by silent mutations of the ORF to disrupt internal GalS and GaIR
27
2.2.2 Range of in vivo expression of gik
repressor binding sites (99, 100) (see Table 2-4 for specific mutations). The resulting strain,
KTSO22, was able to recover over 60% of the parent strain's specific growth rate in glucose
supplemented minimal medium (Figure 2-2) and attain similar final ODs.
When bearing a
plasmid, a typical requirement for the expression of a heterologous pathway, specific growth
rates were indistinguishable between the two strains (Figure 2-2).
0.6-
0A02
DH10B
KT6022
Strain
Figure 2-21 Growth rate recovery of AptsHlcrr galP, mutants
Specific growth rate of PTS~ Glu* strain (KTSO22) and its parent strain (DH10B) with and without a plasmid
(pBAD30) in M9(0.4% glucose). Plasmid bearing strains are indicated with gray bars. pBAD30 is a medium
copy plasmid (p15A) constitutively expressing beta lactamase (AmpR). Graph depicts average ± standard
deviation of duplicate cultures from independent experiments.
2.2.2. Range of in vivo expression ofgik
A family of glucokinase (gik) expression mutants (KTSx22) that enabled the exploration
of the viable range of Glk activity was generated in the PTS~ Glu* background. The native FruR
regulatory binding sequence was disrupted by substitution with a sequence that had little
homology to the noted consensus sequence (101) to remove endogenous regulation (Consensus
sequence: RS I TGAAWC SNTHHW -> Mutated sequence: TA GATTCA AACGGG). The primary
native promoter (P1) (102-104) was also replaced with a member from the Anderson promoter
library to tune Glk expression. These mutations (Figure 2-3 & Table 2-1) were first generated
with PCR site-directed mutagenesis and cloned into pKD13 between Sall sites before being
introduced into the chromosome at the native locus by A-Red mediated recombination (105).
28
CHAPTER 2 | TUNING PRIMARY METABOLISM FOR HETEROLOGOUS PATHWAY PRODUCTIVITY
a
b
2
KTS 022
2,
s
,
DHI1B ded*od with
A1ptsHkcfgaig
4E.
J23xxN
KTS x22
KTS 022 dMW
t)22
2,
57,582
2,
51'o,
2,
507,
GCTGAAACGA TAAAGTAATT GTGTACA
GGCYGTTU'
2"rGCCCCC
;AM;T
AA?1
K?,A,
A
A6TAWAC
A
A
A
Aa'GTGiAAA
AGAAT
A
ATTT
T
TCAcTlA'GC
Av-
TACATTCAAA CGGGGTAATT GTGTGACCA GATCGATATI TACAGGGAGCCTGCC''TTCC
522
GGCX310
aNNu
462 GGAGcAGTTG AAGA
llii
MnUUIDIINOW
NNN.
AA AGAATTATTT
TGACTrTAC
ACA AAGT
Figure 2-31 Glucokinase expression family construction
a) Schematic of mutations introduced. b) Sequence surrounding the 5' UTR of the glucokinase expression
family. The FruR binding site (underlined) was first replaced with a sequence of minimal homology to the
natural consensus sequence (double underlined). The overlapping promoters (italicized) of glk were
disrupted and the primary promoter, P1 (bold), replaced with a promoter from the Anderson library
(J23xxx). The start codon of Glk is noted in lowercase green letters. Chromosomal location relative to a
MG1655 chromosome is noted on the left.
Table 2-1 Glucokinase expression family promoter sequences
Family members and their constitutive promoter part names, relative strengths and sequences as
annotated in the Registry of Standard Biological Parts, grou'ped and ordered by origin and relative
strength. Sequence variations relative to J23100 have been highlighted in bold.
Strain
Part Name
Relative Strength
Sequence
GTTGTTGTTA TGCCCCCAGG TATTACAGTG TGA
KTSO22
Native
N/A
TTGACGGCTA GCTCAGTCCT AGGTACAGTG CTAGC
KTS322
J23100
1.000
TTTACGGCTA GCTCAGTCCT AGGTACAATG CTAGC
KTS522
J23110
0.331
TTTATAGCTA GCTCAGCCCT TGGTACAATG CTAGC
KTS722
J23115
0.152
TTTATGGCTA GCTCAGTCCT AGGTACAATG CTAGC
KTS922
J23114
0.101
TTGACAGCTA GCTCAGTCCT AGGGATTGTG CTAGC
KTS622
J23117
0.064
TTTACAGCTA GCTCAGTCCT AGGGACTGTG CTAGC
KTS822
J23109
0.042
CTGATGGCTA GCTCAGTCCT AGGGATTATG CTAGC
KTS1022
J23113
0.008
CTGATAGCTA GCTCAGTCCT AGGGATTATG CTAGC
KTS1122
J23112
0.004
With the construction of the glucokinase expression family completed, the range of Glk
activities that were achievable with the Anderson library variants was determined.
The full
dynamic range of the KTSx22 family was characterized using glycerol as a carbon source,
minimizing concern for the impact of Glk expression on carbon metabolism and physiology. As
anticipated, increasing relative promoter strength led to an increase in Glk activity (Figure 2-4a)
over a log range. However, at high promoter strengths, the cultures grew very slowly with two
family members (KTS322 and KTS522) requiring twice as much time to reach late exponential
phase (OD600 ~1.0).
Moreover, these constructs also displayed a lower than expected activity
given their promoter strength. Hypothesizing a metabolic burden, the relative mRNA levels of
29
2.2.2 Range of in vivo expression of glk
gik were determined by qRT-PCR. While mRNA levels scaled well with Glk activity at low and
intermediate promoter strengths, the qRT-PCR data was unable to explain the reduced activity
of KTS522 (promoter strength = 0.33) suggesting an issue at the translational level. Possible
explanations for this discrepancy include ribosome saturation or insufficient energy (GTP/ATP)
available to meet the translational demand.
A decrease in mRNA expression for KTS322
(promoter strength = 1.00) relative to KTS522 was also noted (Figure 2-4a), suggesting an
additional transcriptional burden at extremely high expression levels.
a
b
0.7 -
0.6 -
0 RM-
10
m4A Lhvda
0.5-
0.4
0.35
.
0.10.
0.2
1
0.1
C.0.1
-
0.1
0.0 1
0.0
0.00
0.05
0.10
0.15
0.4
0.6
Relative Promoter Strength
0.8
1
0.0
0.1
0.2
0.3
0.4
0.5
Gik Activity (UImg)
Figure 2-41 Expression family activity data
a) Glucokinase activity and mRNA expression relative to the native promoter as a function of promoter
activity in M9(0.4% glycerol). Region boundaries depicted are arbitrary, based on observed trends. b)
Specific growth rate as a function of Glk activity in M9(1.5% glucose) of the healthy glucose viable
mutants. Wildtype activity is indicated by an open triangle in Panel B. Figures depict average ± standard
deviation of triplicate cultures from a representative experiment.
Only relatively low expression levels of Glk are viable when the family is grown in
minimal medium supplemented with glucose (Figure 2-4a). These mutants correspond to the
weakest promoters found in the Anderson promoter library with a promoter strength of only
0.06 (KTS622) being roughly equivalent to wildtype Glk expression. Within the viable regime,
Glk activity spanned a 3-fold range from 0.15-0.48 units per mg total protein (U/mg) or roughly
0.5-2x that of wildtype Glk activity (0.28 U/mg). This range was noticeably smaller than the
corresponding span of 0.01-0.42 U/mg seen in glycerol supplemented medium.
Unlike the
experiments in glycerol, however, there is a direct correlation between Glk activity and specific
30
CHAPTER 2 TUNING PRIMARY METABOLISM FOR HETEROLOGOUS PATHWAY PRODUCTIVITY
growth rate which is approximately linear (Figure 2-4b), suggesting that Glk is indeed controlling
growth and thus glycolytic flux. Despite the range of activities and growth rates in glucose
supplemented medium, all viable mutants attained a similar final OD600 . This observation is in
stark contrast to previous attempts to control central carbon metabolism and/or cellular
phenotype where the final OD600, and not specific growth rate, is affected (86, 106).
In those
studies, however, metabolism was modulated at nodes where alternate, but inefficient, routes
of carbon utilization exist. In such strains, it is anticipated that the biomass yield on glucose is
reduced leading to reduced biomass accumulation.
KTS322 and KTS522, which were noted to be subject to high transcriptional and
translational burdens, were not viable in glucose supplemented medium.
Similarly, a
transitional regime was identified encompassing KTS722 (promoter strength = 0.15) which
showed weak growth and was only able to attain a quarter of the maximum OD600 observed in
the healthier, lower expressing strains (data not shown). Given the ATPase activity of Glk in the
presence of glucose, we hypothesized that at these high levels of glucose phosphorylation,
glycolysis is saturated and the cell consumes ATP much faster than it can be regenerated
aerobically. Indeed, an 02 respiration limit of ~16 mmol/hr/g dcw has been suggested (107),
potentially due to membrane space constraints in accommodating the respiratory chain (108).
Performing a stoichiometric analysis of ATP generation and consumption by central carbon
metabolism, including the energetic costs in energizing the membrane, consumption of some
carbon for biomass rather than energy and the dissipation of excess energy through futile cycles
(109, 110), suggests that the ATPase activity of Glk at these expression levels exceeds the
aerobic ATP regeneration limit of the cell and results in arrested growth.
Factoring in the
increased energy demand for DNA replication, protein synthesis and lipid biosynthesis with
increasing Glk and growth rate (Figure 2-4b) (111) reveals that the transitional regime of cellular
31
2.2.3 Glk expression controls gluconate productivity
health approaches this energetic limit, potentially explaining the reduced biomass production
(see §4.2 for details).
2.2.3.
Glk expression controls gluconate productivity
Having established the dynamic range of Glk activities possible, we then probed the
efficiency with which the glucose viable strains could synthesize a small molecule product that
competes with Glk for glucose as a substrate. The single step oxidation of glucose to gluconic
acid (gluconate), catalyzed by glucose dehydrogenase (Gdh) with the consumption of one
NAD(P)+, was selected. Gdh was expressed in a gluconate catabolism negative background (idnK
gntK, named KTSx221G) (112, 113) to facilitate accumulation of product. Preliminary studies
with a related gntK idnKDO background showed that Gdh catalyzed the generation of gluconate
in E. coli with KTSO22 showing an approximate 100% increase in gluconate productivity relative
to its DH10B PTS* parent (data not shown). In KTSO221G, the accumulation of gluconic acid is
thermodynamically favored (47); however, small amounts may be reversibly reduced to 5-ketogluconic acid.
Gdh was expressed at one of three induction levels from a high copy, IPTG-inducible
pTrc99-derived vector. To minimize issues with synchronization due to the strain family's wide
range of growth rates, these vectors were induced upon inoculation.
Preliminary studies
showed no differences in the results obtained in this manner as compared to more traditional
protocols where cultures are induced prior to the onset of stationary phase. After 72 h, the
cultures were assayed for Gdh activity, growth and titers of several major fermentation products
(Table 2-2). Induction level appeared to have a strong effect on cellular growth, expression and
titers.
We observed an inverse correlation between final OD and induction for mutants
expressing Glk significantly below wildtype levels.
This effect may be attributed to the
metabolic burden of expression and reduced carbon flux through Glk for endogenous activities.
32
CHAPTER 2|TUNING PRIMARY METABOLISM FOR HETEROLOGOUS PATHWAY PRODUCTIVITY
The burden of induction also manifested in the ability of the mutants to express Gdh, with most
strains exhibiting maximum Gdh expression at lower induction levels.
Counterintuitively,
although not unexpectedly (19), the highest Gdh activity observed did not correspond with the
highest product titers. These titers, ~ 5 g/l, were obtained at intermediate Gdh expression levels
at the lowest induction tested (10 pM).
At this induction level, all strains tested produced
significant amounts of gluconate (~1.7- 5 g/I).
Table 2-21 Titers of the major fermentation products
Final OD600, and titers of gluconate (Gnt), its reduced form (5-ketogluconate, 5KG), acetate and glucose
consumed at 72 h as a function of Gdh and ordered by increasing Glk expression. Average values of at
least 2 parallel cultures are reported with standard deviations given in parentheses. 50 ml of culture were
grown in M9(1.5% Glucose) at 37 *C.
Strain
IPTG
M]
KTS8221G
KTS10221G
KTS6221G
10
KTSO221Gb
KTS9221G
KTS8221G
KTS10221G
KTS6221G
25
KTSO22IGb
KTS9221G
KTS8221G
KTS10221G
KTS6221G
100
KTSO221Gb
KTS9221G
a Glk activity of
Final OD6 00
Gdh
Activity
U/mg]
Estimateda
GIk Activity
[U/mg]
1.56 (0.15)
15.2 (2.3)
0.23 (0.04)
1.34 (0.17)
10.9 (0.2)
0.27 (0.03)
1.73 (0.27)
23.5 (1.0)
0.30 (0.02)
1.58 (0.23)
21.2 (0.5)
0.33 (0.02)
1.12 (0.06)
5.7 (0.1)
0.50 (0.03)
0.27 (0.16)
ND
0.23 (0.04)
0.66 (0.31)
0.8 (0.7)
0.27 (0.03)
1.37 (0.04)
33.0 (1.7)
0.30 (0.02)
1.91 (0.36)
31.8 (3.5)
0.33 (0.02)
0.98 (0.07)
22.2 (1.2)
0.50 (0.03)
0.06 (0.02)
ND
0.23 (0.04)
0.10 (0.02)
ND
0.27 (0.03)
1.57 (0.40)
19.2 (1.0)
0.30 (0.02)
1.13 (0.08)
15.2 (0.5)
0.33 (0.02)
1.09 (0.24)
27.2 (0.1)
0.50 (0.03)
the plasmid-free parental strain.
Glucose
Acetate
Consumed
[g/L]
[g/L]
2.61 (0.09)
0.32 (0.01)
ND
6.01 (0.27)
2.69 (0.26)
0.28 (0.03)
ND
5.19 (0.59)
5.00 (0.21)
0.37 (0.07)
0.79 (0.00)
10.05 (0.01)
4.88 (0.31)
0.38 (0.07)
0.41 (0.04)
10.25 (0.03)
1.71 (0.06)
0.20 (0.02)
0.67 (0.09)
7.98 (0.01)
0.16 (0.01)
0.03 (0.02)
0.12 (0.12)
ND
0.21 (0.04)
0.10 (0.01)
0.17 (0.08)
1.34 (0.58)
3.30 (0.11)
0.47 (0.12)
0.71 (0.09)
9.00 (0.09)
4.13 (0.48)
0.31 (0.02)
0.49 (0.09)
10.25 (0.89)
1.72 (0.07)
0.21 (0.08)
0.64 (0.04)
7.51 (0.06)
ND
0.01 (0.00)
ND
ND
ND
0.02 (0.01)
ND
ND
3.04 (0.14)
0.50 (0.01)
0.01 (0.00)
6.57 (0.01)
1.30 (0.04)
0.11 (0.02)
ND
3.35 (0.01)
2.95 (0.55)
0.22 (0.01)
0.22 (0.01)
6.28 (0.35)
Due to the assay chemistry, Glk activity is
Gnt
[g/L]
Titers
5KG
[g/L]
indistinguishable from Gdh when both are present.
bWildtype Glk expression
ND - not detected or statistically insignificant (measured range encompasses zero)
Examining titers more closely at specific induction levels, trends are more difficult to
discern as gluconate production is a non-linear function of cellular phenotype, namely, Gdh and
Glk levels as well as intracellular glucose, NAD(P)* and ATP pools. The high variability in cellular
phenotype across strains and induction levels is echoed in the acetate levels (Table 2-2),
typically indicative of excess carbon flux and/or cellular stress. However, if we evaluate the
conversion of glucose to gluconate and 5-ketogluconate as a function of Glk activity (Figure 2-5),
33
2.2.3 Glk expression controls gluconate productivity
trends begin to emerge despite variations in these other parameters. At the lowest induction
level tested, which generated the highest product titers and gluconate yields, there is a clear
inverse correlation between gluconate yield and Glk activity with gluconate yields doubling to
more than 0.5 mol of gluconate per mol glucose while Glk activity is decreased ~50% going from
KTS9221G to KTS10221G (0.50-0.27 U/mg). However, as we decreased Glk activity further (i.e.,
KTS8221G - 0.23 U/mg), there appeared to be a slight decrease in yield, likely due to the limited
availability of intracellular resources and compromises in cellular health. At the next induction
level tested, 25 p.M, a similar pattern emerges where there is a 50% increase in yield with only
40% decrease in Glk activity (0.5-0.3 U/mg).
Further suppression of Glk compromises
production in KTS10221G (0.27 U/mg) and impairs production and growth of KTS8221G (0.23
U/mg) altogether, presumably due to the increased metabolic burden of Gdh expression with
reduced carbon flux. At the highest induction level tested, the trends disappear altogether and
both KTS8221G and KTS10221G show negligible growth. We hypothesize that at this high level of
induction, the metabolic burden begins to impinge on the health of all the strains tested
resulting in poor growth and/or pathway limitations other than glucose availability.
06
06.
06-
10 yM IPTG
--
25
100pMIPTG
tMIPTG
05-
05.0
04.
04.
04.
03-
03
0 3
1
~02
0.2-02
0A1
01
01.
0,0
0,0
0.0
0.2
03
0'4
O'S
G Actity (UhMg)
Estimated
02
0'3
04
Estmated G&Acdtvy (U/mg)
045
0'2
0'3
04
0'5
Gik Actitiy (Ulmg)
Estimated
Figure 2-51 Gluconate molar yield on glucose at 72 h as a function of Glk activity and IPTG induction
Gluconate molar yield represents the yield of the major gluconate derived species (gluconate (Gnt) and 5ketogluconate (5KG)). Plots represent averages ± standard deviations of at least 2 parallel cultures in a
representative experiment. Due to the similar assay chemistries, Glk activity cannot be measured in the
presence of Gdh. Estimated Glk activities presented here represent those of the host strain in the
absence of any plasmid.
34
CHAPTER 2 TUNING PRIMARY METABOLISM FOR HETEROLOGOUS PATHWAY PRODUCTIVITY
2.3.Summary & Concluding Remarks
In this study, we generated a family of g/k expression mutants through the use of
promoters found in the Anderson promoter library.
Only the weakest promoters in this
collection, with relative strengths ranging from 0.008-0.101 were found to be viable in minimal
medium supplemented with glucose. Within this range, however, growth rate was controlled
from 38-141% of wildtype with a corresponding 3-fold change in Glk activity (0.15 -0.48 U/mg).
Despite the diversity of specific growth rates possible, we were able to achieve comparable
levels of biomass accumulation. Moreover, in this library, we were able to identify an upper
bound to glucose utilization through Glk of at least twice that of wildtype levels. As we moved
beyond this limit, cells failed to grow and/or attain maximal cell density which we hypothesize is
due to the unregulated ATPase activity of Glk. Similarly, we suspect that there is a floor to Glk
activity based on the observed correlation of growth rate and Glk activity.
Within this Glk family, we were then able to examine the ability of the cell to redirect
glucose to a heterologous pathway. Over a subset of Glk activities, there is a clear inverse
correlation between endogenous glucose utilization and the efficiency of a model pathway
suggesting successful redirection of carbon flux. However, reducing endogenous metabolism
also had the deleterious effect of reducing metabolic flexibility, the ability of the cell to support
exogenous loads. At high induction levels, impaired cells failed to grow or express recombinant
protein.
Moreover, while product molar yield increased with decreasing Glk, these increases
were not necessarily reflected in improved titers, suggesting that the metabolic burden imposed
on these metabolically inflexible cells had an uncharacterized impact on other intracellular
parameters such as glucose uptake rate and cofactor pools. The existence of these inferred
effects implies that there exists a more complex relationship than that between carbon flux
redirection and biomass production (specific growth rate). This issue, however, may potentially
35
2.4.1 Strains and plasmids
be addressed by a more dynamic 'valve' or 'switch' (54, 55, 85, 86, 88, 106) which allows cells to
accumulate the necessary endogenous components to support production before flux
redirection is implemented.
Despite these challenges and limitations, we were able to create a microbial host that
allows for the incorporation of unphosphorylated glucose directly into heterologous production
pathways enabling the creation of unnatural pathways. The existence of community sourced
and, more importantly, curated tools, in this case the Anderson Promoter Library, proved to be
invaluable in allowing us to probe the glucose utilization landscape and identify general viability
regimes. In spite of its 2-log range, however, the number of weak promoters available limited
our ability to precisely identify and define phenotypic boundaries.
Nonetheless, with this
library, we were able to demonstrate not only the ability to redirect carbon and control product
conversion yields, but examine the flexibility of cells to tolerate perturbations in central carbon
metabolism without alternate routes of carbon flux.
2.4.Methods
2.4.1.
Strains and plasmids
E. coli strains and plasmids used in this study are listed in Table 2-3.
All molecular
biology manipulations were carried out according to standard practices (114).
Chromosomal
manipulations were achieved through
-Red mediated recombination (105, 115) using pKD46
(105) or a derivative containing recA (see Appendix Al).
Sequence specific mutations (e.g.
promoter replacements, binding site disruption) were introduced either directly through the
primers used for PCR generation of the recombination cassette (Table 2-4) or first cloned into
the suicide vector pKD13 (105) (CGSC, New Haven, CT) between Sall sites using standard
techniques before PCR amplification. In all cases, the kan selection cassette was cured from the
36
CHAPTER 2 TUNING PRIMARY METABOLISM FOR HETEROLOGOUS PATHWAY PRODUCTIVITY
final mutants with FLP recombinase expressed from pCP20 (CGSC) (105, 116).
The plasmid
pTrc99ACm-gdh, containing genes for the expression of glucose dehydrogenase (gdh, EC
1.1.1.47) from B. subtilis (117), was used to enable gluconic acid production.
The vector
pTrc99ACm was generated by first PCR amplifying cat (CmR) from pMMB206 (118) (ATCC,
Manassas, VA), then blunt ligating the fragment with pTrc99A (119) linearized with Scal,
disrupting bla (AmpR). gdh was then amplified from B. subtilis genomic DNA with the addition of
appropriate restriction sites and inserted between the Xbal and HindIll sites of pTrc99ACm. All
mutants were identified and isolated by colony PCR with appropriate primers; knock-ins and
other sequence specific mutations were further verified with sequencing and restriction digests
of colony PCR products. PCR amplifications were performed with Phusion High-Fidelity DNA
Polymerase (NEB, Ipswich, MA) and oligonucleotides from Sigma-Genosys (St. Louis, MO; Table
2-4).
All plasmids were cloned and propagated in E. coli DH1OB (Invitrogen, now Life
Technologies, Carlsbad, CA) with the exception of pKD13-derived
plasmids which were
propagated in EC100D pir-116 (Epicentre Biotechnologies, Madison, WI).
Table 2-31 Main strains and plasmids used
Name
Relevant Genotype
E. coli strains
F mcrA A(mrr-hsdRMS-mcrBC)< p80acZAM15
Electromax DH10B
AlacX74 recAl endAl araD139 N(ara, leu)7697 galU
Source
Invitrogen
galK A rpsL nupG
Transformax EC100D pir-116
As above, pir-116(DHFR)
KTS002
DH10lB LptsH/crr
KTSx22 family
As above,
KTSx221G family
As above, AgntKAidnK
Pgik::Pcon*
Epicentre
Biotechnologies
This study
galP
This study
This study
Plasmids
(120)
pBAD30
p15A, bla (Am pR)
pCP20
0 pr Rep", bla(Am pR) cat(CmR), X c1857 (ts); k Pr FLP
ColE1(pBR322) ori, cat (CmR), lac, Ptrc gdh
CGSC #7629
pTrc99ACm-gdh
pKD13
oriRy; bla(Am pR), kan
CGSC #7633
pKD46
pKD46RecA
t
oriR101, repA101 , bla (Am pR), araC,
As above, recA
2
ParaB AyAjQ exo
This study
CGSC #7739
Appendix Al
con = Anderson library promoter
37
Name
Glk mutant construction
glklFruR
gIk2_J23100
gfk2J23110
glk2J23117
g/k2J23115
glk2_J23109
g/k2_J23114
g/k2J23113
gIk2_J23112
glkkan
J23100_insert
J23110_insert
CTAT TCCTTA TGCGGGGTCAGATACTTAGTTTGCCCAGC7TGCAAAAAGGCATCGCTGCAATTGGTGTGTAGGCTGGAGCTGCTTC
CACATCACCGACTAATGCATACTTTGTCAT TCTTCAACTGCTCCGCTAAAGTCAAAATAATTCTTTGCTAGCACTGTACCTAGGACTGAGCTA
J23117_insert
J23115_insert
CA CATCACCGACTA ATGCA TACTTTGTCATTCTTCAA CTGCTCCGCTAAAGTCAAA ATA ATTC7TTGCTAGCACAATCCCTAGGACTGA
CACATCACCGACTAATGCATACTT TGTCAT TCTTCAACTGCTCCGCTAAAGTCAAAATAATTCTTTGCTAGCATTGTACCAAGGGCT
J23109_inset
CACATCACCGACTAATGCATACTTTGTCATTCTTCAACTGCTCCGCTAAAGTCAAAATAATTCTTTGCTAGCACAGTCCCTAGGACTGA
CA CATCACCGACTA ATGCATACTTTG TCATTCTCA ACTGCTCCGCTAAAGTCAAA ATAAT TCTTTGCTAGCATTGTACCTAGGACTGAGCTA
J23114_insert
J23113_insert
J23112_insert
I)
-o
#A
5t
a,
0
0.D
0
4-
ptsHlcrr deletionb
pts-For-1s'
t
pts-Rev-1"
nd
pts-For-2
pts-Rev-2"d
CACATCACCGACTA ATGCATACT TTGTCA TTCTTCAA CTGCTCCGCTAAAGTCAAA ATAAT TCTTTGCTAGCATTGTACCTAGGACTGAGCTA
CACATCACCGACTAATGCATACTTTGTCATTCTTCAACTGCTCCGCTAAAGTCAAAATAATTCTTTGCTAGCATAATCCCTAGGACTGAGC
TCTTTGCTAGCATAATCCCTAGGACTGAGC
CACATCACCGACTAATGCATACTTTGTCATTCTTCAACTGCTCCGCTAAAGTCAAAATAAT
TTAGTTCCACAACACTAAACCTATAAGTTGGGGAAATACAGTGTAGGCTGGAGCTGCTTC
TGATGCGGATAACCGGGGTTTCACCCACGGTTACGCTACCTCCGTCGACCTGCAGTTCG
GCTAACAATACAGGCTAAAGTCGAACCGCCAGGCTAGACTTTAGTTCCACAACACTAAACCTATAAGTTG
CCGATGGGCGCCATTT1TCACTGCGGCAAGAATTACTTCTTGATGCGGATAACCGGGGT
Qa/P mutationc
U,
U,
Sequence 5' -3''a
& integration
CTATGTCGACTAGATTCAAACGGGGTAATTGTGTGACCCAGA
CTATGTCGACGCTAGCACTGTACCTAGGACTGAGCTAGCCGTCAAGCCGGAAAGGCA
CTATGTCGACGCTAGCATTGTACCTAGGACTGAGCTAGCCGTAAAGCCGGAAAGGCA
CTATGTCGACGCTAGCACAATCCCTAGGACTGAGCTAGCTGTCAAGCCGGAAAGGCA
CTATGTCGACGCTAGCATTGTACCAAGGGCTGAGCTAGCTATAAAGCCGGAAAGGCA
CTATGTCGACGCTAGCACAGTCCCTAGGACTGAGCTAGCTGTAAAGCCGGAAAGGCA
CTATGTCGACGCTAGCATTGTACCTAGGACTGAGCTAGCCATAAAGCCGGAAAGGCA
CTATGTCGACGCTAGCATAATCCCTAGGACTGAGCTAGCCATCAGGCCGGAAAGGCA
CTATGTCGACGCTAGCATAATCCCTAGGACTGAGCTAGCTATCAGGCCGGAAAGGCA
gaIP-For-1
ga/P-Rev-14
ga/P-For-2
ga/P-Rev-2nd
ga/P-For-Rep
ga/P-Rev-Rep
TGGTGCAAAACCTTTCGCGGTATGGCATGATAGCGCCCACAATAAAAAATAACCATATTGGAGGGCATC
ACCTGCGATAACGCCGATATCCAGGCCAAAGAGTAATCCCGCCAGAG
GTCGACTGGTGCAAAACC1TTCGCGG
GTCGACACCTGCGATAACGCCGATAT
AATTATTTTCATGCACTTAAATCATAACTAAGATAAATGTGTGTAGGCTGGAGCTGCTTC
AGTAA TCTGGA TFCA TCTGCAA TAAACGGCAGTGCACCTGCGA TAACGCCGA TAT
00
Gluconate catabolism deletion
didnKF
AAATTATTATGCCGCCAGGCGTAGTATCGCAGCAGGTAAGATGA
TTCAGGAGA TTTTAAAGTGTAGGCTGGAGCTGCTTC
didnKR
CAGCATGTGCGCGACGGTAAGGCGCGTTACCGCGTGGTGTTGAAAGCCGATT17TTGAAAATCCGTCGACCTGCAGTT
gntKF2
CTGATA TTGTCCGGCTGGACAATGTTACCGATAACAGTTACCCGTAACA1111MAATTCTTGTATTGTGGGGGCACCACTGTGTAGGCTGGAGCTGCTTC
gntKR2
ATACGCGCCTTCATGACTAAAAACAGCAGCAGTAAAACAGACCCTACTGCTGTTAAAACAAGCGTTAATGTAGTCACTACTCCGTCGACCTGCAGTT
Gdh cloning
ForCm-pMMB206
Rev Cm pMMB206
For gdhsubtilis
Revgdhsubtilis
TGGTGTCCCTGTTGATACC
CGCTTATGGCAGAGCAG
TACATATAAGTCTAGATAACAAATGGAGGAGGATG
CAAGTAACTAAAGCTTTCATGTCTGGGTCGCT
Colony PCR & Sequencing
002checkF (pts)
GCTAACAATACAGGCTAAAGTCGAAC
002checkR (pts)
CCGATGGGCGCC
022checkF (gaiP)
AATTATTTTCATGCACTTAAATC
022checkR (galP)
GTAATCTGGAATTCATCTGC
glkchkF
AAAAGGCATCGCTGCAAT
gikasfor500
CTATGAATTCCTATTCGGCGCAAAATCAAC
gikasfor1O
CTATGAATTCAATAGGTCTTAGCCTGCGAG
gntchkF
AAGTAGCTCACACTTATACACTTAAG
gntchkR
TGACTAAAAACAGCAGCAG
idnchkF
GTGAAATTATTATGCCGC
idnchkR
TGACTAAAAACAGCAGCAG
Introduced mutations (promoters or other) are in boldface type; homologous sequences for recombination are underlined and italicized; restriction sites used
for cloning are underlined.
aAll oligonucleotides purchased from Sigma-Genosys, St. Louis, MO.
bPCR for the ptsHlcrr deletion cassette performed in two stages where a 2 round of PCR was used to increase the degree of homology to facilitate
recombination.
galP mutations performed in 3 parts: mutations were introduced in the first round before a Sall site was added for cloning into pKD13. The final primer set
(galP-xxx-Rep) was then used to generate the linear DNA cassette for recombination.
Li.)
cG)
2.4.2 Culture conditions
2.4.2.
Culture conditions
For construction, strains were propagated in Luria-Bertani (LB) medium at 370C (30 0 C for
intermediate recombination steps). Temperature sensitive plasmids were cured at 42 0 C. All
experimental cultures were grown at 37*C in M9 minimal medium supplemented with 0.8 mM
L- leucine and either glucose or glycerol as indicated. For all experiments, starter cultures were
grown to an OD measured at a wavelength of 600 nm (OD600 ) of at least 0.2 before being
transferred to a 50 ml culture in a flask of the same media. For experiments involving Gdh, IPTG
was added as indicated to both the starter culture and 50 ml flasks at inoculation. In enzyme
activity and qRT-PCR studies, 50 ml shake flasks were inoculated to a final OD 600 of 0.001 and 5
ml samples were taken at approximately OD600O0.5 for flasks supplemented with glucose or
OD 60 0 ~.0 for flasks supplemented with glycerol. These samples were then pelleted and stored
at -20 0 C until assayed. For gluconate experiments, shake flasks were inoculated to a final OD 600
of 0.005 with M9 medium supplemented with glucose and monitored for 72 h before samples
were taken and clarified by centrifugation. The supernatant was then stored at 4 0C until HPLC
analysis. As appropriate, the antibiotics ampicillin, kanamycin and chloramphenicol were used
at concentrations of 100 pg/ml, 25 ptg/ml and 34 pg/ml, respectively.
2.4.3. Quantification of mRNA levels
Total RNA from culture samples grown in M9 medium supplemented with glycerol were
extracted using the illustra RNAspin Mini Isolation Kit (GE Healthcare Bio-Sciences, Piscataway,
NJ) with an on-column DNasel treatment following the kit protocol.
Using the QuantiTect
Reverse Transcription Kit (Qiagen, Valencia, CA) according to manufacturer's instructions, 0.5 pg
of total RNA was treated to remove trace DNA contamination before cDNA was synthesized with
random primers.
The synthesized cDNA was then amplified with primers g/k qPCR_for (5'-
TTGCGGGCGGTATCGT-3') and g/k qPCR rev (5'-GGAAACCGGAGGCTTTGAAG-3')
40
in a qPCR
CHAPTER 2 TUNING PRIMARY METABOLISM FOR HETEROLOGOUS PATHWAY PRODUCTIVITY
reaction with Brilliant il Sybr Green High ROX QPCR Mix (Agilent Technologies, Santa Clara, CA)
on an ABI 7300 Real Time PCR System instrument (Applied Biosystems, currently Life
Technologies, Carlsbad, CA). Transcript levels were quantified in duplicate with appropriate notemplate and no-RT controls and are relative to that of native g/k expression levels (KTS022) as
determined from a standard curve. Reported levels are the averages of triplicate flasks, each
measured in duplicate.
2.4.4.
Enzyme activity assays
All enzyme activity assays were performed on crude lysates generated by resuspending
frozen cell pellets in 500 pl of 10 mM Tris-HCI (pH=8.0) and sonicating at ~ 8 W in 5 x 5 s pulses
on ice.
The resulting lysates were clarified by centrifugation before being used as follows.
Glucokinase (Glk) activity was measured in a coupled enzymatic assay as first described by
DiPietro and Weinhouse (121). Glk phosphorylates glucose in the presence of ATP to glucose-6phosphate.
Glucose-6-phosphate is in turn oxidized to 6-phospho-D-gluconate by glucose-6-
phosphate dehydrogenase (G6PDH) with the generation of a spectrophotometric NADPH signal
at 340 nm. One unit of Glk activity will phosphorylate 1.0 pmole/min of D-glucose at pH 7.5 and
room temperature in the presence of 3.33 U/ml G6PDH, 60 mM Tris-HCI, 20 mM magnesium
chloride, 8.0 mM ATP, 12.0 mM glucose and 0.9 mM NADP*.
Glucose dehydrogenase (Gdh)
activity was measured directly from the oxidation of glucose to gluconolactone with the
production of a NADH signal at 340 nm (122). One unit of Gdh activity oxidizes 1.0 pmole/min
of D-glucose at pH 7.6 at room temperature in 60 mM potassium phosphate buffer and 0.67
mM NAD*. As Gdh is able to accept NAD* and NADP* as cofactors (123) with the generation of a
common spectrophotometric signal at 340 nm, Glk activity cannot be measured independently
in the presence of Gdh.
All enzyme activities were normalized by total protein levels as
measured in a modified Bradford assay described by Zor & Selinger (124).
41
2.4.5 Metabolite analysis
2.4.5. Metabolite analysis
Culture supernatant was analyzed on an Agilent 1100 series HPLC instrument with
separation on an Aminex* HPX-87 H anion exchange column (Bio-Rad Laboratories, Hercules,
CA) with 5 mM H2SO4 as the mobile phase at 550 C and a constant flow rate of 0.6 ml/min. A
refractive index detector (RID) set to 55 0 C was used to monitor glucose levels while a diode
array detector (DAD) was used to measure acetate and 5-ketogluconate (a gluconic acid derived
species) at 210 nm, and gluconic acid at 230 nm. As gluconic acid and glucose co-elute (- 9 min)
and are both detected by the RID, the DAD signal at 230 nm was used to resolve the two
species. Concentrations were determined from standard curves of each analyte prepared from
commercial standards. Evaporation was corrected for by noting the mass change of culture
flasks between inoculation and sampling.
42
CHAPTER 3
A dynamic metabolite valve for the control of central
carbon metabolism
ABSTRACT
Successful redirection of endogenous resources into heterologous pathways is a central tenet in
the creation of efficient microbial cell factories. This redirection, however, may come at a price
of poor biomass accumulation, reduced cofactor regeneration and low recombinant enzyme
expression.
In this chapter, we propose a metabolite valve to mitigate these issues by
dynamically tuning endogenous processes to balance the demands of cell health and pathway
efficiency. A control node of glucose utilization, Glk, was exogenously manipulated through
either orthogonal engineered antisense RNA or an inverting gene circuit.
Using these
techniques, we were able to directly control glycolytic flux reducing the specific growth rate of
engineered E. coli by up to 50% relative to a PTS- Glue mutant without altering final biomass
accumulation. This modulation was accompanied by successful redirection of glucose into a
model pathway leading to an increase in the pathway yield and reduced carbon waste to
acetate. This work represents one of the first examples of the dynamic redirection of glucose
away from central carbon metabolism and enables the creation of novel intracellular pathways
with glucose used directly as a substrate.
Portions of this chapter to be submitted for publication as:
Solomon, KV, Sanders, TM, Prather, KU. A dynamic metabolite valve for the control of central
carbon metabolism
43
3.1 Introduction
3. 1. Introduction
Microbes are powerful manufacturing platforms for a wide array of chemicals and
compounds from renewable resources. Optimization of these processes for economic viability,
however, remains a large challenge.
In addition to potential issues with catalytic activity,
engineered pathways frequently compete with endogenous ones for common substrates. To
solve this issue, researchers typically knock out the competing pathways or overexpress the
production pathway enzymes to drive the flux towards the target compound.
However, for
pathways that compete with essential and/or catalytically efficient processes, these approaches
are not necessarily feasible and call for tunable control of the competing pathway.
Recent
works by ourselves (Chapter 2) and others (28, 30, 43-45, 47) have demonstrated that tuning
expression of these endogenous competing enzymes is a viable method of improving the yield
and/or titers of desired heterologous pathways.
When glycolysis itself is the competing
pathway, however, unique challenges arise.
Central carbon metabolism is tightly regulated and controls processes integral to cell
growth and proliferation.
Thus, the tuning of carbon flux through central metabolism has
pleiotropic effects. These include foreseeable phenotypic effects such as changes in biomass
accumulation (86) and specific growth rate (85) (§2.2.2) to more subtle and far ranging global
consequences on protein expression (125, 126).
When placed in the context of heterologous
production, these changes may have an unforeseen impact on the ability of the cell to express
the necessary recombinant genes or maintain cofactors at optimal levels (127) (§2.2.3). For all
these reasons, a dynamic tuning system to control central carbon metabolism, which we call a
glucose valve, is needed.
With a valve, one could potentially maximize heterologous
productivity by optimizing pathway efficiency, recombinant protein expression, and biomass
and intracellular resources accumulation with timed manipulations of central metabolism.
44
CHAPTER 3 | A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
To fully realize the potential of dynamic control in such systems, one can imagine
cultures accumulating biomass at wildtype rates and once a target cell density is achieved, the
system downregulates endogenous metabolism redirecting flux of carbon away from growth
into product formation.
In such environments, the traditional paradigm of an inducible
promoter is no longer feasible; such systems would require a costly separation process to
remove the inducer in the shift from biomass to product formation. Similarly, while promoter
systems which repress expression due to chemical addition do exist, they are less well
characterized for metabolic engineering applications and are tied to the regulation of
endogenous processes limiting control orthogonality (128-131). Thus, we explored the potential
of orthogonal, engineered systems with the desired inverse input-output relationship, namely,
inducible antisense RNA and an inverting gene circuit (inverter). Such systems have enjoyed a
number of successes in diverse applications ranging from the control of bioprocesses (27, 28, 34,
35, 42) to the design of complex devices (53, 85, 132, 133).
We recently demonstrated the ability to control carbon flux in a PTS Glu* mutant of E.
coli through control of glucokinase (Glk) activity (Chapter 2). While the system was able to
control the yield of a model glucose consuming pathway, such control was static. Moreover,
due to the lack of alternate routes of carbon utilization, the permanent changes had negative
effects on the ability of the cell to express recombinant enzyme and simultaneously support
growth under certain conditions. In this study, we explore dynamic control of Glk and its impact
on heterologous pathway productivity. Through antisense RNA and inverters, we were able to
inhibit Glk activity by up to 25% and 50% respectively. Furthermore, through dynamic control
with the inverter, we were able to observe improvement in titers and yields of a model
compound and decrease in carbon waste to acetate, not only as a function of induction, but also
timing of induction. These improvements were not accompanied by the limitations seen in the
45
3.2.1 Inducible antisense RNA as a dynamic glucose valve actuator
static system. This work provides further evidence of the feasibility and impact of the control of
glycolysis through Glk as well as illustrates a few advantages and challenges of dynamic control.
3.2.Results
3.2.1.
Inducible antisense RNA as a dynamic glucose valve actuator
As an initial attempt to institute dynamic control of Glk activity, we investigated the
potential of antisense RNA (asRNA) as a valve actuator.
While relatively uncommon in
prokaryotic systems, there has been increasing interest in the use of RNA-based gene expression
control systems for biotechnology applications (27, 28, 30, 34, 35, 42, 53, 85). The few examples
of such natural and engineered systems in the literature are diverse in their approach with
equally diverse design rules for asRNA constructs stressing the strength of the binding
interaction (134, 135), kinetics of hybridization (136, 137) or the structure of the construct itself
(27, 136). Successful designs have been noted to be short (<100 nts) (53, 85, 138) or long with
significant complementarity to the ORF (27, 30, 34, 35, 42), with single hairpins (85, 138) or
more complex architecture for stability (27, 136). Among these conflicting designs, however, is
an emerging consensus that the 5' UTR, in particular the ribosome binding site (RBS), is critical
for functional asRNA (27, 28, 30, 35, 42, 53, 85, 138).
Adopting a compromise of these
conflicting rules, previously demonstrated to be successful (28, 30, 35, 42), we created asRNA
constructs by simply inverting contiguous sequence of the g/k operon. These constructs began
20 bps upstream of the start codon, encompassing the RBS, and extended into the ORF. To
explore issues such as construct length and the effect of asRNA structure on activity, we
truncated these constructs to 4 different lengths (Figure 3-1).
46
CHAPTER 3 | A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
chromosome
R953 -
R756 gk
RBS
R5
R100
-. .
--
RBS
RBS
plasmid
g/k
gk
Figure 3-11 Schematic of the inducible antisense constructs.
The reverse complement of the RBS and the truncated g/k ORF are expressed from a plasmid-based
inducible promoter (e.g. PBAD with |-arabinose or Pitt with anhydrotetracycline). The construct name
specifies the number of bases from the ORF incorporated into the antisense RNA.
Another important parameter to consider is the expression strength of such systems.
With an unknown hybridization efficacy, expression of asRNA from a low copy plasmid may
prove too weak to effectively control Glk activity. Similarly, expression from a high copy plasmid
may overwhelm native mRNA levels and completely suppress Glk expression and/or impose a
significant metabolic burden, killing the cell. As a compromise, the arabinose-inducible medium
copy pBAD30 (120) was initially selected to express the asRNA constructs. Due to the ptsHlcrr
ga/P4 background of the inverter host, catabolite repression of arabinose induction in the
presence of glucose is no longer active (139).
Given the correlation between Glk activity and
specific growth rate (Figure 2-4b), these constructs were first screened for growth in minimal
medium supplemented with 0.4% (w/v) glucose. While some constructs showed repression in
growth when induced with antisense, this effect was marginally more significant than that seen
in an empty vector control which also displayed some arabinose controlled repression of growth
(data not shown).
The Glk activity of these constructs, however, displayed no reproducible
behavior when induced.
Fearing that the inconclusive results were due to high metabolic
47
3.2.1 Inducible antisense RNA as a dynamic glucose valve actuator
burden caused by leaky expression from pBAD30, a tightly regulated tetracycline inducible
vector (140), pKVS45, was constructed.
To test the hypothesis of leaky expression in pBAD30 and validate expression from the
newly generated pKVS45, superfolder green fluorescent protein (sGFP) (141) was cloned into
the multicloning sites of both plasmids as a reporter protein. Cells were grown in minimal
medium supplemented with glucose; sGFP was induced in both constructs at inoculation and
fluorescence measured at exponential phase (Figure 3-2A).
pKVS45 displayed no leaky
expression in uninduced controls relative to an empty vector. Upon induction, however, there
was a 2-fold change in expression by mid-exponential phase (Figure 3-2A). pBAD30, in contrast,
displayed no expression, leaky or otherwise, potentially explaining the inconclusive asRNA
results. Letting the cultures grow longer, however, a tight response was observed for pBAD30 in
stationary phase (Figure 3-2B). Given this delayed response in pBAD30, all subsequent asRNA
testing was performed with pKVS45.
A
B
20,
D
182
pBAD30
1i6
pKVS45
-
20.
1
14.
14
812
812
C
C
110-0
0
8
z
z4
2.
2.--
-
emptyvet0 C401%
4
mducer
+kduce
emptyvector
control
- inducer
+inducer
Figure 3-21 Reporter gene assay of pBAD30 and pKVS45 in KTSO22
KTSO22 cultures were grown in triplicate at 37 *C in M9(0.4% glucose) with empty vector controls and
uninduced/induced sGFP bearing plasmids (A) Normalized fluorescence data of pKVS45 and pBAD30 at
mid exponential phase (OD~0.5). (B) Normalized fluorescence data of pBAD30 at stationary phase
(OD~1.4). pBAD30 was induced with 0.20% L-arabinose while pKVS45 was induced with 100 ng/ml
anhydrotetracycline (aTc). Graphs depict the triplicate mean ± SD.
The asRNA constructs were then rescreened for growth
in minimal
medium
supplemented with 1.5% (w/v) glucose and expressed from pKVS45. Upon induction, 2 of the
48
CHAPTER 3 |A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
constructs displayed significant reductions in specific growth rate (> 5%); R100 and R953 were
observed to reproducibly inhibit growth by 7% and 10%, respectively, when induced with 100
ng/ml anhydrotetracycline (aTc) (Figure 3-3A).
These growth differences, however, did not
affect final biomass accumulation. Screening these constructs further, decreases in Glk activity
of 12% and 25%, respectively, were observed (Figure 3-3B). R953 was further tested in PTS Glu*
mutants of MG1655 (DE3) (named MKTS3) and MG1655 (DE3 endA recA) (named MGER3). Glk
repression of ~15% was observed in all strains tested (Figure 3-3C) confirming the modularity of
the designed glucose valve. Finally, to evaluate the performance of all constructs, a qRT-PCR
assay was used to measure the various RNA species. Due to the complementarity between the
(sense) mRNA and asRNA, primers to detect asRNA will necessarily measure mRNA levels as
well; however, the truncated nature of the asRNA allows selective mRNA amplification of the 3'
end of g/k. A common primer set for all asRNA constructs was designed by targeting a ~60 bp
amplicon near the 5' terminus of the g/k CDS while mRNA was amplified with primers targeting
the 3' end. Using these primers, it was observed that the pKVS45 expression system is fairly
leaky. While this finding contradicts earlier sGFP studies, the discrepancy may be attributable to
the long maturation time of GFP (142) resulting in fluorescence being below the detection limit
at the time of observation despite the presence of leaky transcription products.
However,
induction with 100 ng/ml aTc led to at least a 10-fold change in asRNA+mRNA levels for all
constructs (Figure 3-3D) confirming that the non-functional constructs were expressed.
Examining the effects of the asRNA on gik mRNA, two very different effects were observed for
R100 and R953. R100, shown to have up to a 12% decrease in Glk activity had no effect on
mRNA levels suggesting an RBS occlusion mode of action which reduces translational efficiency.
In contrast, R953 was observed to reduce g/k mRNA by more than 85% and suggests that R953
binding stimulates increased mRNA degradation (Figure 3-3D).
Despite this large change in
49
3.2.1 Inducible antisense RNA as a dynamic glucose valve actuator
mRNA, however, only a 25% maximum decrease in activity was observed. Repeated tests of
these constructs under gluconeogenic and glycolytic conditions confirm the significance and
repeatability of this observation (data not shown).
A
B
0.3
0.250.200.20
0.150.10-
2
0.1
0.050.00
Empty
vector
R100
R500
R756
R953
R100
Empty
0.0
vector
R953
Construct
Construct
C
D
10
1.0-
0.8
.60 .
10'
0.2
0
10,
0.2-
M
MKTS3
MGER3
Strain
KTS622
Empty
R100
R500
R756
R953
vco
Construct
Figure 3-31 Antisense glucose valve
(A) Valve performance as measured by specific growth rate as a function of induction condition. (B) Valve
performance as measured by specific Glk activity of the most promising constructs from growth
characterization as a function of induction condition. (C) Performance of R953 in various hosts. Glk
activity is normalized to that of uninduced R953 in each host. (D) Valve performance at the RNA level.
RNA levels are relative to a single uninduced empty vector control. Constructs were induced with 100
ng/ml aTc (indicated with gray/dark bars). Unfilled bars represent g/k mRNA while hatched bars represent
g/k mRNA and asRNA levels. Uninduced cultures are represented by the white/lighter bars. All cultures
grown in 1.5% (w/v) glucose supplemented M9 medium. Unless noted otherwise, constructs were tested
in KTS622. Graphs in A, B and D depict the duplicate mean ± SD of a representative experiment.
Given the weak correlation between g/k mRNA and Glk activity in R953, we sought to
identify modes of feedback that could potentially explain the discrepancy. In so doing, we noted
that the antisense performance was sensitive to medium composition and transient under
certain conditions (Figure 3-4). In minimal medium supplemented with low levels of glucose
50
CHAPTER 3 |A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
(0.4% w/v), the growth trends are consistent with those grown in 1.5% glucose (Figure 3-3B);
however, activity differences for the active asRNA constructs are barely distinguishable from
noise and disappear over time (Figure 3-4A). When grown in minimal medium supplemented
with glycerol, a clear activity trend is observed well above background that persists with time
(Figure 3-4B).
The literature presented no known genetic mechanisms to explain these
observations; however, we recognized that the coupling between Glk activity and specific
growth rate in glucose, but not glycerol, supplemented medium (§2.2.2), may offer a potential
explanation. Glk, like most bacterial proteins, is fairly stable (143) and essentially only diluted by
growth. This stability passively generates a negative feedback mechanism (125), which we have
termed growth mediated buffering, to somewhat regulate carbon flux and Glk activity. For a
fixed expression level of g/k, Glk activity, specific growth rate and carbon flux are in equilibrium.
If the system is perturbed with an increase in g/k expression, however, Glk activity, carbon flux
and specific growth rate increase. This increase in specific growth rate dilutes Glk activity,
reducing carbon flux, and buffers the cell against large perturbations in carbon flux. Conversely,
lowering g/k expression would decrease the growth rate and concentrate Glk to minimize the
impact on carbon flux. This effect is further convoluted by feedback inhibition of PEP on Glk
(144) and metabolite efficiency (145). Based on these facts and the evidence gathered, we
propose the following model.
asRNA trends
observed
In glycerol, Glk activity is decoupled from cell physiology and
represent their true
potential.
At
high
levels
of glucose
supplementation, carbon flux through the cell is high and the in vivo activity of Glk is inhibited
below its total activity by PEP. asRNA modulation of Glk leads to reduced carbon flow through
glycolysis.
This in turn leads to a reduction of PEP repression and increased metabolite
a in vivo activity is a strong function of metabolite concentration as described by traditional MichaelisMenten kinetics; reduced total enzyme increases the substrate metabolite pool and specific activity of the
remaining enzyme minimizing changes in flux
51
3.2.2 Effect of asRNA expression on product yields
efficiency. However, this is insufficient to counter the gross reduction in flux leading to changes
in the observed specific growth rate, growth mediated buffering and a diminished trend in Glk
For low levels of glucose supplementation, PEP regulation and metabolite
total activity.
efficiency is minimal due to the reduced carbon flux and thus glycolytic flux changes are much
larger than in the high glucose case. This leads to larger growth rate changes and buffering
which concentrate Glk to a point where changes in expression are no longer perceptible (Figure
3-4A). Thus, we offer that the coupling of growth with Glk activity leads to emergent behavior
that regulates carbon flux and Glk activity regardless of g/k expression levels (see §4.3.2 for a
more quantitative discussion of the potential impact).
B
A
0.3 -03
-
0.2-
02
DI
0.04
0.0
R100
(
Construct
R100
)
Construct
Figure 3-41 Effects of medium composition on antisense performance
Performance of an empty vector (pKVS45) and R100 in A) M9(0.4% glucose) and B) M9(0.4% glycerol) at
20 h (no fill pattern) and 48 h (hatched bars). Uninduced samples are white and samples induced with
100 ng/ml aTc represented by gray. Figure represents the mean ± SD of duplicate 50 ml flasks of KTS622
at 37 *C.
3.2.2.
Effect of asRNA expression on product yields
The ability of these asRNA constructs to redirect carbon towards a competing
heterologous pathway was investigated next. The one step conversion of glucose to gluconate,
described previously (§2.3.3), was used. Given the sensitivity of gluconate production to cell
physiology (§2.3.3) we first evaluated the ability of the system to tolerate multiple plasmids
encoding for our pathway of interest and the asRNA construct itself. As an initial study, pTrc-
52
CHAPTER
3| A
DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
99Cm-Gdh and the parental pKVS45 were tested in four mutants from the viable glucokinase
The presence of the additional plasmid was not necessarily deleterious as
library (§2.3.3).
trends with induction level generally agreed with those reported previously (§2.2.3).
At an
induction level of 25 iM IPTG, production of gluconate from the lowest activity Glk strain
(KTS8221G) was observed to increase by at least 300% to titers in excess of 0.5 g/L (Table 3-1)
when compared to a single plasmid context (Table 2-2). More modest changes were noted for
the remaining strains. Evaluating the conversion of glucose into gluconate derived species, we
observed a direct monotonic correlation between Glk activity and gluconate molar yield (Figure
3-5) confirming the observation that Glk may be used to control productivity of a heterologous
pathway (§2.3.3).
However, as reported previously, this effect was highly sensitive to
parameters such as IPTG induction level (see §4.3.1).
Table 3-1| Titers of the major fermentation products as a function of Glk expression.
Final OD600, and titers of gluconate (Gnt), its reduced form (5-ketogluconate or 5KG), acetate and glucose
consumed at 72 h. Average values of triplicate cultures are reported with standard deviations given in
parentheses. Flasks were grown in M9 (1.5% (w/v) glucose) and induced with 25 p.M IPTG. All cultures
contain a plasmid for Gdh expression and an empty vector plasmid as surrogate for the antisense valve.
Titers
Strain
KTS8221G
KTS6221G
KTSO221G
KTS9221G
Final OD 600
Estimatedb
Glk Activity
0.24
0.98
1.01
1.29
[U/mg]
0.23 (0.04)
0.30 (0.02)
0.33 (0.02)
0.50 (0.03)
(0.04)
(0.09)
(0.08)
(0.13)
Gnt
[g/L]
0.53
3.33
3.32
2.25
(0.03)
(0.10)
(0.03)
(0.40)
5KG
[g/L]
Acetate
[g/L]
0.03 (0.01)
0.17 (0.01)
0.19 (0.01)
0.20(0.12)
ND
0.32 (0.14)
0.28 (0.08)
0.82 (0.09)
Glucose
Consumed
[g/L]
0.87 (0.27)
7.45 (0.27)
7.05 (0.09)
7.34 (0.25)
"Mutants spanning roughly 0.5-2x wildtype (KTSO221G) growth rates (§2.2.2)
bDue to similar assay chemistries, Glk cannot be measured in the presence of Gdh. Glk activity values
reported are in the absence of any plasmid.
53
3.2.2 Effect of asRNA expression on product yields
0.7
0.6
00.5
±0.4
~50.3
0.2
0.1
0.0
0.1
0.2
0.3
0.4
0.5
0.6
Estimated Gik Activity (U/mg)
Figure 3-51 Potential impact of valve on productivity at 72 h
Gluconate molar yield plotted as a function of Glk activity in cells bearing a plasmid for Gdh expression
and an empty vector plasmid (pKVS45) as surrogate for the antisense valve. Graph depicts the triplicate
mean ± SD of a representative experiment. Due to the similar assay chemistries, GIk activity cannot be
measured in the presence of Gdh.
Estimated Glk activities presented here represent those of the
KTSx221G parent strain (§2.2.3) in the absence of any plasmid.
Encouraged by these results, we then tested the ability of asRNA constructs R100 and
R953 to inducibly control the productivity of our test pathway. In static tests where the asRNA
was induced at inoculation, we were unable to reliably change pathway efficiency.
The
presence of either asRNA construct occasionally displayed some improved productivity over the
empty vector control under various conditions (Figure 3-6); induction, however, generally
provided no further improvement.
Similarly, testing the constructs in various hosts of the
glucokinase family resulted in inconclusive results (Figure 3-6). Only in a handful of experiments
did induction provide some benefit. The reproducibility and significance of this effect, however,
are unclear and may be a product of a transient asRNA response (Figure 3-4A).
We also
considered the possibility of dynamically inducing the asRNA but were unable to see any
changes in Glk activity unless the asRNA was induced at inoculation. From this, we concluded
that the small nominal dynamic range of the asRNA constructs coupled with the regulatory
54
CHAPTER 3 | A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
mechanisms discussed and other phenotypic changes can easily negate their effects making
them unsuitable for future engineering efforts.
A
B
0.7
0.7
0.6.
0.6
0T
0.5
0.5
Z0.4.
0.4
0.3-
0.3
S0.2-
2
0.1
0.1.
0
0..0
Emptyvector
contrei
0
R100
R953
Emptyvectorontrol
Construct
R100
R53
Construct
Figure 3-61 Effect of asRNA on gluconate yield
Molar yields of gluconate as a function of construct at 72 h and induced with 25 pM IPTG (grey bars) at
inoculation grown in 50 ml M9(1.5% glucose). Hosts are (A) KTS6221G and (B) KTS9221G. Empty vector
control - pKVS45. Data represent duplicate mean ± SD from a representative experiment.
3.2.3.
Inverting gene circuit as a dynamic glucose valve activator
Given the results with asRNA as a valve actuator, we opted to create a system with
better characterized parts, potential for a larger dynamic range and the desired input-output
characteristic. Based on these criteria, an inverting gene circuit, or inverter, was selected. First
demonstrated by Yokobayashi and coworkers (132), an inverter is a basic gene circuit analogous
to a NOT logic gate. In our design, TetR is constitutively expressed and regulates expression of
lacl. Lacl, in turn, controls expression of g/k. In response to aTc supplementation, TetR is
repressed driving expression of lacl which turns OFF g/k expression (Figure 3-7).
We constructed the inverter by harvesting the TetR expression cassette from pKVS45
and using its Tet promoter to drive expression of LVA-tagged Lacl (146). The lacl and tetR
cassettes were oriented opposite to gik to preclude read-through transcription.
degradation
tagged
variant of
Lacl from
the
Registry of Standard
The LVA-
Biological
Parts
(http://partsregistry.org) was selected to reduce steady state levels of Lacl in the system and
55
3.2.3 Inverting gene circuit as a dynamic glucose valve activator
minimize the deleterious effects of leaky lacl expression in the ON configuration. Similarly, no
transcriptional terminators were placed between lac/ and tetR to provide negative feedback for
leaky lac/ expression. To create a Lacl sensitive promoter, a constitutive promoter from the
Anderson Collection in the Registry was fused to a symmetric lacO operator (147) directly
downstream. The circuit was insulated from the genome with a transcriptional terminator from
the Registry placed downstream of tetR. The resulting construct was then integrated into the
genome directly upstream of g/k, replacing the FruR binding site through the primary native
promoter (P1) (102, 103) (Figure 3-7).
S®r
aTc
%
J61048
Figure 3-71 Genetic inverter as a glucose valve
Schematic of genetic inverter (NVS9221G) with a more detailed description of each module. Alphanumeric
codes refer to part numbers within the Registry of Standard Biological Parts (http://partsregistry.org).
J61048 - transcription terminator; C0012 - lacl fused with an LVA degradation tag; B0034 - a ribosome
binding site; J23114 - Anderson promoter library variant (~2x native g/k promoter). TetR cassette refers
to the TetR ORF and promoters from pKVS45 while Ptet is the inducible Tet promoter also found in pKVS45.
Two main variants of the inverter were created: NVS922 and its derivative NVS9221G
(Gnt~), and NVS622 and its derivative NVS622IG (Gnt-). NVS922 was constructed with J23114
(http://partsregistry.org), a promoter corresponding to roughly twice the native g/k promoter
strength (§2.2.2).
56
Similarly, NVS622 was designed with J23117 (http://partsregistry.org), a
CHAPTER 3 | A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
promoter equivalent to the native g/k promoter in strength (§2.2.2). NVS622 (and its derivative
NVS6221G), however, ultimately proved non-viable in minimal medium supplemented with
glucose.
NVS9221G was initially characterized by supplementing cultures with either IPTG
and/or aTc at inoculation and measuring the Glk activity 20 h later in mid-exponential phase
(Figure 3-8). aTc was seen to suppress Glk activity in a dose dependent manner, demonstrating
a successful inversion response. Moreover, IPTG and aTc supplementation was able to recover
Glk activity, confirming the mechanism of action. IPTG induction alone, however, could increase
the ON state expression of Glk by ~40% (Figure 3-8). This evidence indicates significant leaky
lac/ expression is present which may be sufficient to suppress g/k expression in the ON state to
such a degree that NVS622 is unable to attain the Glk levels necessary to sustain growth. IPTG
was also noted to be sufficient to negate the inverting behavior of the aTc induced inverter
(Figure 3-8).
1.60
1.40
1.20
7B1.00
13 0.80
0.60
0.40
0.20
0
100
0
5
10
0
ng/mI aTc
100
100
100
PM IPTG
Figure 3-81 Initial characterization of a genetic inverter
Glk activity of NVS9221G as a function of aTc and IPTG induction level at 20 h in 0.4% (w/v) glycerol
supplemented minimal medium. Activity is relative to an uninduced control.
NVS9221G was characterized more thoroughly in triplicate by growing cells in minimal
medium supplemented with 0.4% (w/v) glycerol and inducing with aTc at one of 6 levels at
inoculation.
After 27 h the cells were at approximately steady state (mid-late exponential
57
3.2.3 Inverting gene circuit as a dynamic glucose valve activator
phase, OD600 0.5-0.9), with the exception of the highest induction level (OD 600 0.1). At this
time samples were taken and assayed for Glk activity and g/k mRNA abundance. The ON, or
uninduced state, had an activity of 0.29 ± 0.01 U/mg, which was significantly lower than the
same promoter in isolation (0.42 U/mg ± 0.02 U/mg - §2.2.2) and is likely attributable to leaky
lacl expression.
The inverter, nonetheless, was observed to control Glk activity in a dose-
dependent manner with a canonical sigmoidal relationship.
achieved at approximately ~100 ng/ml aTc (Figure 3-9).
Full suppression of Glk was
Evaluating the g/k mRNA abundance
relative to the uninduced state, a similar dose-dependent sigmoidal relationship with up to 80%
repression in mRNA levels was observed (Figure 3-9).
1.8
1.6
0.3 -
1.4
0.21.2
tM
0 .2
, , ,
,
,
1 .0
.,
z
E
0.8
0.6
Cu
0.1-A
0.4
0.2
0.0..........................0.0
0.01
0.1
1
10
100
1000
aTc (ng/ml)
Figure 3-91 Genetic inverter performance
Performance of NVS9221G at the protein (filled squares) and RNA (open triangle) levels as a function of
aTc in 0.4% (w/v) glycerol supplemented minimal medium at 27 h. RNA levels are relative to an
uninduced control. Unbroken and dashed lines represent a best fit to a Hill function of the activity and
mRNA data, correspondingly. Uninduced specific activity and relative mRNA are 0.29 ± 0.01 U/mg and
1.06 ± 0.08, respectively. Graph depicts the triplicate mean ± SD.
Dynamic control with NVS9221G and its performance under more relevant glycolytic
growth conditions were evaluated in minimal medium supplemented with 1.5% (w/v) glucose.
Triplicate flasks were inoculated with cells. At early-exponential phase (OD~0.25) each flask was
split into two equal volumes with a flask from each pair induced with 100 ng/ml aTc (Figure
3-10A). The OD600, Glk activity and g/k mRNA of these flasks were tracked over time. As a
58
CHAPTER 3 | A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
control, another flask was grown and induced at inoculation.
Uninduced flasks had specific
growth rates and activities comparable to native levels in glucose supplemented medium (p ~
0.3 h-; activity ~ 0.23 U/mg).
low/inhibited g/k expression.
As anticipated, the induced control failed to grow due to
In the induced split flasks, however, there is an immediate
divergence in specific growth rate from the uninduced flasks with growth persisting for at least
30 h post induction (Figure 3-10B). As noted in Chapter 2, control of Glk in this manner offers
true dynamic control of glucose flux and macroscopic behavior such as specific growth rate,
unlike control at nodes further downstream in glycolysis (86). Evaluating the relative changes of
the induced to uninduced flasks across the macroscopic, protein and RNA levels, a consistent
steady state decrease of ~ 50% is seen (Figure 3-10C). Despite the uniformity of performance
across levels, however, the dynamic performance of the inverter at the protein level is
inconsistent with the behavior noted in glycerol supplemented medium (Figure 3-9) or even
glucose supplemented medium when it is induced at inoculation (Figure 3-10B).
Given these
earlier studies, it was anticipated that induction of the inverter would fully arrest growth.
However, this inconsistency may be understood in light of the growth mediated buffering
presented earlier (§3.2.1): the reduced specific growth rate that accompanies inverter induction
concentrates Glk to a level that allows growth to persist.
Changes are noticeably faster at the protein level, with the system reaching steady state
within 5 h, before being propagated to the macroscopic level (tsteady
state
>20 h). This response
time at the protein level is consistent with theoretical predictions of Glk degradation primarily
due to dilution by growth. With transcription rates and mRNA lifetimes typically on the order of
minutes (111, 148), translation rates of lac/ on the order of minutes (111) and no active
degradation of Glk, the dynamics of Glk levels are anticipated to be controlled by dilution due to
growth. Indeed, analyzing the dynamic protein data an exponential decay time constant of 3.74
59
3.2.4 Effect of inverter expression on product yields
± 1.91 h was obtained which yields a protein half-life (2.59 ± 1.33 h) approaching the doubling
time of uninduced cells (2.23 h).
A
B
100nWMaTC
3.0-
V
V
V
4
2.0
A
A
10
A
0.5
A 0.0
Culture Time
I
-
10
0
20
-
30
Time post induction (h)
20-
20
C20-
to.
to
to.
16.
14.
10
08
to.
10
08.
08
06.
06.
04.
04
04
02-
02
02
12.
*
0
000
0
ino N0
1o
ia
30
rpm (N
t
0
i0
10
30
Titl (h)
Figure 3-101 Dynamic performance of inverter as a glucose valve
(A) Experimental scheme. (B) Optical density of split flasks post induction (open squares - uninduced,
filled squares - induced) and a representative control culture induced at inoculation (filled triangles). (C)
Relative performance of induced:uninduced across macroscopic (growth), protein and mRNA levels. The
graph in B depicts the triplicate mean ± SD with the exception of the representative control which
represents a single flask. The graph in C represents the triplicate mean ± SEM. Cultures were induced
with 100 ng/ml aTc.
3.2.4. Effect of inverter expression on product yields
Finally, we evaluated the potential of our system to redirect carbon into gluconate
production. As the inverter is IPTG responsive (Figure 3-8), pBAD30 (120) rather than the IPTG
inducible pTrc99Cm (§2.2.3 and §3.2.2) was used to express Gdh. Given the long culture times
of these experiments, the delayed induction response of pBAD30 (Figure 3-2) was not
anticipated to be problematic.
NVS9221G
was inoculated in triplicate in M9 medium
supplemented with glucose and Gdh induced with L-arabinose at inoculation. Cells were grown
60
CHAPTER 3 | A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
for at least 19h into early exponential phase before the inverter was triggered with 100 ng/ml
aTc. Titers of major fermentation products and glucose were assayed at 72 h.
Inducing the
inverter at an OD 600 ~ 0.3, there was a 30% improvement in gluconate titers from 6.14 g/l to
8.04 g/l, a 40% reduction in acetate titers from 0.41 g/I to 0.24 g/I and a slight increase in
glucose consumption from 7.3 g/I to 8.2 g/L with negligible change in final biomass
accumulation relative to uninduced controls
(Figure
3-11A).
Normalizing by glucose
consumption on a molar basis, this translates to an 18% improvement in gluconate molar yield
(0.70 mol/mol Glu to 0.83 mol/mol Glu) and a 46% reduction in carbon wastage to acetate (0.14
mol/mol Glu to 0.08 mol/mol Glu) (Figure 3-11B). This improvement was a strong function of
OD 600 at induction (Figure 3-11C). The effect on both gluconate and acetate yields diminished as
induction OD 600 increased and was non-existent for OD 600 >1.0. Such behavior is consistent with
Glk degradation being primarily controlled by dilution due to growth. At early OD600, cells have
sufficient generations to wash out Glk when the inverter is switched to the OFF state, allowing
for the redirection of carbon and yield improvements of the model pathway. As the number of
doublings until stationary phase decreases, delayed induction of the inverter decreases its
dynamic range resulting in reduced carbon redirection and smaller observed improvements.
More complex behavior may potentially be observed in a system with destabilized Glk.
However, care must be taken in the design of these systems as protein degradation is energy
dependent and may introduce unforeseen effects when coupled with reduced carbon flux.
Moreover, as evidenced with preliminary studies with an ASV degradation tag (146) (data not
shown), the inverter ON state will need to be retuned for viable expression.
61
3.2.4 Effect of inverter expression on product yields
A
B
Gkocose
consumed
Gluconete
SKG
Acetate
10-
6-
Acetate
0.8-
2-
0
Gnt + 5KG
1.0-
0.6-
E5
4-
0.4-
2-
0.2-
0(+)
0-
Inverter induction
Inverter Induction
C
1.4
1.2-
~LH
H+~H
-
I~I
(D
M
F-i--I
0.8-
I
I
0.60.4-
4
0.20.0
1
0.0
1
0.2
0.4
r-
0.6
1
1
0.8
.1
1.0
1
I
1.2
1.4
1.6
Induction OD
Figure 3-11l Effect of inverter on gluconate yield
(A) Titers of major fermentation products and (B) Molar yields of gluconate and acetate of a
representative experiment at 72 h for an uninduced control and inverter induced at early exponential
phase (OD ~0.3). (C) Molar yield of acetate (triangles) and gluconate (squares) at 72 h of induced inverter
relative to an uninduced control as a function of OD at induction. Data from independent experiments
are distinguished by fill pattern. In all figures, Gdh was expressed with 0.02% (w/v) L-Ara. The inverter
was induced with 100 ng/ml aTc. Data represent the triplicate mean ± SD.
The expression of Gdh as a function of inverter induction level and timing (OD 60o) was
also evaluated. At early times (~24 h), there was no change in the amount of Gdh measured.
However, for a number of cases at 72 h, the induced inverter had a noticeable increase in Gdh
activity ( 50%) (data not shown), presumably a consequence of the reduced dilution due to
growth. Nonetheless, for the earliest OD600 at which the inverter was induced (~0.1), where this
62
CHAPTER 3
IA DYNAMIC
METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
effect is likely to be greatest, there was no observed increase in Gdh expression (Figure 3-12)
suggesting more complex mechanisms governing expression levels. Moreover, there were still
improvements in gluconate and acetate yields (Figure 3-11) suggesting that the effects observed
are not solely influenced by Gdh expression.
That is, the improvements observed are
attributable in part to carbon flux redirection.
1210 -
-C 8E
6-
0
OFF
CN
Inverter State
Figure 3-12| Gdh activity as a function of Inverter state.
Gdh activity of inverter at 72 h when uninduced (ON) and induced with 100 ng/ml aTc at OD600
(OFF). Data represents the triplicate mean ± SD of the cultures depicted in Figure 9C (no fill).
0.1
3.3.Discussion
The optimization of metabolic phenotypes frequently requires the rerouting of
metabolite flux towards the pathway of interest. While heavily implemented for secondary
metabolites, the engineered diversion of carbon from central carbon metabolism has until
recently been relatively unexplored. Of late, groups have begun to explore the effects of flux
redirection through glycolytic nodes such as
glyceraldehyde-3-phosphate (86), glucose 6-
phosphate and 6-phosphogluconolactone (85).
However, none has sought to address the
feasibility of using glucose itself as a substrate or its impact on cell physiology. In this work and
its companion study (Chapter 2), we have demonstrated the ability to control glucose
distribution through the modulation of Glk and evaluated its impact on cell health.
63
3.2.4 Effect of inverter expression on product yields
In order to redirect glucose flux, E. coli needed to be reengineered to generate a control
node through Glk. This modification made Glk essential for cell growth. Moreover, the direct
changes in carbon flux through Glk had the potential to alter the phenotype of the cell.
However, unlike similar studies where cellular phenotype was directly controlled, only specific
growth rates and not biomass accumulation was affected (86, 106). These growth rate changes,
nonetheless, were accompanied by global changes in protein expression (125, 126) and
metabolite pools which form negative feedback mechanisms to counter the induced changes
(125, 145). The control of central carbon flux also reduced the resources available for the cell to
support heterologous expression (127).
We thus pursued a dynamic control strategy to mitigate these negative consequences by
reducing the culture time over which they operate.
asRNA and inverting gene circuits as
orthogonal engineered control systems were explored. With the designed asRNA constructs, we
achieved limited success in inhibiting Glk activity by up to 25%. These changes, however, did
not translate into controllable gains in pathway productivity.
In contrast, the wealth of
community knowledge in an easily curated form allowed us to rapidly design and prototype a
functional inverter.
This design was able to reproducibly inhibit growth by up to 50% in a
dynamic fashion with up to 20% increases in pathway efficiency and almost 50% reduction in
carbon wasting to acetate. These process gains were not accompanied by limited recombinant
protein expression or reduced biomass accumulation as seen in the static control cases despite
similar reductions in Glk activity (§2.2.3, cf. KTSO221G/KTS8221G). These results represent one of
the first examples of the redirection of glucose away from central carbon metabolism in a
dynamic fashion.
In this manner, we were not only able to demonstrate the potential to
improve yield and productivity of a heterologous pathway with a dynamic valve, but also the
ability to reduce carbon waste to acetate and potentially alleviate the demands of pH control in
64
CHAPTER 3 |A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
bioprocessing.
Furthermore, the redirection of glucose away from central metabolism,
especially in a dynamic fashion, enables the generation of new classes of pathways where
glucose may be used directly as a substrate.
3.4.Materials and Methods
3.4.1.
Strains and plasmids
E. coli strains and plasmids used in this chapter are listed in Table 3-2. All molecular
biology manipulations were carried out according to standard practices (114).
Molecular
cloning and vector propagation were completed in E. coli DH1OB (Invitrogen/Life Technologies,
Carlsbad, CA). Antisense control of glycolytic flux was tested in the PTS Glu* mutants KTS622
(§2.2.2), MGER3 and MKTS3. MGER3 and MKTS3 were constructed using X-Red recombination
(105) from MG1655 (DE3).
Primers as indicated in Table 3-3 were used to generate linear DNA
cassettes for recombination from colony PCRs of kan variants of KTS022 intermediates (§2.2.1).
Reporter gene assays were conducted in the PTS~ Glu* mutant KTS022 (§2.2.2).
Gluconate
production in strains bearing multiple plasmids was tested in KTSO221G, KTS6221G, KTS8221G and
KTS9221G of the glucokinase expression family (§2.2.3). A medium-copy tet-inducible vector for
antisense induction was created by PCR amplifying the pBAD30 backbone (120) (ATCC,
Manassas, VA) and TetR cassette of pWW308 (149) (provided by John Deuber, UC Berkeley), and
ligating them together at introduced Avrll and Xhol sites.
sGFP (141) was subcloned from
pTrcsGFP (150) (provided by Christine Santos, Stephanopoulos Lab) into pKVS45 and pBAD30
using Hindlll and Kpnl restriction sites. Antisense constructs were amplified with HotStarTaq
(Qiagen, Valencia, CA)from DH10B genomic DNA and cloned into pBAD30 using the introduced
Sall and EcoRI cloning sites. These constructs were later subcloned into pKVS45 using the same
sites. Subsequent sequencing revealed the constructs had more than 99% complementarity
65
3.4.1 Strains and plasmids
with the target mRNA except a single point mutation in each construct except R756. These
point mutations were unique to each construct.
Gdh was subcloned from pTrc99Cm-Gdh
(§2.2.3) into pBAD30 using Xbal and Hindill restriction sites. Mutants were verified by colony
PCR and selection markers cured. Unless otherwise noted, PCR amplifications were performed
with Phusion High-Fidelity DNA Polymerase (NEB, Ipswich, MA) and oligonucleotides from
Sigma-Genosys (St. Louis, MO; Table 3-3). Restriction enzymes and T4 ligase were also obtained
from NEB.
Table 3-21 Main strains and plasmids used
Relevant Genotype
Name
E. coli strains
Source
MG1655(DE3)
F mcrA A(mrr-hsdRMS-mcrBC) p80facZLM15 LlacX74
recAl endAl araD139 A(ara, leu)7697 galU galK A rpsL
nupG
(F- ilvG rfb-50 rph-1 DE3)
MG1655(DE3 endA recA)
MG1655 (DE3)
AendA ArecA
MKTS3
MG1655 (DE3)
AptsH/crr galP
MGER3
MKTS3 zendA
Electromax DH10B
Invitrogen
(3)
(151)
This study
This study
KTSO22
ArecA
DH10B AptsH/crr galP4
KTS622
KTSO22 Pgtk::J23117
Chapter2
KTSx221G family
DH10lB AptsH/crr galP4 Pgik: :Pca
NVS622|G
NVS9221G
Chapter 2
galP4AidnK AgntK
D1H101B AptsHicrr galP4 Pglk::(tetR Pet lacl-LVA J23117
lacO) AidnKAgntK
DH 101B AptsHlcrr galP4 Pglk::(tetR Ptet lacI-LVA J23114
lacO) AidnKAgntK
Chapter 2
This study
This study
Plasmids
pCP20
k c1857 (ts),A Pr Rep t , bla(AmpR), Cat(CmR), k pr FLP
pTrc99ACm-gdh
ColE1(pBR322) ori, cat (CmR), laC, Ptrc gdh
pBAD30
p15A, bla(Am pR) tetR
pBAD30-gdh
p15A, bla (Am pR) araC, ParaB gdh
This study
pBAD30-sGFP
p15A, bla (Am pR) araC, ParaB sGFP
This study
pKVS45
p15A, bla(Am pR) tetR
This study
pKVS45-sGFP
p15A, bla(Amp R) tetR Ptet sGFP
This study
pBAD30-Rxxx
p15A, bla (AmpR) araC, ParaB RXXX (asRNA)
This study
pKVS45-Rxxx
p15A, bla(Am pR) tetR Pret RXXX (asRNA)
This study
pTrcsGFP
pTrcHis2B sGFP
Chapter 2
(120)
(150)
t
pT KRE D
oriR101, repA101 s aadA (SpecR), araC, Piac AriApAexo',
lacl, Piac I-Scel
(152)
PTKIP-neo
pMB1, b/a(Am pR) neo (KanR)
p15A, cat ( Cm ), Piaciq tetA
(152)
pTKS/CS
con = Anderson library promoter
66
CGSC #7629
(152)
--I
w
0
Name
pKVS45 construction
TetR_F
TetRR2
pB30_F
pB30-R2
Antisense RNA construction
g/k as rev -20
g/k as for 100
gik as for 500
gik as for 756
gik as for 953
Sequence 5'->3'
CTATCTCGAGTTAAGACCCACTTTCACATTTAAGTTG
CTATCCTAGGGTGCTCAGTATCTCTATCACTGATAGG
CTATCTCGAGGATAAGCTGTCAAACATGAGCAG
CTATCCTAGGTTTCTCCATACCCGTTFTTTTG
0
0
n
H
0
M
C
CTATGTCGAC GATATCTTTAGCGGAGCAGTTGAAGA
CTATGAATTCAATAGGTCTTAGCCTGCGAG
CTATGAATTCTATTCGGCGCAAAATCAAC
CTATGAATTCGAGATTGAGCGCCAGATTG
CTATGAATTCCCTAAGGTCTGGCGTAAATG
(A
0
Inverter cloning
TetR5'Hindl/
TetR3'EcoR/I
KTS9/ac/5'Kpn/
KTS6Iac/5' Kpnl
lac/3'Hind///
CTATAAGCTTTCGACTCTAGTGGATCTGCTA
CTATGAATTCTTTCTTTTGGGTATAGCGTC
CTATGGTACCTGTGGAATTGTGAGCGCTCACAATTCCACAGCTAGCATTGTACCTAGGACTGAGCTAGCCATAAACTAAGAAACCATTATTATCATGACATTAAC
CTATGGTACCTGTGGAATTGTGAGCGCTCACAATTCCACAGCTAGCACAATCCCTAGGACTGAGCTAGCTGTCAACTAAGAAACCATTATTATCATGACATTAAC
CTATAAGCTTTTATTAAGCTACTAAAGCGTAGTTTTCG
Inverter Integration
GlkLP2X
Glk LP1Xv2
CTATTCCTTATGCGGGGTCAGATACTTAGTTTGCCCAGCTTGCAAAAAGGCATCGCTGCAATTGGTTTGGCTTCAGGGATGAG
CACATCACCGACTAATGCATACTTTGTCATTCTTCAACTGCTCCGCTAAAGTCAAAATAATTCTT1TACGGCCCCAAGGT
MKTS/MGER3 construction
002check_F (pts)
002checkR (pts)
022check_F (gaIP)
022check_R (gaIP)
H
m
0
z
0
0
GCTAACAATACAGGCTAAAGTCGAAC
CCGATGGGCGCC
AATTATTTTCATGCACTTAAATC
GTAATCTGGAATTCATCTGC
Cm
z
H
Restriction sites for cloning are underlined. Homologous sequences for recombination are italicized.
Introduced mutations are bolded (promoter) or bolded & doubled underlined (jasg).
0)
0
3.4.2 Inverter construction
3.4.2.
Inverter construction
tetR and lac/ cassettes as described in Section 3.2.3 were synthesized by GenScript
(Piscataway,
NJ).
The
GTCGACTCGACTCTAGTGG
transcriptional
tetR
cassette
through the
terminator
from
comprised
Xhol site
the
the
region
spanning
inclusive of pKVS45
Registry
of
Standard
bases
5'
with J61048, a
Biological
Parts
(http://partsregistry.org), fused directly to the 3' end. The lac/ cassette consisted of Ptet from
pKVS45 (bases spanning 5' - CTAAGAAACCATTATT through 5' - ATACTGAGCACCTAGGT inclusive)
directly fused to B0034, a ribosome binding site, lacl-LVA (C0012) and B0015, a transcriptional
terminator, all from the Registry. To facilitate modular construction of several inverter designs,
the tetR and lac/ cassettes were then amplified with the primers indicated in Table 3-3 and
assembled with standard cloning techniques. This strategy allowed us to directly incorporate
the promoters used to drive g/k expression and the lacO operator site into the primers for lac/
amplification. The final designs were generated without the terminator B0015 and used either
the promoter J23114 or J23117 (http://partsregistry.org) for expression of g/k as discussed in
§3.2.3. Assembled constructs were cloned into the plasmid pTKIP-neo (152).
Final constructs were integrated into the genome of KTSO22 and KTSO221G (Gnt~) (§2.2.1
and §2.2.3) directly upstream of g/k replacing the FruR binding site through the primary
promoter using the method of Kuhlman and Cox and helper plasmids pTKRED, pTKS/CS and
pCP20 (152). Briefly, a TetA landing pad, amplified from pTKS/CS, was integrated into the 5'UTR
of g/k in KTSO22 and KTSO221G with pTKRED.
Successful integrants were selected for with
tetracycline and verified by colony PCR. pTKIP-neo plasmids containing the completed inverters
were then used to transform KTS022 and KTSO221G bearing the landing pad and pTKRED.
Homing endonuclease, 1-Scel, and X-Red recombinases were expressed from pTKRED to
integrate the completed inverter. The integrated design was selected for with kanamycin. FLP
68
CHAPTER 3 A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
recombinase expressed from pCP20 (105) was used to cure the neo marker. Finally, all helper
plasmids were cured from the final constructs of NVS622 (NVS6221G - Gnt-) and NVS922
(NVS9221G - Gnt), (J23117 and J23114, respectively). All mutants were identified and isolated
by colony PCR with appropriate primers. The final integrated constructs were further verified
with sequencing of colony PCR products.
3.4.3.
Culture conditions
Strains were propagated in Luria-Bertani (LB) medium at 37*C for daily manipulations
(30"C for intermediate recombination steps). Temperature sensitive plasmids were cured at
420 C on non-selective medium. All experimental cultures were grown at 37 0C in M9 minimal
medium supplemented with 0.8 mM L- leucine and either glucose or glycerol as indicated. For
all experiments, starter cultures were grown to an OD measured at a wavelength of 600 nm
(OD60 0 ) of at least 0.2 before being transferred to a 50 ml flask of the same media.
For
experiments involving pTrc99Cm-Gdh, IPTG was added as indicated to both the starter culture
and 50 ml flasks at inoculation. For gluconate experiments involving pBAD30-Gdh, 0.02% Larabinose was added to the 50 ml flasks at inoculation. pBAD30-sGFP and pKVS45-sGFP were
induced with 0.2% L-arabinose and 100 ng/ml aTc respectively at inoculation of the 50 ml flasks
for the reporter gene assay. In enzyme activity and qRT-PCR studies, 50 ml shake flasks were
inoculated to a final OD600 of ~0.001. 5 ml samples were taken periodically as indicated. These
samples were then pelleted and stored at -20"C until assayed. For gluconate experiments, shake
flasks were inoculated to a final OD600 of ~0.005 with M9 medium supplemented with 1.5%
(w/v) glucose and monitored for 72 h before samples were taken and clarified by centrifugation.
The supernatant was then stored at 40C until HPLC analysis. As appropriate, the antibiotics
ampicillin,
kanamycin, chloramphenicol,
spectinomycin
and
tetracycline were
used
at
concentrations of 100, 25, 34, 100, and 10 ig/ml, respectively.
69
3.4.4 Enzyme activity levels
3.4.4. Enzyme activity levels
All enzyme activity assays were performed on crude lysates generated by sonication of 5
ml frozen cell pellets resuspended in 500 pl of 10 mM Tris-HCI (pH=8.0) at ~ 8 W in 5 x 5 s pulses
on ice. The resulting lysates were clarified by centrifugation before being assayed for either Glk
or Gdh activity as described previously (§2.4.4).
In brief, one unit of Glk activity will
phosphorylate 1.0 ptmole/min of D-glucose at pH 7.5 and room temperature in the presence of
3.33 U/ml G6PDH, 60 mM Tris-HCI, 20 mM magnesium chloride, 8.0 mM ATP, 12.0 mM glucose
and 0.9 mM NADP*. Similarly, one unit of Gdh activity oxidizes 1.0 ptmole/min of D-glucose at
pH 7.6 at room temperature in 60 mM potassium phosphate buffer and 0.67 mM NAD*. In both
assays, reaction progress is measured at a wavelength of 340 nm. Due to the assay chemistry,
Glk activity cannot be measured independently of Gdh activity. Enzyme activities reported are
normalized by total protein levels as measured in a modified Bradford Assay (124) .
3.4.5. Reporter gene assay
sGFP fluorescence was measured on a Thermo Scientific VarioSkan Flash plate reader
(provided by Prof. Narendra Maheshri). 100 Iil of culture was excited with light at a wavelength
of 488 ± 5 nm before being measured at 530 ± 5 nm. OD600 was also measured on the plate
reader to normalize the fluorescence data.
3.4.6.
Quantification of mRNA levels
Total RNA was isolated from 5 ml of culture at the indicated times using the illustra
RNAspin Mini Isolation Kit (GE Healthcare Bio-Sciences, Piscataway, NJ) with an on-column
DNasel treatment according to the kit protocol. Using the QuantiTect Reverse Transcription Kit
(Qiagen, Valencia, CA) according to manufacturer's instructions, 0.5 jag of total RNA was treated
to remove trace DNA contamination before cDNA was synthesized with random primers. To
70
CHAPTER 3 |A DYNAMIC METABOLITE VALVE FOR THE CONTROL OF CENTRAL CARBON METABOLISM
quantify g/k mRNA, synthesized cDNA was amplified with primers g/k_disambigF2 (5'
TGCCGCATTTGAAGATAAAGG) and g/k disambigR (5'GATATAAAAGGAAGGATTTACAGAATGTGA)
at 300 nM final concentration.
g/k antisense + sense mRNA were quantified with primers
g/kR100_for (5' - TTGGTGCCGCCCACAT) and g/k R100 rev (5' - GCGGAGCAGTTGAAGAATGAC)
at 50 nM final concentration. qPCR reactions were performed with 3 ll of cDNA product in a 25
d reaction with Brilliant 11Sybr Green High ROX QPCR Mix (Agilent Technologies, Santa Clara,
CA) and the requisite primers on an ABI 7300 Real Time PCR System instrument (Applied
Biosystems, currently Life Technologies, Carlsbad, CA).
Transcript levels were quantified in
duplicate with appropriate no-template and no-RT controls and are relative to that of the
indicated control sample as determined from a standard curve.
Reported levels are the
averages of replicate flasks, each measured in duplicate.
3.4.7.
Metabolite analysis
Glucose, gluconate (Gnt), 5-ketogluconate (5KG) and acetate levels were measured by
HPLC analysis using an Aminex* HPX-87H anion exchange column (Bio-Rad Laboratories,
Hercules, CA) in an Agilent 1100 series HPLC instrument (Santa Clara, CA) as described
previously (§2.4.5).
71
CHAPTER 4
Glucose valve performance potential and
impact on physiology
ABSTRACT
In redirecting carbon flux by Glk modulation, several key features have been noted: high levels
of Glk expression are deleterious for growth, recombinant protein expression (and perhaps
other important cofactors) is tied to Glk expression, and the coupling of Glk activity and growth
provides negative feedback to resist Glk modulation. However, the mechanisms through which
these arise and the quantitative impact of such phenomena are unclear.
In this study,
stoichiometric and mass action kinetic models were employed to study them.
phenotypic features observed were recapitulated using these models.
The major
Furthermore, it was
demonstrated how some of these multifaceted responses may emerge from relatively minor
changes to Glk activity and their impact on a model heterologous production pathway is
quantified.
72
4.1 Introduction
4. 1. Introduction
Microbial production systems have extraordinary potential to help alleviate the energy,
health and material demands of the world (153). This promise, however, is tempered by the
technological issues that hinder economic viability and widespread adoption.
While these
systems are capable of producing stereospecific compounds (46, 154), structures (8) and
complex behaviors (133, 155), they are first and foremost optimized for survival and not the
requirements of human designers. Thus, a key challenge in the engineering of these systems is
the ability to redirect metabolites away from endogenous processes at will to accomplish set
design specifications.
One solution to this problem is to tune expression of endogenous processes as needed.
A number of tools have been developed to address this need (27, 51, 53, 132). However, until
recently, the performance of these tools and their global impact on cell physiology have not
been evaluated in more practical or demanding environments.
A number of researchers,
including ourselves, have begun to explore their application in the context of controlling
essential genes (85, 86, 106). As discussed in Chapters 2 & 3, we have studied this problem with
the goal of redirecting glucose from glycolysis. We have also taken the additional step of
evaluating its impact on a heterologous production pathway. Despite the successes achieved,
results point to a potential challenge of such tools: they alter the physiology of their hosts. This
has most notably taken the form of reduced biomass accumulation (86, 106) or altered specific
growth rates (85) (§2.2.2, §3.2.1 and §3.2.3). However, our experience also points to deeper
phenotypic changes affecting recombinant protein
expression and potentially cofactor
availability (§2.2.3). Thus, theoretical models were sought to explore the potential magnitude
and impact of such effects.
73
4.2 Effects of Glk on physiology: ATP Regeneration and Cell Health
Theoretical
models
have
enjoyed enormous success
in
metabolic engineering
applications. While rooted in simplified representations and basic precepts, models such as Flux
Balance Analysis (FBA) (156), Metabolic Control Analysis (MCA) (157), and mass action kinetics
are successful at recapitulating and predicting complex cellular behavior despite the intrinsic
stochastic nature of such processes (109, 110, 158, 159). In this chapter, similar stoichiometric
and mass action kinetic based approaches are used to evaluate the effect of Glk activity on
carbon flux and physiology. The impact of these perturbations on the potential of engineered
devices was also examined. While these were not implemented in a global fashion to study the
total sum of metabolism, observed cellular behavior was recapitulated. Through these models,
insights into the growth characteristics of cells and boundaries to the potential of glucose flux
manipulation are established.
4.2.Effects of Glk on physiology: ATP Regeneration and Cell Health
In Chapters 1-3, flux through glycolysis and Glk has been primarily viewed as the source
of physical resources the cell needs for both growth (biomass formation) and product formation.
However, its physiological purpose is much greater than the material; it is the mechanism by
which the cell oxidizes glucose and stores that electrochemical potential for endogenous
processes such as biosynthetic reactions, DNA replication, transcription and translation. Thus,
changes in Glk activity will necessitate changes in this energy supply which may hinder
endogenous processes, growth and development.
To evaluate the effect of Glk activity on ATP production, it was necessary to determine
the stoichiometric coefficient of ATP (x) on glucose from aerobic respiration as described by
Equation 1.
C6H12 0 6 + 6 02 -> 6 CO2 + 6 H20 (+ x ATP)
74
(1)
CHAPTER 4 | GLUCOSE VALVE PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
While aerobic respiration produces 4 ATP/glucose directly (Figure 4-1), the majority of
ATP generation is mediated through the reduced cofactors NADH and FADH 2 in the process
known as oxidative phosphorylation. Thus, the electron flow of glycolysis and the citric acid
cycle needed to be recapitulated (Figure 4-1). The reduced cofactor NADPH, also produced by
aerobic respiration and typically used for biosynthesis, is considered to be equivalent to NADH
in this analysis. To convert the electrochemical potential of the reduced cofactors, bacteria
typically use a two-step process.
First, the electron pairs of the reduced cofactor are
transported to an electron acceptor outside the cell through an electron transport chain.
aerobic fermentation, this transport is to
02
In
which is reduced to water. This electron transport
is coupled to the extrusion of protons from the cell generating a proton gradient of potential
energy. Then, this electrochemical gradient is dissipated by the re-entry of protons into the cell.
This physical movement of charge is coupled to the generation of ATP through membrane
bound ATP synthases (160).
The regulation and efficiency of these processes are highly environment dependent.
Proton extrusion has been measured to vary from 2-8 H*/NADH (161). Similarly, 2-4 H*/FADH 2
may be extrudedb (161).
This results in up to 88 H' being extruded per mole of glucose
consumed with 6 moles of oxygen being reduced. The amount of protons needed to generate
ATP is also highly variable.
Based on the free energy available from the proton gradient,
dependent on environmental conditions such as pH and temperature, a consensus value of ~3
H4 is nominally needed per ATP synthesized (160). Thus, each molecule of glucose can generate
up to 29.3 molecules of ATP from the reduced cofactors or a total of 33.3 moles ATP per mole
glucose with the consumption of 6 moles of molecular oxygen.
The electrons of FADH 2 enter the electron transport chain directly after the NADH dehydrogenase
lowering the proton extrusion count by 0-4 H*/FADH 2 .
b
75
4.2 Effects of Glk on physiology: ATP Regeneration and Cell Health
G
GIJON
ATP
LADP
Y
C
SS
0
Payba
phs
JG'"'''=2PGALD
1
CADP
DP
T
AD*
ATPD
NADH, CO,
acety CoA
N DH
NAD*
oxaloacetate
mlt
matatec
FADH,
fumarate
isocitrate
NADP*
Citric Acid Cycle
NADPH
FAD sucnte
oxalos cci.nate
ATP
ADP succinyl-C A
CO 2, NADH
a-ketoglutarate
CO 2
AD
Step
Net yield/glucose
Notes
Preparatory Phase
Payback Phase
Decarboxylation
Citric Acid Cycle
AEROBic RESPIRATION
- 2 ATP
4 ATP + 2 NADH
2 NADH
2 ATP + 4 NADH +2 FADH 2 + 2 NADPH
4 ATP + 8 NADH + 2 FADH 2 + 2NADPH
Glucose 4F6P
PGALD- pyruvate
Pyruvate ->acetyl-CoA
Citric Acid cycle
Glucose -CO
2
Figure 4-1I Overview of aerobic respiration
The major reactions and electron flow of glycolysis and the citric acid cycle with an accounting of the net
energy and electron flow. Abbreviations: GFP - glucose-6-phosphate; FGP - fructose-6-phosphate; FDP fructose-1,6- bispphosphate; PGALD - 3-phosphoglyceraldehyde; PEP - phosphoenolpyruvate; ATP adenosine triphosphate; NAD* - oxidized nicotinamide adenine dinucleotide; NADH - reduced
nicotinamide adenine dinucleotide; FAD - oxidized flavin adenine dinucleotide; FADH 2 - reduced flavin
adenine dinucleotide. Adapted from (160).
While cells maximizing ATP production would appear to be optimal, such ATP
stoichiometric coefficients are inconsistent with the experimental evidence collected to date
(110).
FBA models have been proposed with an objective of maximizing intracellular pools of
key metabolites for biomass formation or maximizing the flux towards them.
With such an
objective function, the stoichiometric coefficient of ATP drops to ~ 19 moles ATP per mole
76
CHAPTER 4 | GLUCOSE VALVE PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
glucose but provides much better agreement with experimentally measured biomass yields on
glucose (109, 110). Physically, this reduced efficiency may be realized by futile cycling of key
metabolites that dissipate energy to maintain metabolite pools (109, 162, 163). Accounting for
maintenance requirements such as macromolecule turnover and maintenance of membrane
potentials, a stoichiometric coefficient of 13-15 moles ATP/mole glucose is obtained which
corresponds well with experimental observations of biomass yield (110).
ATP production scales with glucose flux; however, there appears to be an upper bound
to the aerobic regeneration capacity as dictated by a respiration limit of ~ 16 mmol 0 2/h/ g dcw
(107). This limit has been suggested to be a biophysical constraint of the cell membrane being
only able to support a finite number of proton extrusion sites (108). Regardless of physiological
mechanism, however, taking a consumption rate of 6 moles 0 2/mole glucose and 14 moles ATP/
mole glucose net (15 moles ATP/mole glucose gross without accounting for Glk activity), a limit
of aerobic ATP production is established at ~40 mmol ATP/h/g dcw.
Having calculated an upper bound to the ATP synthesis rate, it was then necessary to
determine the amount of glucose processed by Glk and ATP consumed by Glk and endogenous
processes. Using experimentally measured in vitro total activities of Glk, the maximum glucose
flux as a function of Glk was determined as described in Equation 2:
F =60Aa
(2)
where F is the maximum flux of glucose through Glk ([=]mmol/h/ g dcw), A is the total specific in
vitro activity ([=]
U/mg total soluble (cytoplasmic) protein = ptmol glucose/min/mg total
cytoplasmic protein ), a is the fraction of dry cell weight that is cytoplasmic protein (0.38) (164).
Assuming 15 moles of ATP gross are produced per mole of glucose, this leads to an aerobic ATP
synthesis rate (P [=] mmol/h/g dcw) given by Equation 3:
77
4.2 Effects of Glk on physiology: ATP Regeneration and Cell Health
P = 900Aa;
A
0.117 U
P = 40;
A > 0.117 U
(
mg
(3)
mg
This result suggests that aerobic ATP generation saturates rather quickly (0.12 U/mg Glk) and
that this capacity is exceeded even with wildtype expression levels of Glk (0.28 U/mg - §2.2.2).
ATP consumption by Glk, however, is expected to scale linearly with Glk activity and is equal to
glucose flux (Equation 2), due to 1 ATP being used to phosphorylate each glucose molecule.
Glk activity was also seen to scale directly with the specific growth rate of the cell
(Figure 2-4b) with an expected attendant increase in the energetic demand necessary to support
the accelerated growth and division. Experimental measurements of the DNA, RNA, and protein
content of the cell suggests that they each vary quadratically with specific growth rate (Figure
4-2) (111). It was assumed that the cell dimensions would not change significantly as specific
growth rate increased (165) leading to a fixed phospholipid content of each cell. Salmonella
typhimurium, a bacterium of comparable size to E. coli (165), typically contains 8.7x10 6
phospholipids/cell (166) or ~5.9x10 19 phospholipids/g dcw. Phosphatidylethanolamine is the
primary phospholipid component of E. coli cell membranes (167) and is assumed to be made
primarily through decarboxylation of phosphatidylserine (160)
Using the correlations of specific growth rate and Glk activity (Figure 2-4b), and growth
rate and intracellular content (Figure 4-2), with the assumption that the new intracellular
content is generated within one doubling, the rates of protein, DNA, and RNA synthesis as a
function of GIk activity were calculated. These were then multiplied by the ATP stoichiometric
coefficients of Table 4-1 to determine their energetic cost. Finally, the ATP synthesis rate and
Glk ATPase activity as given by Equations 2 and 3 were computed. For non-viable and unhealthy
strains, Glk activity was extrapolated based on g/k mRNA abundance data (Figure 2-4a) and the
above calculations repeated.
78
The results are summarized in Figure 4-3.
Despite the many
CHAPTER 4 1 GLUCOSE VALVE PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
simplifying assumptions inherent in the model, its results are consistent with experimental
observations and provide some mechanistic insight.
For most strains tested (Chapter 2), the
ATPase activity of Glk and total ATP demand are below the aerobic ATP regeneration capacity
resulting in healthy viable cells.
KTS322 and KTS522, the strongest expressing g/k strains
(§2.2.2), fail to grow in minimal medium supplemented with glucose.
Linearly extrapolating
their Glk activity in glucose supplemented medium, it is clear that the ATPase activity of Glk
alone matches or exceeds the aerobic regeneration capacity of the cell and is one potential
explanation for their non-viability. KTS722, which was noted to be sick accumulating one fourth
the biomass of the healthy mutants (§2.2.2), is predicted to have a total energetic demand
exceeding that of the aerobic capacity of the cell. Given that the correlations of intracellular
content are for exponentially dividing cells, it is possible that cells, initially in lag phase, produce
intracellular components at a slow rate allowing the cell to satisfy the energetic demand.
However, as the cell enters exponential phase and increases production to the predicted levels,
the cell can no longer meet the demand resulting in arrested growth and the reduced biomass
accumulation.
40.0
A RNA
N Protein
+DNA
35.0
i30.0
4
25.0
E 15.0
RNA = 0.1857x 2 + 0.4683x+ 2.1629
R=S9
Protein = 3.5638x 2 - 15.575x+ 45.788
.9896
10
DNA 0.0849x2 - 0.3793x+ 0.6241
S20.0
10.0=0.9936
5.0
-
0
0.5
1
1.5
Specific Growth Rate (1/h)
2
2.5
Figure 4-21 Protein, RNA and DNA content as a function of specific growth rate
Data is adapted from (111).
79
4.2 Effects of Glk on physiology: ATP Regeneration and Cell Health
Table 4-1 Approximate energetic costs of transcription, translation and replication
Notes
Process
ATP equivalent/monomera
6
~4 ATPb for nucleoside biosynthesis + 2 ATP for
RNA
transcription
conversion from monophosphate to triphosphate
Protein
4d
1 ATP for tRNA charging; 1 GTP for tRNA binding; 1
translationc
GTP for chain translocation
DNA
6
~4 ATP for nucleoside biosynthesis + 2 ATP for
replication
conversion from monophosphate to triphosphate
Decarboxylation from phosphotidylserine; 1 CTP
Phospholipid
used to activate diacylglycerol for reaction with L1
synthesis
serinef
a Unless explicitly accounted in the literature elsewhere, count represents the
number of energetic
phosphorylated compounds involved (e.g. CTP) from a central carbon metabolism intermediate. Each
phosphorylated compound is assumed to be equivalent to 1 ATP.
b Purine biosynthesis requires 8 ATP while pyrimidines requires 2-3
ATP (2 for UTP/TTP, 3 for CTP).
Assuming slightly higher pyrimidine abundance in the genome, " 4 ATP required per base (168)
Neglects the energetic cost of amino acid biosynthesis
d (169)
e ATP is cleaved to AMP. To regenerate ADP for ATP synthesis, consumption
of 1 more ATP is required.
(160)
SpedcficGrowth Rats (1/h)
0
0.5
1
1.5
2
2.5
200
-
ATP Production Rate
- - - Gik ATPase activity
Total ATP Demand
160
140
Mae I
g120
100
Health
sick
-
80
60
40
ep
20
0
0
0.5
1
1.5
2
2.5
Glk in vitro actvty (U/mg)
3
3.5
4
Figure 4-31 ATP consumption rates as a function of specific growth rate and Glk in vitro total activity in
minimal medium supplemented with glucose.
Total ATP demand represents the ATP required for biosynthesis of protein, RNA, DNA and phospholipids
in the cell as well as the ATPase activity of Glk. ATP production rates represent gross ATP production
without accounting for Glk activity (15 moles ATP/mole glucose). Correlation between specific growth
rate and Glk activity is based on experimental data (Figure 2-4b) and is represented by the scales of both
x-axes. ATP production rate and consumption rates are based on correlations between ATP demand and
specific growth rate as described. Experimentally observed specific growth rates and GIk activities of g/k
expression library mutants (§2.2.2) were used in these models to determine the total ATP demand (filled
blue squares) and GIk ATPase activity (filled triangles) and plotted against measured Glk in vitro activity.
Open blue squares and red triangles are total ATP demand and Glk ATPase activity, respectively, based on
extrapolated Glk activities given the relative mRNA abundance noted in glycerol supplemented medium
(Figure 2-4a). Region boundaries depicted are arbitrary but based on observed trends
80
CHAPTER 4
| GLUCOSE VALVE
PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
Apart from the stated assumptions, several others have been made implicitly in this model:
1.
Concentrations of reagents are irrelevant and all reactions occur at maximalflux
In reality, all the reactions represented are concentration dependent and will
operate at suboptimal reaction rates when reagents are limiting (e.g. ATP). Thus,
the cell could potentially adapt to excessive ATP consumption by reduced carbon
flux due to ATP limitation. However, given the wide range of
Km, ATPvalues
for
endogenous processes (170), it is unlikely that all processes will be affected
equally. Thus, while it may be inaccurate to describe an intermediate regime of
reduced growth as a consequence of total ATP demand exceeding the synthesis
limit, for such strains ATP will be at critically low levels with some reactions at
maximal flux and others at suboptimal flux; this imbalance will lead to deleterious
accumulation/deficiency of some metabolites and arrested growth (cell death).
For cells with critically high Glk ATPase activity, other processes cannot proceed at
all and no growth will occur.
2.
The stoichiometry of ATP on glucose remains constant.
While the stoichiometry of ATP on glucose is highly variable and can be regulated
by the cell, it is unlikely to vary in a manner that materially affects the conclusions
presented here.
Decreases in the stoichiometric coefficient would lower the
energy generating potential, reinforcing the trends predicted and shifting the
predicted regime boundaries to the left.
Increases in the stoichiometric
coefficient, particularly for high levels of Glk expression, are unsustainable and
would also reinforce the trends predicted here.
Firstly, increases in the
stoichiometric coefficient by a cell would imply a reduction in
accumulation and specific growth rate (171).
biomass
This in turn would create positive
81
4.2 Effects of Glk on physiology: ATP Regeneration and Cell Health
feedback which would concentrate Glk through growth mediated buffering (§3.2.1
and §4.3.2) increasing the ATPase activity of Glk. Thus, the improved respiration
capacity would be mostly offset by increased Gik ATPase activity and ultimately
support the trends observed. Furthermore, the increased Glk activity and carbon
flux would lead to increased biomass accumulation as observed (Figure 2-4b) and
counter the increases in the stoichiometric coefficient resulting in minor
differences from the current proposed model.
3.
There are no anaerobic orfermentative routes of ATP production to supplement
the aerobic capacity
For simplicity, fermentation was not considered in this model.
However, such
processes would not increase the energy production rate of the cell much beyond
the 2 ATP produced by glycolysis per glucose. In fact, including such processes
would reduce the energy generation and biomass accumulation rates (due to
carbon loss to these fermentation processes) leading to the negative feedback
described previously and reinforcement of the observed trends.
4.
ATP equivalents in Table 1 sufficiently describe the energetic costs of intracellular
production.
Energy, as defined in this model, refers to phosphorylated energy bearing
compounds such as ATP, CTP and GTP.
Unless otherwise noted, all have been
considered energetically equivalent, potentially underestimating the energetic
costs.
Furthermore, this model ignores the reducing power needed for
biosynthesis which could otherwise be used for energy production. Thus, this
simplifying assumption
processes.
82
underestimates the true
energetic costs of these
CHAPTER 4
| GLUCOSE VALVE
PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
Altogether, these assumptions form a more conservative estimate of the physiological
consequences of Glk expression on ATP utilization in E. coli and do not materially impact the
prediction of three regimes of cell health. More importantly, however, from this rather simple
framework we are able to bound the range of viable expression of Glk in agreement with
experimental observation and understand this phenomenon in the context of cellular energetic
constraints.
4.3.Effects of Physiology on Valve Performance
4.3.1.
Sensitivity of gluconate productivity to system parameters
As discussed in Chapters 2 & 3, changes in Glk activity have led to phenotypic changes in the
health. Given these pleiotropic effects, two questions naturally come to mind:
*
What is the sensitivity of gluconate productivity to system parameters?
*
Are these effects more significant than the relationship anticipated to exist
between productivity and Glk activity (carbon flux)?
To begin to address these questions, a kinetic model of the competing reactions of Glk and Gdh
for glucose (Figure 4-4) was pursued. Implicit in this formulation are the following assumptions:
1. Species concentrations are continuous and uniformly distributed.
2.
Concentrations of cofactors such as NAD*, Mg* and ATP are saturating and
constant.
3. Substrate binding to enzyme is rapid (i.e. the quasi steady state assumption).
4.
Growth and cell death are ignored.
5.
Enzyme concentrations are constant. One interpretation of this assumption is that
the cell is immortal and at steady state. Thus, in conjunction with assumption 4,
83
4.3.1 Sensitivity of gluconate productivity to system parameters
the results of a single cell are representative of a culture and can be linearly scaled
by cell number for aggregate un-normalized behavior.
1-5 suggest Michaelis-Menten
Taken together, assumptions
type kinetics.
Additional
assumptions are:
6.
Glucose transport into the cell occurs at a constant rate independent of
intracellular concentration and Glk activity.
7.
Intracellular products are transported instantaneously to the extracellular
environment.
8.
Gluconate is not further metabolized once generated.
Glucose
ATP
ADP
NAD(P)*
NAD(P)H
Gdh
Glk
D-Gluconate
Glucose - 6 Phosphate
Figure 4-41 Competing reactions for glucose
Based on these assumptions, one can derive the model described in Equation 4 with
parameters and species as described in Table 4-2.
dG
dt
=G
-
k2,gGK
k2,GT
Kg +G
K,,,t +G
dN
k 2 :GT
dt
Km +G
=NG
Y G
84
2
(4)
CHAPTER 4 | GLUCOSE VALVE PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
Table 4-21 Glk-Gdh competition model parameter definitions and values used
Notes
Reference
.i
concentration
Based on experimentally observed range
Figure 2-4b
M
Gdh
concentration
Based on experimentally observed range
Figure 4-6A
M/s
Glucose (GalP)
Glucoe Gate
transport rate
Order of magnitude estimate of kct from4
homologous sugar transporters (5x10
molecules/s); expression level unknown
(172)
Units
Definition
G
M
glucose
concentration
N
M
gluconate
concentration
Molar yield of
gluconate
M
Symbol
Value
K
0-c0.134
T
0 - 0.32x10~
Gi,
10
6
-0.1
s
k2,
410
1/s
Glk turnover
rate
(173)
Kmg
7.6x10 5
M
Glk Michaelis
constant
(173)
k2t
0.772
1/s
Gdh turnover
rate
Km, t
4. 8x10 3
M
Gdh Michaelis
constant
Estimated from total activity of purified
enzyme
(174)
(174)
Simulating the system of Equation 4 in Matlab (m-file: kineticyieldexplore.m,
Appendix A3.1) for various glucose transport rates, and Glk and Gdh concentrations with a
simulation time of 72 h, the model recapitulates expected trends (Figure 4-5).
The model
predicts a strong dependence of gluconate on all 3 parameters. At low glucose transport rates,
the gradient of gluconate yield is oriented diagonally towards the y-axis with a greater xcomponent (Glk activity) than y. This trend is anticipated given the large differences in catalytic
specificity (k2/Km) of the two enzymes: Glk utilization of glucose is more efficient and thus a lot
of glucose is liberated and available for conversion to gluconate with relatively small decreases
in Glk activity. Conversely, due to this efficiency mismatch, increasing Gdh expression by a
comparable amount has a much smaller impact on shifting the balance of flux in its favor. Very
high yields of gluconate conversion can only be obtained with low levels of Glk. As the glucose
transport rate increases, glucose accumulates and is no longer limiting; the gradient rotates to
become parallel with the y-axis at which point, gluconate yield is limited by Gdh activity.
85
4.3.1 Sensitivity of gluconate productivity to system parameters
A
B
90
0.9
0.8
0.8
80
0.7
0.7
70
S0.6
0.6
60
0.5
0.5
50
0.4
04
0
0.3
0.3
0
0.2
0.2
20
0.1
0
0.1
00
0.6
0.4
Relative131k
Level
0.2
101
11
0.6
0
0.2
0.6
0,4
ReiativeGlk Level
0.8
1
D
C
0.9
0.9
0.8
0.8
80
07
0.7
70
0 6
90
.
0.6
60
0.5
0.5
0
0.4
0.4
0
0.3
0.3
30
0.2
0.2
20
0.1
0.1
10
0
0.2
0.4
0.6
RelativeGIkLevel
0,8
1
00
0,2
0.4
0.6
Relative
GikLevel
0.6
1
Figure 4-51 Gluconate yield as a function of Gik and Gdh activity, and glucose transport rate
Relative Glk and Gdh activity are plotted on the x and y axes respectively. Glucose transport rates were
varied across the 4 plots and are A) 106 M/s B) 3x10 4 M/s C) 10- M/s and D) 102 M/s. Gluconate molar
yield % on glucose is represented by the heat map with conversion increasing from dark red to white as
indicated by the colorbar on the right.
In trying to redirect carbon flux into a production pathway, it has been implicitly
assumed that the heterologous genes were expressed at the same level (given similar induction
conditions) as GIk activity was varied. However, experiments have proven this assumption to be
patently false (§2.2.3).
Thus, the effect of Glk activity on Gdh expression and gluconate
productivity was explored. Under conditions similar to Section 3.2.2, Gdh was expressed from
an IPTG inducible promoter in 4 members of the KTSx221G gik expression library (§2.2.3) in the
presence of an additional empty vector. After 72 h, gluconate yields and Gdh expression were
measured. As expected, increasing the amount of IPTG used to induce expression led to an
increase in the maximum amount of Gdh measured (Figure 4-6).
However, there was also a
strong dependence of Gdh expression on Glk activity. Gdh expression varied approximately
86
CHAPTER 4 | GLUCOSE VALVE PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
sigmoidally with Glk activity, likely due to the increased carbon flux to support heterologous
expression. Once a certain threshold of carbon flux was met, however, further improvements
are not met with increased Gdh expression. Of note, the weakest Glk expressing strain only
grew under the lowest induction level tested (25 pM IPTG). Under such conditions, the cell was
able to produce small amounts of Gdh (~2.3 U/mg total protein). Nonetheless, it was able to
achieve the highest gluconate yields (Figure 4-7A).
35.
30
a
A
25 uM IPTG
50 uM IPTG
100 uM IPTG
25
zD 2015-a 1050,
0.0
0.1
0.2
0.3
0.4
0.5
Estimated Glk Activity (U/mg)
Figure 4-61 Impact of GIk activity on Gdh expression
Gdh expressed at one of three IPTG induction levels in a Glk expression family mutant (KTSx221G - §2.2.3)
bearing a plasmid for Gdh expression and an empty vector plasmid as surrogate for the antisense valve
(§3.2.2). Due to the similar assay chemistries, Glk activity cannot be measured in the presence of Gdh.
Estimated Glk activities presented here represent those of the KTSx22 parental strain (§2.2.2) in the
absence of any plasmid. Graph depicts the mean ± SD of at least 3 culture flasks. Curves are fit to a Hill
expression of the form y = V"
Examining gluconate yields (Figure 4-7A), the system only performs as anticipated with
low IPTG induction.
At higher levels of IPTG, there is no trend or an unpredicted direct
correlation of gluconate yield and Glk. Given the counterintuitive nature of these results, the
competition model presented was used to explain the behavior observed.
If high GalP
expression (~1000 molecules/cell -> Gin= 10- M/s) is assumed and the relationship between Glk
and Gdh activities (Figure 4-6) accounted for, the experimental trends are captured (Figure 4-7).
Using the KTSx221G library, the Gluconate-Glk-Gdh landscape (Figure 4-5) may only be sampled
87
4.3.1 Sensitivity of gluconate productivity to system parameters
discretely and is constrained by the relationship between Glk and Gdh (Figure 4-6). Evaluating
the predicted yields at the sampled points, namely relative Glk levels of ~0.3,~ 0.5,~ 0.6 and
~0.9, both the anticipated and counterintuitive relative trends at each induction level are
recapitulated (Figure 4-7B). However, the relative yields between induction levels at a given Glk
expression level and absolute yields are not captured by this model. The model displays a
consistent positive bias in its prediction for gluconate yield suggesting that yields are limited by
another factor currently not considered.
Preliminary experiments suggest that this factor is
NAD*, presumed to be constant in the model.
B
A
1
1.0
_--
5
-.-A-
O08 -
25 uM IPTG
50 uM IPTG
100 uM IPTG
0.80.6-
+ 0.6-s
_
C
0
c
E 0.4-
0.4
75
-6- 25 pM IPTG
.2 0.2-
0.2
8
-
F50.0
0.0
0.1
0.4
0.3
0.2
Estimated Glk Activity (U/mg)
0.5
00
IPTG
-+-
50 pM
-A-
100 pM IPTG
0.2
0.6
0.4
Relative Glk activity
0.8
1
Figure 4-71 Gluconate yield as a function of Glk and Gdh expression A) Experimentally measured
B)Model predictions.
A) Gluconate molar yield as a function of GIk activity in a GIk expression mutant bearing a plasmid for Gdh
expression and an empty vector plasmid as surrogate for the antisense valve. Due to the similar assay
chemistries, Glk activity cannot be measured in the presence of Gdh. Estimated GIk activities presented
here represent those of the KTSx22 parental strain (§2.2.2) in the absence of any plasmid. Graph depicts
3
the mean ± SD of at least 3 cultures. B) Glucose transport rate was set at 10 M/s. Yields as predicted by
the model with estimated GIk activity values of Panel A. Relative GIk activity spans 0 - 0.5 U/mg Glk.
Despite
its simplicity, major features of the system performance
have been
recapitulated by this model. The model predicts yield improvements under glucose limiting
conditions with decreased Glk activity, and increased Gdh activity otherwise.
Taking into
account the constraints of Gdh and GIk expression and the discrete sampling of the landscape,
its trend predictions at individual induction levels qualitatively agree with those from
88
CHAPTER 4 GLUCOSE VALVE PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
experiment and further support the assertion that there exists a more complex tradeoff
between pathway efficiency, cell health and pathway expression (§2.2.3). The model, however,
is unable to accurately predict absolute trends across induction levels. While some of the error
may be attributable to demonstrably false assumptions (e.g. constant glucose transport macroscopic measurements of glucose titers suggest a lower utilization rate for the lowest Glk
expressing strain), the consistent bias suggests a fourth (or more) unexplored variable.
Nonetheless, with only 3 variables we are able to recreate general trends, explore the complex
relationship between gluconate productivity and cell phenotype and begin to elucidate the
interplay between these factors.
4.3.2.
Significance of growth mediated buffering as a regulatory mechanism
As observed in Figure 2-4b, Glk activity is correlated with the specific growth rate of PTS
Glu* mutants. This effect, however, has consequences for the ability to control Glk itself. Most
bacterial proteins, including Glk, are intrinsically stable and diluted only by growth (143). This in
turn generates negative feedback which can potentially buffer the cell against some of the
effects of Glk manipulation (125) (§3.2.1). However, it is still unclear as to the magnitude of this
effect. To answer this question, a model was generated with the following assumptions:
1. Species concentrations are continuous and uniformly distributed.
2.
Growth and Glk activity are related by a monotonic linear increasing function as
described by Figure 2-4b.
3.
Reaction participants (polymerases, nucleotides, etc.) bind quickly and are in
excess.
4.
Growth rate changes, when accounted for, are instantaneous.
From these assumptions, transcription and translation of Glk may be represented by Equation 5
with species and parameter values as defined in Table 4-3.
89
4.3.2 Significance of growth mediated buffering as a regulatory mechanism
dm
dm=a-(kg
(k1 +pup
dt
dP/
P
dP= flm - PP
(5)
dt
P = 7P + so
Table 4-31 Growth mediated buffering model parameter definitions and values used
Symbol
Value
Units
Definition
Notes
Reference
g/k mRNA
m
molecules/cell
P
molecules/cell
Glk concentration
p
1/min
specific growth rate
0.189
molecules/cell/
min
transcription rate
6
proteins/
transcript/min
translation rate
1/min
mRNA
rate
degradation
Calculated from half-life
of 3.3 min
(175)
1/min/
protein
of growth-Glk
slope
correlation
Fit to experimental data
(Figure 2-4b)
§2.2.2
1/min
intercept of growthGlk correlation
Fit to experimental data
(Figure 2-4b)
§2.2.2
a
kd
0.21
y
4.1x10.n
(p
2.2x10~5
concena
concentration
Number of transcripts
per
division/doubling
time
Fit to experimental data
for
KTSO22;
nominal
range ~3-6
(175)
(111)
To illustrate the effect of growth mediated buffering on the dynamic range of control, c
was varied.
In the test case (Figure 4-8), the mRNA transcript abundance was halved,
representing asRNA or inverter control, and the system simulated for 1200+ mins. When the
specific growth rate is assumed to be constant at 0.29 h-1 (0.005 min-'), the relative change in
Glk activity is equal to the relative change in mRNA levels. If growth mediated buffering is
included, however, it is observed that Glk activity only decreases by 25% despite the 50%
reduction in mRNA levels. Some of the potential reduction in activity is offset by the reduction
in growth rate which concentrates the remaining Glk. In this manner, the negative feedback
between growth and Glk activity buffers the cell against changes in g/k expression.
90
CHAPTER 4 | GLUCOSE VALVE PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
0.3
0.25
'
'
0.2
0.15
0.1
-
0.05--
Initial steady state
0 ----- growth mediated buffering
constant ,
200
400
600
800
time [min]
1000
1200
Figure 4-81 Predicted Glk activity as a function of time when mRNA levels are halved at t = 0.
Predicted Gik activity is calculated with p constant at 0.28 h1 and p variable as described in Figure 2-4b
(growth mediated buffering).
This phenomenon was then further investigated by exploring this buffering effect as a
function of changes in g/k expression.
The gain (K) or steady state change in protein
concentration for a given step change in (free) mRNA concentration may be defined for the two
cases as follows.
Constant p
At steady state, dx/dt = 0. Thus, the steady state concentrations (xss) are given by setting the
differential equations of Equation 5 to 0:
0 = a-(k, +pm,
kd >> p
(6)
a
d
0 =&Qs
#3m,,
p
pP
_
f8a
(7)
kdp
91
4.3.2 Significance of growth mediated buffering as a regulatory mechanism
For a negative step change in the transcription rate of (1-f)a, the gain (Kconstant) may be defined
as:
K=AY
AX
Iss,final
ss,initial
final
initial
/Jfa
Pa
kdP
kd
(8)
fa-a
(f-1)a
(f-1)a
Kconstant
kdp
As expected, the gain is constant leading to a linear relationship between mRNA and protein
levels.
Variable p
The steady state concentrations of gik mRNA is the same as that as described by Equation 6.
The steady state protein levels, however, is given by the following:
0=
p8m , -
0 = fla
(yP +(P)P
kd
0
pP,
= yP + pP
/Ja
k
-
(9)
kd
-+
--
92
_
42+
Jay
2
kd
P'S
27
d
CHAPTER 4 1 GLUCOSE VALVE PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
For physical solutions, the steady state protein level is described by Equation 10.
43ay
2
k
Psd
(10)
27
SS
The gain is then given by:
K
AY
AX
s,initial
,final ~
afinal
+
ainitial
2
+
+4fay
2
4ay
(11)
_k__k_
2y/
27
fa-a
(2k
Kvariable
=
4/fay
k7
2(f
-1)ya
2
4,6ay
kC
In the case of growth mediated buffering, the gain is variable and a non-linear function of the
input change.
The buffering capacity, defined as (Kconstant
- Kvariable)/Kconstant, was
then evaluated and
plotted as a function of the final g/k mRNA concentration after perturbation (Figure 4-9) (see mfile bufferevalf, Appendix 3.2). This parameter represents the relative amount of change in
Glk activity not realized due to the coupling of activity and growth.
As g/k expression is
increased from wildtype levels by a perturbation, at least half of the perturbation is rejected.
Moreover, as this disturbance grows, the increased growth rate offsets the increase in
expression leading to the buffering capacity of the cell growing. While much of these changes
are not realizable due to the ATP limitation presented in Section 4.2.1, if such scenarios were
possible, up to 90% of the perturbation would be rejected through growth mediated buffering.
Conversely, as disturbances decrease g/k mRNA levels, less than 50% of the disturbance is
93
4.3.2 Significance of growth mediated buffering as a regulatory mechanism
rejected.
This capacity diminishes as transcription is suppressed further.
This result is
counterintuitive as in such scenarios the cell would seemingly be better served by rejecting as
much of the disturbance as possible. However, to do so, the cell would have to sacrifice growth
to a much larger extent leading to this superficially suboptimal solution.
This behavior also
begins to explain the relative performance differences between the inverter and asRNA valve
actuators. Minimal downregulation of g/k expression (i.e. asRNA) can be compensated for more
efficiently than the significant repression of expression (i.e. inverter).
1
0.93 0.8
c 0.7
.20.6
Downregulation
D
0.5
- XUpregulation
6 '
0.4
0 0.30.2
0.1
0
-2
102
1
10'
0
100
2
10
102
Relative gik mRNA
Figure 4-91 Buffering capacity of cells perturbed from the initial steady state (Relative gik mRNA
a function of the final mRNA level
=
1) as
Growth mediated buffering, as presented here, has significant potential to minimize the
impact of differential expression of g/k expression. At increased expression levels, the increase
in specific growth rate has the potential to counter the majority of this disturbance. Conversely,
maximum relative changes in Glk activity may be realized with significant inhibition of Glk
expression. While seemingly suboptimal, by opting to ignore such changes the cell may still
maintain growth rates that allow the cell to proliferate and potentially remain competitive in the
94
CHAPTER 4 | GLUCOSE VALVE PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
environment.
Coupled with metabolite efficiency (145) and PEP substrate inhibition (144)
discussed in detail elsewhere, the cell is equipped to passively and effectively respond to most
perturbations in carbon flux without energy intensive and slower genetic regulation, reducing
the efficiency of an engineered valve.
4.4.Perspectives
As engineers attempt to design ever more complex biological systems to solve the
problems of tomorrow, they will need both a rich toolbox of parts with which to work and
detailed information about their performance. Synthetic biology has taken up this mantle with
the goal of developing well defined and characterized parts (60, 62, 73, 74).
To date, this
characterization has been primarily empirical datasets at fixed experimental conditions (73, 74).
Organizations such as the BIOFAB (http:// biofab.org) and the Registry of Standard Biological
Parts (http://partsregistry.org)
characterization
have also lent their support to the development
of such parts with rich community discussions and
characterization of their performance.
and
high throughput
Indeed, it is through the existence and use of these
resources that much of the work in this dissertation has been completed. With this vision in
mind, a framework of frequency domain based transfer functions to characterize such systems
was proposed.
Frequency domain based transfer functions are a mainstay of classical control
engineering (176). Central to this framework are three key parameters: gain (input-output dose
response slope), time constant (measure of the response time), and dead/lag time (time taken
by the input signal to propagate before the system begins to respond). These parameters are
sufficient to accurately represent the information content of current datasheets (Figure 4-10)
while being rooted in physical models. Moreover, like the modular physical assembly of such
parts, the properties of their related transfer functions would allow the prediction of the
95
4.3.2 Significance of growth mediated buffering as a regulatory mechanism
assembled system by simple algebraic manipulation. Thus, they not only offer the advantage of
simplifying characterization and its representation, but allow for some prediction of system
behavior in novel contexts. However, the properties of such systems that make this approach
tractable are not present normally in biological systems.
Parameters are neither necessarily
time invariant nor are system responses linear. Moreover, integration of multiple inputs and
differential regulation in response to various stimuli make linearization schemes tenuous. More
sophisticated frameworks from control theory which can accommodate such features, however,
lose the elegance and simplicity of frequency domain based transfer functions making them
undesirable for routine use.
B
A
pBbE5
BBaF2620
30C.HSL -POPS Recshw
a
w
wm e
scenen
sa
F
afte
eenbe
L. a.h.
k-wa
.
a
a
I_____
O4SL
An
Lne s~t
pean IVeMW
Mo
g
MW
se toppoducea baduenAND
amesNw.scmos
mq~eq
chast Va$
coft
WON
Apodft
tm~siom
wwoa
e dM4am
sas
(3Osousepg.*edc0
Mif
150
POLVd30np
"pgmeE
)EI
sibmb
mawmI
canetm
Figure 4-10|I Prototypical characterization datasheets
(A)Canton & Endy, 2008 (73). (B)Lee et. al, 2011 (74)
Nonetheless, the results and models developed through this work illustrate a need for a
framework with greater appreciation of the context dependence of such systems. While some
96
CHAPTER 4 1GLUCOSE VALVE PERFORMANCE POTENTIAL AND IMPACT ON PHYSIOLOGY
effort has been directed towards relatively linear context dependent effects (medium
composition, strain, etc) (74), the net sum of these effects and their interaction on related
systems has been relatively unexplored. In this work, the impacts of some of these interactions
have been tested and examined theoretically for the designed glucose valves. Firstly, the effects
of glycolytic modulation on cellular health were explored and boundaries on realizable levels of
Glk expression established.
Secondly, some of the attendant global changes of this
manipulation on performance of a heterologous pathway were explored. Finally, the impact of
existing passive negative feedback mechanisms on the control potential of such systems was
examined.
Underlying this work is the theme that there exist emergent and complex cellular
responses to seemingly minor physiological changes. While these challenges may hinder the
progress of synthetic biology, they are not insurmountable. They do, however, point to a need
for more sophisticated design tools such as computer aided design tools or
BioCAD
environments (68-72) that may capture and predict these phenomena a priori.
4.5.Materials and Methods
4.5.1.
Strains and plasmids
Gluconate production in strains bearing multiple plasmids was tested in KTSO221G,
KTS6221G, KTS8221G and KTS922IG of the glucokinase expression family (§2.2.3).
Gdh was
expressed from pTrc99Cm-Gdh (§2.2.3) with IPTG as indicated. The surrogate plasmid used in
all cases was pKVS45 (§3.2.1).
4.5.2.
Culture conditions
Experimental cultures were grown at 37'C in M9 minimal medium supplemented with
0.8 mM L- leucine and 1.5% (w/v) glucose. For all experiments, starter cultures were grown to
an OD measured at a wavelength of 600 nm (OD600 ) of at least 0.2 before being transferred to a
97
4.5.4 Metabolite analysis
50 ml flask of the same media. IPTG was added as indicated to both the starter culture and 50
ml flasks at inoculation. Shake flasks were inoculated to a final OD600 of ~0.005 and monitored
for 72 h before samples were taken and clarified by centrifugation. The supernatant was then
stored at 4'C until HPLC analysis. Ampicillin, and chloramphenicol were used at concentrations
of 100 and 34 pg/ml, respectively.
4.5.3. Enzyme activity levels
All enzyme activity assays were performed on crude lysates generated by sonication of 5
ml frozen cell pellets resuspended in 500 pI of 10 mM Tris-HCI (pH=8.0) at ~ 8 W in 5 x 5 s pulses
on ice.
The resulting lysates were clarified by centrifugation before being assayed for Gdh
activity as described previously (§2.4.4).
In brief, one unit of Gdh activity oxidizes 1.0
gmole/min of D-glucose at pH 7.6 at room temperature in 60 mM potassium phosphate buffer
and 0.67 mM NAD*.
Reaction progress is measured at a wavelength of 340 nm.
Due to the
assay chemistry, Glk activity cannot be measured independently of Gdh activity.
Enzyme
activities reported are normalized by total protein levels as measured in a modified Bradford
Assay (124) .
4.5.4.
Metabolite analysis
Glucose, gluconate (Gnt), and 5-ketogluconate (5KG) were measured by HPLC analysis
using an Aminex* HPX-87H anion exchange column (Bio-Rad Laboratories, Hercules, CA) in an
Agilent 1100 series HPLC instrument (Santa Clara, CA) as described previously (§2.4.5)
98
CHAPTER
5
Conclusions and Future Directions
99
5.1 Thesis Summary
5.1.Thesis Summary
This thesis has focused on the development of glucose valves for metabolic engineering
in E. coli. Such a device has the potential to balance the needs of cell health with the demands
of a heterologous pathway and enable the design of new classes of unnatural pathways where
glucose is the sole carbon source used directly as a substrate for production and growth. The
primary findings of the thesis are as follows.
In Chapter 2, a novel E. coli host was engineered to allow for the uptake of
unphosphorylated glucose. Once in the cell, commitment of glucose to endogenous processes is
accomplished through glucokinase (Glk) making it a natural glucose valve.
To demonstrate
potential control of Glk, an expression library was made using promoters from the Registry of
Standard Biological Parts. Glk expression was controlled between half to twice that of native
expression with an attendant change in the specific growth rate, suggesting successful
redirection of intracellular carbon flux. This capability was then confirmed by the redirection of
carbon into a model heterologous pathway where pathway yield was increased by 2 fold with
decreases in Glk activity. This control, however, was limited by the ability of the cell to meet
endogenous demands and express recombinant enzyme.
Dynamic control of Glk activity could potentially alleviate this limitation by allowing the
cell to accumulate biomass and needed intracellular resources before carbon flux was
redirected towards heterologous production. Exploring this issue, antisense RNA (asRNA) and
gene circuit devices (inverters) were constructed in Chapter 3. Using asRNA, Glk activity could
be inhibited by up to 25%; however, this did not lead to controllable increases in pathway
productivity.
Inverters, on the other hand, were able to quickly, reliably and dynamically
decrease Glk activity by up to 50% with a concomitant increase in pathway productivity and
decrease in carbon waste to acetate.
100
CHAPTER 5 | CONCLUSIONS AND FUTURE DIRECTIONS
To explore the full potential of a Glk glucose valve, models describing various facets of
the Glk system were developed in Chapter 4. These models were able to recapitulate the basic
system phenomena observed and provide insight into how they arise. Specifically, the range of
Glk expression possible was bounded due to energy production constraints of the cell. Then,
the interplay between various physiological features and their impact on the output of a
heterologous pathway was explored. Finally, the emergence of negative feedback through the
coupling of Glk activity and growth was examined and its effect on the controllability of Glk
studied.
5.2.Future Directions
This thesis represents but the first iteration through the design process. Naturally, the
lessons learnt here suggest several points of refinement. As discussed, the coupling of growth,
Glk activity and Glk degradation presents a significant regulatory mechanism resulting in
variable gain. Thus, to decouple the negative feedback and realize the full potential of control
actions against g/k mRNA, Glk should be engineered with a half-life significantly lower than the
cell doubling time. This may be achieved through the use of tunable protein degradation tags
(146, 177).
Another viable line of inquiry is an exploration of alternative valve actuator
mechanisms. In this study, antisense RNA and an inverting gene circuit were explored. These
designs represent but one permutation possible.
For instance, the existing inverter may be
tuned by varying the expression and binding strengths of its constituents (by substituting the
RBS, operators, promoters, etc) to potentially find a more optimal solution.
Moreover, the
control elements themselves (tetR, lad) may be swapped with other regulators such as araC,
luxR and so on for a potentially more responsive system.
Similarly, over the course of this
thesis, a number of tools and methodologies for the design of antisense RNA have been
published; most notable of these is the advent of modular antisense RNA control (178).
101
5.2 Future Directions
Integrating these new techniques may result in better performing constructs. Moving beyond
these designs, diverse alternative valve actuation mechanisms may be possible including
riboregulators (53), recombination switches (88, 179) and controlled expression of orthogonal
protein translation (180).
Finally, post-transcriptional methods of control, such as engineered
allostery (51) or inducible degradation (50) may be explored for systems with better dynamic
performance.
Glucose valves, as presented in this dissertation, are exclusively manually controlled;
however, they may be integrated into larger gene circuits for autonomous control. For instance,
they may be engineered to recognize factors such as cell density (181), intracellular metabolites
(56) or even light (133).
In such contexts, a valve would be more economically attractive as it
would no longer be associated with the cost of inducer.
Moreover, such systems may
potentially be more stable as they would be responsive to environmental perturbations. Finally,
the concept of a valve is readily extensible to any number of metabolites where deletion of the
competing pathway is deleterious and there is a significant difference in the catalytic specificity
of competing reactions at the node of interest.
The work presented here has been enabled by the tools of synthetic biology. Among
these were the promoter libraries and inverter components from the Registry of Standard
Biological Parts. These resources have extraordinary potential in furthering the engineering of
biological systems and would benefit from the addition of richer and more diverse part libraries.
For example, while the Anderson promoter library proved invaluable in probing the boundaries
of Glk control, work was ultimately limited by the resolution of such resources. The BIOFAB,
based in Silicon Valley, is beginning to address this issue. Nonetheless, there is a clear and
present need for standardization in characterization across such resources.
Attempts were
made over the course of this thesis to fill gaps by experimenting with parts from both databases.
102
CHAPTER 5 1CONCLUSIONS AND FUTURE DIRECTIONS
However, the context dependence of the characterization between these resources led to
unpredictable results which were, ultimately, not developed in this dissertation (see Appendix
A2). Finally, incorporating modules that can predict the interactions of engineered systems with
host physiology and the consequences of this on device performance into existing CAD tools,
such as BioJADE (72) and Clotho (182), would be of much utility to the field.
5.3.Concluding Remarks
In this thesis, global phenotypes were controlled while simultaneously allowing the
regulation of metabolite flux into a heterologous pathway.
More importantly, however, this
work represents one of the first reports of glucose redirection in E. coli and enables the creation
of unnatural pathways utilizing glucose directly as a substrate for growth and production; a
capability which heretofore was unavailable.
Underpinning this work is use of various
paradigms and tools of synthetic biology which have the potential to empower the creation of
ever more complex biological devices. This work, however, acknowledges the importance of the
physiological response to such devices and, hopefully, underscores the need for the improved
understanding of such effects.
103
REFERENCES
1.
Pitera, D. J., Paddon, C. J., Newman, J. D., and Keasling, J. D. (2007) Balancing a
heterologous mevalonate pathway for improved isoprenoid production in Escherichia
coli, Metab. Eng. 9, 193-207.
2.
Lee, S.-H., Park, S., Lee, S., and Hong, S. (2008) Biosynthesis of enantiopure (S)-3hydroxybutyric acid in metabolically engineered Escherichia coli, Appl. Microb.
Biotechnol. 79, 633-641.
3.
Tseng, H.-C., Harwell, C., Martin, C., and Prather, K. (2010) Biosynthesis of chiral 3hydroxyvalerate from single propionate-unrelated carbon sources in metabolically
engineered E. coli, Microb. Cell Factories 9, 96.
4.
Martin, V. J. J., Pitera, D. J., Withers, S. T., Newman, J. D., and Keasling, J. D. (2003)
Engineering a mevalonate pathway in Escherichia coli for production of terpenoids, Nat.
Biotech. 21, 796-802.
5.
Ajikumar, P. K., Xiao, W.-H., Tyo, K. E. J., Wang, Y., Simeon, F., Leonard, E., Mucha, 0.,
Phon, T. H., Pfeifer, B., and Stephanopoulos, G. (2010) Isoprenoid pathway optimization
for taxol precursor overproduction in Escherichia coli, Science 330, 70-74.
6.
McDaniel, R., Ebert-Khosla, S., Hopwood, D. A., and Khosla, C. (1993) Engineered
biosynthesis of novel polyketides, Science 262, 1546-1550.
7.
Pfeifer, B. A., Admiraal, S. J., Gramajo, H., Cane, D. E., and Khosla, C. (2001) Biosynthesis
of complex polyketides in a metabolically engineered strain of E. coli, Science 291, 17901792.
8.
Xia, X.-X., Qian, Z.-G., Ki, C. S., Park, Y. H., Kaplan, D. L., and Lee, S. Y. (2010) Native-sized
recombinant spider silk protein produced in metabolically engineered Escherichia coli
results in a strong fiber, Proc. Natl. Acad. Sci. U. S. A. 107,14059-14063.
9.
Wang, F., and Lee, S. Y. (1997) Production of poly(3-hydroxybutyrate) by fed-batch
culture of filamentation-suppressed recombinant Escherichia coli, Appl. Environ.
Microbiol. 63, 4765-4769.
104
REFERENCES
10.
Singh, 0., and Kumar, R. (2007) Biotechnological production of gluconic acid: future
implications, Appl. Microb. Biotechnol. 75, 713-722.
11.
Kurian, J. V. (2005) A new polymer platform for the future derived 1,3-propanediol, J. Polym. Environ. 13, 159-167.
12.
Van Noorden, R. (2010) Demand for malaria drug soars, Nature 466, 672-673.
13.
Varma, A., and Palsson, B. 0. (1994) Metabolic flux balancing: Basic concepts, scientific
and practical use, Nat. Biotech. 12, 994-998.
14.
Varma, A., and Palsson, B. 0. (1994) Stoichiometric flux balance models quantitatively
predict growth and metabolic by-product secretion in wild-type Escherichia coli W3110,
Appl. Environ. Microbiol. 60, 3724-3731.
15.
Burgard, A. P., Pharkya, P., and Maranas, C. D. (2003) Optknock: A bilevel programming
framework for identifying gene knockout strategies for microbial strain optimization,
Biotechnol. Bioeng. 84, 647-657.
16.
Pharkya, P., and Maranas, C. D. (2006) An optimization framework for identifying
reaction activation/inhibition or elimination candidates for overproduction in microbial
systems, Metab. Eng. 8, 1-13.
17.
Ranganathan, S., Suthers, P. F., and Maranas, C. D. (2010) OptForce: An optimization
procedure for identifying all genetic manipulations leading to targeted overproductions,
PLoS Comput. Biol. 6, e1000744.
18.
Glick, B. R. (1995) Metabolic load and heterologous gene expression, Biotechnol. Adv.
13, 247-261.
19.
Jones, K. L., Kim, S.-W., and Keasling, J. D. (2000) Low-copy plasmids can perform as well
as or better than high-copy plasmids for metabolic engineering of bacteria, Metab. Eng.
2, 328-338.
20.
Dong, H., Nilsson, L., and Kurland, C. G. (1995) Gratuitous overexpression of genes in
Escherichia coli leads to growth inhibition and ribosome destruction, J. Bacteriol. 177,
1497-1504.
Sorona(R) from corn
105
REFERENCES
21.
Harcum, S. W., and Bentley, W. E. (1993) Response dynamics of 26-, 34-, 39-, 54-, and
80-kDa proteases in induced cultures of recombinant Escherichia coli, Biotechnol.
Bioeng. 42, 675-685.
22.
Goff, S. A., and Goldberg, A. L. (1985) Production of abnormal proteins in E. coli
stimulates transcription of Ion and other heat shock genes, Cell 41, 587-595.
23.
Berry, A., Dodge, T. C., Pepsin, M., and Weyler, W. (2002) Application of metabolic
engineering to improve both the production and use of biotech indigo, J. Ind. Microbiol.
Biotechnol. 28, 127-133.
24.
Zhu, M. M., Skraly, F. A., and Cameron, D. C. (2001) Accumulation of methylglyoxal in
anaerobically grown Escherichia coli and its detoxification by expression of the
Pseudomonas putida glyoxalase I gene, Metab. Eng. 3, 218-225.
25.
Barbirato, F., Grivet, J. P., Soucaille, P., and Bories, A. (1996) 3-Hydroxypropionaldehyde,
an inhibitory metabolite of glycerol fermentation to 1,3-propanediol by enterobacterial
species, Appl Environ Microbiol 62, 1448-1451.
26.
Fong, S. S., Burgard, A. P., Herring, C. D., Knight, E. M., Blattner, F. R., Maranas, C. D.,
and Palsson, B. 0. (2005) In silico design and adaptive evolution of Escherichia coli for
production of lactic acid, Biotechnol. Bioeng. 91, 643-648.
27.
Tummala, S. B., Welker, N. E., and Papoutsakis, E. T. (2003) Design of antisense RNA
constructs for downregulation of the acetone formation pathway of Clostridium
acetobutylicum, J. Bacteriol. 185, 1923-1934.
28.
Kim, J., Hirasawa, T., Sato, Y., Nagahisa, K., Furusawa, C., and Shimizu, H. (2009) Effect of
odhA overexpression and odhA antisense RNA expression on Tween-40-triggered
glutamate production by Corynebacterium glutamicum, Appl. Microb. Biotechnol. 81,
1097-1106.
29.
Sillers, R., AI-Hinai, M. A., and Papoutsakis, E. T. (2009) Aldehyde-alcohol
dehydrogenase and/or thiolase overexpression coupled with CoA transferase
downregulation lead to higher alcohol titers and selectivity in Clostridium
acetobutylicum fermentations, Biotechnol. Bioeng. 102, 38-49.
30.
Biedendieck, R., Malten, M., Barg, H., Bunk, B., Martens, J.-H., Deery, E., Leech, H.,
Warren, M. J., and Jahn, D. (2009) Metabolic engineering of cobalamin (vitamin B12)
production in Bacillus megaterium, Microb. Biotechnol. 3, 24-37.
106
REFERENCES
31.
Gibbs, D. F. (1983) The rise and fall (and rise?) of acetone/butanol fermentations, Trends
Biotechnol. 1, 12-15.
32.
Grupe, H., and Gottschalk, G. (1992) Physiological events in Clostridium acetobutylicum
during the shift from acidogenesis to solventogenesis in continuous culture and
presentation of a model for shift induction, Appl Environ Microbiol 58, 3896-3902.
33.
Terracciano, J. S., and Kashket, E. R. (1986) Intracellular conditions required for initiation
of solvent production by Clostridium acetobutylicum, Appl. Environ. Microbiol. 52, 8691.
34.
Tummala, S. B., Junne, S. G., and Papoutsakis, E. T. (2003) Antisense RNA
downregulation of coenzyme A transferase combined with alcohol-aldehyde
dehydrogenase overexpression leads to predominantly alcohologenic Clostridium
acetobutylicum fermentations, J. Bacteriol. 185, 3644-3653.
35.
Desai, R. P., and Papoutsakis, E. T. (1999) Antisense RNA strategies for metabolic
engineering of Clostridium acetobutylicum, Appl. Environ. Microbiol. 65, 936-945.
36.
Eggeling, L. (1994) Biology of L-lysine overproduction by Corynebacterium glutamicum,
Amino Acids 6, 261-272.
37.
Jensen, E. B., and Carlsen, S. (1990) Production of recombinant human growth hormone
in Escherichia coli: Expression of different precursors and physiological effects of
glucose, acetate, and salts, Biotechnol. Bioeng. 36, 1-11.
38.
Bauer, K. A., Ben-Bassat, A., Dawson, M., de la Puente, V. T., and Neway, J. 0. (1990)
Improved expression of human interleukin-2 in high-cell-density fermentor cultures of
Escherichia coli K-12 by a phosphotransacetylase mutant, Appl. Environ. Microbiol. 56,
1296-1302.
39.
Aristidou, A. A., San, K.-Y., and Bennett, G. N. (1995) Metabolic engineering of
Escherichia coli to enhance recombinant protein production through acetate reduction,
Biotechnol. Prog. 11, 475-478.
40.
Hahm, D. H., Pan, J., and Rhee, J. S. (1994) Characterization and evaluation of a pta
(phosphotransacetylase) negative mutant of Escherichia coli HB101 as production host
of foreign lipase, Appl. Microb. Biotechnol. 42, 100-107.
107
REFERENCES
41.
Chang, D.-E., Shin, S., Rhee, J.-S., and Pan, J.-G. (1999) Acetate metabolism in a pta
Mutant of Escherichia coli W3110: Importance of maintaining acetyl coenzyme A flux for
growth and survival, J. Bacteriol. 181, 6656-6663.
42.
Kim, J. Y. H., and Cha, H. J. (2003) Down-regulation of acetate pathway through
antisense strategy in Escherichia coli: Improved foreign protein production, Biotechnol.
Bioeng. 83, 841-853.
43.
Becker, J., Klopprogge, C., Schroder, H., and Wittmann, C. (2009) TCA cycle engineering
for improved lysine production in Corynebacterium glutamicum, Appl. Environ.
Microbiol. 75, 7866-7869.
44.
Becker, J., Buschke, N., Bucker, R., and Wittmann, C. (2010) Systems level engineering of
Corynebacterium glutamicum - Reprogramming translational efficiency for superior
production, Eng. Life Sci. 10, 430-438.
45.
Asadollahi, M. A., Maury, J., Moller, K., Nielsen, K. F., Schalk, M., Clark, A., and Nielsen, J.
(2008) Production of plant sesquiterpenes in Saccharomyces cerevisiae: Effect of ERG9
repression on sesquiterpene biosynthesis, Biotechnol. Bioeng. 99, 666-677.
46.
Ro, D.-K., Paradise, E. M., Ouellet, M., Fisher, K. J., Newman, K. L., Ndungu, J. M., Ho, K.
A., Eachus, R. A., Ham, T. S., Kirby, J., Chang, M. C. Y., Withers, S. T., Shiba, Y., Sarpong,
R., and Keasling, J. D. (2006) Production of the antimalarial drug precursor artemisinic
acid in engineered yeast, Nature 440, 940-943.
47.
Asadollahi, M. A., Maury, J., Schalk, M., Clark, A., and Nielsen, J. (2010) Enhancement of
farnesyl diphosphate pool as direct precursor of sesquiterpenes through metabolic
engineering of the mevalonate pathway in Saccharomyces cerevisiae, Biotechnol.
Bioeng. 106, 86-96.
48.
Kim, S.-W., and Keasling, J. D. (2001) Metabolic engineering of the nonmevalonate
isopentenyl diphosphate synthesis pathway in Escherichia coli enhances lycopene
production, Biotechnol. Bioeng. 72, 408-415.
49.
Herring, C. D., Glasner, J. D., and Blattner, F. R. (2003) Gene replacement without
selection: regulated suppression of amber mutations in Escherichia coli, Gene 311, 153163.
50.
McGinness, K. E., Baker, T. A., and Sauer, R. T. (2006) Engineering controllable protein
degradation, Mo. Cell 22, 701-707.
108
REFERENCES
51.
Radley, T. L., Markowska, A. I., Bettinger, B. T., Ha, J.-H., and Loh, S. N. (2003) Allosteric
switching by mutually exclusive folding of protein domains, J. Mol. Biol. 332, 529-536.
52.
Win, M. N., and Smolke, C. D. (2007) A modular and extensible RNA-based generegulatory platform for engineering cellular function, PNAS 104, 14283-14288.
53.
Isaacs, F. J., Dwyer, D. J., Ding, C. M., Pervouchine, D. D., Cantor, C. R., and Collins, J. J.
(2004) Engineered riboregulators enable post-transcriptional control of gene expression,
Nat. Biotechnol. 22, 841-847.
54.
Gadkar, K. G., Doyle lii, F. J., Edwards, J. S., and Mahadevan, R. (2005) Estimating optimal
profiles of genetic alterations using constraint-based models, Biotechnol. Bioeng. 89,
243-251.
55.
Anesiadis, N., Cluett, W. R., and Mahadevan, R. (2008) Dynamic metabolic engineering
for increasing bioprocess productivity, Metab. Eng. 10, 255-266.
56.
Farmer, W. R., and Liao, J. C. (2000) Improving lycopene production in Escherichia coli by
engineering metabolic control, Nat. Biotechnol. 18, 533-537.
57.
Kobayashi, H., KA rn, M., Araki, M., Chung, K., Gardner, T. S., Cantor, C. R., and Collins, J.
J. (2004) Programmable cells: Interfacing natural and engineered gene networks, Proc.
Nati. Acad. Sci. U. S. A. 101, 8414-8419.
58.
Tsao, C.-Y., Hooshangi, S., Wu, H.-C., Valdes, J. J., and Bentley, W. E. (2010) Autonomous
induction of recombinant proteins by minimally rewiring native quorum sensing regulon
of E. coli, Metab. Eng. 12, 291-297.
59.
DeLisa, M. P., Valdes, J. J., and Bentley, W. E. (2001) Quorum signaling via AI-2
communicates the "Metabolic Burden" associated with heterologous protein production
in Escherichia coli, Biotechnol. Bioeng. 75, 439-450.
60.
Leonard, E., Nielsen, D., Solomon, K., and Prather, K. J. (2008) Engineering microbes with
synthetic biology frameworks, Trends Biotechnol. 26, 674-681.
61.
Andrianantoandro, E., Basu, S., Karig, D. K., and Weiss, R. (2006) Synthetic biology: new
engineering rules for an emerging discipline, Mol. Syst. Biol. 2.
62.
Endy, D. (2005) Foundations for engineering biology, Nature 438, 449-453.
109
REFERENCES
63.
Alper, H., Fischer, C., Nevoigt, E., and Stephanopoulos, G. (2005) Tuning genetic control
through promoter engineering, Proc. Natl. Acad. Sci. U. S. A. 102, 12678-12683.
64.
Hammer, K., Mijakovic, I., and Jensen, P. R. (2006) Synthetic promoter libraries - tuning
of gene expression, Trends Biotechnol. 24, 53-55.
65.
Cox, R. S., Surette, M. G., and Elowitz, M. B. (2007) Programming gene expression with
combinatorial promoters, Mol. Syst. Biol. 3.
66.
Anderson, J. C., Voigt, C. A., and Arkin, A. P. (2007) Environmental signal integration by a
modular AND gate, Mol. Syst. Biol. 3.
67.
Feng, J., Atkinson, M. R., McCleary, W., Stock, J. B., Wanner, B. L., and Ninfa, A. J. (1992)
Role of phosphorylated metabolic intermediates in the regulation of glutamine
synthetase synthesis in Escherichia coli, J. Bacteriol. 174, 6061-6070.
68.
Chandran, D., Bergmann, F. T., and Sauro, H. M. (2009) TinkerCell: modular CAD tool for
synthetic biology, Journal of biological engineering 3, 19.
69.
Hill, A. D., Tomshine, J. R., Weeding, E. M. B., Sotiropoulos, V., and Kaznessis, Y. N.
(2008) SynBioSS: the synthetic biology modeling suite, Bioinformatics 24, 2551-2553.
70.
Myers, C. J., Barker, N., Jones, K., Kuwahara, H., Madsen, C., and Nguyen, N. P. (2009)
iBioSim: a tool for the analysis and design of genetic circuits, Bioinformatics (Oxford,
England) 25, 2848-2849.
71.
Cai, Y., Wilson, M. L., and Peccoud, J. (2010) GenoCAD for iGEM: a grammatical
approach to the design of standard-compliant constructs, Nucl. Acids Res. 38, 26372644.
72.
Goler, J. A. (2004) BioJADE: a design and simulation tool for synthetic biological systems,
doi:1721.1/30475.
73.
Canton, B., Labno, A., and Endy, D. (2008) Refinement and standardization of synthetic
biological parts and devices, Nat. Biotech. 26, 787-793.
74.
Lee, T. S., Krupa, R. A., Zhang, F., Hajimorad, M., Holtz, W. J., Prasad, N., Lee, S. K., and
Keasling, J. D. (2011) BglBrick vectors and datasheets: A synthetic biology platform for
gene expression, J. Biol. Eng. 5, 12.
110
REFERENCES
75.
Del Vecchio, D., Ninfa, A. J., and Sontag, E. D. (2008) Modular cell biology: retroactivity
and insulation, Mol. Syst. Biol. 4.
76.
Salis, H. M., Mirsky, E. A., and Voigt, C. A. (2009) Automated design of synthetic
ribosome binding sites to control protein expression, Nat. Biotech. 27, 946-950.
77.
Pfleger, B. F., Pitera, D. J., Smolke, C. D., and Keasling, J. D. (2006) Combinatorial
engineering of intergenic regions in operons tunes expression of multiple genes, Nat
Biotech 24, 1027-1032.
78.
Ellis, T., Wang, X., and Collins, J. J. (2009) Diversity-based, model-guided construction of
synthetic gene networks with predicted functions, Nat. Biotechnol. 27, 465-471.
79.
Skretas, G., and Wood, D. W. (2005) Regulation of protein activity with small-moleculecontrolled inteins, Protein Sci 14, 523-532.
80.
Barbirato, F., Himmi, E. H., Conte, T., and Bories, A. (1998) 1,3-propanediol production
by fermentation: An interesting way to valorize glycerin from the ester and ethanol
industries, Ind. Crops Prod. 7, 281-289.
81.
Nielsen, D. R., Leonard, E., Yoon, S.-H., Tseng, H.-C., Yuan, C., and Prather, K. L. J. (2009)
Engineering alternative butanol production platforms in heterologous bacteria, Metab.
Eng. 11, 262-273.
82.
Lan, E. I., and Liao, J. C. (2012) Metabolic engineering of cyanobacteria for 1-butanol
production from carbon dioxide, Metab. Eng. 13, 353-363.
83.
Dekishima, Y., Lan, E. I., Shen, C. R., Cho, K. M., and Liao, J. C. (2011) Extending carbon
chain length of 1-butanol pathway for 1-hexanol synthesis from glucose by engineered
Escherichia coli, J. Am. Chem. Soc. 133, 11399-11401.
84.
Steen, E. J., Kang, Y., Bokinsky, G., Hu, Z., Schirmer, A., McClure, A., del Cardayre, S. B.,
and Keasling, J. D. (2009) Microbial production of fatty-acid-derived fuels and chemicals
from plant biomass, Nature 463, 559-562.
85.
Callura, J. M., Cantor, C. R., and Collins, J. J. (2012) Genetic switchboard for synthetic
biology applications, Proc. Natl. Acad. Sci. U. S. A. 109, 5850-5855.
111
REFERENCES
86.
Cho, H.-S., Seo, S. W., Kim, Y. M., Jung, G. Y., and Park, J. M. (2012) Engineering
glyceraldehyde-3-phosphate dehydrogenase for switching control of glycolysis in
Escherichia coli, Biotechnol. Bioeng., doi:10.1002/bit.24532.
87.
Lu, J., Tang, J., Liu, Y., Zhu, X., Zhang, T., and Zhang, X. (2012) Combinatorial modulation
of galP and gik gene expression for improved alternative glucose utilization, Appl.
Microbiol. Biotechnol. 93, 2455-2462.
88.
Yamanishi, M., and Matsuyama, T. (2012) A modified cre-lox genetic switch to
dynamically control metabolic flow in Saccharomyces cerevisiae, ACS Synth. Biol. 1, 172180.
89.
Curtis, S. J., and Epstein, W. (1975) Phosphorylation of D-glucose in Escherichia coli
mutants defective in glucosephosphotransferase, mannosephosphotransferase, and
glucokinase, J. Bacteriol. 122, 1189-1199.
90.
Gosset, G. (2005) Improvement of Escherichia coli production strains by modification of
the phosphoenolpyruvate : sugar phosphotransferase system, Microb. Cell Factories 4,
11.
91.
Postma,
P.
W.,
Lengeler,
J. W.,
and
Jacobson,
G.
R.
(1993)
Phosphoenolpyruvate:ca rbohyd rate phosphotransferase systems of bacteria, Microbiol.
Rev. 57, 543-594.
92.
Henderson, P. J., Giddens, R. A., and Jones-Mortimer, M. C. (1977) Transport of
galactose, glucose and their molecular analogues by Escherichia coli K12, Biochem. J.
162, 309-320.
93.
Fraenkel, D. G., Falcoz-Kelly, F., and Horecker, B. L. (1964) The utilization of glucose 6phosphate by glucokinaseless and wild-type strains of Escherichia coli, Proc. Natl. Acad.
Sci. U. S. A. 52, 1207-1213.
94.
De Anda, R., Lara, A. R., Hernandez, V., Hernandez-Montalvo, V., Gosset, G., Bolivar, F.,
and Ramirez, 0. T. (2006) Replacement of the glucose phosphotransferase transport
system by galactose permease reduces acetate accumulation and improves process
performance of Escherichia coli for recombinant protein production without impairment
of growth rate, Metab. Eng. 8, 281-290.
95.
Flores, N., Xiao, J., Berry, A., Bolivar, F., and Valle, F. (1996) Pathway engineering for the
production of aromatic compounds in Escherichia coli, Nat. Biotechnol. 14, 620-623.
112
REFERENCES
96.
Hernandez-Montalvo, V., Martinez, A., Hernandez-Chavez, G., Bolivar, F., Valle, F., and
Gosset, G. (2003) Expression of galP and gik in a Escherichia coli PTS mutant restores
glucose transport and increases glycolytic flux to fermentation products, Biotechnol.
Bioeng. 83, 687-694.
97.
PNNL, and NREL. (2004) Top value added chemicals from biomass volume I-Results of
screening for potential candidates from sugars and synthesis gas, (Werpy, T., and
Peterson, G., Eds.), EERE.
98.
Ramachandran, S., Fontanille, P., Pandey, A., and Larroche, C. (2006) Gluconic acid:
properties, applications and microbial production, Food Technol. Biotechnol 44, 185195.
99.
Weickert, M. J., and Adhya, S. (1993) The galactose regulon of Escherichia coli, Mol.
Microbiol. 10, 245-251.
100.
Semsey, S., Krishna, S., Sneppen, K., and Adhya, S. (2007) Signal integration in the
galactose network of Escherichia coli, Mol. Microbiol. 65, 465-476.
101.
Saier Jr, M. H., and Ramseier, T. M. (1996) The catabolite repressor/activator(Cra)
protein of enteric bacteria, J. Bacteriol. 178, 3411-3417.
102.
Mendoza-Vargas, A., Olvera, L., Olvera, M., Grande, R., Vega-Alvarado, L., Taboada, B.,
Jimenez-Jacinto, V., Salgado, H., Juarez, K., Contreras-Moreira, B., Huerta, A. M.,
Collado-Vides, J., and Morett, E. (2009) Genome-wide identification of transcription
start sites, promoters and transcription factor binding sites in E. coli, PLoS One 4, e7526.
103.
Meyer, D., Schneider-Fresenius, C., Horlacher, R., Peist, R., and Boos, W. (1997)
Molecular characterization of glucokinase from Escherichia coli K-12, J. Bacteriol. 179,
1298-1306.
104.
Olvera, L., Mendoza-Vargas, A., Flores, N., Olvera, M., Sigala, J. C., Gosset, G., Morett, E.,
and Bolivar, F. (2009) Transcription analysis of central metabolism genes in Escherichia
coli. Possible roles of sigma38 in their expression, as a response to carbon limitation,
PLoS One 4, e7466.
105.
Datsenko, K. A., and Wanner, B. L. (2000) One-step inactivation of chromosomal genes
in Escherichia coli K-12 using PCR products, Proc. Natl. Acad. Sci. U. S. A. 97, 6640-6645.
113
REFERENCES
106.
Min, B. E., Seo, S. W., and Jung, G. Y. (2012) Switching control of an essential gene for
reprogramming of cellular phenotypes in Escherichia coli, Biotechnol. Bioeng. 109, 18751880.
107.
Paalme, T., Elken, R., Kahru, A., Vanatalu, K., and Vilu, R. (1997) The growth rate control
in Escherichia coli at near to maximum growth rates: the A-stat approach, Antonie van
Leeuwenhoek 71, 217-230.
108.
Zhuang, K., Vemuri, G. N., and Mahadevan, R. (2011) Economics of membrane
occupancy and respiro-fermentation, Mol. Syst. Biol. 7, 500.
109.
Varma, A., and Palsson, B. 0. (1993) Metabolic capabilities of Escherichia coli: 1.
synthesis of biosynthetic precursors and cofactors, Journal of Theoretical Biology 165,
477-502.
110.
Varma, A., and Palsson, B. 0. (1993) Metabolic Capabilities of Escherichia coli 11.Optimal
Growth Patterns, Journal of Theoretical Biology 165, 503-522.
111.
Bremer, H., and Dennis, P. P. (2008) Modulation of Chemical Composition and Other
Parameters of the Cell at Different Exponential Growth Rates, in EcoSal-Escherichia coli
and Salmonella: Cellular and Molecular Biology (A. B6ck, R. C. I., J. B. Kaper, P. D. Karp, F.
C. Neidhardt, T. Nystr6m, J. M. Slauch, C. L. Squires, and D. Ussery, Ed.), ASM Press,
Washington, DC.
112.
Bausch, C., Peekhaus, N., Utz, C., Blais, T., Murray, E., Lowary, T., and Conway, T. (1998)
Sequence analysis of the Gntil (subsidiary) system for gluconate metabolism reveals a
novel pathway for L-idonic acid catabolism in Escherichia coli, J. Bacteriol. 180, 37043710.
113.
Isturiz, T., Palmero, E., and Vitelli-Flores, J. (1986) Mutations affecting gluconate
catabolism in Escherichia coli. Genetic mapping of the locus for the thermosensitive
gluconokinase, J. Gen. Microbiol. 132, 3209-3219.
114.
Sambrook, J., and Russell, D. W. (2001) Molecular cloning: a laboratory manual, Cold
Spring Harbor Laboratory Press, Cold Spring Harbor.
115.
Yuan, L. Z., Rouviere, P. E., LaRossa, R. A., and Suh, W. (2006) Chromosomal promoter
replacement of the isoprenoid pathway for enhancing carotenoid production in E. coli,
Metab. Eng. 8, 79-90.
114
REFERENCES
116.
Cherepanov, P. P., and Wackernagel, W. (1995) Gene disruption in Escherichia coli: TcR
and KmR cassettes with the option of Flp-catalyzed excision of the antibiotic-resistance
determinant, Gene 158, 9-14.
117.
Lampel, K. A., Uratani, B., Chaudhry, G. R., Ramaley, R. F., and Rudikoff, S. (1986)
Characterization of the developmentally regulated Bacillus subtilis glucose
dehydrogenase gene, J. Bacteriol. 166, 238-243.
118.
Morales, V. M., Bsckman, A., and Bagdasarian, M. (1991) A series of wide-host-range
low-copy-number vectors that allow direct screening for recombinants, Gene 97, 39-47.
119.
Amann, E., and Brosius, J. (1985) "ATG vectors' for regulated high-level expression of
cloned genes in Escherichia coli, Gene 40, 183-190.
120.
Guzman, L. M., Belin, D., Carson, M. J., and Beckwith, J. (1995) Tight regulation,
modulation, and high-level expression by vectors containing the arabinose PBAD
promoter, J. Bacteriol. 177, 4121-4130.
121.
DiPietro, D. L., and Weinhouse, S. (1960) Hepatic glucokinase in the fed, fasted, and
alloxan-diabetic Rat, J. Biol. Chem. 235, 2542-2545.
122.
Strecker, H. J. (1955) [44] Glucose dehydrogenase from liver, in Methods Enzymol., pp
335-339, Academic Press.
123.
Fujita, Y., Ramaley, R., and Freese, E. (1977) Location and properties of glucose
dehydrogenase in sporulating cells and spores of Bacillus subtilis, J. Bacteriol. 132, 282293.
124.
Zor, T., and Selinger, Z. (1996) Linearization of the Bradford protein assay increases its
sensitivity: theoretical and experimental studies, Anal. Biochem. 236, 302-308.
125.
Klumpp, S., Zhang, Z., and Hwa, T. (2009) Growth rate-dependent global effects on gene
expression in bacteria, Cell 139, 1366-1375.
126.
Scott, M., Gunderson, C. W., Mateescu, E. M., Zhang, Z., and Hwa, T. (2010)
Interdependence of cell growth and gene expression: origins and consequences, Science
330, 1099-1102.
115
REFERENCES
127.
Heyland, J., Blank, L. M., and Schmid, A. (2011) Quantification of metabolic limitations
during recombinant protein production in Escherichia coli, J. Biotechnol. 155, 178-184.
128.
Shimizu, Y., Shimizu, N., and Hayashi, M. (1973) In vitro repression of transcription of
the tryptophan operon by trp repressor, Proc. Natl. Acad. Sci. U. S. A. 70, 1990-1994.
129.
Shoeman, R., Redfield, B., Coleman, T., Greene, R. C., Smith, A. A., Brot, N., and
Weissbach, H. (1985) Regulation of methionine synthesis in Escherichia coli: Effect of
metJ gene product and S-adenosylmethionine on the expression of the metF gene, Proc.
Nat/. Acad. Sc.i U. S. A. 82, 3601-3605.
130.
Lim, D. B., Oppenheim, J. D., Eckhardt, T., and Maas, W. K. (1987) Nucleotide sequence
of the argR gene of Escherichia coli K-12 and isolation of its product, the arginine
repressor, Proc. Natl. Acad. Sci. U. S. A. 84, 6697-6701.
131.
Stojiljkovic, I., Baumler, A. J., and Hantke, K. (1994) Fur regulon in gram-negative
bacteria. Identification and characterization of new iron-regulated Escherichia coli genes
by a fur titration assay, J. Mol. Biol. 236, 531-545.
132.
Yokobayashi, Y., Weiss, R., and Arnold, F. H. (2002) Directed evolution of a genetic
circuit, Proc. Natl. Acad. Sci. U. S. A. 99, 16587-16591.
133.
Tabor, J. J., Salis, H. M., Simpson, Z. B., Chevalier, A. A., Levskaya, A., Marcotte, E. M.,
Voigt, C. A., and Ellington, A. D. (2009) A synthetic genetic edge detection program, Cell
137, 1272-1281.
134.
Lu, Z. J., and Mathews, D. H. (2008) OligoWalk: an online siRNA design tool utilizing
hybridization thermodynamics, Nucleic Acids Res. 36, W104-108.
135.
Mutalik, V. K., Qi, L., Guimaraes, J. C., Lucks, J. B., and Arkin, A. P. (2012) Rationally
designed families of orthogonal RNA regulators of translation, Nat. Chem. Biol. 8, 447454.
136.
Patzel, V., and Sczakiel, G. (1998) Theoretical design of antisense RNA structures
substantially improves annealing kinetics and efficacy in human cells, Nat. Biotechnol.
16, 64-68.
137.
Wang, J. Y., and Drlica, K. (2003) Modeling hybridization kinetics, Math. Biosci. 183, 3747.
116
REFERENCES
138.
Good, L. (2003) Translation repression by antisense sequences, Cell. Mol. Life Sci. 60,
854-861.
139.
Hernandez-Montalvo, V., Valle, F., Bolivar, F., and Gosset, G. (2001) Characterization of
sugar mixtures utilization by an Escherichia coli mutant devoid of the
phosphotransferase system, Appl. Microbiol. Biotechnol. 57, 186-191.
140.
Skerra, A. (1994) Use of the tetracycline promoter for the tightly regulated production of
a murine antibody fragment in Escherichia coli, Gene 151, 131-135.
141.
Pedelacq, J. D., Cabantous, S., Tran, T., Terwilliger, T. C., and Waldo, G. S. (2005)
Engineering and characterization of a superfolder green fluorescent protein, Nat.
Biotechnol. 24, 79-88.
142.
Reid, B. G., and Flynn, G. C. (1997) Chromophore formation in green fluorescent protein,
Biochemistry 36, 6786-6791.
143.
Wilkins, M. R., Gasteiger, E., Bairoch, A., Sanchez, J. C., Williams, K. L., Appel, R. D., and
Hochstrasser, D. F. (1999) Protein identification and analysis tools in the ExPASy server,
Methods in molecular biology (Clifton, N.J 112, 531-552.
144.
Ogawa, T., Mori, H., Tomita, M., and Yoshino, M. (2007) Inhibitory effect of
phosphoenolpyruvate on glycolytic enzymes in Escherichia coli, Res. Microbiol. 158, 159163.
145.
Fendt, S.-M., Buescher, J. M., Rudroff, F., Picotti, P., Zamboni, N., and Sauer, U. (2010)
Tradeoff between enzyme and metabolite efficiency maintains metabolic homeostasis
upon perturbations in enzyme capacity, Molecular systems biology 6.
146.
Andersen, J. B., Sternberg, C., Poulsen, L. K., Bjorn, S. P., Givskov, M., and Molin, S.
(1998) New unstable variants of green fluorescent protein for studies of transient gene
expression in bacteria, Appl. Environ. Microbiol. 64, 2240-2246.
147.
Sadler, J. R., Sasmor, H., and Betz, J. L. (1983) A perfectly symmetric lac operator binds
the lac repressor very tightly, Proc. Natl. Acad. Sci. U. S. A. 80, 6785-6789.
148.
Vogel, U., and Jensen, K. F. (1994) The RNA chain elongation rate in Escherichia coli
depends on the growth rate, J. Bacteriol. 176, 2807-2813.
117
REFERENCES
149.
Moon, T. S., Dueber, J. E., Shiue, E., and Prather, K. L. J. (2010) Use of modular, synthetic
scaffolds for improved production of glucaric acid in engineered E. coli, Metab. Eng. 12,
298-305.
150.
Santos, C. N., Koffas, M., and Stephanopoulos, G. (2011) Optimization of a heterologous
pathway for the production of flavonoids from glucose, Metab. Eng. 13, 392-400.
151.
Leonard, E., Ajikumar, P. K., Thayer, K., Xiao, W. H., Mo, J. D., Tidor, B., Stephanopoulos,
G., and Prather, K. L. (2010) Combining metabolic and protein engineering of a
terpenoid biosynthetic pathway for overproduction and selectivity control, Proc. Natl.
Acad. Sci. U. S. A. 107, 13654-13659.
152.
Kuhlman, T. E., and Cox, E. C. (2010) Site-specific chromosomal integration of large
synthetic constructs, Nucleic Acids Res. 38, e92.
153.
Keasling, J. D. (2008) Synthetic biology for synthetic chemistry, ACS Chem. Biol. 3, 64-76.
154.
Tseng, H. C., Martin, C. H., Nielsen, D. R., and Prather, K. L. (2009) Metabolic engineering
of Escherichia coli for enhanced production of (R)- and (S)-3-hydroxybutyrate, Appl.
Environ. Microbiol. 75, 3137-3145.
155.
Fung, E., Wong, W. W., Suen, J. K., Bulter, T., Lee, S. G., and Liao, J. C. (2005) A synthetic
gene-metabolic oscillator, Nature 435, 118-122.
156.
Orth, J. D., Thiele, I., and Palsson, B. 0. (2010) What is flux balance analysis?, Nat.
Biotechnol. 28, 245-248.
157.
Fell, D. A. (1992) Metabolic control analysis: a survey of its theoretical and experimental
development, Biochem. J. 286 (Pt 2), 313-330.
158.
Elowitz, M. B., and Leibler, S. (2000) A synthetic oscillatory network of transcriptional
regulators, Nature 403, 335-338.
159.
Lu, J. L., and Liao, J. C. (1997) Metabolic engineering and control analysis for production
of aromatics: Role of transaldolase, Biotechnology and bioengineering 53, 132-138.
160.
White, D. (2007) The physiology and biochemistry of prokaryotes, 3rd ed., Oxford
University Press, New York, NY.
118
REFERENCES
161.
Unden, G., and Bongaerts, J. (1997) Alternative respiratory pathways of Escherichia coli:
energetics and transcriptional regulation in response to electron acceptors, Biochim.
Biophys. Acta. 1320, 217-234.
162.
Patnaik, R., Roof, W. D., Young, R. F., and Liao, J. C. (1992) Stimulation of glucose
catabolism in Escherichia coli by a potential futile cycle, J. Bacteriol. 174, 7527-7532.
163.
Chao, Y. P., and Liao, J. C. (1994) Metabolic responses to substrate futile cycling in
Escherichia coli, J. Biol. Chem. 269, 5122-5126.
164.
Goodsell, D. S. (1991) Inside a living cell, Trends Biochem. Sci. 16, 203-206.
165.
Prats, R., and de Pedro, M. A. (1989) Normal growth and division of Escherichia coli with
a reduced amount of murein, J. Bacteriol. 171, 3740-3745.
166.
Smit, J., Kamio, Y., and Nikaido, H. (1975) Outer membrane of Salmonella typhimurium:
chemical analysis and freeze-fracture studies with lipopolysaccharide mutants, J.
Bacteriol. 124, 942-958.
167.
Ganong, B. R., and Raetz, C. R. (1982) Massive accumulation of phosphatidic acid in
conditionally lethal CDP-diglyceride synthetase mutants and cytidine auxotrophs of
Escherichia coli, J. Biol. Chem. 257, 389-394.
168.
Voet, D., Voet, J. G., and Pratt, C. W. (1998) Fundamentals of Biochemistry, John Wiley &
Sons, Inc, New York, NY.
169.
Neidhardt, F., Ingraham, J., and Schaechter, M. (1990) Physiology of the Bacterial Cell: A
Molecular Approach, Sinauer Associates, Sunderland, MA.
170.
Bennett, B. D., Kimball, E. H., Gao, M., Osterhout, R., Van Dien, S. J., and Rabinowitz, J.
D. (2009) Absolute metabolite concentrations and implied enzyme active site occupancy
in Escherichia coli, Nat. Chem. Biol. 5, 593-599.
171.
Holm, A. K., Blank, L. M., Oldiges, M., Schmid, A., Solem, C., Jensen, P. R., and Vemuri, G.
N. (2010) Metabolic and transcriptional response to cofactor perturbations in
Escherichia coli, J. Biol. Chem. 285, 17498-17506.
119
REFERENCES
172.
Walmsley, A. R., Barrett, M. P., Bringaud, F., and Gould, G. W. (1998) Sugar transporters
from bacteria, parasites and mammals: structure-activity relationships, Trends Biochem.
Sci. 23, 476-48 1.
173.
Miller, B. G., and Raines, R. T. (2004) Identifying latent enzyme activities: substrate
ambiguity within modern bacterial sugar kinases, Biochemistry 43, 6387-6392.
174.
Ramaley, R. F., and Vasantha, N. (1983) Glycerol protection and purification of Bacillus
subtilis glucose dehydrogenase, J. Biol. Chem. 258, 12558-12565.
175.
Bon, M., McGowan, S. J., and Cook, P. R. (2006) Many expressed genes in bacteria and
yeast are transcribed only once per cell cycle, FASEB J 20, 1721-1723.
176.
Marlin, T. E. (2000) Process control : designing processes and control systems for
dynamic performance, 2nd ed., McGraw-Hill, Boston.
177.
Keiler, K. C., Waller, P. R., and Sauer, R. T. (1996) Role of a peptide tagging system in
degradation of proteins synthesized from damaged messenger RNA, Science 271, 990993.
178.
Lucks, J. B., Qi, L., Mutalik, V. K., Wang, D., and Arkin, A. P. (2011) Versatile RNA-sensing
transcriptional regulators for engineering genetic networks, Proc. Nati. Acad. Sci. U. S. A.
108, 8617-8622.
179.
Ham, T. S., Lee, S. K., Keasling, J. D., and Arkin, A. P. (2006) A tightly regulated inducible
expression system utilizing the fim inversion recombination switch, Biotechnol. Bioeng.
94,1-4.
180.
Rackham, 0., and Chin, J. W. (2005) A network of orthogonal-ribosome mRNA pairs, Nat.
Chem. Biol. 1, 159-166.
181.
You, L. C., Cox, R. S., Weiss, R., and Arnold, F. H. (2004) Programmed population control
by cell-cell communication and regulated killing, Nature 428, 868-871.
182.
Xia, B., Bhatia, S., Bubenheim, B., Dadgar, M., Densmore, D., and Anderson, J. C. (2011)
Developer's and user's guide to Clotho v2.0 A software platform for the creation of
synthetic biological systems, Methods Enzymol. 498, 97-135.
120
APPENDICES
121
Al| Generating a recA expression system for chromosomal manipulation
Al| Generating a recA expression system for chromosomal
manipulation
A1.1
Introduction
The work performed in this dissertation was primarily performed in recA E. coli mutants.
However, the methods of chromosomal manipulations needed to engineer these strains, such as
A-Red recombination and P1 transduction, require homologous recombination which benefit
from recA expression (Datsenko and Wanner, 2000; Sandri and Berger, 1980; Wang et al., 2006;
Zabrovitz et al., 1977).
Commercial
plasmids that express recA as a part of a A-Red
recombination system exist (Wang et al., 2006); these systems, however, are encumbered by
intellectual property rights that restrict their use for routine operations. Thus, in-house recA
expression systems were generated for routine manipulations.
A1.2
Materials & methods
Cloning operations conducted according to standard practices, unless otherwise noted
(Sambrook and Russell, 2001). PCR amplifications were completed with Phusion High Fidelity
DNA Polymerase (NEB, Ipswich, MA) and oligonucleotides from Sigma-Genosys (St. Louis, MO,
Table A - 1).
Plasmids were selected for and maintained with 100 pg/ml ampicillin in LB
medium at 30 0 C to prevent curing. Plates were irradiated with UV at 254 nm and 10 J/cm
2
as
indicated in a Stratalinker* UV Crosslinker (Stragene, now Agilent Technologies, Santa Clara, CA)
graciously provided by Prof. Natalie Kuldell.
As these expression systems are needed only transiently for routine operation, curable
vectors were constructed.
The temperature sensitive origin of replication from pKD46 was
selected (Datsenko and Wanner, 2000).
To facilitate inducible expression of recA for P1
transduction protocols, the origin of pKD46 and the Tet cassette and promoters of pKVS45
122
APPENDICES
(Chapter 3) were amplified with 40 nt overhangs to facilitate homologous recombination with
themselves and recA. Similarly, for X-Red recombination protocols, pKD46 was amplified with
40 nts overhangs homologous to recA.
genome using PCR.
recA from E. coli MG1655 was amplified from the
These were assembled in an equimolar one pot isothermal in vitro
recombination reaction as described by Gibson et al (2009) to make pKVSRecA, a tet-inducible
recA
plasmid,
and
pKD46RecA, an
arabinose inducible operon containing
the A-Red
recombinases and recA, respectively. Assembly products were then used to transform E. coli
DH10B. The resulting colonies were screened and isolated using standard practices (Sambrook
and Russell, 2001).
Table A - II Oligonucleotides used
Sequence ( 5 '->3')a
Primer
pKVSRecA
SCTerm
tgatagcgaaggcgtagcagaaactaacgaagatttttaaAATGGCGATGACGC
SCAmp
ATAAATCGATGCAGGTGG
KVSTetR
gcgcacatttccccgaaaagtgccacctgcatcgatttatTTAAGACCCACTTTCACATTT
KVSR0040
tgttttcgtcgatagccatttttactcctgtcatgccgGGTAAGTGCTCAGTATCTCTATCACTGATA
pKD46RecA
pKD46F
tgatagcgaaggcgtagcagaaactaacgaagatttttaaCGCATCCTCACGATAATATC
pKD46R
tgttttcgtcgatagccatttttactcctgtcatgccgGGTAATCATCGCCATTGCTC
RecA
RecAF
cggcatgacaggagtaa
RecAR
ttaaaaatcttcgttagtttctg
aHomologous sequences for recombination are indicated in lowercase.
Introduced spacer sequence is indicated with ITALICS.
A1.3
Results & Discussion
To confirm conferral of a RecA* phenotype by these plasmids, DH1OB (RecA) was
transformed with isolated preparations of pKD46RecA and pKVSRecA and irradiated with UV at
254 nm. UV radiation introduces double strand breaks into the chromosome of the host which
may only be repaired effectively in the presence of recA (Krasin and Hutchinson, 1977). Cells
unable to repair these breaks cannot replicate and die. As a RecA control, DH10B with pKD46
was evaluated. Similarly, MG1655 (RecA*) with pKD46 was tested as a positive control. 10 pl of
123
Al I Generating a recA expression system for chromosomal manipulation
culture grown overnight in selective LB medium was spot plated in triplicate on selective LB
plates and exposed to UV radiation before being induced with 250 ng/ml anhydrodtetracycline
(aTc) or 0.5% L-arabinose (L-Ara) and incubated overnight at 30 *C. After approximately 20 h,
the plates were photographed (Figure A - 1). As expected, the unexposed control plate had
growth under all conditions. Similarly, on the exposed plate, the RecA* MG1655 grew under all
conditions while the DH10B RecA~ [pKD46] control failed to grow. pKD46RecA conferred a RecA*
phenotype solely when induced with L-Ara as anticipated and enabled chromosomal
manipulation as desired (Chapters 2-3). pKVSRecA, however, seemed to constitutively express
recA leading to growth when exposed to UV regardless of inducer used. Subsequent sequencing
of the plasmid revealed a point mutation in tetR which may prevent its binding and regulation of
the tet-promoter resulting in the phenotype seen.
Despite this, pKVSRecA still yields viable
colonies suitable for P1 transduction applications (Geisa Lopes, personal communication).
10 J/cm 2 @ 254 nm
No UV
MG1655
[pKD46] (+ve)
DHIOB
[pKD46) (-ve)
DHIOB
[pKVSRecA]
DHIOB
[pKD46RecA+]
+aTc
-
+
-
-
+
.
+L-Ara
-
-
+
-
-
+
Figure A - 11 Photograph of exposed and unexposed plates after 20 h.
124
APPENDICES
A1.4
Works Cited
Datsenko, K. A., Wanner, B. L., 2000. One-step inactivation of chromosomal genes in Escherichia
coli K-12 using PCR products. Proceedings of the National Academy of Sciences of the
United States of America. 97, 6640-6645.
Gibson, D. G., et al., 2009. Enzymatic assembly of DNA molecules up to several hundred
kilobases. Nat Meth. 6, 343-345.
Krasin, F., Hutchinson, F., 1977. Repair of DNA double-strand breaks in Escherichia coli, which
requires recA function and the presence of a duplicate genome. J. Mol. Biol. 116, 81-98.
Sambrook, J., Russell, D. W., 2001. Molecular cloning: a laboratory manual. Cold Spring Harbor
Laboratory Press, Cold Spring Harbor.
Sandri, R. M., Berger, H., 1980. Bacteriophage P1-mediated generalized transduction in
Escherichia coli: fate of transduced DNA in rec+ and recA- recipients. Virology. 106, 1429.
Wang, J., et al., 2006. An improved recombineering approach by adding RecA to lambda Red
recombination. Mol. Biotechnol. 32, 43-53.
Zabrovitz, S., et al., 1977. Growth of bacteriophage P1 in recombination-deficient hosts of
Escherichia coli. Virology. 80, 233-248.
125
A21 Extending the Glk expression family
A2| Extending the Glk expression family
A2.1
Introduction
Despite the successes with the Anderson promoter library, its crude resolution (20
promoters) hindered attempts to probe more precisely the phenotypic boundaries observed
and develop a richer dataset.
The BIOFAB through their C - dog project, however, offers
hundreds of promoters which may fill this need. More importantly, included in their dataset is
one member of the Anderson promoter library which allows for some degree of performance
mapping between characterizations of the two datasets.
A2.2
Materials and Methods
As an initial study, 3 promoters from the BIOFAB collection were integrated into
KTSO221G as described in Section 2.4.1 with primers as described in Table A - 2 (Sigma Genosys,
St. Louis, MO). Culture conditions and enzymes assays as described in Sections 2.4.2 and 2.4.4.
A2.3
Results and Discussion
Within the BIOFAB dataset lies the promoter J23101 (promoter strength = 0.70). Using
this as a basis, several BIOFAB promoters were selected to extend the Glk expression family that
lay both within the existing points as described in Table A - 3. Initially, KTS14221G, KTS16221G
and KTS19221G were generated.
126
CI
0
0
C
0
CL
Name
Sequence 5 '- 3 ,a
Glk mutant construction & integration
g/k1FruR
CTATGTCGACTAGATTCAAACGGGGTAATTGTGTGACCCAGA
glk2K1422
CTATGTCGACTCCACACATTATACGAGCCGGATGATTAACTGGGAACCTAGGGCCGGAAAGGCA
g1k2K1622
CTATGTCGACATCCACACATTATACGAGCCGGATGATTAATTGTCAACCTAGGGCCGGAAAGGCA
g/k2K1922
CTATGTCGACTCCACACATTATACGAGCCGGATGATTAACATTCAACCTAGGGCCGGAAAGGCA
glk kan
CTATTCCTTA TGCGGGGTCAGATACTTAGTTTGCCCAGCTTGCAAAA AGGCATCGCTGCAATTGGTGTGTAGGCTGGAGCTGCTTC
K1422/1922 insert
K1622 insert
CACA TCACCGACTAA TGCATACTTTGTCATTCTTCAACTGCTCCGCTAAAGTCAAAA
TAATTCTTTTCCACACATTATACGAGCCG
CACATCACCGACTAATGCATACTTTGTCATTCTTCAACTGCTCCGCTAAAGTCAAAATAATTCTTTATCCACACATTATACGAGCCG
(A
C-
Introduced mutations (promoters or other) are in boldface type; homologous sequences for recombination are
underlined and italicized; restriction sites used for cloning are underlined.
aAll oligonucleotides purchased from Sigma-Genosys,
St. Louis, MO.
z
N)
A2| Extending the Glk expression family
Table A - 31 BIOFAB promoters tested
BioFab Part Name
Strain Name
KTS9221G
KTS14221G
Strength relative to J23100
apFAB263
KTS16221G
KTS62216
apFAB60
KTS19221G
apFAB253
0.10
0.09
0.08
0.06
KTS8221G
0.05
0.04
As an initial assessment, the mutants were grown in triplicate in minimal medium
supplemented with glycerol and assayed for Glk activity in mid-exponential phase (OD 600 ~ 0.5).
In spite of previous characterization at the BIOFAB, the new context of the GIk system provided
wholly unpredicted results (Figure A - 2A). Given that an infinite number of straight lines may
be plotted through
a single datapoint, it is unsurprising that the BIOFAB
promoter
characterization does not map precisely to the Anderson promoter dataset. What is interesting,
however, is that even between the BIOFAB promoters themselves, relative trends do not persist
when switched to a novel context (Figure A - 2A). Such observations, however, have been
noted previously (Vivek Mutalik, personal communication) and genetic motifs have been
developed to control this issue (Mutalik, V et al., unpublished manuscript).
Evaluating these
constructs in minimal medium supplemented with glucose, only KTS14221G of the BIOFAB
promoters is viable (Figure A - 2B). Here, its performance also represents a significant outlier of
the Glk activity vs specific growth rate trend noted previously (§2.2.2) which cannot be as easily
explained.
Moreover, KTS16221G which should lie within the viable range of Glk (§2.2.2) as
predicted by its glycerol performance,
was not viable.
However, given the relative
underperformance of KTS14221G in going from glycerol to glucose supplemented medium
(weakest activity in glucose but
unanticipated.
2 nd
highest activity after KTS9221G in glycerol), this is not wholly
One potential hypothesis for the drastic differences in performance is
differences in the physical architecture of the two promoter libraries. Nevertheless, given the
128
APPENDICES
challenges which arise from the context dependence of biological systems, efforts with the
BIOFAB promoters were abandoned.
A
1.21.1
B
*
Anderson Promoters
-
BIOFAB Promoters
Fit of Anderson data
A
0-45.
1.0-
040-
= Anderson Promoters
A BIOFAB Promoter
Fit of Anderson data
0.90.35-
0.8-
0.300.6
05
i
..
2
0.25-
0.20-
0,3
0.1
0.02
0.04
0.10-
A
e
0.00
0.15-
A
0,2
0.6
008
Predicted promoter strength
010
0.2
0.3
0.4
0.5
Gik Activity (U/mg)
Figure A - 21 Performance of selected BIOFAB promoters vs the Anderson library
A) Glk activity as a function of predicted promoter strength from cultures grown in 50 ml M9(0.4%
glycerol). B) Specific growth rate vs. Glk activity of cultures grown in 50 ml M9(1.5% glucose). Data
represents the triplicate mean ± SD.
Predicted promoter strength is relative to J23100
(http://partsregistry.org).
129
A3.1
Gdh-Glk competition model
A31 MATLAB Code
A3.1 Gdh-Glk competition model
function kinetic-yield explore
tic
clc
clf
%Simulates the role of the relative kinetics of Gdh and Glk on
Gluconate
%%%%%O-%%%%%0%%%%0%%%%%%%.KINETIC
PARAMETERS%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%00000
%Glk @ R/T, pH = 7.5, Miller 2004
%(Miller has Km order of magnitude lower than Meyer 1997)
%[=] M
Kmg = 7.6e-5;
%[=] 1/s
k2_g = 410;
%Gdh @ 23 C, pH =6.5, Ramaly, 1983
%[=] M
Kmh = 4.8e-3;
%[=] 1/s
k2_h = 0.772;
PARAMETERS (NEB)%%%%%%%%%%%%%%O%%%%%%%%%%%%%%%%%%
%http://www.neb.com/nebecomm/tech-reference/generaldata/proteindata.a
sp
protcell = 1.55e-13;
volcell = le-15;
%[=]g,
%[=]L,
mass of protein per cell
cell volume
000000000000%0%000000000000000000000000000000000000000000000000000000000
%%SCAL ING%%%
%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%% %%%%%%%%%%%%%%%
%%%M
Glkmax = 0.5;
Gdh-max = 40;
%[=]
% [=]
U/mg; experimental
U/mg; experimental
(approx)
(approx)
k = linspace(0,1, 50);
%dimensionless Gdh expression
h = 0.8;
Gin = le-3; % le-6; le-4, 2e-4, 3e-4, 7e-4, le-3, le-2, le-1, 1, 10];
%[=] M/s; Assuming GalP turnover of - 50 000 molecules/min==>le-6;
Walmsley, TIBS, 1998
% %Defining Gdh regimes
Gdhreal(:,4)
=
27.51299*((k*Glkmax).A8.2896)./(0.24167A8.2896+(k*Glkmax).^8.2896);
%100 uM induction
Gdhreal(:,3) =
22.9759*((k*Glkmax) .^6.1879) ./(0.2667^6.1879+(k*Glkmax) . 6.1879);
%50 uM induction
=
Gdhreal(:,2)
11.69235*((k*Glkmax) . 7.41262) ./(0.1812^7.41262+(k*Glkmax) . 7.41262);
%25 uM induction
130
APPENDICES
Gdhreal(:,
Gdh
1) = h*Gdh max;
%constant
Gdh = Units2Mole(k*Gdh max, k2_h, prot cell, volcell);
%Defining Glk values
Glk = Units2Mole(k*Glkmax, k2_g, prot cell, volcell);
[GDH, GLK]
= meshgrid(Gdh, Glk);
for 1 = 1:
size(Gin, 2
for i =
1:50
clear Glu;
clear t;
clear x;
clear y;
clear
titers
for
j
= 1:50
[t Glu] = ode23s(@(t, g) dGdtau(t, g , Gin(l), k2_g,
Km_g, Kmh, GLK(i,j), GDH(i,j)), [0 72*60*60], [0 0 0 0]');
[x y]
k2_h,
= size(Glu);
titers(1:4)
= Glu(x,
1:4);
yield(i,j,l) = titers(2)./titers(4)*100;
Gnt(i,j,1)= titers(2);
end
end
end
Gdhreal = Gdhreal/Gdhmax;
[relGdh relGlk] = meshgrid(k,k);
figure(1); clf(1)
hold on
surf(relGlk, relGdh, yield(:,:,1)); colorbar
shading interp;
colormap(jet);
plot3(k, Gdhreal(:, 1),100*ones(50,1),'k', k,
Gdhreal(:,2),100*ones(50,1), '-.k', k, Gdhreal(:, 3),100*ones(50,1),
':k',k, Gdh real(:, 4),100*ones(50,1), '--k');
legend('Yield','Constant Gdh', '25 uM','50 uM', '100 uM', 'Location',
'SouthOutside',
'Orientation',
'horizontal')
xlabel('Relative Glk Level');
title(['G_i n ='
xlim([0 1]);
num2str(Gin)
ylim([0 1]);
'
ylabel('Relative Gdh Activity');
M/s'])
hold off
toc
S%in silico prediction
clear Gdhreal
clear yield
k = [0.9332 0.56446 0.49844 0.29724];
Gdhreal(:,3)
=
27.51299*((k*Glkmax).^8.2896)./(0.24167A8.2896+(k*Glk max) .^8.2896);
%100 uM induction
Gdhreal(:,2)
=
22.9759*((k*Glkmax).A6.1879)./(0.2667A6.1879+(k*Glkmax).A6.1879);
%50 uM induction
131
Gdh-Glk competition model
A3.1
=
Gdhreal(:,1)
11.69235*((k*Glkmax).A7.41262)./(0.1812^7.41262+(k*Glkmax) .^7.41262);
%25 uM induction
Gdh = Units2Mole(Gdhreal, k2_h, prot cell, volcell);
Glk = Units2Mole(k*Glkmax, k2_g, prot cell, volcell);
for 1 =
1:
size(Gin, 2)
for i = 1:3
for j = 1:size(k,2)
clear Glu;
c lear t;
clear x;
clear y;
clear
titers
[t Glu] = ode23s(@(t, g) dGdtau(t, g , Gin(l), k2_g,
Km_g, Kmh, Glk(j), Gdh(j,i)), [0 72*60*60], [0 0 0 0]');
[x y]
=
k2_h,
size(Glu);
titers(1:4)
yield(j,i)
=
=
Glu(x, 1:4);
titers(2)./titers(4);
end
end
end
figure(2); clf(2)
'o-', k(1:3),
'x-',
k(1:3), yield(1:3,2),
plot(k, yield(:,l),
yield(1:3,3), 's-');
legend('25 \muM IPTG','50 \muM IPTG', '100 \muM IPTG', 'Location',
'SouthOutside', 'Orientation', 'horizontal')
xlabel('Relative Glk activity'); ylabel('Gluconate yield');
xlim([O
1]);
ylim([0 1]);
return;
function dy = dGdtau(t, G, Gin, k2_g, k2_h, Km_g, Kmh, Glk, Gdh)
Glu = G(1);
%System Balances
Gludot = Gin-k2_g*Glk*Glu/(Km g+Glu)-k2_h*Gdh*Glu/(Kmh+Glu);
Gntdot = k2_h*Gdh*Glu/(Kmh+Glu);
%Verification for Yield Calcs
G6Pdot = k2_g*Glk*Glu/(Km g+Glu);
Uptake
= Gin;
%Func return
dy =
[Gludot; Gntdot; G6Pdot; Uptake];
return;
function Act = Units2Mole(Units, k2, prot_cell, volcell)
%
Converting from U/mg to moles/second converted to mole enz/cell
%
%
U = 1 umol/min
k2 converts activity
(Vmax) to enzyme conc.
Act = Units*le-6/60*1000/k2*prot cell/volcell;
return;
132
APPENDICES
A3.2 Effect of Growth Mediated Buffering
function buffereval_f
%Illustrates the magnitude of buffering effect on changes in mRNA
abundance
clc
alpha
S.,
kd
=
17.018/90;
and Cook, P.R.
= log(2)/3.3;
McGowan, S.,
%glk transcripts/min in M9
(2006) FASEB J.
%Bon, M.,
McGowan,
1721-1723
%based on half life of 3.3 mins from %Bon, M.,
and Cook, P.R.
%Glk @ R/T, pH =
k2_g
= 410;
20,
(2006) FASEB J.
20,
1721-1723
7.5, Miller 2004
%[=] 1/s
%http://www.neb.com/nebecomm/techreference/generaldata/protein-data.a
sp
prot cell = 1.55e-13;
%[=]g, mass of protein per cell
% Calculating Steady State Concentrations
[t Xss] = ode23s(@(t,x) glk_prof(t, x, alpha, k2_g, prot cell, kd,
0.29, 0),
[0 10000],
[0 0]');
Pssval = Xss(size(Xss,l), 1);
mRNAval = Xss(size(Xss,l), 2);
Glki=mol2Units(Pssval, k2_g, prot cell);
clear Xss;
[tl Xss] = ode23s(@(t,x) glkprof(t, x, alpha*.5,
0,1),
[0 10000],
[Pssval mRNAval] ');
Glkvary = mol2Units(Xss(:,l), k2_g, prot cell);
k2_g, protcell, kd,
clear Xss;
[t2 Xss] = ode23s(@(t,x) glkprof(t,
x, alpha*.5, k2_g,
0.29,0), [0 10000],
[Pssval mRNAval] ');
Glkconst = mol2Units(Xss(:,l), k2_g, prot cell);
figure(1);
plot([0
prot cell,
kd,
clf(1);
t1(size(t1,1))],
[Glki Glki],
tl,
Glkvary,
'-.',
t2,
Glkconst,'--')
legend('Initial
steady state',
'growth mediated buffering',
'constant
\mu', 'Location', 'southwest')
xlabel
('time [min]'); ylabel ('Glk activity
[U/mg]'); ylim ([0 0.31]);
figure(2);
clf(2);
f = logspace(-2, 2, 100);
const-gain = 6e15*alpha/(kd+0.29/60)/(0.29/60)/alpha;
varygain = (sqrt(2.16144E-05A2+4*6el5.*alpha.*4.10174E21./kd) ./2./4.10174E-21-sqrt(2.16144E-05^2+4*6e5.*alpha*f.*4.10174E21./kd)./2./4.10174E-21)./(l-f)/alpha;
semilogx(f, (const_gain-vary-gain)/constgain);
ylim([0 1]); xlim([0 100])
133
A3.2
Effect of Growth Mediated Buffering
Buffering capacity
xlabel('Relative {\itglk} mRNA'); ylabel('
\newline [(K_c_o n s t a_n_t - K v a r i_a_b_1_e)/K_c_o_n s t a n_t] ')
return;
function F = glkprof(t, x, alpha, k2_g, prot cell, kd, mu, flag)
%flag - sign to calculate growth rate == 1
%Calculates dynamic profile of glk protein and mRNA concentration
= zeros(2,1);
F
%Correlation between activity and growth - my data
%growth = gamma*Activity (U/mg) + phi
= 0.933962003;
gamma
= 0.001296861;
phi
%calculates growthrate if experimental data not supplied
if flag == 1
= gamma*mol2Units(x(l),k2_g, prot_cell)+phi;
mu
end
F(1) = 6*x(2)-mu/60*x(1);
F(2) = alpha - (kd + mu/60)*x(2);
return;
function Units = mol2Units (conc, k2, prot_cell)
Converting from molecules enz/cell to units/mg
%
U = 1 umol/min
%
k2 converts activity (Vmax) to enzyme conc.
%
Units = conc/6.023e23/prot-cell*k2/1000*60*le6;
return;
134
[=]
1/h