Tertiary packing effects on secondary structure
formation in proteins
by
Daniel Louis Minor, Jr.
B.A. Biochemistry and Biophysics
University of Pennsylvania, 1989
Submitted to the Department of Chemistry
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
in Chemistry
at the
Massachusetts Institute of Technology
February 1996
© 1996 by Daniel L. Minor, Jr. All rights reserved.
The author hereby grants to MIT permission to reproduce
and to distribute publicly paper and electronic
copies of this thesis document in whole or in part.
(N
Signature of Author
Department of Chemist ry
January 10, 19196
Certified by
V
Peter S. Kim
Professor, Department of Biology
Thesis Supervisor
Accepted by
/ I
.....
;.,-,.s-.-
::
OF TECHNOLOGY'
•
"Dietmar Seyferth
Chairman, Departmental Committee on Graduate Students
MAR 04 1996 .:..:
LIBRARIES
#
_
This doctoral thesis has been examined by a Committee of the Department of
Chemistry as follows:
Professor Daniel S. Kemp
Professor Peter S. Kim
Professor Robert T. Sauer
Professor James R. Williamson
Chairman
Thesis Supervisor
Dedication
To Mom and Dad
Thank you for your unbounded love and support
and for instilling in me the values of hard work and humility
Acknowledgments
The work presented in this thesis has consumed the better part of my adult life
to this point. During this time, I have been blessed with the support and
encouragement of tremendous friends, outstanding colleagues and a wonderful
family. Whatever I write here cannot possibly convey the deep gratitude, respect
and love I have for these people.
I would like to thank those people who encouraged me in the beginning of my
academic career to follow my curiosity, Mr. Adam Edmondson, a superb teacher and
human being and Dr. Alan F. Horwitz, who took me in to his lab as a very naive but
persistent freshman, let me share his office, and gave me a place to work.
During my time in Peter's lab I have been the beneficiary of the instruction, in
direct ways and by example, of a superb group of scientists. The way I approach
science has been profoundly affected by Erin O'Shea, a friend, skiing partner and
scientific role model. She has set the standard for the type of scientist I hope to be.
Brenda Schulman has been a constant source of encouragement, advice (scientific,
personal and culinary), information and support. I thank her for her willingness to
listen to my ideas and problems. I owe much of my education in scientific writing
to Pete Petillo, who patiently read, edited, reread, re-edited, etc. many drafts of some
of the work that comprises this thesis. To Pete I also owe much gratitude for
believing in me at the times when I did not, numerous good times on the squash
court, and for many hours of therapy, computer generated and otherwise. Ton
Schumacher has been a great friend, collaborator, worthy dogfighting opponent and
my moral compass. His clarity of thought and steady demeanor were often the
perfect guide for pulling me through difficult times. Lorenz Mayr has been a terrific
friend, and a good sport as a target for my, Ton and Pete's jokes. I am forever
grateful to Zheng-yu Peng for teaching me about NMR, for his friendship and for
the great meals supplied by his wife Li. I thank Rheba Rutkowski for setting the
standard of efficiency and courtesy by which the lab still runs. Her integrity and
drive make her one of my personal role models. I also thank her for coining my
nickname, Major Minor.
Pehr Harbury was a tremendous labmate, not only because of his nonstop
curiosity and willingness to kick around ideas, but also because he shared his neverending quest to find new ways to make the lab a fun place to be. I thank Jon Staley
for his friendship and for saving a number of the lab from mountaineering near
disaster. I am very lucky to have shared a lab with Martha Oakley, Andrea Cochran,
Lawren Wu, Debbie Fass, Masaru Ueno, Steve Blacklow, Jonathan Weissman, Dave
Lockhart, Chave Carr, Bob Talanian, Kevin Lumb, Lawrence McIntosh and Jamie
McKnight. These people were integral parts of making the science in the Kim lab
first rate and my time here very enjoyable.
I thank Stefan Boresch for being a great friend and for sharing with me many
adventures on the mountains of the New England and elsewhere, as well as many
pints of Boston's best beers. Brian Krostenko is a terrific friend with whom I shared
many things including a common background and view of life. I thank him for the
many beers, hours of philosophical discussions, Wally's blues nights and for sharing
in my love of the North End. I thank, Melissa Wilbricht and Jenny Strimmel for
sharing their never-ending optimism and love of a good competitive game of cards.
I owe a special debt to Meg Moloney, Caroline Wittenberg and Bridey Maxwell.
They seemed to have endless patience for waiting for me while I ran back to lab to
do one more thing or needed to sit by the HPLC for one more hour. I thank Meg
and Caroline, in particular, for remaining loyal friends.
I owe very much to some of the companions I have had through most of my
life. Sean Donovan, Bill O'Donnell and Paul X. Tolerico are my brothers, if not in
genetic make up, than in the other ways in which it really counts. They are great
friends and a tremendous source of inspiration and support. I thank them for being
able to take the time to rescue me from the insanity of constant work, for the
occasional road trip to the North Country or ear splitting concert.
I thank my mother and father for their unending love and support. They
never stopped encouraging me to pursue my dreams or giving me the freedom and
opportunities to do so. They are the role models for the type of person I hope to be
kind, caring, humble and giving. I owe a special debt of gratitude to another family
member, Mr. Hobbes. He has been my loyal friend and companion for an number
of my years and always has the ability to lift my spirits. I also thank Jessica Downs
for the pivotal role she has played in the course of my life, not the least of which has
been helping me to figure out what to do next. I hope I can live up to all of the good
things she manages to see in me.
Finally, I will be forever grateful to Peter Kim. Peter has the knack for cutting
through to the heart of a question, even when surrounded by a sea of confusing
data. He has a tremendous nose for figuring out what is important and what is not.
He is also a tremendous writer. I hope I have gained some small fraction of his
talents as a scientist while I have been in his lab and as a writer during our late night
fax wars. I also thank him for providing me with something that I often needed to
spur my work, an adversary. Most of all, I have benefited beyond measure from the
outstanding group of scientists he as been able to organize in his group. I am
grateful for the opportunity to have been a part of a premier research lab at an
outstanding institution.
Tertiary Packing Effects on Secondary Structure
Formation in Proteins
by
Daniel Louis Minor, Jr.
Submitted to the Department of Chemistry
on December 29, 1995 in Partial Fulfillment of the
Requirements for the Degree of
Doctor of Philosophy in Chemistry
ABSTRACT
This thesis describes studies of the factors that determine secondary structure
formation in proteins using the IgG-binding domain of protein G (GB1) as a model
system. Biophysical studies of substitution of the twenty naturally occurring amino
acids at a guest position at a central p-sheet position establishes a scale for B-sheet
propensity (Chapter 2). Similar experiments conducted on an edge P-sheet position
reveal that the dominant factor determining B-sheet propensity is tertiary
interactions rather than local intrinsic conformational preferences (Chapter 3).
The limits of the effects of tertiary interactions on secondary structure formation
were explored by the design of an eleven residue peptide sequence capable of context
dependent secondary structure formation. This sequence (the 'Chameleon
sequence') folds as an a-helix or as a p-sheet when placed at positions 23-33 or 42-52
of the primary sequence of GB1 respectively. These experiments demonstrate that
context can affect not only the conformation of individual amino acids, but the
conformation of entire secondary structures. This study as well as the experiments
investigating P-sheet propensity underscore the importance of tertiary interactions
in determining protein structure.
Thesis supervisor: Peter S. Kim
Title: Professor of Biology
TABLE OF CONTENTS
Page
Dedication
Acknowledgements
Abstract
Table of Contents
Chapter 1
Introduction: Overview of Protein Secondary Structure
Chapter 2
Chapter 2 has been published as D. L. Minor, Jr. and P.S. Kim
'A Thermodynamic scale for P-sheet formation', Nature 367 660-663
(1994) @ Macmillan Magazines, Ltd. London
Chapter 3
Chapter 3 has been published as D. L. Minor, Jr. and P.S. Kim
'Context is a major determinant of 3-sheet formation',
Nature 367 660-663 (1994) @ Macmillan Magazines, Ltd. London
Chapter 4
Design of a Protein Sequence That Shows Context Dependent
Secondary Structure Formation
Chapter 5
The Role of Tertiary Interactions in Establishing Protein
Secondary Structure
Appendix I
92
Comparisons of experimental and statistical
3-sheet formation scales
Biographical Note
105
CHAPTER 1
INTRODUCTION
'The living cell is like a symphony orchestra without a conductor: its score is laid
down in its DNA and proteins are its performing instruments.' M. F. Perutz 1
Information transfer and the protein folding problem
The folded structure of a protein dictates its function. The information
determining the structure of a protein originates in the genes of a cell's DNA. The
transfer of this information from DNA into active protein culminates with the
process of protein folding. In this final step, the one dimensional primary sequence
of a protein directs the establishment of the folded three dimensional conformation
of a protein's polypeptide chain. Although the one dimensional code specifying
how protein sequences are made from the instructions in DNA has been
understood for quite some time 2 , the code for the final step remains imperfectly
understood. The challenge of current protein chemistry is to unlock this step of the
information flow from DNA to protein.
Many of the important problems in biology, such as oncogenesis and the
development of drug resistant viruses, bacteria and cells, have their root in small
changes in the DNA 'score' that lead to protein 'instruments' with altered functions
that are sometimes deleterious to the life of a cell. Elucidating how the sequence of
a protein determines its three dimensional structure should be a step towards
understanding how changes in the sequence of a protein affect its folding and
function.
Investigation of the physical chemistry that determines protein structure is
aimed at providing a set of 'rules' for understanding how sequence determines
structure. Understanding these rules is particularly important as we rapidly
approach the complete sequencing of the human genome; a project that will
provide an enormous amount of protein sequences for which there will be no
structural data. Information regarding the rules of protein folding will aid the
development of methods for predicting protein structure and function from
primary sequence information. Additionally, the rules for protein folding can be
tested and applied to the design of new proteins with novel folds and functions 3
Initial hypotheses for the nature of the protein folding code were driven by the
observation that non-random distributions of amino acids were found in the
a-helix and f-sheet secondary structures in proteins 4 . It was presumed that this bias
in some way reflected an intrinsic propensity of each amino acid for a particular set
of 0 and x dihedral angles consistent with the conformation of a particular
secondary structure. These biases were believed to originate from short range
interactions between the amino acid sidechain and the peptide backbone.
Interpretation of the individual propensities in terms of energetics suggested that
the conformational biases were relatively weak (ranges of +0.4 to -0.4 kcal mol-1).
Nonetheless, summed over a number of contiguous residues, the inferred small but
distinct conformational preferences could be energetically significant for establishing
a particular protein fold. These observations provided a tractable approach for
trying to understand protein structure in terms of local interactions. From this
perspective, one primarily needed to determine which parts of the sequence coded
for the secondary structure 'building blocks'. Once these were identified, it was
thought they could be arranged to give the overall fold of a protein.
Competing Models for Folding, Local vs. Nonlocal forces
Two models of protein folding, the framework model and the collapse model,
represent the extremes of how the information in the protein sequence might be
organized. The framework model assumes that local interactions among near
neighbors in the protein sequence are the main determinants of protein structure .
Thus, much of the information directing the final fold of a protein is contained in
the parts of the sequence that code for secondary structures and turns. These
secondary structure 'building blocks' fold and then assemble via tertiary interactions.
The information hierarchy in this model is 'linear'. Primary sequence encodes
secondary structures that are then assembled in a particular way to make tertiary
structures. This view of protein structure is consistent with the interpretation of
secondary structure propensities as local phenomena.
An alternative view of folding is given by the collapse model 6 , 7. In this case
the code is not localized within windows of secondary structure, but is distributed
throughout the sequence in the hydrophobic/hydrophilic (H/P) patterning of the
sidechains. The primary driving force for folding in this case is nonlocal. Secondary
structures result as a consequence of the optimal ways in which a polypeptide chain
can pack against itself to minimize the contact of its hydrophobic sidechains with
water. Thus, secondary structures are a consequence rather than the cause of tertiary
structure. The information for folding is not really organized in a hierarchy in this
case. Both secondary structures and tertiary interactions can form as a result of
nonlocal interactions. For this view, secondary structure propensities reflect the
tertiary environment provided by the secondary structure for the burial of
hydrophobic surface area.
Experimental studies of secondary structure propensities
a-helices
Pioneering work was done by Scheraga and colleagues to investigate the helix
propensities of the amino acids using a host-guest system based on alkyl-glutamine
amino acid polymer hosts 8 . Incorporation of small amounts of guest amino acids
into the random copolymer affected the helix-coil transition. From this work, the
values for helix propagation by each of the amino acids were determined. In line
with what was expected from the statistical database surveys, the relative measured
ac-helix propensities of the amino acids were small, having equilibrium constants
very near values equal to 1. Thus, most amino acids were largely helix indifferent.
These results suggested that the formation of isolated ca-helices would be
energetically favorable only for relatively long peptides (-100 amino acids).
The discovery of short, stable, isolated ca-helices contradicted the predictions
from the helix-coil studies. These findings facilitated the development of a number
of short peptide systems for the study of ca-helix propensity. The investigation of
ca-helix propensity in these systems indicated that the differences in the helix
forming propensities of the amino acids were significantly larger than previously
indicated by the studies of Scheraga 9
In addition to work on peptide model systems, a number of protein model
systems have been developed to investigate cc-helix propensity: coiled coils 1 0
T4 lysozyme 1 1, 12 and barnasel3. In agreement with the results from studies of
helix formation in short peptides, these investigations indicated that there were
significant energetic differences among the amino acids for forming ca-helices.
Together, these studies suggested that (a-helixformation was largely a local
phenomena governed, to a first approximation, by the favorable or unfavorable
interactions of the sidechain with the local helical backbone geometry. From the
summation of the peptide and protein work, a consensus a-helix propensity scale
has emerged 9 . In this sense, a-helix formation is consistent with the framework
model as it seems to be governed largely by local interactions.
/-sheets
Although a-helix propensity was extensively studied, similar good model
systems were not available for experimental investigation of P-sheet propensity.
This seemed largely due to the tertiary nature of the 3-sheet structure. It was
recognized that 3-sheets were both secondary structure and tertiary structure as their
formation involved interactions between 3-strands that could be quite distant in
primary sequence 4 . Since individual 3-strands assemble through a lateral network
of hydrogen bonds, peptides that formed 1-sheets often suffered from uncontrolled
propagation. This property generally lead to systems that were not uniquely defined
molecular species which also had a strong tendency to aggregatel 5' 16
Progress has been made in the development of strategies to overcome the
aggregation tendency of individual 3-sheet forming peptides. This has been
accomplished through the use of small organic molecule templates to define strand
orientation, to control strand register or to serve as hydrogen bonding templates for
17-21
1-strands 7 - 2 1 Two basic strategies have been used. In one, an organic molecule is
used to nucleate the growth of two anti-parallel 3-strands 1 8-21 . This strategy has
yielded molecules capable of forming non-aggregated P-sheet structures in aqueous
environments. A second strategy has been to use an organic molecule as a hydrogen
bonding template to provide a scaffold of 3-sheet hydrogen bonding sites for a
covalently attached peptidel 7 .
The organic molecule template systems offer the possibility for precise control
of the local 1-sheet environment. For this reason, they may serve as ideal systems
for understanding the local conformational preferences separate from significant
tertiary effects. Reports on studies of 3-sheet formation in these systems in aqueous
environments are eagerly awaited.
This thesis deals with the identification and development of a suitable model
system for the investigation of the factors involved in determining the 3-sheet
propensities of the twenty naturally occurring amino acids. The principle questions
asked are: Are there significant energetic differences among the 3-sheet propensities
of the twenty naturally occurring amino acids? Is 3-sheet propensity governed by
local or nonlocal interactions?
The general approach used for these studies is described by the work in
Chapter 2. This chapter details the choice of the IgG-binding domain of protein
(GB1) as a model system and the investigation of the 1-sheet propensities of the
naturally occurring amino acids at a central P-strand position. GB1 is a 56 amino
acid 1-sheet containing protein having many characteristics that make it suitable as
a model system for studying 1-sheet formation. It is folded into an ubiquitin-type
xp roll2 2 with a four stranded 1-sheet on top of a single a-helix2 3 . GB1 undergoes
reversible two-state thermal denaturation with a very high transition temperature,
Tm = 87 0 C. It contains no prolines, disulfide bonds or cofactors, is monomeric, very
amenable to NMR studies and can be expressed in large quantities (> 100 mg 1-1) in
E. coli.
1-sheet residues can occur in two distinct tertiary environments, center and
edge 1-strands 14 . The investigation of the effects of tertiary context on the 1-sheet
propensities of the naturally occurring amino acids at an edge P-sheet position is
described by Chapter 3. These studies address the origin of 1-sheet propensity as
local or nonlocal and indicate that 1-sheet propensity is largely a property of tertiary
interactions. Thus, the formation of 1-sheet structure is more consistent with the
picture of folding given by the collapse model in that it is dominated by nonlocal
factors that include hydrophobic interactions. It also probably includes
consideration of detailed sidechain packing interactions.
Since nonlocal factors were found to be important in determining the 1-sheet
propensities of individual amino acids, experiments were conducted to test the
extent to which nonlocal factors could control the formation of long stretches of
both a-helix and 1-sheet secondary structures. An eleven amino acid sequence, the
'Chameleon' sequence was designed. This sequence folds as an a-helix when in one
position of the primary sequence of GB1 but as a P-sheet in another position. The
design of the Chameleon sequence and characterization of the proteins containing it
are described in Chapter 4.
Together, the studies of 1-sheet formation and the design of the Chameleon
sequence presented here provide a picture of protein structure that is most
consistent with the view described by the collapse model. The critical information
directing both j-sheet and a-helix secondary structure primarily involves nonlocal
interactions, in these cases. One significant difference from the collapse model is
that specific packing interactions seem to be important in addition to
hydrophobicity. The possible importance of packing interactions in P-sheet
structures is examined in Chapter 5.
References
1.
Perutz, M.F. Protein Structure: New Approaches to Disease and Therapy (W. H.
2.
3.
Freeman and company, New York, 1992).
Stryer, L. Biochemistry (W.H. Freeman and Company, New York, 1995).
Bryson, J.W., et al. Science 270, 935-941 (1995).
4.
5.
6.
7.
8.
9.
10.
11.
12.
Chou, P.Y. & Fasman, G.D. Biochemistry 13, 211-222 (1973).
Kim, P.S. & Baldwin, R.L. Ann. Rev. Biochem. 51, 459-89 (1982).
Dill, K.A. Biochemistry 29, 7133-7155 (1990).
Dill, K.A., et al. Prot. Sci. 4, 561-602 (1995).
Scheraga, H.A. Pure & Appl. Chem. 50, 315-324 (1978).
Chakrabartty, A. & Baldwin, R.L. Adv. Prot. Chem. 46, 141-176 (1995).
O'Neil, K.T. & DeGrado, W.F. Science 250, 646-651 (1990).
Blaber, M., Zhang, X. & Matthews, B.W. Science 260, 1637-1640 (1993).
Blaber, M., et al. J. Mol. Biol. 235, 600-624 (1994).
13. Horovitz, A., Matthews, J.M. & Fersht, A.R. J. Mol. Biol. 227, 560-568 (1992).
14. Chothia, C. J. Mol. Biol. 105, 1-14 (1976).
15. Mandel, R. & Fasman, G.D. Biopolymers 14, 1633-1649 (1975).
16. Walter, B. & Fasman, G.D. Biopolymers 16, 17-32 (1977).
17. Kemp, D.S. & Bowen, B.R. in Protein Folding (eds. Geirasch, L.M. & King, J.)
293-303 (AAAS, Washington, D.C., 1990).
18. Diaz, H., Tsang, K.Y., Choo, D., Espina, J.R. & Kelly, J.W. J. Am. Chem. Soc. 115,
3790-3791 (1993).
19. Tsang, K.Y., Diaz, H. & Kelly, J.W. J. Am. Chem. Soc. 116, 3988-4005 (1994).
20. Kemp, D.S. & Li, Z.Q. Tet. Lett. 24, 4179-4180 (1995).
21. Schneider, J.P. & Kelly, J.W. J. Am. Chem. Soc. 117, 2533-2546 (1995).
22. Orengo, C.A., Jones, D.T. & Thornton, J.M. Nature 372, 631-634 (1994).
23. Gronenborn, A.M., et al. Science 253, 657-661 (1991).
CHAPTER 2
A THERMODYNAMIC SCALE FOR P-SHEET PROPENSITY
Several model systems have been used to evaluate the a-helical propensities of
different amino acidsl-
7.
In contrast, experimental quantitation of P-sheet
preferences has been addressed in only one model system, a zinc finger peptide 8
Here we measure the relative propensity for P-sheet formation of the twenty
naturally occurring amino acids in a variant of the small, monomeric, f-sheet rich,
IgG binding domain from protein G. Amino acids substitutions were made at a
guest site on the solvent-exposed surface of the P-sheet. Several criteria were used
to establish that the mutations did not cause significant structural changes: binding
to the Fc domain of IgG, calorimetric unfolding and nuclear magnetic resonance
(NMR) spectroscopy. Characterization of the thermal stabilities of these proteins
leads to a thermodynamic scale for P-sheet propensities that spans a range of
-2 kcal mo1-1 for the naturally occurring amino acids, excluding proline. The
magnitude of the differences suggests that 1-sheet preferences can be important
determinants of protein stability.
The immunoglobin-binding domain B1 from protein G (denoted GB1) is a
small, soluble, monomeric, highly stable protein comprising a four stranded P-sheet
and a single ao-helix (Fig. la). GB1 has been shown to undergo a two-state,
reversible, thermal unfolding transition 9 ' 10 These properties make GB1 an
attractive model system for studying 1-sheet propensities. A 'guest site', into which
all twenty natural amino acids could be substituted, was chosen at a solvent-exposed
position (residue 53) of a central 3-strand of the protein. This position was chosen
in order to avoid potential effects from the edges or the ends of the sheet (Fig. la).
As we wanted to measure the intrinsic propensities for 1-sheet formation, we
constructed potential host molecules that minimized local interactions to the guest
site. Three potential host systems were constructed in which the guest position was
surrounded by small, neutral amino acids. The majority of interresidue contacts to
the guest site arise from the two residues on each of the immediately adjacent
P-strands (Fig. la). These residues, Ile 6 and Thr 44, were replaced by alanine. The
residues i-2 and i+2 from the guest site (residues 51 and 55) make fewer, more
distant contacts and were replaced simultaneously with either alanine, serine, or
threonine.
8'
Glycine, an amino acid that is expected to have a low 1-sheet propensity 11
was substituted at the guest site in order to evaluate the potential host systems. In
the system containing alanine substitutions at the i-2 and i+2 positions, with glycine
at the guest position, the circular dichroism (CD) thermal unfolding curve lacks a
clearly defined folded baseline (unpublished results). A well defined folded
baseline, however, is obtained with the glycine mutant in the systems containing
serine or threonine substitutions at the i-2 and i+2 positions. The host containing
serine, the smaller side chain substitution, at the i-2 and i+2 positions
(I6A/T44A/T51S/T55S; denoted AASS) was chosen for further study.
The twenty naturally occurring amino acids were substituted for residue 53 in
the AASS background by site directed mutagenesis. The stability of each protein
(designated AASS-53Xaa) was measured by thermal unfolding as monitored by CD
(Fig. ib). AAG values for 1-sheet formation were obtained by assuming that changes
in global stability result entirely from changes in the ability of the residue at the
guest site to adopt a P-sheet conformation. ACp for unfolding, a necessary parameter
for calculation of the temperature dependence of the free energy of unfolding, was
determined by combining data from the pH dependence of stability for AASS-53Thr,
AASS-53Phe, and AASS-53Val with the stability data obtained at pH 5.4 for each of
the guest site mutants (Fig. ic). The resultant value (ACp = 624 + 62 cal/mol-deg) is
in excellent agreement with the previously reported value for wild-type GB1
(621 ± 71 cal mol-1 deg-1) 9 and these data illustrate that the mutations made here
have little effect on ACp, as might be expected for mutations at solvent-exposed
positions 1 2 . The relative free energy differences for 1-sheet formation are listed in
Table 1.
There are differences among the mutant proteins in the magnitude of the CD
signal at 218 nm (Fig. ib). However, all the proteins bind Fc with the same affinity
as wild-type GB1, with the exception of the unfolded mutant AASS-53Pro (Table 1).
This result suggests that none of the folded molecules has undergone substantial
changes in overall conformation. In light of this result, it is likely that differences in
the baseline values of the molar ellipticity at 218 nm ([]0218 nm) arise from
contributions to the CD signal from aromatic side chains (compare refs. 13, 14) at this
wavelength (see also Fig. 1 legend).
Two molecules representing each end of the stability scale, AASS-53Thr and
AASS-53Ala, were chosen for further characterization. For both molecules, AHcal
was found to be equal to AHvan't Hoff (Fig. 1 legend), suggesting that the two-state
nature of the unfolding reaction for the wild-type GB1 protein 9 is preserved in the
variants studied here.
AASS-53Thr and AASS-53Ala were also investigated by NMR. The 1H- 15 N
correlation spectra of these two molecules are very similar except for resonances
from residues in the immediate vicinity of the guest site (Fig. 2a, b). The NMR
spectra of both proteins also contain cross-strand nuclear Overhauser effect (NOE)
patterns which are present in wild-type GB1 and are indicative of P-sheet structure
at the guest site (Fig. 2c). In particular, these data suggest that any structural changes
between the mutants are small and do not involve significant disruption of the
backbone 1-sheet structure, such as fraying of residues 53-56.
Our results indicate that there are significant differences in 1-sheet forming
propensities among the naturally occurring amino acids. The propensity scale
suggests that P-branching of the side chain favors 1-sheet formation, and the rank
order shows a modest correlation with 1-sheet preferences measured by statistical
methods 1 1 and with the rank order measured in the zinc finger system 8 (Fig. 3). It
is striking, however, that the range of AAG values for f-sheet formation obtained
here is an order of magnitude larger than the values obtained with the zinc finger
host (2.05 kcal mol-1 vs. 0.21 kcal mol-1 full scale, excluding proline and glycine).
Finally, the magnitude of P-sheet preferences obtained here is comparable to
those measured for oa-helices 1-6 , suggesting that f-sheet propensities can be as
important as a-helix propensities in determining protein stability. The
thermodynamic preferences for P-sheet formation reported here should be useful
for deciphering the rules involved in protein stability, protein folding and protein
design.
Amino acid
AAG (kcal mol - 1 )
Tm (OC)
Thr
Ile
Tyr
Phe
Val
1.1
1.0
53.7
53.0
52.5
Met
Ser
Trp
Cys
Leu
Arg
Lys
Gln
Glu
Ala
His
Asn
Asp
Gly
Pro
Ka/KaAASS
51.6
51.2
1.0
1.0
1.0
1.0
1.1
50.2
1.0
50.1
48.7
48.5
48.4
0.23
47.9
46.3
45.8
1.1
1.1
1.1
1.0
1.0
1.1
1.2
0.01
44.0
1.1
0.00
-0.02
-0.08
-0.94
-1.2
-- 3
43.8
43.3
42.9
1.1
1.0
1.0
1.1
1.0
0
0.96
0.86
0.82
0.72
0.70
0.54
0.52
0.51
0.45
0.27
35.2
30.2
<0
-5 3
Thr
Table 1 AAG values for 3-sheet formation relative to alanine, thermal melting
temperatures (Tm), and relative binding constants to human Fc for the AASS-53Xaa
proteins normalized to the binding of AASS-53Thr.
METHODS. The Ka value for wild-type GB1 is 1.4 x 108 M-1 (ref. 15). The
Ka/KaAASS-53 Thr value for wild-type GB1 is 1.1. AG values for unfolding of the
AASS-Xaa proteins were calculated using the Gibbs-Helmholtz equation,
AG(T) = AHm (1-T/Tm)-ACp [(Tm-T) + Tln(T/Tm)], assuming ACp = 624 cal mol1 deg-1 for all molecules (see Fig. ic). AAG values were calculated at 321K, the mean
of the Tm's excluding the glycine mutant, to minimize extrapolations in the
calculation of AG(T). A positive value of AAG indicates an increase in stability. The
estimated errors in determination of Tm and AG are ± 0.5 'C and ± 0.06 kcal mol-1
respectively. Fc was made by proteolytic digestion of Human IgG (Sigma 1-4506)
with papain and was purified from other digestion products by affinity (Protein G) 1 6
and gel filtration (G75) chromatography in a buffer of 10 mM phosphate, 150 mM
NaCL, pH 7.3, followed by FPLC purification using a linear gradient of NaCl (0.05 M
to 1 M) in a buffer of 50 mM sodium Acetate, pH 4.75 on Mono S resin (Pharmacia).
The purified material was >98% pure as judged by SDS polyacrylamide gel
electrophoresis. The binding assay was carried out using standard methodsl6 on 96
well microtiterplates (Costar), coated for two hours with a 100 pgg m1-1 solution of Fc
and then blocked for two hours with 3% bovine serum albumin (BSA). The assay
was carried out at 50 C using a fixed quantity of protein G-alkaline phosphatase (100
gg ml-1) (Pierce 31399X) and a variable amount of competitor protein (0.6-3.5 mM) in
the presence of 1.5% BSA in a buffer of 50 mM sodium Acetate, 150 mM NaCl, pH
5.4. Samples were allowed to equilibrate for 39-48 hrs. Control experiments
indicated that equilibrium was reached after -25 hrs. The wells were developed
with a fresh solution of 1 mg/ml p-nitrophenol phosphate (Pierce 34045X) and the
absorbance at 410 nm was read using a microplate reader (Dynatech). The ratio of
affinity constants of the competitor molecule with protein G was determined using
the equation: ( Y * Y ) / (Y - 1 ) = - KaProtein G / Kacompetitor [ competitor ], where Y is
the fraction of protein G-alkaline phosphatase bound. The ratio of affinity constants
for the mutants was normalized to AASS-53Thr, which was run as an internal
standard on each plate.
Fig. 1 a, Ribbon drawing of GB1 (based on a diagram produced with the program
RIBBON 1 7 using the coordinate set 2GB1 1 0 ). The positions of the guest site and the
surrounding residues are indicated. The arrows represent the interresidue contacts
to the guest site made by the surrounding residues. In the native structureof GB1,
the number of contacts to residue 53 from residues 6, 44, 51 and 55 is twelve, seven,
three, and two respectively. In the AASS background the number would be reduced
to three, two, one and two respectively, assuming that the backbone structure is
unperturbed from that of GB1. b, Temperature dependence of the CD signal from
representative GB1 mutants. O AASS-53Thr, 0 AASS-53Ser, X AASS-53Gln,
+ AASS-53Asp, 0 AASS-53Trp, A AASS-53Pro. The folded baselines of all
molecules fall within the bounds of the range displayed here, with the exception of
AASS-53Gly which had a [01218 nm value of -13100 deg cm 2 dmol -1 at 00 C. The
differences in folded baseline [01218 nm values probably result largely from different
aromatic contributions to the CD signal. Residue 53 is situated above W43 which is
in the hydrophobic core of the protein. Molecules containing the mutation W43F
show changes in both the folded and unfolded [01218nm baseline values of 45% and
35% respectively (data not shown), indicating that W43 is making a substantial
contribution to the CD signal at 218 nm (see also text and refs. 13, 14) c, Plot of AHm
vs. temperature for the AASS mutants. Open circles indicate Tm and AHm values
from the AASS-53Xaa mutants. Filled symbols indicate Tm and AHm for
OAASS-53Thr, A AASS-53 Phe, and U AASS-53Val at different pH values. The
slope of the line is ACp (624 cal mol -1 deg-1).
METHODS. Calculations of nearest-neighbor side chain contacts were made using a
program written by T.G. Oas 1 8 . Significant contacts were identified from the
coordinate set 2GB1 and were defined as any atom pair having a center to center
distance less than or equal to 150% of the sum of the van der Walls radii. A
synthetic gene for the GB1 sequencel 9 was constructed and cloned into a T7
expression plasmid 2 0 , pAED4 (ref. 21), using standard cloning procedures22. All
proteins were purified by affinity purification 9 on IgG 6 Fast Flow Sepharose
(Pharmacia, 17-0969-01) followed by reverse phase high pressure liquid
chromatography (HPLC) purification on a Vydac preparative C18 column using a
linear H20-acetonitrile gradient in the presence of 0.1% TFA. This purification
scheme yielded two species of GB1, a 56-residue protein containing an N-terminal
methionine and a 55-residue protein in which the N-terminal methionine had been
removed. The 55-residue species was substantially less stable (by -1.7 kcal mol-1)
than the 56-residue species. The mutation Metl-)Thr was made to produce a
56-residue, N-terminally processed protein that began with threonine at position 1.
This protein had the same stability as the 56-residue GB1 containing methionine at
position 1. All mutant genes were made in the GB1 background containing
threonine at position 1 by site directed mutagenesis23. Proteins were purified by
affinity and HPLC chromatography as described above with the exception of AASS53Pro for which the IgG affinity column step was replaced by G75 chromatography in
a buffer of 10 mM phosphate, 150 mM NaCI, pH 7.3. Mutant genes were sequenced
completely (SequenaseTN, USB) and the identity of each HPLC purified protein was
confirmed by laser desorption mass spectrometry (Finnegan Mat Lasermat). All
molecular weights were within 3 daltons of the expected mass. CD measurements
were made at a concentration of 10 gM using a 1 cm pathlength cell with an Aviv
model 62DS circular dichroism spectrometer equipped with a thermoelectric
temperature controller. The sample buffer contained 50 mM sodium Acetate,
150 mM NaCI, pH 5.4. The melting curves were fit (Kaliedagraph, Abelbeck
Software) to the equation:
e=eu+(ef-eu)/(1
+ e ((-AHm/R)(1/T-1/Tm) + (ACp/R)[((Tm/T)-1)+ln(T/Tm)])), using a
fixed ACp of 624 cal mol- 1-deg-1, where 0 is the measured CD signal, Of and Ou are
linear functions representing the folded and unfolded baselines respectively, Tm is
the melting temperature, and AHm is the enthalpy of unfolding at Tm2 4 . The two
state unfolding model was tested for AASS-53Thr and AASS-53Ala by measuring
the enthalpy of unfolding by differential scanning calorimetry. The results are
AHcal = 39.9 kcal mol-1 and 32.8 kcal mol-1, AHcal/AHvan't Hoff = 0.99 and 0.97, for
AASS-53Thr and AASS-53Ala respectively. Differential scanning calorimetry
experiments were performed using a Microcal MC-2 scanning calorimeter
(Northampton, MA). Calorimetry samples contained -3 mg ml-1 protein in
50 mM sodium Acetate, 150 mM NaC1, pH 5.4 and were dialyzed extensively versus
this buffer prior to the experiment. Samples were degassed under vacuum prior to
loading into the sample chamber and were heated from 40 -950 C using a scanning
rate of 10 minute- 1. The pH dependence of Tm was measured in a buffer of 1 mM
each phosphate/acetate/citrate/borate, 196 mM NaCl. Protein concentration was
determined for all samples by measuring absorbance of the unfolded protein25
Fig. 2 a, 1H- 15N correlation spectra 2 6 of AASS-53Thr and b, AASS-53Ala proteins at
5OC in 150 mM NaCi, pH 5.4 (10% D 20). The labels indicate resonance assignments.
The side chain amide resonances from the one Gln and three Asn residues are
unassigned and are labeled as N8/QE. Observed backbone and side-chain NOE's
diagnostic for P-sheet structure at the guest site include: HN5 1-HN46, Ha 51-HN 4 ,
HN5 2-Ha5 , HN5 2-HN 6, HN52-HN4, H%52-HN 46, HN52-HP'5, HN53-HN44 (not seen for
AASS-53Ala due to overlap), H%53-HN 6, HN54 H 7, HN 54 --HN8 , HC5 4-Ha 43 ,
Ha54-HN 44, HN55-HN42, HN55 -H%43, HY254 -H 843,HY254-HE4 3, HY254-Hý243,
HY2 54-Hý343 , HY254-Hr143, HY254-HNE43 , HY15 4-HE 4 3,HY154
-H8 43, H'Y154-Hý243 ,
H054-H8 43 , HP5 4-Hý243 , H05-Hr143 and H 35-Hý343.
METHODS. Escherichia coli harboring the expression plasmid for AASS-53Thr or
AASS-53Ala were grown in M9 media supplemented with (15NH 4)2SO4 (99.7% 15N,
Isotec, Miamisburg, OH) to obtain uniformly (2 95 %) 15N-labeled protein. Data
were collected on a Bruker AMX 500 MHz NMR spectrometer. Spectral widths in
the 1H(F 2) and 15 N(F1) dimensions were 14.08 ppm and 31.98 ppm, respectively.
Spectra were referenced to the carrier in F1 (118.5 ppm) and trimethylsilylpropionate
(TMSP) in F 2 corrected for the pH dependence of TMSP 2 7 . Resonance assignments
were made using standard methods 28,29 and were consistent with the assignments
for wild-type GB1(ref. 10).
Fig. 3 a, Correlation of the AAG values for P-sheet formation measured in the
AASS system (Table 1) with the P-sheet forming propensities (Pp) of Chou and
Fasman 1 1. b, Correlation of the AAG values for P-sheet formation obtained in this
work and the zinc finger system 8 . Note the difference in scale between the two
systems. The value for proline is omitted.
References
1. Scheraga, H.A. Pure & Appl. Chem. 50, 315-324 (1978).
2. Padmanabhan, S., Marqusee, S., Ridgeway, T., Laue, T.M. & Baldwin, R.L.
Nature 344, 268-270 (1990).
3. O'Neil, K.T. & DeGrado, W.F. Science 250, 646-651 (1990).
4. Lyu, P.C., Liff, M.I., Marky, L.A. & Kallenbach, N.R. Science 250, 669-673 (1990).
5. Horovitz, A., Matthews, J.M. & Fersht, A.R. J. Mol. Biol. 227, 560-568 (1992).
6. Blaber, M., Zhang, X. & Matthews, B.W. Science 260, 1637-1640 (1993).
7. Wojcik, J., Altman, K.-H. & Scheraga, H.A. Biopolymers 30, 121-134 (1990).
8. Kim, C.A. & Berg, J.M. Nature 362, 267-270 (1993).
9. Alexander, P., Fahnestock, S., Lee, T., Orban, J. & Bryan, P. Biochemistry 31, 35973603 (1992).
10. Gronenborn, A.M., et al. Science 253, 657-661 (1991).
11. Chou, P.Y. & Fasman, G.D. Biochemistry 13, 211-222 (1973).
12. Livingstone, J.R., Spolar, R.S. & Record, M.T., Jr. Biochemistry 30, 4237-4244
(1991).
13. Chakrabartty, A., Kortemme, T.S., Padmanabhan, S. & Baldwin, R.L.
Biochemistry 32, 5560-5565 (1993).
14. Vuilleumier, S., Sancho, J., Loewenthal, R. & Fersht, A.R. Biochemistry 32,
10303-10313 (1993).
15. Fahnestock, S.R., Alexander, P., Filpula, D. & Nagle, J. in Bacterial
Immunoglobin-Binding Proteins (eds. Boyle, M.D.P.) (Academic Press, Inc., San
Diego, 1990).
16. Harlow, E. & Lane, D. Antibodies: a laboratory manual (Cold Spring Harbor
Laboratory, Cold Spring Harbor, 1988).
17. Priestle, J.P. J. Appl. Cryst. 21, 572-576 (1988).
18. Oas, T.G. & Kim, P.S. Nature 336, 42-48 (1988).
19. Fahnestock, S.R., Alexander, P., Nagle, J. & Filpula, D. J. Bact. 167, 870-880 (1986).
20. Studier, F.W. & Moffat, B.A. J. Mol. Biol. 189, 113-130 (1986).
21. Doering, D.S. Functional and Structural Studies of a Small f-actin Binding
Domain (Massachusetts Institute of Technology, 1992).
22. Sambrook, J., Fritsch, E.F. & Maniatis, T. Molecular Cloning: a laboratory
manual (Cold Spring Harbor Laboratories, Cold Spring Harbor, 1989).
23. Kunkel, T.A., Roberts, J.D. & Zakour, R.A. Methods in Enzymology 154, 367-382
(1987).
24. Becktel, W.J. & Schellman, J.A. Biopolymers 26, 1859-1877 (1987).
25. Edelhoch, H. Biochemistry 6, 1948-1954 (1967).
26. Bodenhausen, G. & Ruben, D.J. Chem. Phys. Lett. 69, 185-189 (1980).
27. DeMarco, A. J. Mag. Res. 26, 527-528 (1977).
28. Wilthrich, K. NMR of Proteins and Nucleic Acids (John Wiley & Sons, Inc.,
New York, 1986).
29. McIntosh, L.P., Wand, A.J., Lowry, D.F., Redfield, A.G. & Dahlquist, F.W.
Biochem. 29, 6341-6362 (1990).
ACKNOWLEDGMENTS. We thank Z.-Y. Peng and P.A. Petillo for critical reading
of the manuscript, K. J. Lumb, C. J. McKnight, Z.-Y. Peng, B. A. Schulman and J. P.
Staley for expert NMR assistance, and members of the Kim lab for helpful advice
and discussions.
. la
B
Fig. 1
0
E
-5
0E
-10
co
E
C\
01
N1"
-15
-20
0
15
30
45
Temperature
C
60
75
90
325
330
(0C)
50
40
0
O
30
E
20
10
300
305
310
315
Tm (K)
320
15 N
(p.p.m.)
C
3:
I ,
?3
23
L/·
k)
15 N(p.p.m.)
-
0-I
o
/
X-2
0=o
0=0
0=0
wP
wp
O -I
o=o/
0=--0
z
0=0
-- =
z__I
~
L
=
0 =0
0
-=Z
0=0
-_
I-0-
0
I-r
Z0
-2I
e
z
So
-z I
0=0
\
0=
oo -M++X-C
N
X- 0 4M P
S=-Z
0=0
/~I--0
Z -I
0 -0
0=O0
I-Z
a0-I
/
o=0
I-Z
Z-zw
071
0=0
A
/
Z-I
A
Fig. 3
VO
cz
E
*1
1.5
W
Lt
LL
0Q
1
oe5
c-
* M
N4H O R
A
OK OS
GO
r)
0v
0.5
C
-DO
OE
I
-2
i
-1
AAG (kcal
B
mol 1 )
0.5
-'
H
0
e.l
o
C
DO
0
E*
NA
A
RC M
QK
S
S-0.25
(:e
-0.5
I
-2
AAG (GB1)
13
*T
CHAPTER 3
CONTEXT IS A MAJOR DETERMINANT OF f3-SHEET PROPENSITY
Residues in f-sheets occur in two distinct tertiary contexts: central strands,
bordered on both sides by other 3-strands, and edge strands, bordered on only a
single side by another P-strand 1 . The AAG values for P-sheet formation measured at
an edge P-strand of the IgG binding domain of protein G (GB1) are quite different
from those obtained previously2, 3 at a central position in the same protein. In
particular, there is no correlation at the edge position with statistically determined
3-sheet forming preferences 4 . The differences between P-sheet propensities
measured at central and edge P-strands, AAAG values, correlate with the values of
water/octanol transfer free energies 5 and sidechain non-polar surface area for the
amino acids 6 . These results strongly suggest that, unlike (a-helix formation, P-sheet
formation is determined in large part by tertiary context, even at solvent-accessible
sites, and not by intrinsic secondary structure preferences.
The twenty naturally occurring amino acids were substituted at a solvent
exposed edge n-strand position, residue 44, by site-directed mutagenesis in a host
molecule in which local interactions to the guest site had been minimized by
replacing the nearest neighbors with alanine (see Fig. 1 legend). This edge P-strand
is bordered on one side by another f-strand and on the other side by solvent
(Fig. la). The stability of each protein (denoted AAA-44Xaa) was measured by
thermal unfolding as monitored by circular dichroism (CD) at 218 nm (Fig. Ib). AAG
values for P-sheet formation, referenced to alanine, were obtained by assuming that
changes in global stability result entirely from changes in the ability of the residue at
the guest site to adopt a P-sheet conformation. In support of this assumption,
molecules representative of the entire AAG range were found to have
AHvan't Hoff /AHcal ratios near unity, indicating that the two-state nature of the
equilibrium unfolding transition observed for wild-type GB1 (ref. 7) remains intact
(see Fig. 1 legend). The relative free energy differences for f-sheet formation at the
edge position are listed in Table 1.
All the proteins tested, with the exception of unfolded AAA-44Pro, bind Fc
with around sevenfold reduced affinity relative to wild-type GB1 (Table 1). There is
some variation in binding constants but this is not correlated with protein stability.
As residues 42-46 have been identified as participants in the Fc-binding interface 8' 9
it seems likely that the observed affinity differences reflect direct effects on binding
by substitutions at residue 44.
As a more detailed check on the conformation at the guest site of each
molecule, the chemical shifts of the aromatic ring protons of Trp43 were measured
in each of the twenty variants. Trp43 is expected to be sensitive to changes in
structure since it immediately precedes the guest site and is part of the hydrophobic
core of the molecule. The chemical shifts of the Trp43 ring protons are similar in all
of the variants, with the exception of unfolded AAA-44Pro, and are significantly
different from the chemical shifts for free tryptophan (see Fig. 2 legend).
Additionally, unambiguous cross-strand NOE's between the guest strand and its
71
y2
neighboring strand can be found between upfield shifted I-4 and H5 4 protons of
2 H 3
5
E3 r12 NE
Val54 and the H 8H
H43, H43, H43 ring protons of Trp43 in each folded
H43,
43,the43,
Va54 and
protein.
Two molecules with substantially different stability, AAA-44Thr and
AAA-44Ala, which also have different Fc binding affinities, were characterized
further by NMR. The 1H-1 5N correlation spectra of these two molecules are very
similar except for resonances from residues in the immediate vicinity of the guest
site (Fig. 2a, b). The NMR spectra of both proteins also contain cross-strand nuclear
Overhauser effect (NOE) patterns near the guest site that are present in wild-type
GB1 (Fig. 2c). Taken together with the Fc binding and Trp43 chemical shift results,
these data indicate that any structural changes between the variants are small and do
not involve significant disruption of the backbone 1-sheet structure.
The dramatic effect that context can have on 1-sheet propensities is seen in the
comparison of the thermodynamic preferences for 3-sheet formation measured at
the central and edge 1-sheet positions (Fig. 3a). While the overall magnitude of the
scale for 1-sheet formation at the edge position (-2 kcal/mole, excluding proline) is
similar to that measured at the central 1-sheet position2, 3, there is no apparent
correlation with the statistically measured 1-sheet frequencies of Chou and Fasman 4
(Fig. 3b). In addition, unlike the results obtained at the central position, for which
the AAG values are distributed fairly evenly over a wide range, more than half
(13/20) of the AAG values measured at the edge position fall within a small range
(-0.4 kcal/mole).
The context dependence of 1-sheet formation suggests that two components
contribute to P-sheet propensity: (1) intrinsic ability to form a local extended
1-strand structure; and (2) ability to interact with the surrounding tertiary 1-sheet
structure. The small range of 1-sheet propensities measured at the edge position,
taken together with the relative lack of preference for particular sidechain rotamers
in P-strand residues 1 0 , suggests that (1) has only a minor role in determining
1-sheet propensity. This situation contrasts with a-helix formation where a
significantly biased sidechain rotamer distribution 1 0 and a well-distributed range of
a-helix propensity values are found11-18
The difference in 1-sheet propensity between the edge and center sites,
S-Xaa
Xaa
designated as AAAG = AAG dge - AAGce nter, correlates with both the free energy of
transfer for the amino acids from octanol to water 5 and with the non-polar
accessible surface area of the sidechain 6 (Fig. 3c, d). These correlations are striking
since the amount of buried surface area differs substantially at center and edge
f-sheet positions 1 . Thus, these results suggest that interaction with the
surrounding n-sheet structure (that is, component, (2) above) is the dominant term
for determining n-sheet propensity.
Our experiments indicate that P-sheet propensity is modulated strongly by
tertiary context, even at solvent-exposed positions. This result provides an
explanation for the differences in apparent P-sheet propensity measured in the zinc
fingerl9 and GB1 (refs. 2, 3) model systems. More generally, our results emphasize
that P-sheets are elements of both secondary and tertiary structure 2
Amino acid
AAG (kcal mol- 1)
Tm (oC)
Ka/KaAAA-44Thr
AAAG (kcal mol-1)
Thr
Ser
Glu
Val
Phe
Tyr
Cys
Gin
Ile
Ala
His
Met
Asp
Trp
Asn
Leu
0.83
0.63
0.31
0.17
0.16
0.11
0.08
0.04
0.02
0
-0.01
-0.02
-0.10
-0.17
-0.24
-0.24
-0.40
-0.43
-0.85
< -4
60.2
59.4
57.0
56.1
56.1
55.9
55.1
54.7
54.8
54.7
54.6
54.2
53.7
53.3
52.6
52.8
51.4
51.2
47.6
<0
1.0
3.1
1.5
1.5
1.1
1.1
0.7
1.9
1.6
2.9
1.8
1.8
1.2
3.0
1.0
1.0
0.9
1.0
1.4
0
-0.27
-0.07
0.30
-0.65
-0.70
-0.75
-0.44
-0.19
-0.98
0
0.01
-0.74
0.84
-0.77
-0.16
-0.75
-0.67
-0.88
0.35
Lys
Arg
Gly
Pro
Table 1 AAG values for 3-sheet formation at an edge position relative to alanine,
thermal melting temperatures (Tm) for the AAA-44Xaa proteins, relative binding
constants (Ka) to human Fc for the AAA-44Xaa proteins, and AAAG values for
3-sheet formation, comparing AAGedge - AAGcenter values (this work and ref. 2,
respectively)
METHODS The Ka for wild-type GB1 is 1.4 x 108 M -1 (ref. 20). The Ka/KaAAA -44Thr
value for wild-type GB1 is 7.1. AG values for unfolding at 321K were calculated as
described previously 2 using data obtained from CD thermal unfolding
measurements and the Gibbs-Helmholz equation. A positive AAG value indicates
an increase in stability relative to alanine. Estimated errors in determination of Tm
and AG are + 0.5 'C and ± 0.06 kcal mol- 1 respectively. Fc binding was measured as
described previously 2 by adding variable amounts of competitor protein to a fixed
quantity of protein G-alkaline phosphatase at 50 C in 96 well plates. The ratio of
affinity constants for the mutants was normalized to AAA-44Thr. GB1-Thrl (see
Fig. 1 legend) was included as an internal standard on each plate.
Figure 1 a, Ribbon drawing 2 1 of GB1 based on the coordinate set 2GB1 2 2 . The
positions of the guest site and the surrounding residues are indicated. The arrows
represent the inter-residue contacts to the guest site made by the surrounding
residues. b, Temperature dependence of the CD signal from representative GB1
variants AAA-44Thr (0), AAA-44Val (+), AAA-44Ala (0), AAA-44Asn (A),
AAA-44Gly ( * ) and AAA-44Pro (A) in 150 mM NaC1, 50 mM Na Acetate, pH 5.4.
METHODS. As described previously 2 , major inter-residue contacts were identified 2 3
from the coordinate set 2GB1 and were defined as any atom pair having a center to
center distance less than or equal to 150% of the sum of the van der Waals radii.
Major contacts to residue 44 arise from residue 53 on the adjacent strand with fewer,
more distant contacts being made by residues at positions i+2 and i-2 from the guest
site. These residues were changed to alanine to create the host molecule
E42A/D46A/T53A (denoted AAA). To test for possible effects from the i+2 and i-2
residues, which had been serine in our central position study 2 , we also created the
host molecule E42S/D46S/T53A (denoted SSA). Substitution at the guest position
with threonine, an amino acid expected to be a very good 3-sheet former and
glycine, an amino acid expected to be a very poor P-sheet former, yields identical
AAG values relative to alanine for P-sheet formation in both the AAA and SSA
backgrounds. The host molecule bearing the smaller amino acid substitutions,
AAA, was chosen for further study. Recombinant GB1 mutants were expressed
from a synthetic gene bearing the mutation Metl-+Thr (GB1-Thrl) described
previously 2 . Mutations were generated by single strand mutagenesis 2 4 and were
verified by dideoxynucleotide sequencing (SequenaseT M , USB) of the entire mutant
gene. Proteins were expressed in E. coli (BL21 (DE3) pLysS) and were induced at an
O.D. at 600 nm of -0.5-0.8 with a final concentration of IPTG of 0.4 mM for 2-4 hours.
All folded proteins were purified by affinity chromatography 7 with IgG 6 Fast Flow
Sepharose (Pharmacia) followed by reverse phase high pressure liquid
chromatography (HPLC) purification on a Vydac preparative C18 column using a
linear H20-acetonitrile gradient in the presence of 0.1% TFA. The proteins are
expressed as mixtures of N-terminally processed protein beginning with threonine
at position 1 (56 residues) and non-processed protein (57 residues) beginning with
methionine at position 0. The ratio of processed to unprocessed protein appears to
depend on the stability of the molecule (data not shown). The HPLC purification
step separates these two species and in all cases the 56-residue protein was used. For
the unfolded molecule AAA-44Pro, the IgG affinity column step was replaced by G75
Sephadex chromatography in a buffer of 10 mM phosphate, 150 mM NaC1, pH 7.3.
The identity of each HPLC purified protein was confirmed by laser desorption mass
spectrometry (Finnigan Mat Lasermat). All measured molecular weights were
within 3 daltons of the expected mass. CD measurements were made and thermal
unfolding curves were fit as described previously 2 . ACp values for unfolding (data
not shown) were similar (± 15 %) to those measured for the center site AASS-Xaa
molecules 2 . Protein concentration was determined by measuring absorbance of the
unfolded protein 2 5 . Differential scanning calorimetry was performed with a
Microcal MC-2 scanning calorimeter (Northampton, MA) as described previously 2
AAA-44Thr, AAA-44Ser, AAA-44Gln, AAA-44Ala and AAA-44Gly were found to
have AHvan't Hoff/ Hcal ratios of 0.98, 0.94, 1.03, 0.98 and 0.99, respectively.
Figure 2 1H- 15N correlation spectra of a, AAA-44Thr and b, AAA-44Ala proteins at
5°C in 150 mM NaC1, pH 5.4 (10% D20). Labels indicate resonance assignments. c,
diagram of NOE's, present in both AAA-44Thr and AAA-44Ala, that are diagnostic
for n-sheet structure at the guest site. Each arrow represents one or more NOE's.
The position of the guest site is shaded grey. The range of chemical shifts (in p.p.m.,
see methods) for all of the folded variants measured for the 8H, ENH, ý2H, 712H, ý3H,
protons of Trp43 were respectively: 7.30-7.34, 10.32-10.38, 7.07-7.11, 6.45-6.50,
6.35-6.40. These chemical shifts are substantially different from the respective
chemical shifts for free L-Trptophan: 7.01, 9.98, 7.24, 6.98, 6.90 and the respective
chemical shifts for unfolded AAA-44Pro: 6.90, 10.00/9.96, 7.20, 6.91, NA. (Two
chemical shifts are seen for the eNH proton reflecting the influence of cis and trans
isomers of Pro44 2 6 . The ý3H resonance in AAA-44Pro could not be assigned due to
proximity with the diagonal).
METHODS. 1H- 1H DQF-COSY and NOESY spectra were collected at 5°C, 150 mM
NaC1, pH 5.4, 10% D20 for all molecules. Spectral widths for 1H- 1H experiments
were 14.08 ppm in both dimensions. E. coli harboring the expression plasmid for
AAA-44Thr or AAA-44Ala were grown in M9 media supplemented with
(15NH 4 )2SO 4 (99.7% 1 5 N, Isotec, Miamisburg, OH) as the sole nitrogen source to
obtain uniformly (2 95 %) 15N-labeled protein 2 7 . Data were collected on a Bruker
AMX 500 MHz NMR spectrometer. For heteronuclear experiments the spectral
widths for the 1H(F2) and 15N(F1) dimensions were 14.08 ppm and 31.98 ppm,
respectively. Proton spectra were referenced to the carrier in both dimensions (4.76
ppm). Heteronuclear spectra were referenced to the carrier in F1 (118.5 ppm) and F2
(4.76 ppm). Resonance assignments were made using standard methods 2 7' 28 and
were consistent with the assignments for wild-type GB1 2 2 . Observed cross strand
Nc
N N
Na
aN
NN
Na
NN
NOE's include: H4 -H 5 1 , H6 - H5 2 , H 6 -H53, 7-H54, H 8 -H54,H 8 -H55,H42-H55
a
C
N
N
N
a
N N p
N N N 8 T3 IT8
H 4 3 -H5 4 , H 4 4 -H5 3 , H4 4 -H5 4 , H46-H51 H46-H51' H5 2 -H4 .H52 -43' H52
HS -HT3
52
43'
52
-H43'
-2 HH
• H2
•3H
H
54 43
54
7r12
43'
-H8
H
H
_H2
HP _HH3
43' 54 43' 54 43 H54- 43' 54 43'
2 H•yl HN
N H2-H H -H 2
H 1 -H HTl-Hý2 H_1 -H_3 HY1 -H 3 ,71-H1
- 43 H54 43' 54 43'
54 43' 54 43' 54 43' 54 43' 54 43' 54
H -H 3 H12_He3 H -Hr 2 H 2_ NE N -H
54 43' 54 43' 54 43' 54 43 'H5 5 -H4 3
Figure 3 a, Comparison of f-sheet forming propensities at central 2 and edge P-sheet
positions. b, Comparison of the edge site P-sheet propensities with the n-sheet
forming frequencies of Chou and Fasman 4 . The data are not correlated (r=0.15,
p>0.25).
c, Correlation of the difference in f-sheet forming propensity measured at edge and
center f-sheet positions, AAAG, with AG of transfer for the amino acid sidechains
from octanol to water 5 (r=0.57, p<0.005; r=0.80, p<0.0005, excluding values of R and
K, see methods) and
d, sidechain non-polar surface area 6 (r=0.72, p<0.0005). Amino acids are identified
by the labels. Shaded areas are meant only to emphasize the overall trends in the
data.
METHODS. Values for amino acid sidechain non-polar surface areas are for "set 1"
from ref. 14. The values from "set 2" show a similar trend. It should be noted that
with respect to the correlation between AAAG and AG of transfer there are two clear
outlying points, arginine and lysine. The apparent anomalous behavior of these
two residues most likely reflects the large distance between the sidechain charge and
the backbone 1-sheet structure. Correlation coefficients were calculated for the (xi,yi)
pairs using the formula 2 9 :
r
[
N
i=l(xi- R)(Yi2 NN__1
[j(xi-j)
i=1
2
(i-
i=1
r)
)21/2
2
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
Chothia, C. J. Mol. Biol. 105, 1-14 (1976).
Minor, D.L., Jr. & Kim, P.S. Nature 367, 660-663 (1994).
Smith, C.K., Withka, J.M. & Regan, L. Biochemistry 33, 5510-5517 (1994).
Chou, P.Y. & Fasman, G.D. Biochemistry 13, 211-222 (1973).
Fauchere, J.-L. & Pliska, V. Eur. J. Med. Chem. - Chim. Ther. 18, 369-375 (1983).
Livingstone, J.R., Spolar, R.S. & Record, M.T., Jr. Biochemistry 30, 4237-4244
(1991).
Alexander, P., Fahnestock, S., Lee, T., Orban, J. & Bryan, P. Biochemistry 31, 35973603 (1992).
Frick, I.-M., et al. Proc. Natl. Acad. Sci. U.S.A. 89, 8532-8536 (1992).
Gronenborn, A.M. & Clore, G.M. J. Mol. Biol. 223, 331-335 (1993).
McGregor, M.J., Islam, S.A. & Sternberg, M.J.E. J. Mol. Biol. 198, 295-310 (1987).
Lyu, P.C., Liff, M.I., Marky, L.A. & Kallenbach, N.R. Science 250, 669-673 (1990).
O'Neil, K.T. & DeGrado, W.F. Science 250, 646-651 (1990).
Padmanabhan, S., Marqusee, S., Ridgeway, T., Laue, T.M. & Baldwin, R.L.
Nature 344, 268-270 (1990).
Wojcik, J., Altman, K.-H. & Scheraga, H.A. Biopolymers 30, 121-134 (1990).
Chakrabartty, A., Schellman, J.A. & Baldwin, R.L. Nature 351, 586-588 (1991).
Horovitz, A., Matthews, J.M. & Fersht, A.R. J.Mol.Biol. 227, 560-568 (1992).
Serrano, 1., Neira, J.-L., Sancho, J. & Fersht, A.R. Nature 356, 453-455 (1992).
Blaber, M., Zhang, X. & Matthews, B.W. Science 260, 1637-1640 (1993).
Kim, C.A. & Berg, J.M. Nature 362, 267-270 (1993).
Lifson, S. & Sander, C. J. Mol. Biol. 139, 627-639 (1980).
Priestle, J.P. J. Appl. Cryst. 21, 572-576 (1988).
Gronenborn, A.M., et al. Science 253, 657-661 (1991).
Oas, T.G. & Kim, P.S. Nature 336, 42-48 (1988).
Kunkel, T.A., Roberts, J.D. & Zakour, R.A. Methods in Enzymology 154, 367-382
(1987).
Edelhoch, H. Biochemistry 6, 1948-1954 (1967).
Grathwohl, C. & Wiithrich, K. Biopolymers 20, 2623-2633 (1981).
McIntosh, L.P., Wand, A.J., Lowry, D.F., Redfield, A.G. & Dahlquist, F.W.
Biochemistry 29, 6341-6362 (1990).
28. Wilthrich, K. NMR of Proteins and Nucleic Acids (John Wiley & Sons, Inc.,
New York, 1986).
29. Rosner, B. Fundamentals of Biostatistics (Duxbury Press, Boston, 1982).
Acknowledgements. We thank Z.-Y. Peng for helpful advice and suggestions and
L.M. Mayr, M.G. Oakley and P.A. Petillo critical reading of the manuscript. This
work was supported by the Howard Hughes Medical Institute. D.L.M. is supported
by a National Institutes of Health Biophysical Training Grant.
Fig. la
B
Fig. 1
-5
c1J
E
C.)
cv0-10
E
C
co
-15
-20
0
15
30
45
60
Temperature (oC)
75
90
r
03-
XX
'°
0O
/m
o
0CA
0
~
--
SZ -L
/
'
2,, 1 /: o ,
Co
"
T1*
0
G)G
G) coC%
-th
-
Lr-
o
<
Z,
z x
u
<~v
>,~ .~..ut
c.D
CO
)
<
(i
/
_z
N
-4
Ul
-
<
I
I1
I
126.0
+-
zN.
Q
9
O
2
:r,
N~ .I
I
I
120.0
114.0
108.0
15 N(p.p.m.)
0-
-n
ru
>P
0
-4
-<.0
-0
C)
0z
0
o
a
M 0<S
oX
o
N)
I
-4
G)
0)
*0
rn z o<0C
)
a"
-H
S0
C-
O
P
CC·
VI
*
N),
CA)C
C)
Co
/l,
U
*
L
L?
~
v
O
.-.O-
.*Q-
00..
1.
126.0
1
120.0
15 N
coo
o~
co
1
11,4.0
(p.p.m.)
108.0
_-0 C
S0--'r
0=0
0~
0-=0
o=0
-
/ -I
2-
)-P
O=
O0
2=
Ao -I
I-z
I--0
z -M
I-0
/R
h
0=0
I-z
=
-M
Pe0=0O
0=0
M-0
)
0= 0
z -I
z -M
I-_0
Z -M "X-0
I-
A
o=0
Ir
"-z
-
0=0
--4
I =0
S0=-z
0=0
M-0 it
z -M
I-z
e
8 0-I
P 0=0
Z-I
I
,T-l
A
rig. .
1
•0 Y F
0
E
ST
M $V
Re
C
O
QO
K
O A OE
NOHH
0
ca
So
LW
0.5
& -0.5
D
-1
OG
t"
-4
-1.5
I
III
I
-0.5
-1
0
0.5
1
1.5
AAG, edge (kcal mol 1 )
Ov
1.6
I*
*Y
WO
w
E
(
Fe
1.2
O C
OT
U-LL
0
CO
0.8
GO
RO
H
KN
A
OS
DO
c•
0.4
Ir
-2
OE
I
I
I
-1
0
1
AAG, edge (kcal mol1 )
2
Fig. 3
C
OW
0
E
F
L
*C
M**V
r-0
Y
e,,,
,<
A
OOH
**A
TO
ON
*E
*K
RO
I
-1.5
D
OD
I
-1.0
0
-0.5
0.5
1.0
1.5
AAAG, edge - center (kcal mol-1)
d
200
OW
OF
Cz
0<
ao 150
Ie
OL
E
Ye
0
OK
100
OH
M
T
R
I3
O
0z
50
CO
,I
I
!I
-1.5
-1.0
-0.5
oA
*S
QO
OND
I
0
E
o
G
OD
I
I
0.5
1.0
AAAG, edge - center (kcal mol-1)
1.5
CHAPTER 4
DESIGN OF A PROTEIN SEQUENCE THAT SHOWS CONTEXT
DEPENDENT SECONDARY STRUCTURE FORMATION
Protein secondary structures have been viewed as fundamental building blocks
for protein folding, structure and design. Experimental studies indicate that the
secondary structure propensities of individual amino acids are the result of a
combination of local conformational preferences 1 1 1 and non-local factors 2 - 1 5 . In
order to examine the extent to which non-local factors can influence the formation
of secondary structural elements, we have designed an 11-amino acid sequence (the
Chameleon sequence) which folds as an (o-helixwhen in one position, but as a
P-sheet when in another position of the primary sequence of the IgG-binding
domain of protein G (GB1). Both proteins, Chameleons and ChameleonP, are
folded into structures similar to native GB1 as judged by a series of biophysical
criteria including cooperative thermal denaturation, slow amide hydrogen
exchange, diagnostic nuclear Overhauser effects (NOE's) and IgG binding. These
results demonstrate directly that non-local interactions can determine the secondary
structure of peptide sequences of substantial length and provide explicit support for
views of protein folding that emphasize the role of tertiary interactions as
determinants of structure (e.g., refs. 16, 17).
The Chameleon sequence was designed to replace a-helix residues 23-33 and
3-sheet residues 42-52 (Fig. la) of GB1. Our guiding design principle was to preserve
the hydrophobic nature of the residues that constitute the interface between each of
these secondary structure elements and the core of GB1 (Fig. ib). Since the
characteristic hydrophobic/hydrophilic patterning of buried residues in c-helices
and 1-sheets are very different, with periodicities of 3.6 and 2 residues respectively,
creation of a sequence consistent with both patterns posed a number of design
problems.
In comparing the structural environments of positions 23-33 and 42-52, we
encountered three major types of environments based on accessible surface area:
class I, sites where a residue was buried in one secondary structure but exposed in
the other, (pairs 23/42; 24/43- 29/48; 30/49 and 32/51- the buried residue is
underlined); class II, sites where a residue occupied a buried position in both
structures but was very different in size or polarity in each structure (pairs 26/45 and
33/52); and class III, sites that had no conflicts in terms of size, polarity or burial
(pairs 25/44; 27/46; 28/47 and 31/50). For class I sites, the buried residue from each
pair was used for the Chameleon sequence, with the exception of pair 29/48 (See Fig.
1 legend). For class II sites, a series of hydrophobic residues were substituted at the
positions of the residue pairs in the wild-type GB1 background in order to find a
common residue for both environments that did not disrupt protein stability too
drastically (See Fig. 1 legend). After determining the residues that would enable a
single sequence to fulfill the tertiary requirements of both positions 23-33 and 42-52
in GB1 using this procedure, the sequence AWTVEKAFKTF was introduced at these
positions by site-directed mutagenesis, thereby creating the proteins Chameleona
and ChameleonP, respectively (Fig. ic).
Chameleone and ChameleonP each display cooperative reversible thermal
unfolding as measured by circular dichroism spectroscopy (CD) (Fig. 2a) and
differential scanning calorimetry (DSC) experiments (Fig. 2 legend). This type of
thermal unfolding behavior is a hallmark of compact single-domain globular
proteins possessing uniquely packed hydrophobic coresl8
The nuclear magnetic resonance (NMR) spectra of Chameleons and
ChameleonP have substantial chemical-shift dispersion (Fig. 2b). In addition,
NMR-detected hydrogen exchange experiments indicate that both proteins contain a
set of backbone amides with exchange rates equal to or slower than those expected if
exchange occurred only from global unfolding (Fig. 2c, d). Together with the
thermal unfolding data, these results strongly suggest that both Chameleon proteins
have unique folded structures with well-packed hydrophobic cores (see discussion
in ref. 19).
NOE spectra indicate that, as designed, the Chameleon sequence is folded into
an a-helix in Chameleona (Fig 3a) and a j-strand/turn/p-strand in ChameleonP
(Fig. 3b). In both proteins, amide protons from residues in the Chameleon sequence
are significantly protected from hydrogen exchange, indicating that the hydrogen
bonds in each secondary structure formed by the Chameleon sequence are stable.
NOE patterns throughout each protein are consistent with those present in
wild-type GB1 (Fig. 3c). Furthermore, Chameleona and Chameleon1 are capable of
competing for Fc binding with wild-type GB1 (Fig. 3 legend), although with
somewhat reduced affinities. The amino acid changes in Chameleons and
Chameleon1 occur in regions of the protein that have been identified as part of the
Fc binding interface 2 0 , so it is likely that much of the affinity differences are the
result of mutation of interface residues. The observation that Fc binding can occur,
taken together with the NMR data, strongly suggests that both Chameleon proteins
are folded into structures similar to wild-type GB1.
As some short isolated peptides have been shown to form significant amounts
of secondary structure2 1' 22, an eleven-residue peptide corresponding to the
Chameleon sequence (Ac-AWTVEKAFKTF-NH 2) was synthesized to investigate
whether this sequence has any intrinsic conformational preferences. Both CD and
NMR experiments indicate that the Chameleon peptide is unfolded in isolation (see
Fig. 3 legend), suggesting that the Chameleon primary sequence itself has no strong
preference for either ca-helix or P-sheet conformation. Thus, the secondary structure
formed by the Chameleon sequence in both Chameleons and Chameleon1 proteins
is specified by tertiary interactions.
The secondary structure of peptide sequences in some natural proteins, such as
hemagglutinin and the serpins, have been observed to undergo major
conformational alterations following tertiary rearrangements induced by pH
changes or proteolytic cleavage events23 - 25 . Tertiary interactions have also been
found to play a dominant role in determining the f-sheet propensities of individual
amino acids 1 3 ' 15 and have been observed to affect the secondary structure
conformation of identical short peptide sequences (up to 6 amino acids long) in
proteins in the structural database 2 6 -2 9 . Our design of the Chameleon sequence
demonstrates directly that the information specifying cx-helix or P-sheet secondary
structures of substantial length can be entirely non-local. Taken together, these
results underscore the importance of tertiary interactions in establishing protein
secondary structure.
Fig. 1 Design of a Chameleon sequence in GB1. a, Schematic diagram of GB1
secondary structure. The primary sequences of GB1, Chameleona and Chameleonf
are positioned below the corresponding secondary structures. b, Schematic
representations of positions 23-33 and 42-52 of GB1. Both the wild-type and
Chameleon sequences are shown. Changes from the wild-type sequence are
indicated in bold. Residues involved in the interface between each secondary
structure and the remainder of the protein are shaded gray. c, Ribbon diagram
indicating the positions of the Chameleon sequence within Chameleon a (left) and
ChameleonI (right), respectively. The position of the Chameleon sequence is
indicated in yellow. The diagram was drawn using the coordinates of wild-type
GB1 3 0
METHODS. All GB1 mutants were derived from a synthentic gene for GB1 by
site-directed mutagenesis, verified by dideoxyribonucleotide sequencing, expressed
in E. coli and purified by IgG affinity chromatography, followed by reverse-phase
HPLC, as described previously 1 0 . The crystal structure of the GB1 homolog, GB2 3 1,
indicates that residue 57 is incorporated as part of the fourth f-strand. Creation of
the 57-residue protein, GB1*, by addition of residue K57 to the previously described
56-residue construct GB1-Thrl 1 0 , was found to increase the Tm by - 20 C. Thus, GB1*
was used as the parent construct for all Chameleon proteins. All molecular
identities were verified by matrix assisted laser desorption/ionization (MALDI)
mass spectrometry (Finnigan MAT Lasermat or PerSeptive Biosystems Voyager) and
found to be within 2 daltons of the expected mass for the 57-residue protein. The
Chameleon sequence positions 23, 26, 30 and 33 were judged to be important for the
helix-protein interface, while positions 43, 45, 51 and 52 were judged to be important
for the P-sheet-protein interface (see text). For all class I sites, the buried residue was
used, with the exception of the 29/48 pair. Because T49F was a necessary change in
the turn between strands III and IV of ChameleonP to preserve the buried Phe from
the 30/49 class I pair, we did not also use the buried residue (valine) from the 29/48
(Val/Ala) class I pair, as this would have caused two sequential amino acid changes
to be made in the turn. For the class II pair 26/45, we tested a number of
substitutions in GBI* at these positions. The Tm (AHm) values from the thermal
unfolding of these proteins were: A26Y, <0 0 C (nd); A26I, 59.6 0 C (42.5 kcal mol-1);
A26V, 66.5 0 C (50.9 kcal mol-1); Y45A, 53.4'C (40.4 kcal mol-); Y45V, 61.1 0C
(47.5 kcal mol-1); Y45I, 61.8 0C (44.5 kcal mol-1). From these data Val was chosen for
the 26-45 pair. Initial constructs containing a 10-residue Chameleon sequence at
positions 23-32 and 42-51 of GB1* were made. As both of these molecules were
folded, the 33/52 pair was examined. The Y33F substitution was found to lower the
Tm of the ox-helix Chameleon molecule by only 0.6 0C whereas the substitution F52Y
was found to lower the Tm of the f-sheet Chameleon molecule by 8.7 0C; therefore,
Phe was chosen for the 33/52 pair to generate Chameleona and ChameleonS,
respectively.
Fig. 2 Folding of Chameleon proteins, a, Thermal denaturation of Chameleona O
and ChameleonP M in 150 mM NaC1, 50 mM sodium acetate, pH 5.4. The Tm
(AHm) values for Chameleona and ChameleonP are 61.4°C (41.5 kcal mol-1) and
39.2 0 C (27.7 kcal mol-1), respectively. The results of differential scanning calorimetry
(DSC) experiments with Chameleona are AHcal = 44.1 kcal mol-1
(AHcal/AHvan't-Hoff = 1.06), consistent with two-state unfolding behavior.
Quantitative analysis of DSC experiments with Chameleon1 was not possible, owing
to difficulty in obtaining a satisfactory folded baseline between 0-10 0 C. b, 1H- 15 N
correlation spectra of Chameleona and ChameleonP proteins at 50 C, pH 5.4 (10%
D 20). Labels indicate resonance assignments. The sidechain resonances are labeled
as NS/QE. At lower contours, the HSQC spectra of Chameleon3 contain a set of
small amide peaks at chemical shifts characteristic of unfolded polypeptides 3 2
These peaks increase in size as the temperature is raised, suggesting that they arise
from the unfolded state, in slow exchange with the folded state on the NMR
timescale. Measurement of the amide proton peak volumes indicates that
approximately 1-3% of the molecules are in the unfolded state at 5°C, in agreement
with the population of unfolded molecules expected from the AG of folding for
Chameleon1 at this temperature. c, Amide proton protection patterns for
Chameleon C and d, ChameleonP. The data are plotted as log (P) for each residue,
where P is the protection factor from exchange, defined as krc/kobs, where krc is the
hydrogen exchange (HX) rate expected in a random coil and corrected for primary
structure effects 3 3 , and kobs is the measured HX rate. Horizontal lines indicate the
log (P) value expected from the global stability of each protein at 5°C.
METHODS. CD thermal unfolding and DSC measurements were performed as
described previously 1 0 . NMR experiments and resonance assignments were made
with 15N-labeled protein as described previously 10 . For hydrogen exchange
experiments, fully protonated Chameleona and Chameleon1 samples were adjusted
to pH 5.4 and lyophilized from water prior to initiation of hydrogen exchange.
Exchange was initiated by dissolution of the lyophilized samples into an exchange
buffer of 50 mM acetic-d 3-acid-d (Aldrich 99.5 atom % D) in D 20 (Aldrich 100.0 atom
% D), pH* 5.4 at 50 C (pH* refers here to meter readings in D20 using a glass pH
electrode, without correction for isotope effects). Proton occupancy was measured by
collecting 1H- 15 N HSQC spectra with 2, 4, 16 or 32 transients per increment with 100
T1 increments. Amide proton decays were followed by measuring peak volumes in
1H- 15 N HSQC spectra using the program FELIX230 (Biosym).
Amide proton
exchange rates were calculated using the equation I(t) = e(-kt) + I(oo), where I(t) is
intensity at time t and k is the measured exchange rate. The protection factor P was
calculated to account for primary sequence effects, as described in ref. 30. The global
stability of each protein at 50C was calculated using data obtained from CD thermal
unfolding measurements and the Gibbs-Helmholtz equation as described
previously 1 0
Fig. 3 Diagram of observed NOE's indicative of the a, c-helical and b, f3-sheet
conformation of the Chameleon sequence in Chameleona and Chameleon 3 . The
position of the Chameleon sequence in each molecule is highlighted in gray. NOE's
observed in Chameleon" indicative of helical structure are indicated by the bars.
N N N N N N
Cross-strand NOE's observed in Chameleonf include: H4 4 -H5 3 , H4 4 -H5 5 , H4 -H5 2,
N N
N N N N cc cc
cc cc
cc
N
H 8 -H5 4 , H 4 6 -H5 1 , H9 -H12 , H4 5 -H5 2 , H43-H 54, H6-H 15 H4 -H17, 45-H53
N cc
N N
N N
cc N N c
N c
c N
o
H 5 -H17, H6 -H53, H4 3 -H5 5 , H 6 -H1 6 , H4 -H5 1 , H7 -H1 5 , H6 -H5 2 , H 7 -H1 4 ,
N N N U N N a N N ca
c c Y
H 4 2 -H5 5 , H4 2 -H5 6 , H3 -H 18 , H 4 1-H5 6 , H4 7 -HN
5 1 , H5 -H1 6 , H4 -H51 H4 -H17
N
c N N c
N cac
N
a
N y
ENI l -H
H 1
2
H7-H 5 4 , H 8 -H5 5 ' H4 4 -H5 4, H9-H56, H47-H 5 0 , H54-H43 , H54-H943 H54-H43
N
1
HY
yH
N H 1 N H2 NH 2 N , H/-H
H72-H 2 Assays for
54-H42 H54-H41' H54-H44, H54-H52, H5 4 -H44, H54 43 54 43Assays
native tertiary structure. c, Diagram of tertiary NOE's observed between the helix
and sheet in both Chameleone and ChameleonP. Tertiary NOE's represented by the
arrows include: H [ -H , HP -H7 1 H, -H72 , H 1 -H H -H71H
H'2 -H H 2 H N
34 54' 34 54' 34 54' 26 3' 26 3' 26 3' 4 3 -H3 1 '
43
31/
HP34-H 7
34
54'
43
H 34-H
34
30
34 -43'
2
H3•-H2
43'
3
26'
34
43
H32-H26 for Chameleonf.
3
26
43
31'
43- 30'
Competitive Fc binding
experiments with GB1* indicate that both Chameleoncx and Chameleonf bind Fc.
Chameleonu and Chameleon 3 have relative dissociation constants (Kd/KdGB1) for
Fc binding of 12.0 and 36.0, respectively. The Kd for wild-type GB1 is 7.1 nM3 4 . The
competitive binding assay was performed as described previously 10 . Sedimentation
equilibrium experiments were performed to assess the association state of
Chameleona and ChameleonP. Molecular weights of 6800 (calculated 6366) and
6500 (calculated 6246) daltons were obtained for Chameleonu and ChameleonP,
respectively, indicating that both proteins are monomeric. The residuals from fits of
the data to a single ideal species model were random, and the observed molecular
weight had no dependence on protein concentration (10-100 gtM). A peptide,
Ac-AWTVEKAFKTF-NH 2, corresponding to the Chameleon sequence with blocked
ends, was examined by CD and NMR spectroscopy. The CD spectrum of this peptide
at O'C had features typical of unfolded polypeptides 3 5 and was concentration
independent (10-200 jiM). The temperature dependence of the CD signal of this
peptide at 222 nm was linear from 0-70'C (data not shown). NMR spectra of this
peptide showed chemical shift dispersion typical of that expected for unstructured
6
peptides 3 2 . Calculation of the Chou/Fasman 3 values for the Chameleon sequence
(<Pa> =1.15 and <Pp> = 1.05) indicates that this sequence has no strong statistical
preference for forming either a-helix or f-sheet structure.
METHODS. Competitive Fc binding assays were performed at 40C, as described
previously 1 0 . Sedimentation equilibrium experiments were performed using a
Beckman Optima XL-A analytical ultracentrifuge operating at 40C with an An-Ti
rotor and six-sectored equilibrium centrifugation centerpieces. Data were collected
at rotor speeds of 35,000 and 42,000 rpm for both Chameleona and ChameleonP.
Samples were prepared in a buffer of 150 mM NaC1, 50 mM sodium acetate, pH 5.4,
and were dialyzed exhaustively against the same buffer. Sample concentrations
were determined by absorbance 3 7 . Dilutions were made into the dialysate to obtain
10-100 gM protein samples and dialysate was used to fill the reference cells.
Apparent molecular weights were calculated by fitting data sets from each sector to a
single ideal species model using Kaleidagraph (Abelbeck Software). Partial specific
volumes of 0.7323 and 0.7406 ml g-1 were used for Chameleona and ChameleonP,
respectively, and were corrected for temperature 3 8. The Chameleon peptide was
synthesized by solid phase Fmoc methods on an Applied Biosystems model 431A
peptide synthesizer and purified by Sephadex G25 size exclusion chromatography in
5% acetic acid, followed by reverse-phase high pressure liquid chromatography
(HPLC) purification on a Vydac C18 column with a linear H20-acetonitrile gradient
in the presence of 0.1% trifluoroacetic acid. Peptide identity was confirmed by
MALDI mass spectrometry, measured 1367 (expected 1368) daltons. CD studies of the
Chameleon peptide were carried out in a buffer of 150 mM NaC1, 50 mM
sodium acetate, pH 5.4. NMR studies of this peptide were carried out in 10% D 20,
pH 5.4.
References
4.
5.
Lyu, P.C., Liff, M.I., Marky, L.A. & Kallenbach, N.R. Science 250, 669-673 (1990).
O'Neil, K.T. & DeGrado, W.F. Science 250, 646-651 (1990).
Padmanabhan, S., Marqusee, S., Ridgeway, T., Laue, T.M. & Baldwin, R.L.
Nature 344, 268-270 (1990).
Wojcik, J., Altman, K.-H. & Scheraga, H.A. Biopolymers 30, 121-134 (1990).
Chakrabartty, A., Schellman, J.A. & Baldwin, R.L. Nature 351, 586-588 (1991).
6.
7.
Horovitz, A., Matthews, J.M. & Fersht, A.R. J.Mol.Biol. 227, 560-568 (1992).
Serrano, L., Sancho, J., Hirshberg, M. & Fersht, A.R. J. Mol. Biol. 227, 544-559
8.
9.
10.
11.
12.
(1992).
Blaber, M., Zhang, X. & Matthews, B.W. Science 260, 1637-1640 (1993).
Kim, C.A. & Berg, J.M. Nature 362, 267-270 (1993).
Minor, D.L., Jr. & Kim, P.S. Nature 367, 660-663 (1994).
Smith, C.K., Withka, J.M. & Regan, L. Biochemistry 33, 5510-5517 (1994).
Heinz, D.W., Baase, W.A., Dahlquist, F.W. & Matthews, B.W. Nature 361, 561-
1.
2.
3.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
564 (1993).
Minor, D.L., Jr. & Kim, P.S. Nature 371, 264-267 (1994).
Xiong, H., Buckwalter, B.L., Shieh, H.-M. & Hecht, M.H. Proc. Natl. Acad. Sci.
USA 92, 6349-6353 (1995).
Smith, C.K. & Regan, L. Science 270, 980-982 (1995).
Dill, K.A., et al. Prot. Sci. 4, 561-602 (1995).
Lumb, K.J., Carr, C.M. & Kim, P.S. Biochemistry 33, 7361-7367 (1994).
Privalov, P.L. Annu. Rev. Biophys. Biophys. Chem. 18, 47-69 (1989).
Lumb, K.J. & Kim, P.S. Biochemistry 34, 8642-8648 (1995).
Gronenborn, A.M. & Clore, G.M. J. Mol. Biol. 223, 331-335 (1993).
Dyson, H.J. & Wright, P.E. Annu. Rev. Biophys. Biophys. Chem. 20, 519-538
(1991).
Scholtz, J.M. & Baldwin, R.L. Annu. Rev. Biophys. Biomol. Struct. 21, 95-118
(1992).
Carr, C.M. & Kim, P.S. Cell 73, 823-832 (1993).
Bullough, P.A., Hughson, F.M., Skehel, J.J. & Wiley, D.C. Nature 371, 37-43
(1994).
Goldsmith, E.J. & Mottonen, J. Structure 2, 241-244 (1994).
Kabsch, W. & Sander, C. Proc. Natl. Acad. Sci. USA 81, 1075-1078 (1984).
27. Wilson, I.A., Haft, D.H., Getzoff, E.D., Tainer, J.A. & Lerner, R.A. Proc. Natl.
Acad. Sci. USA 82, 5255-5259 (1985).
28. Argos, P. J. Mol. Biol. 197, 331-348 (1987).
29. Cohen, B.I., Presnell, S.R. & Cohen, F.E. Prot. Sci. 2, 2134-2145 (1993).
30. Gronenborn, A.M., et al. Science 253, 657-661 (1991).
31. Achari, A., et al. Biochemistry 31, 10449-10457 (1991).
32. Wiithrich, K. NMR of Proteins and Nucleic Acids (John Wiley & Sons, Inc.,
New York, 1986).
33. Bai, Y., Milne, J.S., Mayne, L. & Englander, S.W. Proteins 17, 75-86 (1993).
34. Fahnestock, S.R., Alexander, P., Filpula, D. & Nagle, J. in Bacterial
Immunoglobin-Binding Proteins (eds. Boyle, M.D.P.) (Academic Press, Inc., San
Diego, 1990).
35. Woody, R.W. in The Peptides (Academic Press, Inc., New York, 1985).
36. Chou, P.Y. & Fasman, G.D. Biochemistry 13, 211-222 (1973).
37. Edelhoch, H. Biochemistry 6, 1948-1954 (1967).
38. Schuster, T.M. & Laue, T.M. Modern Analytical Ultracentrifugation
(Birkhdiuser, Boston, MA, 1994).
ACKNOWLEDGEMENTS. We thank L.M. Mayr, C.J. McKnight, Z.Y. Peng, P.A.
Petillo, B.A. Schulman and T.N.M. Schumacher for very valuable advice and
discussions at various stages of this work and A.G. Cochran, B.A. Schulman, T.N.M.
Schumacher, R. Rutkowski and R.T. Sauer for comments on the manuscript.
t-3
tFH
H
03
H
ta
P
tH
H
c
-
II
H
tH
t0t
t0
0
H]
thj
hr
z
z
U
0
Ul
0
2]
tlI
H]
U~r
U
t-
tL
U
U
H]
H
H]
H]
I'
;r
e,
=r
r
Cb
Cf,
O
3
Pc
f3
t3
Fig. 2
A
0
-5
O
E
CMl
****15155551·r~
-5
E
co0
-10
E
co
CO
rCM
MlNM
-15
OOoOOO
o0o0
am....**** gMI
......
0000
***********··
*** ,*
·
-20
15
30
45
Temperature
60
(oC)
75
90
3t
Qn
W
u.ig
u1
C)
N
n
-
-4
(D
S
-4
C
cn
--4
m;
0
a3
-4l
Z
.6
•Z
CAI)
En
--
cn
°-0-
(D
--
~o.
U'z e
-4 (A
U'
ca
N
C
O)
C)o
r-
"-
0
.6. G
-(
-\
0-
40
(0
40
>
N0
O
*4(0 0 0,~c
, CW
a
-4
!-4
(0
I
I
I
I
126.0
I
132.0
120.0
114.0
108.0
15N (p.p.m.)
O_
.
(I
O
(D
;-4O
a
_-
9
--4
0
Cr
*
o,
C~9
°
-N
-1
*
Z
0.
o
g*
0
_--
CM
N
C
132.0
126.0
120.0
15 N
114.0
(p.p.m.)
108.0
C
Chameleon
Fig. 2
3.5
3
2.5
£.
0,
2
1.5
1
0.5
0
1
10
20
30
Residue
D
Chameleon
3.5
3
2.5
2
O
0)
0
-
1.5
1
0.5
0
1 1
21
3esi 1
Residue
40
50
Fig. 3a
-sscOC-'
/N~
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38
dNN (i,i+l)
daoN (i,i+3)
mm
m
n
dNN (i,i+2)
daN (i,i+4)
sa~a
Ir..........
..
0
O
-I: ....
. ....
7,
: _::::
.z~
0=
t0-M
z
X-0
-- --- I
OCI4
M-0o o
I-~
0 =0
i:
-O
46
"-zo
z-0o
0=--0
=I=z
0=0
0-=0
0=0w
z
z-z
0=0
zoa/
\-0
\
/
z-z
0=0
0=0
/-z
0 =0
= o -x M
/-zM
o
z -x
0=0
/
O=--
Z-z
,,
-0 o=0 F"Ov,z--
~I
C0=0=
z-o n=
U
-z
Z -.
0=0
/
A&-z4..
o0-x
z\-z
0=0o
----
0==o
z--
& 0 -M
z--c
PS'
OX-0Z
E-0
X-0 i v
z
I
P
o--
-o
o0=o
L 0 -m~~4mmo
/
z-o
-
00 --:
Z/
X-z
0=0
0o=0
X--Z
x-o
-z
-
4 -+M-
CI
z --
0=0
0=o
z
0=
46
-I
--
0=0
%\ 0-=0
X-Z
F
Fig. 3c
CHAPTER 5
THE ROLE OF TERTIARY INTERACTIONS IN
PROTEIN SECONDARY STRUCTURE FORMATION
A new definition for P-sheet propensity
'Secondary structure propensity' has been traditionally used to describe intrinsic
conformational preferences of amino acids arising from interactions of the amino
acid sidechain with the local protein backbone. This definition seems to be accurate
for a-helices. Studies of ca-helix formation in a variety of systems have resulted in a
consensus scale for ca-helix propensity that is largely context independentl. The
experimental studies of 3-sheet propensity suggest that the term 'p-sheet propensity'
is in large part a misnomer. Whether it is energetically favorable or unfavorable for
a given amino acid to form part of a 1-sheet largely depends on the tertiary context
provided by the surrounding 3-sheet structure 2 . Thus, for 1-sheets, local
'propensity' in the traditional sense plays a minor role for most amino acids,
compared to the dominant effects of tertiary interactions.
Because of the importance of tertiary context, the continued use of the term
'p-sheet propensity' may cause some confusion. For this reason, we propose a new
set of terms to describe the preference for a given amino acid to be in a P-sheet. For
the remainder of the chapter, P-sheet preference will be used to describe the overall
ability of a residue to fit a 3-sheet environment. P-sheet preference comprises the
sum of two contributing factors, intrinsic and extrinsic f-sheet bias. Intrinsic f-sheet
bias denotes a context independent property of an amino acid resulting from local
interactions of the sidechain with the polypeptide backbone that influence the
amino acid toward the 4 and x values for the extended 1-strand conformation (This
term is equivalent to the traditional meaning of 'propensity'). Extrinsic f-sheet bias
denotes context effects such as the preferences for a given residue to pack into the
tertiary environment of a 3-sheet.
What does extrinsic P-sheet bias really represent?
Experimental investigation of P-sheet propensity 2- 5 in the two tertiary
environments possible for 3-strands, central and edge 1-sheet positions, has led to
the conclusion that context is the major determinant of P-sheet propensity 2 . This
conclusion has been corroborated by recent theoretical studies of 3-sheet formation 6
'Context', however, is a vague term encompassing many different specific physical
effects such as hydrophobicity, packing interactions and electrostatic interactions.
These specific factors may play roles of varying importance and it remains to be
determined whether some are of more general importance than others.
The correlation of the differences between j-sheet preference scales from the
center and edge P-sheet studies 2 with measures of sidechain hydrophobicity suggests
that burial of sidechain hydrophobic surface is important. The average P-sheet
position is quite hydrophobic 7 even at solvent exposed f-sheet positions. For
example, the solvent exposed 3-sheet position in the GB1 central strand host,
AASS 4 , provides a tertiary environment that buries on average 70% of the
hydrophobic surface area of a residue. This occurs even though the adjacent
residues are small, alanine and serine, and indicates that much of the burial
involves interactions of the sidechain with the surrounding f-sheet scaffold. The
average surface area buried at xo-helical positions is - 45%6.
Although burial of hydrophobic surface appears to be important, isoleucine and
leucine, two residues with very similar hydrophobicites 8 , differ significantly in their
measured P-sheet forming preferences at both center and edge P-sheet positions.
This observation suggests that there are factors that are more specific than simple
hydrophobicity playing important roles in determining extrinsic P-sheet bias. Since
a sidechain in a P-sheet buries a substantial portion of its surface against the
surrounding basic P-sheet architecture, it seems likely that part of extrinsic P-sheet
bias arises from the specific ways a given residue can pack into the P-sheet scaffold.
The high 1-sheet preferences of 1-branched residues relative to non-P-branched
residues at central 1-sheet positions may reflect a predisposition for P-branched
residues to pack into this backbone environment.
Pairwise interactions between cross-strand neighbors
Another factor that is likely to affect extrinsic P-sheet bias is the specific
environment provided by neighboring sidechains from adjacent 1-strands.
Theoretical studies of 1-sheet formation have suggested that cross-strand sheet
interactions between sidechains on adjacent strands are crucial components of sheet
stability6, 9. Significant nonrandom distributions have been observed in studies of
the pairwise neighbors of amino acids in P-sheets 9 - 1 1 . These pairwise statistical
preferences do not only reflect general interactions between amino acids of the same
type, such as hydrophobic amino acid with hydrophobic amino acid, but appear to
involve specific interactions between residues 9 ' 11
A recent experimental study has directly addressed the energetics of a set of 32
of the possible 400 interstrand pair possibilities at positions 44 and 53 of the GB1
model system 12. Positions 44 and 53 are cross-strand neighbors and are the edge and
center 3-sheet positions used previously to investigate the 0- sheet preferences of
individual amino acids. The pairwise study demonstrates that sidechain-sidechain
interactions may modulate the extrinsic 3-sheet bias of a given residue. On its own,
threonine has the highest 3-sheet preference at both edge and center 3-sheet
positions 2 , 4, 5. However, the measured stability for the Thr-Thr cross-strand pair is
lower than the expected stability calculated from the sum of the measured
individual P-sheet preferences at positions 44 and 53. This indicates that the
presence of one threonine cross-strand partner affects the extrinsic P-sheet bias of the
other threonine in a negative way. The extrinsic 3-sheet bias of other residues are
affected in a positive way when an appropriate cross-strand partner is present. For
example, the Glu-Arg, Glu-Lys and Phe-Phe pairs are - 1 kcal mol- 1 more stable than
expected from the individual 3-sheet preferences for those pairs in this system.
Nonadditive cross-strand interactions are common in this study. The pairing
interaction energies that are found are as large as the center n-sheet preference
values 4 , 5 and appear to correlate with some of the trends seen in the statistical
surveys of pairwise 3-sheet preferences. The exact source of the positive and
negative interaction energies has not been directly addressed and would require
detailed structural information about each molecule. Nonetheless, these
experiments provide further evidence of the importance of context in determining
the P-sheet preference of an amino acid and indicate that extrinsic 3-sheet bias is
affected by very specific interactions.
Close packing of sidechains in P-sheets
Close packing of sidechain residues across adjacent strands occurs in all j-sheet
structures 13 - 1 5. Very different sidechain-sidechain interstrand packing geometries
are found depending on the direction of the neighboring strands, which can be
either parallel or anti-parallel. For parallel P-strands, the basic geometric
arrangement of the a-0 vectors of the sidechains of adjacent amino acids is parallel
(Fig. 1). In anti-parallel 3-sheets, there are two distinct geometries. These
geometries are specific to the size of the ring of backbone atoms enclosed by
successive cross-stand backbone hydrogen bonds 1 3' 16 The a-3 vectors are
convergent for residues occurring in the large hydrogen bond ring positions of
antiparallel P-strands (Fig. 2), and divergent for residues found in the small
hydrogen bond ring positions of antiparallel P-strands (Fig. 3).
The measurement of energetic differences between amino acid pairs examined
in both orientations at positions 53 and 44 of GB112 suggests that the specific
orientation of cross-strand neighbors can have consequences for P-sheet stability.
The 44-53 pair occupies a small hydrogen bond ring in anti-parallel P-strands. The
differences in free energy, AAAG, between three sets of residue pairs, (denoted,
residue 44/residue 53), Thr/Ile - Ile/Thr, Thr/Phe - Phe/Thr and Ile/Phe - Phe/Ile,
examined in both orientations are the order of 0.1 to 0.2 kcal-mol-1 indicating an
asymmetry in the interactions. While these values are small, a limited number of
pairs were examined. It is possible that there could be large differences for other
pairs particularly those that could interact via highly directional interactions like
hydrogen bonds or electrostatic interactions.
There do not appear to be any major restrictions on sidechain rotamers in
1-sheets when the structural database is surveyed as a whole1 7. However, the close
packing of sidechains in 3-sheets may place distinct restrictions at particular 3-sheet
sites in a case by case fashion9, 18 Perhaps the most striking example of this is seen
in the parallel 3-sheets of the 3-helices of the pectate lyase structures 19 . These
proteins are composed of all parallel 3-strands wound into a large right handed coil.
The residues on successive adjacent parallel strands form distinct amino acid 'stacks'
involving cross-strand sidechain packing of adjacent residues. All residue types
seem to be able to pack in this general fashion since individual stacks of aliphatic
residues, aromatic residues, as well as serine and asparagine are found. In
particular, the stacks of aromatic residues show how distinct rotamers are used in
3-sheet structures to allow for the establishment of specific cross-strand
sidechain-sidechain interactions. The sidechain rings of aromatic residues on
adjacent 3-strands are oriented in a manner similar to the observed base pair
stacking in DNA, in which the aromatic rings are all approximately parallel with
slightly offset centers and have an interplane distance of 3.4A 19 .
Examples of the preferred pairwise packing arrangements are also observed in
anti-parallel sheets. Distinct pairwise statistics are seen for both large and small
hydrogen bond ring sites. The study by Wouters and Curmill shows that particular
residue pairs are not only preferred for a given type of P-sheet site, but within the
preferred pairs, specific sidechain rotamers are frequently seen. For example, in the
small hydrogen bonded ring positions Phe-Phe pairs are often found in sidechain
rotamers that facilitate aromatic stacking interactions. Similarly, in the large
hydrogen bond ring positions, small 0-branched residues like Val, Ile and Thr are
frequently found in the trans rotamer which facilitates sidechain-sidechain
packing11. The specific sidechain packing arrangements observed for the three basic
P-sheet geometries, suggest that specific packing interactions are important
contributors to the extrinsic P-sheet biases of amino acids.
Possible directions for the study of 1-sheet cross-strand interactions
Ultimately, for protein engineering and design purposes as well as protein
structure prediction, one would like to understand the 'rules' for the basic
cross-strand packing geometries. The cross-strand interaction study of Smith and
Reganl2 demonstrates that these interactions can significantly affect 1-sheet stability.
However, the combinatorial nature of this problem makes direct experimental study
of all of the possibilities for the three basic packing arrangements a daunting
prospect (1200 total residue pairs to cover all 400 possible residue pairs for each of the
three basic cross-strand geometries, parallel, anti-parallel small hydrogen bond rings
and anti-parallel large hydrogen bond rings). It seems that this problem might be
best approached using computational methods that would allow for the rapid
enumeration and energetic evaluation of all possibilities.
The problem of understanding the pairwise interactions in 3-sheets is similar
to the 'inverse protein folding' problem 2 0, 21 There is a basic, well defined
backbone environment in which to evaluate many amino acid combinations.
Specific P-sheet positions from a known protein, such as specific sites in the 1-sheet
of GB1, could be used as models for each of the basic 1-sheet geometries. Each of
these tertiary template models could be used to evaluate the relative interaction
energies of the residue pairs. The advantage of this approach is that the calculations
could be both calibrated and tested by measurements of the actual interaction
energies for some of the pairs in the experimental system.
Substitutions at any set of guest positions are likely to result in small changes
in the exact conformation of the local r-sheet backbone. For this reason, a method
that not only evaluated the suitability of a residue for a given P-sheet environment,
but also would allow for some degree of backbone adjustment would be quite useful.
It has been demonstrated that tertiary packing in coiled-coil structures can be
predicted very accurately if the motion of the backbone is taken into account 2 2. This
was dependent on the availability of an algebraic description of the coiled-coil
backbone. Recently, a mathematical parameterization has been derived for f-barrel
protein structures 2 3 , 24. The analysis of these structures indicates that there are a
very limited set of defined architectures for this class of proteins and provides a
possible framework for experiments directed at understanding P-sheet packing
arrangements analogous to those of Harbury et. al. for coiled coils.
Tertiary effects on larger units of secondary structures
It is clear from the 3-sheet preference work that context effects play a very
important role in determining the favorability of a given amino acid for a f-sheet
secondary structure. To what extent can nonlocal factors determine the
conformation of entire secondary structures?
Studies of the protein structural database find examples of identical
pentapeptide and hexapeptide sequences in different secondary structure
conformations in different proteins 2 5 -2 8 . These observations suggest that tertiary
interactions can affect the conformation of short peptide sequences. Different
solvent systems such as trifluoroethanol or sodium dodecylsulfate have been
observed to affect secondary structure changes between a-helix and 3-sheet
conformations in peptides. These experiments suggest that tertiary interactions can
affect the conformation of longer peptide sequences but do not give direct
proof 2 9 , 30. Two recent experiments have directly addressed the question of the
importance of nonlocal interactions in establishing secondary structure. These
experiments have directly demonstrated that tertiary interactions can govern
secondary structure formation in peptide sequences of substantial length.
In one set of experiments, peptides with the hydrophobic/hydrophilic (H/P)
patterning of either helices or sheets were made 3 1 . Two peptides were made for
each pattern. In one case, the peptide was composed of residues with strong
secondary structure preferences. In the other case, the peptide was composed of
residues with poor secondary structure preferences. The H/P patterning was found
to dominate over the local secondary structure tendencies for determining the
secondary structure of the peptides in all cases. The peptides with helical H/P
pattering formed self-associated helices and those with the H/P patterning of
3-sheets formed self-associated 3-sheets. These experiments show that the
hydrophobic/hydrophilic pattern of a sequence can provide enough information to
direct secondary structure of self-associating peptides formation regardless of
intrinsic secondary structure preferences although it is not clear whether these
structures are discrete or well packed.
Using a protein design approach, an 11 amino acid sequence (the Chameleon
sequence) that folds as an a-helix in one position of the primary sequence of GB1 but
as a 3-sheet in another was designed 3 2 . The guiding principle for this work was to
create a sequence that could satisfy the tertiary packing requirements of the positions
critical for forming the interface between each secondary structure and the core of
the protein. The proteins containing the Chameleon sequence in the a-helix and
1-sheet conformations, Chameleona and ChameleonP respectively, both fold into
structures similar to wild-type GB1. Moreover, both proteins maintain the physical
features of the wild-type protein, such as disperse NMR spectra, slow amide proton
exchange, reversible thermal unfolding and IgG-binding, indicating that they are
well packed, native-like proteins. This set of experiments provides a striking
example of the extent to which tertiary interactions can determine secondary
structure formation in a discrete, well packed tertiary context.
Both the peptide and the Chameleon experiments provide support for a view
of protein structure that is dramatically different from the local view of protein
structure exemplified by a recent protein structure prediction algorithm, named
LINUS (Local Independently Nucleated Units of Structure) 3 3 . The underlying
assumption in the approach used by LINUS is that protein structure is organized in
a hierarchical fashion with secondary structures providing the building blocks for
higher order protein structure. This suggests that long range, nonlocal interactions
will have minimal consequences for secondary structure formation.
The Chameleon and H/P peptide experiments suggest an inverted hierarchy
for protein structure where secondary structures result from tertiary interactions
rather than local interactions. These results indicate that at least in some cases
secondary structures may result from long range interactions such packing effects
and/or hydrophobic collapse in a manner consistent with the models of proteins
folding suggested by Dill and coworkers 34 35
The structural factors that cause a protein sequence to adopt a given fold are
still not well understood. The information within a sequence that specifies the
three dimensional structure of a protein is highly redundant, as evident in the high
tolerance of proteins for many types of mutations 3 6 . Much effort has been made
recently to understand protein structure in terms of the packing of the polypeptide
chain 3 7 . Along this line of research, methods for protein structure prediction that
address the inverse protein folding problem have been developed 2 1, 38. The work
described in this thesis indicates that P-sheet secondary structure needs to be
understood in terms of the details of polypeptide chain packing and tertiary
interactions.
Figure 1 Schematic diagram of the relative sidechain orientations of amino acids in
a parallel P-sheet, a, presents the view from above the plane of the sheet. Ca
hydrogen atoms are omitted for clarity. b, presents a view of the parallel
orientation of the sidechain a-3 vectors as seen parallel to the plane of the sheet
for the residue pair indicated by the box in a.
Figure 2 Schematic diagram of the relative sidechain orientations of amino acids in
an anti-parallel 3-sheet highlighting the residue pairs of the large hydrogen
bond rings 1 3 ' 16. a, presents the view from above the plane of the sheet. Ca
hydrogen atoms are omitted for clarity. b, presents a view of the •a-3 sidechain
vectors as seen parallel to the plane of the sheet for the residue pair indicated by
the box in a.
Figure 3 Schematic diagram of the relative sidechain orientations of amino acids in
an anti-parallel 1-sheet highlighting the residue pairs of the large hydrogen
bond rings 13, 16 . a, presents the view from above the plane of the sheet. Ca
hydrogen atoms are omitted for clarity. b, presents a view of the Xa-3 sidechain
vectors as seen parallel to the plane of the sheet for the residue pair indicated by
the boxes in a.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
Chakrabartty, A. & Baldwin, R.L. Adv. Prot. Chem. 46, 141-176 (1995).
Minor, D.L., Jr. & Kim, P.S. Nature 371, 264-267 (1994).
Kim, C.A. & Berg, J.M. Nature 362, 267-270 (1993).
Minor, D.L., Jr. & Kim, P.S. Nature 367, 660-663 (1994).
Smith, C.K., Withka, J.M. & Regan, L. Biochemistry 33, 5510-5517 (1994).
Yang, A.S. & Honig, B. J. Mol. Biol. 252, 366-376 (1995).
Sternberg, M.J.E. & Thornton, J.M. J. Mol. Biol. 115, 1-17 (1977).
Fauchere, J.-L. & Pliska, V. Eur. J. Med. Chem. - Chim. Ther. 18, 369-375 (1983).
Lifson, S. & Sander, C. J. Mol. Biol. 139, 627-639 (1980).
von Heijne, G. & Blomberg, C. J. Mol. Biol. 117, 821-824 (1977).
Wouters, M.A. & Curmi, P.M. Proteins 22, 119-131 (1995).
Smith, C.K. & Regan, L. Science 270, 980-982 (1995).
Salemme, F.R. Prog. Biophys. Molec. Biol. 42, 95-133 (1983).
Chothia, C. Ann. Rev. Biochem. 53, 537-572 (1984).
Chothia, C. & Finkelstein, A.V. Ann. Rev. Biochem. 59, 1007-1039 (1990).
Pauling, L. & Corey, R.B. Proc. Natl. Acad. Sci. USA 37, 729-740 (1951).
McGregor, M.J., Islam, S.A. & Sternberg, M.J.E. J. Mol. Biol. 198, 295-310 (1987).
Dunbrack, R.L., Jr. & Karplus, M. Nature Struct. Biol. 1, 334-340 (1994).
Yoder, M.D. & Lietzke, S.E. Structure 1, 241-251 (1993).
Bashford, D., Chothia, C. & Lesk, A.M. J. Mol. Biol. 196, 199-216 (1987).
Bowie, J.U., Ltithy, R. & Eisenberg, D. Science 253, 164-170 (1991).
Harbury, P.B., Tidor, B. & Kim, P.S. Proc. Natl. Acad. Sci. USA 92, 8408-8412
(1995).
Murzin, A.G., Lesk, A.M. & Chothia, C. J. Mol. Biol. 236, 1369-1381 (1994).
Murzin, A.G., Lesk, A.M. & Chothia, C. J. Mol. Biol. 236, 1382-1400 (1994).
Kabsch, W. & Sander, C. Proc. Natl. Acad. Sci. USA 81, 1075-1078 (1984).
Wilson, I.A., Haft, D.H., Getzoff, E.D., Tainer, J.A. & Lerner, R.A. Proc. Natl.
Acad. Sci. USA 82, 5255-5259 (1985).
Argos, P. J. Mol. Biol. 197, 331-348 (1987).
Cohen, B.I., Presnell, S.R. & Cohen, F.E. Prot. Sci. 2, 2134-2145 (1993).
Zhong, L. & Johnson, W.C.J. Proc. Natl. Acad. Sci. USA 89, 4462-4465 (1992).
Waterhous, D.V. & Johnson, W.C.J. Biochemistry 33, 2121-2128 (1994).
Xiong, H., Buckwalter, B.L., Shieh, H.-M. & Hecht, M.H. Proc. Natl. Acad. Sci.
USA 92, 6349-6353 (1995).
32.
33.
34.
Minor, D.L., Jr. & Kim, P.S. Nature submitted
Srinivasan, R. & Rose, G.D. Proteins 22, 81-99 (1995).
35.
36.
Dill, K.A. Biochemistry 29, 7133-7155 (1990).
Dill, K.A., et al. Prot. Sci. 4, 561-602 (1995).
Matthews, B. Adv. Prot. Chem. 46, 249-278 (1995).
37.
38.
Richards, F.M. & Lim, W.A. Quart. Rev. Biophys. 26, 423-498 (1994).
Jones, J.T., Taylor, W.R. & Thornton, J.M. Nature 358, 86-89 (1992).
Fig. 1
Parallel 3-sheet
H
C/
C
II
C,*
J
'
SC
II
c
"I
H
C
II
0
0
f3
R
fi
N
/,Cs
C,
II
0
v
a, View from above sheet
b, View of cross-strand neighbor
(-p3 vectors
Fig. 2
Anti-parallel 3-sheet
'small hydrogen bond rings'
R
H
I
0
-CN.
/N
C
f3
cc
II
•,A
C
f
N
-~
/
cx
y
R
a, View from above sheet
b, View of cross-strand neighbor
o-P vectors
Fig. 3
Anti-parallel 3-sheet
'large hydrogen bond rings'
I
A
O
II
.C\
-C
II
O
H
H
I
O
II
/N ~,C
R
-N/C'C
N
O
H
H
Oc
j13
U
U
,N\ C/C
R
a, View from above sheet
b, View of cross-strand neighbor
oa-3 vectors
Appendix I
COMPARISONS OF EXPERIMENTAL AND STATISTICAL
P-SHEET FORMATION SCALES
Appendix I
Many statistical scales for secondary structure formation are available. This
appendix provides a comparison of the experimental AAG values for f-sheet
formation measured in derivatives of the IgG-binding domain of protein G (GB1) at
both central 1' 2 and edge P-sheet positions 3 with the more common statistical scales.
It also contains comparisons of the f-sheet forming scales from GB1 with those
derived from a second experimental system, a zinc finger peptide 4
Central P-sheet positions
Investigations of 3-sheet propensity in derivatives of GB1 by two independent
groups lead to a thermodynamic scale for 3-sheet formation 1 , 2. Both studies used
position 53, which lies on the solvent exposed side of the f-sheet at a central P-sheet
position, as the guest position into which all twenty naturally occurring amino acids
were substituted. In each case, the nearest neighbor residues were changed to small
neutral amino acids in order to minimize contributions from local sidechain
context. The construction of the guest site was slightly different in each case. Both
groups changed the nearest neighbor residues on 3-strands adjacent to the guest site
to alanine. Either serinel or threonine 2 was used for the nearest neighbor residues
at positions i+2 and i-2 to the guest site. These changes created host molecules
AASS and AATT respectively. The changes in stability caused by substitution of the
twenty naturally occurring amino acids at the guest position were measured by
thermal unfolding.
The thermodynamic scales for 3-sheet propensity measured in AASS and
AATT indicated that there were significant differences among the 3-sheet
propensities of the amino acids. Alanine was used as the reference amino acid for
these scales as it has a structural feature, a 1-carbon, common to all other amino
acids, except glycine. There were some minor variations between the scales, (Table
1, Fig. la), most likely due to small context effects from the differences in the
residues at i+2 and i-2 to the guest site. Amino acids with P-branched sidechains
were found to be the best 3-sheet formers, whereas those with unbranched
sidechains were not. This basic distinction in terms of sidechain shape, is the
converse of that seen for ca-helix propensity.
The overall rank order obtained from these studies was similar to those seen in
structural database surveys, (Table 1, Fig.lb-e). In particular, a pseudoenergy scale for
3-sheet propensity derived from O-xV matrices by Mufioz and Serrano 5 , published
subsequent to the experimental studies, shows remarkable agreement the GB1 host
AASS and AATT values 1' 2. The agreement with the pseudoenergy scale is as good
as the agreement between the two experimental GB1 host systems AASS and AATT.
Edge P-sheet positions
Examination of P-sheet architecture indicates that P-strands occur in two
distinct tertiary environments, central and edge P-sheet positions which differ
substantially in the average amount of amino acid surface area buried6,' 7. Separate
1-sheet propensity scales have been derived for each type of position from either
energetic parameters 8' 9 or from separate statistical analyses of the occurrence of
residues in 1-strands with one (edge) or two (center) P-strand partners 1 0' 11
In order to explore the possibility of a different 3-sheet propensity scale for edge
1-sheet positions, a host molecule having the guest 1-sheet position on an edge
1-strand of GB1 was studied. Position 44 on the solvent exposed edge 3-strand of
GB1 was used as a guest position. The nearest neighbor residue from the adjacent
1-strand as well as the i+2 and i-2 nearest neighbor positions on the 1-strand
containing the guest position were changed to alanine to create a minimal P-sheet
position (host molecule AAA).
A second thermodynamic scale for 3-sheet formation was derived for the edge
guest position in AAA in an analogous manner to the initial AASS study. Unlike
the previous results, this scale indicated many of the amino acids were P-sheet
indifferent, having AAG values near 0 kcal mol -1 relative to alanine. Residue 44 has
a mobility similar to that of the central 1-strand positions of GB1 1 2 . Therefore, it is
unlikely that the differences in AAG values reflect differences in backbone dynamics
between the center and edge positions. This scale was not at all similar to the zinc
finger scale, which had also been measured at an edge 3-sheet position 4 (Table 2).
The measured edge P-sheet propensity values did not correlate with the
statistically derived edge P-sheet propensities of Garratt et al. 10 or the energetic scale
for edge P-sheet positions of Finkelstein 8' 9, (Fig. 2a and b, Table 2). There is a weak
correlation with the more recently reported values for edge 1-sheet propensities
reported by Swindells et. al. 1 1 (Fig 2c, Table 2).
Hydrogen exchange as a measure of P-sheet propensity
Hydrogen exchange factors for the individual amino acids in peptides have
been suggested as a measure of intrinsic 1-sheet propensityl3 (the intrinsic p-sheet
bias of Chapter 5). Comparison of the results of the hydrogen exchange factors with
the zinc finger data 14 showed some similar trends although there was a significant
difference in the energy scales. The energetic values from the hydrogen exchange
measurements also show some correlation with the center site GB1 1-sheet
propensity values (R = 0.45, 0.05 < p < 0.1) with a good agreement in the magnitude
of the effect. However, comparison with the edge site GB1 data indicates no
significant correlation (R = 0.07, p>0.25). Thus, whether the differences in hydrogen
exchange factors measured in peptides have similar physical origins to intrinsic
3-sheet bias remains unclear.
On the whole, the central P-sheet preferences show better correlations with
statistical P-sheet scales than the edge 1-sheet preferences do. The reason for this is
unclear. One possible explanation is that there is a predominance of central 1-sheet
positions in the database and the average tertiary environments of central P-sheet
positions are more or less similar. This is consistent with the observation that the
1-sheet backbone structure buries a substantial fraction of the surface area of a
residue on a central strand 6
Much less area is buried by the backbone at edge positions (-70% of the central
strand value) 6 . The poor correlations of the experimental data with the statistical
edge P-sheet scales suggests that the tertiary environment of edge positions in the
database is more varied. Since less area is buried by the backbone, these positions
have greater potential to be influenced by other types of tertiary interactions.
Correlation
Coefficient (R)
P value
Finkelstein FP 8, 9
0.79
p < 0.001
Chou & Fasman PP 15
0.70
p < 0.001
Swindells et. al. PiP 11
0.70
p < 0.001
Mufioz & Serrano Pp 5
0.96
p < 0.0005
Zinc finger AAGP 4
0.56
p < 0.01
AATT AAGPcenter 2
0.96
p < 0.0005
Parameter
Table 1. Comparisons of the f-sheet propensity scale from the AASS model
system (AAGPcenter) 1 values with other P-sheet scales.
PP and FP are statistical n-sheet propensity values. Pi P are statistical 3-sheet
propensity values derived specifically for central P-strand positions. AAGP and
AAGPcenter are experimentally derived scale for f-sheet formation from the zinc
finger and GB1 systems respectively. Correlations exclude the value for proline
except for the zinc finger derived AAGP values which also exclude the value for
glycine. Correlation coefficients (R) and p values were derived as described in
ref. 16
Correlation
Coefficient (R)
P value
Finkelstein Fe 8, 9
0.13
p > 0 .5
Garratt et. al. PeP 10
0.37
0.2 > p > 0.1
Swindells et. al. PeP 11
0.49
0.05 > p > 0.02
Mufioz & Serrano Pp5
0.58
0.02 > p > 0.01
Chou & Fasman Pp 15
0.15
p > 0.25
Zinc finger AAGP 4
0.12
p > 0 .5
AASS AAGPcenter
0.61
0.01 > p 0.001
Parameter
Table 2. Comparisons of the P-sheet propensity values measured at an edge
n-sheet position in GB1 (AAGBedge) 3 with other 3-sheet scales.
PP are statistical 3-sheet propensity values. FeP and PeP are statistical P-sheet
propensity values derived specifically for edge strand positions. AAGP and
AAGPcenter are experimentally derived values for 3-sheet formation derived
from the zinc finger and GB1 systems respectively. Correlations exclude the
value for proline except for AAGP which also excludes the value for glycine.
Correlation coefficients (R) and p values were derived as described in ref. 16
Figure 1 Comparison of central strand J-sheet propensities derived from host
AASS with a, f-sheet propensities from GB1 host AATT 2 , b, 1-sheet propensity
values of Chou and Fasman 1 5, c, P-sheet formation values of Finkelstein 8' 9, d,
interior P-strand propensities of Swindells et. al. 11 and e, P-sheet propensity
pseudo energy scale of Mufioz and Serrano 5 .
Figure 2 Comparison of 1-sheet propensities derived form edge site host AAA with
a, edge strand 1-sheet propensites of Garratt et. al. 10 b, 1-sheet formation values
for edge P-sheet positions of Finkelstein 8' 9, c, edge strand P-sheet propensities
of Swindells et. al.1 1 and d, 1-sheet propensity pseudoenergy scale of MuiXoz
and Serrano 5
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
Minor, D.L., Jr. & Kim, P.S. Nature 367, 660-663 (1994).
Smith, C.K., Withka, J.M. & Regan, L. Biochemistry 33, 5510-5517 (1994).
Minor, D.L., Jr. & Kim, P.S. Nature 371, 264-267 (1994).
Kim, C.A. & Berg, J.M. Nature 362, 267-270 (1993).
Mufioz, V. & Serrano, L. Proteins 20, 301-311 (1994).
Chothia, C. J. Mol. Biol. 105, 1-14 (1976).
Sternberg, M.J.E. & Thornton, J.M. J. Mol. Biol. 115, 1-17 (1977).
Finkelstein, A.V. & Reva, B.A. Nature 351, 497-499 (1991).
Finkelstein, A.V. Prot. Eng. 8, 207-209 (1995).
12.
Garratt, R.C., Taylor, W.R. & Thornton, J.M. FEBS Lett. 188, 59-62 (1985).
Swindells, M.B., MacArthur, M.W. & Thornton, J.M. Nature Struct. Biol. 2, 596603 (1995).
Barchi, J.J., Jr., Brasberger, B., Gronenborn, A.M. & Clore, G.M. Prot. Sci. 3, 15-21
13.
14.
15.
16.
(1994).
Bai, Y., Milne, J.S., Mayne, L. & Englander, S.W. Proteins 17, 75-86 (1993).
Bai, Y. & Englander, S.W. Proteins 18, 262-266 (1994).
Chou, P.Y. & Fasman, G.D. Biochemistry 13, 211-222 (1973).
Rosner, B. Fundamentals of Biostatistics (PWS Publishers, Boston, 1982).
A
'
Fig. 1
' I
F
•
OT
WoM
C.
cao
I
V
S
No
HOE K R
*A
O)H
Qu
ui
-
Do
Ge
f)
I
-2
B
I
I
2
-1
0
1
AAG center (kcal mo1-1 )
AASS host
2
V*
1.5 *
Le
•
o,
4•
Q
1
GO
0.5 -
*
OY
F
T
C
*M
OM
A OK *S
DO
OE
-2
-1
0
AAG (kcal mo- 1 )
1
2
1
.Fig.
r.
*G
0.5
D
No
K R
0S
0
AoQ*
HO
L CeM*T
-0.5
w
F
V
*01
-1
-2
D
-1
0
1
AAG center (kcal mol 1 )
2
3
2.5
el
V
2
Veoy
MO
1.5
L
SW
Ao
K
G
0.5
DO
0
-2
1
1
No
*
S
*S
E QR
I
I
0
1
AAG center (kcal mol 1)
2
Fig. 1
E
0.8
V.
0.4
w mo°T
C *
0
F
RO *S
He
OI -L
AS
K
E
S-Qd
Y
0
No
-0.4
De
Ge
I
-0.8
-2
I
I
-1
AAG center (kcal mo1-1 )
A
Fig.
2
v
2L
oV
1.5
M
E
le*Y*
oF
L
go
H
C
Q
K
1
GO
F
R*
S
*D
0.5
NO
I
n
-1.5
-1
I
I
I
I
-0.5
0
0.5
1
1.5
AAG edge (kcal mol 1)
B 0.8
Go
L
5 0.4
D
0
* O
W M
T
Yw F
-0.4
-0.8
-1.5
-1
I
I
-0.5
0
0.5
AAG edge (kcal mol 1 )
I
1
1.5
C
Fig. 2
I-'
v
1.5
Io
We
eT
F
ye
•C
KO LMe
10.5
*E
Re
NO A*
Ge
0.5
I
0n
-1 .5
-1
I
I
I
I
-0.5
0
0.5
AAG edge (kcal mol ')
1 .5
D 0.8
0.4
0
1-1
R*O
0
K
*
WM
I,*Y
*F
C
* T
*H
L A*
S
*E
NO
-0.4
De
G,
I
-0.8
-1.5
-1
I
I
I
-0.5
0
0.5
AAG edge (kcal mol -1 )
I
1.5
Biographical note
Daniel Louis Minor, Jr.
Education
1989 - present
Doctoral candidate
Massachusetts Institute of Technology,
Department of Chemistry,
Thesis advisor: Professor Peter S. Kim, Ph.D.
1989
University of Pennsylvania,
Biochemistry and Biophysics,
Magna cum Laude,
Honors in Biochemistry, Honors in Biophysics
B.A.
Awards
Phi Beta Kappa, 1989
Helix Prize in Biochemistry, University of Pennsylvania, 1989
Dean's list, University of Pennsylvania, 1986, 1987, 1988
Penn Student Agencies Scholarship, University of Pennsylvania, 1988-89
ILGWU Scholarship 1985
Publications
Minor, D. L., Jr. and Kim P. S."Measurement of the 3-sheet forming propensities of amino
acids" Nature 367 660-663 (1994)
Minor D.L., Jr. and Kim P.S. "Context is a major determinant of p-sheet propensity"
Nature 371 264-267 (1994)
Schumacher, T.N.M., Mayr, L.M., Minor, D.L., Jr., Milhollen, M.A., Burgess, M.W.and
Kim, P.S. "Applications of chirality: identification of (D)-amino acid peptide ligands
through Mirror-Image phage display' (Science, in press)
Minor D.L., Jr. and Kim P.S. "Design of a protein sequence that shows context-dependent
secondary structure formation" (Nature, submitted)
Personal
Date of Birth:
Place of Birth:
Citizenship:
August 21, 1967
Hazleton, PA
USA
105