BB49CH13_Ozkan ARjats.

cls April 28, 2020 19:22

Annual Review of Biophysics

The Role of Conformational
Dynamics and Allostery in
Modulating Protein Evolution
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

Paul Campitelli,1 Tushar Modi,1 Sudhir Kumar,2,3,4

and S. Banu Ozkan1
1
Center for Biological Physics, Department of Physics, Arizona State University, Tempe,
Arizona 85281, USA; email: [email protected], [email protected], [email protected]
2
Institute for Genomics and Evolutionary Medicine, Temple University, Philadelphia,
Pennsylvania 19122, USA; email: [email protected]
3
Department of Biology, Temple University, Philadelphia, Pennsylvania 19122, USA
4
Center for Excellence in Genome Medicine and Research, King Abdulaziz University,
Jeddah 21589, Saudi Arabia

Annu. Rev. Biophys. 2020. 49:267–88 Keywords

First published as a Review in Advance on
conformational dynamics, allostery, protein evolution, protein flexibility,
February 19, 2020
molecular dynamics simulation, protein design
The Annual Review of Biophysics is online at
biophys.annualreviews.org Abstract
https://doi.org/10.1146/annurev-biophys-052118-
Advances in sequencing techniques and statistical methods have made it pos-
115517
sible not only to predict sequences of ancestral proteins but also to identify
Copyright © 2020 by Annual Reviews.
thousands of mutations in the human exome, some of which are disease as-
All rights reserved
sociated. These developments have motivated numerous theories and raised
many questions regarding the fundamental principles behind protein evolu-
tion, which have been traditionally investigated horizontally using the tip of
the phylogenetic tree through comparative studies of extant proteins within
a family. In this article, we review a vertical comparison of the modern and
resurrected ancestral proteins. We focus mainly on the dynamical properties
responsible for a protein’s ability to adapt new functions in response to envi-
ronmental changes. Using the Dynamic Flexibility Index and the Dynamic
Coupling Index to quantify the relative flexibility and dynamic coupling at
a site-specific, single-amino-acid level, we provide evidence that the migra-
tion of hinges, which are often functionally critical rigid sites, is a mecha-
nism through which proteins can rapidly evolve. Additionally, we show that
disease-associated mutations in proteins often result in flexibility changes
even at positions distal from mutational sites, particularly in the modulation
of active site dynamics.

267
BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

Contents
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
NATURE UTILIZES CONFORMATIONAL DYNAMICS
FOR PROTEIN EVOLUTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
First Dominant Collective Modes Are Conserved to Maintain Function . . . . . . . . . . . 269
Sequence Conservation Correlates with the Flexibility of a Position . . . . . . . . . . . . . . . 270
Nature Uses a Hinge-Shift Mechanism to Create New Dynamics
through Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
DYNAMICALLY COUPLED ALLOSTERIC POSITIONS PLAY
A CRITICAL ROLE IN MODULATING DYNAMICS . . . . . . . . . . . . . . . . . . . . . . . . . 276
The Distal, Non-Conserved Sites Coupled to Active Sites Are Used
in Evolution to Modulate Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

Disease-Associated Variants in the Human Exome Also Modulate

Conformational Dynamics and Use Allostery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
CONCLUSIONS AND FUTURE PERSPECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283

INTRODUCTION
Proteins are incredible machines in living systems on the molecular level. Not only are they accu-
rate, proficient, and specific in their function, but they also can acquire new functions and struc-
tures. The capability for protein evolution is profound when we consider the fact that a vast major-
ity of today’s proteins diverged from but a few common ancestors. Moreover, recent evolutionary
events like the emergence of drug resistance or enzymes with the capacity to degrade new chem-
icals emphasize the need to expand our understanding of these distinctive and admirable features
of proteins.
Modern proteins have evolved through a series of small changes from ancient times. Computa-
tional analysis of the evolutionary record of proteins offers a tractable and highly effective solution
to better understand protein function and diversification. Evolution has been a single, massive, on-
going experiment in diversification and optimization of the protein sequence–structure–function
relationship over billions of years, the outcomes of which are present in the sequences, structures,
and functions of modern-day (i.e., extant) protein families. Comparative analysis of sequences in
protein families sheds light on mutations that lead to changes in functions and provides insight
into the sequence–structure–function relationship. Thus, one can work backward to uncover pro-
tein evolution as encoded in the present-day proteomes; this process is also key to working forward
to genetically engineer new modified proteins by substantially speeding up future evolution.
Indeed, comparative analyses of proteins (29, 38, 91, 92, 102, 106) and protein engineering
studies (70, 73) were the first attempts to provide insight into the diversity of protein structures,
along with their sequence variation, to understand the sequence–structure–function relationship
in protein evolution. Recent advances in sequencing, along with high-resolution structure deter-
mination, have enhanced the ability to predict ancestral sequences and their 3D structures, thus
allowing us to integrate evolutionary history encoded in sequences into the study of proteins to
identify the mechanism by which mutations generate new functions (2, 17, 20, 34, 67, 68, 85, 103).
Furthermore, enhanced computational and experimental techniques in protein dynamics high-
light the crucial role of structure-encoded dynamics in function (18, 53, 66–68, 95) and are poised
to answer questions about the role that protein dynamics play in protein evolution.

268 Campitelli et al.

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

NATURE UTILIZES CONFORMATIONAL DYNAMICS

FOR PROTEIN EVOLUTION
The obsolete view of the single native structure has been long replaced by the view of an ensem-
ble of states that accurately represent the native state (61, 62, 96, 108). In the ensemble model,
a protein samples a variety of conformations through local changes such as loop motions, side-
chain rotations, or global changes through domain rearrangement. Indeed, what makes proteins
uniquely different from other macromolecules in nature is their vast conformational diversity. Al-
lostery, commonly known as regulation at a distance, is a widely used emergent property of this
ensemble picture. Rather than forming a new structure, a ligand binding to a remote site promotes
a shift in the dynamics of all residues in the protein, changing the distribution of accessible con-
formational states in the ensemble and, thus, promoting easier access to certain conformers for
allosteric regulations while restricting others (1, 8, 16, 28, 31, 32, 55, 61, 62, 74, 87, 100, 108, 109).
Furthermore, the ensemble view also matches with evolution, in which the same conserved 3D
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

native fold can adopt new functions by modulating the conformational sample space (26, 61, 89).
However, identifying substitutions that modify conformational dynamics toward a new function
or adaptation to a new environment remains a challenge. Many questions are left to explore: How
do conformational dynamics evolve as proteins evolve? Can we relate evolutionary conservation
(substitution rates per position) to conformational dynamics? If functionally critical sites are more
conserved throughout evolution, then how do proteins adapt to a new environment or evolve to
perform new functions? Do directed laboratory and natural evolution share the same physical
principles? We explore some of these questions in this article.

First Dominant Collective Modes Are Conserved to Maintain Function

The relationship between protein evolution and conformational dynamics is complex (5, 6, 30,
47, 76, 96, 105). Protein engineering studies have shown some principles for the emergence of
a new function. Notably, laboratory-directed evolution studies have demonstrated that stabiliz-
ing mutations are crucial for compensating function-altering mutations (10, 40). In addition to
stabilizing mutations, the vital role of conformational dynamics has also been reported (15, 43,
96, 109). Moreover, many advances have been made through the study of homologous proteins
corresponding to a family unified by a common fold but possessing different (sometimes wildly
so) function(s) from one another (44, 82). Experimental and computational studies have shown
that the timescales and motions of enzymatic activity can be widely different among enzyme ho-
mologs of different species, indicating that these enzymes possess fundamentally different confor-
mational dynamics while maintaining similar folds (9, 37, 59). These studies also highlight drastic
differences in specific key regions, which exhibit differences in flexibility that may be mostly re-
sponsible for functional divergence. Additionally, a correlation between coevolving residues and
the local dynamics of substrate recognition sites has also been presented, highlighting the role of
local dynamics in the design of substrate interactions (56).
In addition, the large-scale, global motions of a protein are determined by its specific 3D ar-
chitecture. Thus, when the fold is conserved among members of a protein family, these collec-
tive, low-frequency modes should be, as well (21, 59, 60, 112, 114). These low-frequency (i.e.,
slow) global fluctuations of residues in a protein play an important role in synchronizing coherent
motions at more considerable distances, aiding in long-distance interactions, which are allosteric
in nature (114). It follows that the functional differences among structural homologs relate to
changes in nonglobal modes, modes that are primarily dominated by movements in subsets of the
structure rather than by the entire protein chain in an intermediate frequency range. In fact, for a
majority of protein enzyme families, the signature collective motion (i.e., functional dynamics)

www.annualreviews.org • Protein Dynamics and Evolution 269

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

associated with catalysis is conserved and can be identified through common, low-frequency
modes (59, 60, 112, 114). Moreover, the specific functional changes among these homologous pro-
teins could arise from the differences in motions in the range of low-to-intermediate frequency
modes that are specific to individual proteins (63, 112).
This type of horizontal approach, based on a comparison of modern-day proteins on the tips
of the phylogenetic tree, far away from their ancestors, is insightful but incomplete. The pro-
tein functions have evolved vertically, as mutations throughout their history have accumulated
in their ancestral protein lineages. Therefore, it is essential to incorporate this historical back-
ground, which contains both neutral and function-altering mutations. As correctly predicted by
Pauling, Zuckerkandl, and their colleagues (78) in early 1960s, recent advances in statistical meth-
ods, along with the sequencing of full genomes, have made it possible to obtain ancestral sequences
through protein sequence alignments in a phylogenetic framework using a variety of statistical
frameworks (33, 85, 86). Most probable ancestral sequences are now constructed by synthesizing
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

DNA molecules, and the protein is subsequently expressed, which allows for robust experimental
and computational characterization of ancient proteins (86).
When a similar dynamical analysis is performed on these ancestrally reconstructed enzymes,
the principle that global dynamics of a protein are conserved in evolution still holds (112). The
comparison of slow global fluctuations of ancestral enzymes with their extant homologs indicates
that the root mean square fluctuations obtained from slower modes of extant enzymes signifi-
cantly correlate with those of their 3-billion-year-old ancestral counterparts: (a) The Escherichia
coli thioredoxin (Thrx) and its last bacterial common ancestor (LBCA) show a correlation of 0.76,
and (b) extant TEM-1 β-lactamase and its Gram-negative common ancestor (GNCA) show a cor-
relation of 0.79. Indeed, in both cases, the fold is strictly conserved, yet the ancestral proteins
are much more stable and function completely differently than their extant homologs. However,
the intermediate-to-higher-frequency modes exhibit different behavior, with limited correlations
of 0.12 and 0.37 for Thrx and β-lactamase, respectively, highlighting the shift in conformational
ensemble during evolution (Figure 1).

Sequence Conservation Correlates with the Flexibility of a Position

Functional specificity among structural homologs can be connected to the evolution of interme-
diate frequency modes governed by the motion of local regions within a protein structure. This
means that a relationship should exist between the variation in conformational dynamics of specific
positions and evolutionary rates. Studies involving specific protein families and subsets of enzymes
have shown that residues that act as hinges (i.e., sites with low flexibility) are generally more evo-
lutionarily conserved than other positions for specific protein families or a subset of enzymes (23,
56, 59, 60, 63).
While the above analyses suggest that dynamics play a role in evolution, we understand that,
ultimately, evolution of proteins involves a continuous accumulation of amino acid substitution at
different positions, some of which are functionally critical. This begs the question of how we can
relate the role of dynamics at the level of individual residue positions to evolution. A significant
reason for the lack of methods incorporating dynamical changes with amino acid substitutions in
evolution, despite growing realization of their importance, has been the absence of amino acid site–
specific measures that can statistically quantify each position’s contribution to and its substitution’s
impact on the structural dynamics of the protein.
Many techniques revolve around an attempt to reconstruct the networking maps of com-
munication between regions of a given protein (31, 61, 65, 98); one such technique is the al-
losteric wiring diagram (AWD), which captures the most relevant residue networks participating in

270 Campitelli et al.

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

S TR U C TU R E S LO W E R M O D E S FA S T E R M O D E S
0.225
E. coli E. coli
2.5 0.200
LBCA LBCA
0.175
Thioredoxin

2.0 0.150

RMSF (Å)
0.125
1.5 0.100
0.075
1.0
0.050
0.5 0.025

0 20 40 60 80 100 0 20 40 60 80 100

2.25 GNCA 0.18 GNCA

2.0 TEM-1 TEM-1
0.16
Access provided by Temple University Libraries on 06/09/20. For personal use only.

β - l a c ta m a s e

1.75
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

RMSF (Å)

1.5 0.14
1.25
0.12
1.0
0.75 0.10

0.5 0.08

0 50 100 150 200 250 0 50 100 150 200 250

Residue number Residue number
Figure 1
Slow, global motions in proteins are evolutionarily conserved, while faster modes can change dramatically between ancestral and extant
proteins. Although the modern versions of thioredoxin and β-lactamase enzymes have evolved from their ancestors toward a new
function, comparison of their 3D folds and the RMSF from slower modes show striking similarity, as evident from a high correlation
between them (0.76 and 0.79, respectively). However, there are significant differences in the RMSF from intermediate to fast modes,
indicating that the function has been fine-tuned through the modulation of intermediate or fast mode dynamics. Abbreviations: E. coli,
Escherichia coli; GNCA, Gram-negative common ancestor; LBCA, last bacterial common ancestor; RMSF, root mean square fluctuation.

allosteric signal processing. Some of these methods incorporate simulated perturbations and per-
turbation responses to amino acids within a protein structure, such as the structural perturbation
method (SPM) (98). The SPM can employ a variety of general energy functions, often using the
elastic network model (ENM), in which the protein is treated as a network of nodes (residues)
connected by elastic springs with harmonic potentials. In this model, the magnitude of response
of a specific amino acid to a perturbation will be proportional to the elastic energy of a given
mode as a result of the springs connected to that amino acid residue or position. Amino acids that
respond strongly are residues that are functionally important in allosteric signaling. The SPM has
been used to successfully identify AWDs in a variety of structures, including bacterial chaperonin
(114), molecular motors, and DNA polymerase (113).
The success of methods such as the one mentioned above emphasizes the importance of utiliz-
ing simulated forces for exploring protein conformational dynamics. Forces are used ubiquitously
in biology for important protein–protein and protein–ligand interactions, as well as in protein
chaperones, which can assist in the folding and unfolding of other proteins or biological macro-
molecules and even facilitate the refolding of misfolded proteins (54, 57, 79). To emulate the effect
of such forces, a technique has been developed that employs explicit forces to capture protein dy-
namics at the amino acid level: the dynamic flexibility index (DFI), which combines the ENM and
linear response theory (LRT) (4, 24, 25) (see Equation 3 below). In the DFI, force is used as an
additional probe to estimate the fluctuation response profile of a protein upon exertion of directed

www.annualreviews.org • Protein Dynamics and Evolution 271

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

random forces on selected residues; it allows one to sample the native ensemble efficiently and to
identify long-range dynamics that propagate or modulate the allosteric communication.
The protein network of interactions is modeled either as an elastic network, as described above,
or by incorporating dynamics from all-atomistic force fields for estimating the DFI. A unit force
perturbation is applied to the representative node of each amino acid, acting as a random Brownian
kick. This creates a response to the perturbation that then propagates through the rest of the
structure and causes other positions to fluctuate through the interaction network. The fluctuation
response, R, of each position can be calculated through LRT (see Equation 1 below), from which
a response vector is constructed to measure the magnitude and direction (x, y, z) of displacement
of every residue from its equilibrium position in the native state. Averaging this response over
multiple unit forces in different directions simulates an isotropic perturbation. This approach,
under the harmonic approximation, closely mimics the response of a protein to an approaching
substrate (11, 12, 51):
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

[R]3N×1 = [H]−1
3N×3N [F]3N×1 . 1.

In the above equation, H is the Hessian, a 3N × 3N matrix that can be constructed from atomic
coordinates and is composed of the second derivatives of the harmonic potential from the ENM
with respect to the components of the Cartesian position vectors of length 3N. In the DFI, per-
turbations are introduced as random external forces exerted on selected residues, in contrast other
approaches such as modifying the distances between pairs of nodes or spring constants (98). This
enables us to analyze residues affected by the perturbation in a manner similar to naturally occur-
ring regulatory motions that regulate dynamics in the cell. This approach also allows us to capture
the coupling between different amino acids measured by response at one amino acid as a result of
perturbations at another site.
Additionally, H can be extracted directly from molecular dynamics simulations as the inverse of
the covariance matrix, which implicitly captures specific physiochemical properties of amino acids
and more accurate residue–residue interactions via atomistic force fields and subsequent all-atom
simulation data. Each position in the structure is perturbed sequentially, repeating the process
above, for generating a perturbation response matrix A,
⎡ ⎤
|R1 |1 · · · |RN |1
⎢ . .. ⎥
AN×N = ⎢ ⎣ ..
..
. .
⎥,
⎦ 2.
|R1 |N · · · |RN |N

where |Rj |i = (R)2 is the magnitude of fluctuation response at position i due to the pertur-
bations at position j. Subsequently, the DFI value of position i is calculated as the displacement
response of position i relative to the net displacement response of the entire protein:
N
j=1 |Rj |i
DFIi = N N
. 3.
i=1 j=1 |Rj |i

The DFI quantifies the resilience of a position to perturbations exerted at other parts of the chain
through simulated mechanical force perturbations to residues in the chain. By repeatedly applying
these random perturbations to each of the positions in the chain one at a time, we can compute the
normalized response profile (i.e., DFI value) for every residue in the protein (Figure 2a). Thus,
the DFI is a relative value, being higher or lower than the average response to perturbations ob-
served at any position. Residues with very low DFI are dynamically stable; they do not exhibit

272 Campitelli et al.

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

a DFI of protein Pin1 b Trends over 100 different protein structures

1.4
Flexible
R = 0.85
1.2

Average rate of evolution

1.0

Rigid

1.0 0.8
Site flexibility

0.8
by %DFI

0.6
0.6
Access provided by Temple University Libraries on 06/09/20. For personal use only.

0.4
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

0.2
0.0 0.4
6 26 46 66 86 106 126 146 20 40 60 80 100
Residue number %DFI bin ranges
Figure 2
The dynamic flexibility index (DFI) captures amino acid flexibility at every position and correlates with evolutionary conservation. (a,
top) DFI values mapped onto protein Pin1 (Protein Data Bank ID: 1PIN), where it is colored within a spectrum of red (flexible) to blue
(rigid). (Bottom) DFI values ranked as a percentile, plotted against the residue index. Positions with %DFI values under 0.2 are
considered hinge regions and are often functionally important sites. (b) Evolutionary rates calculated over 100 sequences from the
human proteome versus %DFI values. Amino acid flexibility and evolutionary conservation are highly correlated, with more rigid
positions often being more conserved.

large fluctuations upon external force perturbations, but they do play a pivotal role in transferring
these perturbations throughout the chain in a cascading fashion. Thus, they will often be treated
as the hinge parts of the protein that control and mediate the motion, similar to joints in a skele-
ton. In contrast, sites with very high DFI are structurally flexible, are prone to amino acid chain
perturbations, and can play an important role in biochemical functions such as ligand recognition.
The DFI has been used successfully in a variety of contexts, including the establishment of a
broad relationship between the structural dynamics of individual amino acids and their evolution-
ary conservation (13). In a large-scale study, the flexibility of 39,813 residues from 100 different
proteins was analyzed using the DFI. It was observed that %DFI (DFI scores of residues ranked
with their percentiles) strongly correlates with position-specific rates of evolutionary change ob-
tained from multispecies sequence analysis. This result indicates that positions that are more im-
portant to protein dynamics, such as hinges, are in fact under stronger natural selection, and thus
that nature permits fewer amino acid substitutions at these positions (Figure 2b).

Nature Uses a Hinge-Shift Mechanism to Create New Dynamics

through Evolution
While there is a correlation with the flexibility of a position and its evolutionary rate, the question
of how flexibility profiles evolve throughout evolution needs further investigation. Ancestrally re-
constructed proteins allow us to explore the typical behavior associated with adaptation to a new
environment or the emergence of a new function in different species. For example, Thrxs are
ubiquitous oxidoreductase enzymes present in all living organisms, from Archaebacteria to hu-
mans. Recently, the ancestral forms of Thrxs (i.e., Precambrian Thrxs) have been resurrected and

www.annualreviews.org • Protein Dynamics and Evolution 273

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

analyzed (80, 83, 84, 88). In agreement with ancestral protein resurrection studies, ancestral Thrxs
share the same canonical 3D structure, and even similar chemical mechanisms of reduction, with
modern enzymes (Figure 3a,b). However, in accordance with their melting temperatures, they
are 32°C more stable than the modern extant proteins and also have higher catalytic efficiency
(i.e., higher activity) at a pH of 5. In other words, Thrxs evolved toward both lower stability and
lower activity to adapt to the changes in temperature and ocean acidity, which occurred through
environmental evolution from the ancient ambient conditions to the current conditions on Earth.
This brings up the question of how they achieved these adaptations through sequence variation
while conserving their 3D fold. To answer this question, we first performed all-atom molecular
dynamics simulations and then obtained the DFI profiles of each Thrx. Comparison of the distri-
bution of flexibility of residues in each protein reveals the differences between the ancestral and
extant Thrxs for both human and E. coli branches. Notably, as Thrxs evolved, there was a redistri-
bution of residues with medium flexibilities with a gain in rigid and highly flexible sites, suggesting
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

that proteins fine-tune their activity following the functional requirement. This characteristic pat-
tern of increasing distribution width with evolution is further supported by the high correlation
between the variance of DFI distributions over time (a correlation of R = 0.77; p = 1.6 × 10−2 )
(Figure 3c). In addition, since melting temperature correlates negatively with evolutionary time,
it correlates significantly with the variance of DFI distributions (a correlation of R = −0.86; p =
3.2 × 10−3 ) (67) (Figure 3d). Moreover, projecting the hierarchical clustering of the DFI profiles
of all nine Thrxs onto the 2D map of their stability and catalytic rates indicates that these bio-
physical attributes are indeed associated with the flexibility of their residues (Figure 3e,f ). Thus,
nature sculpts the native ensemble to adapt and alter function, a result that is in agreement with
our earlier work on protein evolution (41, 115).
Comparison of how the DFI scores of the residues differ between ancestral and extant Thrxs
on human and E. coli branches provides a plausible molecular mechanism for their adaptation to
lower temperatures. In the bacterial branch of Thrxs, the increased flexibility in the α3 region
(a hinge loss), which contributes most to stability, is compensated by a loss in the flexibility of
α4, which is critical for folding (Figure 4a). This may explain the fact that modern Thrxs have
decreased stability while maintaining their canonical 3D fold (67). This mechanism of hinge shift,
that is, the migration of a hinge (a position with low flexibility) from one region to another within
a protein, has not been observed or investigated in other ancestral studies of Thrxs.
The same hinge-shift mechanism has also been witnessed in the evolution of TEM-1 β-
lactamase (Figure 4b). Unlike the modern TEM-1, which can only degrade penicillin, the last
Gram-positive and Gram-negative common ancestor (PNCA) GNCA bacteria could degrade
both penicillin and second-generation antibiotics with similar efficiency. The enhanced substrate
promiscuity of ancestral enzymes (i.e., GNCAs and PNCAs) is not accompanied by significant
changes in the active site region; the 3D crystal structure is also conserved throughout their evolu-
tion. However, the substrate-promiscuous ancestral β-lactamases exhibit high flexibility around re-
gions close to the active site, which is significantly more rigid in penicillin-specific extant TEM-1,
emphasizing the flexibility required for the binding of different ligands (115). The decreased flex-
ibility of this region is compensated for by the rigidity of the N-terminal helix, indicating the
migration of hinges in functional evolution of TEM-1 β-lactamase.
Finally, a study using reconstructed ancestors of green fluorescent protein (GFP) shows that
the evolution of red color from a green ancestor was a result of migration of the hinge posi-
tions from the active site diagonally across the β-barrel fold, making sites near the chromophore
more flexible and subsequently accommodating the sizeable conformational change necessary for
red chromophores (41) (Figure 4c). This dynamics-driven evolutionary mechanism modifies the
flexibility profile of the barrel-shaped β-protein with a large number of tertiary contacts and a

274 Campitelli et al.

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

a E. coli b

BACTERIA ARCHAEA EUKARYOTA

LGPCA
LPBCA
LBCA

LUCA LACA

AECA
LECA

LAFCA
Human

4 3 2 1 0
Time before present (billion years)
Access provided by Temple University Libraries on 06/09/20. For personal use only.

c × 10 –5 d × 10 –5
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

3.5 3.5
LAFCA
R = 0.77, p = 0.001617 R = –0.86, p = 0.00324
Variance in DFI distribution

Variance in DFI distribution

LAFCA
E. coli E. coli
3.0 3.0

2.5 2.5
Human
Human
2.0 2.0
LGPCA LPBCA
LGPCA LBCA
1.5 LACA
1.5 LBCA
LPBCA
LACA LECA
LECA
1.0 AECA
1.0 AECA

0 1 2 3 4 90 95 100 105 110 115 120

Time (billion years) Melting temperature (°C)

4
e f
3.0
AECA LACA
2.5
3
MARGINALLY
STABLE
Distance

2.0 MODERATELY HIGHLY

ln k cat

STABLE STABLE
1.5
2 LECA

1.0
LAFCA
Human LPBCA LGPCA
0.5 LBCA
1 E. coli
0.0
AECA
LAFCA

Human

LECA

E. coli

LACA

LGPCA

LPBCA

LBCA

5 15 10
ΔG(kT)

Figure 3
Proteins can evolve new functions by keeping the same 3D fold but changing the dynamical properties, which are quantified by the
dynamic flexibility index (DFI). (a) Phylogenic tree showing human and Escherichia coli thioredoxin (Thrx) back to their common
ancestor. (b) Structural overlay of modern Thrxs and their ancestors. While they exhibit significant sequence variations, the 3D folds
are conserved. Variance in DFI distributions of ancestral and modern-day Thrx proteins correlates strongly with both (c) evolutionary
time (R = 0.77) and (d) melting temperature (R = −0.86). (e) Clustering the DFI profiles of all nine Thrx variants and ( f ) projecting
them onto a 2D map of catalytic rates and stability show that the DFI accurately groups the variants by these biophysical properties.

www.annualreviews.org • Protein Dynamics and Evolution 275

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

a ANCESTRAL PROTEIN EXTANT PROTEIN

β-lactamase
More
flexible

b
Thioredoxin
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

c Less
flexible
GFP

Figure 4
Nature can modulate function by relocating hinges within a protein through evolutionary changes in
dynamical properties without altering the 3D structure. The hinge-shift mechanism is illustrated through
the comparison of a modern protein with a reconstructed ancestral version for three different proteins,
(a) β-lactamase, (b) thioredoxin, and (c) green fluorescent protein (GFP), and the dynamic flexibility index
values are mapped onto the structures. In all cases, the location of some significant, highly flexible and highly
rigid sites shifts dramatically. For each protein, the structure is conserved throughout evolution, but
properties such as (a) ligand-binding promiscuity, (b) stability and catalytic rates, and (c) photochromatic
activity all change substantially as a result of a hinge shift.

relatively closed topology, harboring a buried active site during the evolution of green color to
red.
Taken together, these ancestral studies indicate that a hinge shift, i.e., enhancement of the flex-
ibility of some rigid sites (loss of hinges at some sites) compensated for by decreased flexibility
of some other distal sites (gain of hinges at some other sites), leading to change in dynamics, is a
common mechanism in evolution of different protein systems with different folds. These obser-
vations prompt the next question: Do the mutational sites themselves exhibit the most substantial
changes in flexibility, or does nature utilize another mechanism?

DYNAMICALLY COUPLED ALLOSTERIC POSITIONS PLAY

A CRITICAL ROLE IN MODULATING DYNAMICS
Functionally critical amino acids within a protein, such as those directly involved with ligand bind-
ing, are very often highly evolutionarily conserved positions because mutations at these positions
often result in dramatic and deleterious changes (49, 94). However, studies have shown that, even

276 Campitelli et al.

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

ALL-Q62H LEA

More Less
deformable deformable
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

Figure 5
Mutations can induce significant changes in flexibility at distal regions, while the sites of mutations
themselves can remain relatively unchanged. (Top) Structures of green fluorescent protein common ancestor
variant ALL-Q62H (green) and its LEA (red) colored by %DFI values. Residues with significantly different
flexibility are represented as space-filling spheres using main and side-chain atoms. (Bottom) The top 15% of
DFI (difference of DFI values between the LEA and ALL-Q62H) values mapped onto the ALL-Q62H
structure (right), where red residues are more flexible and blue residues are more rigid in the LEA as
compared to the ancestral homolog; the sites that do not exhibit significant change in flexibility are colored
gray. This mapping presents the formation of new hinge sites that may regulate changes in function between
the two enzymes. The sites of sequence variation are marked as spheres (left). A majority of mutations are
colored gray, indicating that mutations do not significantly impact the flexibility of these positions. Note that
the most significant changes in flexibility occur at regions that were not directly mutated. Abbreviations:
DFI, dynamic flexibility index; LEA, least evolved ancestor.

at regions distal to binding sites or catalytic regions, point mutations may have a dramatic effect
on the function of a protein (19, 32, 35, 68, 89, 100). In the absence of structural changes or signif-
icant changes to local dynamics at the region surrounding the mutation site, these distal mutations
are a reliable indicator of the presence of allostery or allosteric regulation.
The importance of these allosteric mutations has been observed when the ancestral proteins
are incorporated in studies (52, 89). This is particularly pronounced, as explained above, in the
study of GFPs. This study showed that the mutational sites exhibit relatively small changes in
flexibility, as measured by the DFI, yet these mutations significantly impact the flexibility of posi-
tions distal from these mutational sites (Figure 5). This indicates that the changes in amino acid
sequence for green to red chromophore evolution usually did not occur at sites directly involved
with photochromatic activity (i.e., at functionally critical catalytic sites).
In fact, amino acids that possess strong dynamic allosteric residue coupling (DARC) spots to
other regions of the protein can affect protein function, regardless of the separation distance.
As these sites are often less conserved than other, more crucial positions, it appears that nature

www.annualreviews.org • Protein Dynamics and Evolution 277

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

can take advantage of functional modulation via DARC spot mutations that can impact pro-
tein dynamics through changes in allosteric networking. The dynamic coupling index (DCI) (see
Equation 4 below) is one metric that captures the complex effects of allosteric interactions reg-
ulated by DARC spots, as well as helping to describe the emergent changes in the functional
behavior of DARC spot mutations (13, 14, 25, 47, 68).
Similar to the DFI, the DCI captures the strength of the displacement response of a given
position i upon perturbation to a single functionally important position (or subset of positions) j,
relative to the average fluctuation response of position i calculated using perturbations to all other
positions within a structure:
Nfunctional
j |Rj |i /Nfunctional
DCIi = N
. 4.
j=1 |R |i /N
j
Access provided by Temple University Libraries on 06/09/20. For personal use only.

As such, the DCI can be considered a measure of the dynamic coupling between residues i and j
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

upon perturbation to residue j.

The Distal, Non-Conserved Sites Coupled to Active Sites Are Used

in Evolution to Modulate Function
The DCI has been able to identify critical allosteric interactions in a wide variety of systems (46,
47, 50, 68). For example, in a comparison study of the modern extant enzyme TEM-1, it was
revealed that the majority of the mutations that contribute to resistance are distally located from
the catalytic sites (68) yet strongly dynamically coupled to the active sites, as indicated by higher
DCI scores. Comparison of the DFI analysis of the antibiotic-resistant mutations with that of
wild-type TEM-1 revealed that these distal mutations remotely alter the flexibility of the active
site to accommodate the hydrolysis of newer antibiotics (68), as observed in antibiotic-resistant
ancestral β-lactamases (115). Additionally, the analysis of dynamic coupling of an exhaustive set
of approximately 5,000 mutations in TEM-1 has shown that the mutations that contribute most
to the emergence of a function for the degradation of a new antibiotic are those that exhibit mid-
range flexibility and high dynamic coupling with the active sites. While the medium flexibility of
these sites allows for substitutions, their higher coupling with the active site creates a cascading
set of changes in the interaction network, leading to a change in the flexibility profile of regions
that play a critical role in function.
Additionally, a DARC spot analysis of ancestral Thrxs, comparing the coupling of residues to
the catalytic site residues in an ancestor with that of an extant enzyme, showed that regions far from
functionally important catalytic sites contributed significantly to enzymatic activity and overall
stability. Specifically, one of the α helices (α3), previously shown to impact structural stability
when its formation was disrupted, exhibits much weaker dynamic coupling to these catalytic sites
in extant Thrx than in the ancestors, which suggests that dynamic coupling networks could fine-
tune function and stability during evolution.
Overall, small, subtle perturbations in such distally coupled sites cascade a set of changes to-
ward the functionally active sites. Nature utilizes this principle of minimum perturbation with
maximum response by allosterically altering the dynamics of the functionally critical sites, rather
than acquiring new, large-effect mutations.
While allosteric mutations fine-tune the dynamics of functionally critical positions toward
an evolved function, one must also wonder if the sped-up version of evolution practiced in di-
rected evolution experiments follows the same principle. Indeed, enzymes engineered through di-
rected evolution have yielded many cases in which distal mutations in regions that were previously

278 Campitelli et al.

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

Stronger
a coupling
b
1.1

expected distributions
Ratio of observed to
1.0

0.9

Weaker
coupling
0.2 0.4 0.6 0.8
DCI with catalytic sites (%)
Figure 6
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

Function-altering mutations often occur at amino acid positions dynamically coupled to critical regions, such as catalytic sites,
regardless of the structural separation distance between them. (a) The dynamic coupling index (DCI) values of the substitutions
observed in the directed evolution of wild-type bacterial phosphotriesterases (R0) to R22 are color-coded within a spectrum of blue–
white–red, where blue indicates weaker coupling, and red indicates stronger coupling. The catalytic sites are shown in black. R22 has
evolved to exhibit a 109 -fold change in the ratio of its activity to the hydrolysis of different organophosphates. We observe that a large
number of acquired mutational positions exhibit higher coupling with the catalytic sites regardless of their distance, indicating the
critical role of allosteric interactions of dynamic allosteric residue coupling (DARC) spots. (b) The observed-to-expected ratio obtained
over the DCI distribution of the remote mutational sites with the DCI distribution of all 5,200 amino acid positions across 18 proteins.
The distal positions that were acquired throughout the directed evolution experiments are overabundant at high DCI values (ratio =
1.11), whereas those with lower DCI values show a significant deficit (ratio = 0.84), suggesting that the majority of function-altering
mutations can be DARC spots. Furthermore, it also emphasizes the importance of identifying DARC spots when designing proteins
engineered toward specific functions.

thought not to affect function have actually been functionally beneficial. Some of these mutations
improve thermal stability and protein expression (7, 39, 69, 99), while others improve catalytic effi-
ciency by modulation of conformational space, thus impacting active site dynamics (15, 18, 36, 76,
97, 110). In fact, a large number of function-altering mutations that are distally positioned from
catalytic sites are indeed dynamically coupled to these sites. For example, the promiscuous, low
activity of bacterial phosphotriesterase (PTE) for arylester hydrolysis was carried out through 22
generations of directed evolution experiments toward arylesterase (AE) activity. A majority of mu-
tations leading to an approximately 40,000-fold increase in AE activity and 40,000-fold decrease
in the activity of PTE were far from the active site (15).
Upon analysis of the dynamic coupling of the substituted residue positions with the catalytic
site via DCI analysis, the mutations selected for the emergence of new function exhibited a higher
coupling with the catalytic site despite considerable separation distance (Figure 6a). Addition-
ally, in a similar analysis performed on all 18 engineered proteins containing over 100 remote and
function-altering mutations (constructed from the data set in Reference 107), it was observed that
a large number of distal mutations impacting function occur at residues that are highly coupled
to active sites, suggesting that they are DARC spots. In a robust statistical analysis comparing
the DCI distribution of the remote mutational sites with the DCI distribution of all positions
(5,200 amino acid positions across 18 proteins), the observed-to-expected ratio was calculated for
the remote mutations by categorizing the DCI into five bins (Figure 6b). Under the null hy-
pothesis of no effect, the ratio of the expected to observed numbers of residue positions hosting
function-altering remote mutations should be close to 1.0 for each category, but the null hypoth-
esis is soundly rejected (Figure 6b). Positions exhibiting the strongest dynamic coupling show the
highest enrichment of remote functions altering mutations (ratio = 1.11), whereas those with the

www.annualreviews.org • Protein Dynamics and Evolution 279

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

lower DCI values show a deficit of these variants (ratio = 0.84). This analysis also highlights the
importance of DARC spots in engineering enzymes toward a desired function.

Disease-Associated Variants in the Human Exome Also Modulate

Conformational Dynamics and Use Allostery
With advancements in genome sequencing efforts, there has been an exponential growth in the
number of known nonsynonymous single nucleotide variants (nSNVs). Indeed, new mutations
occur randomly in nature and are constantly subjected to natural selection. While many of the
mutations that significantly impact organismal fitness (owing to the disruption of protein function)
manifest themselves in the form of diseases in populations, mutations with small or insignificant
fitness effects are found as polymorphisms (48, 49). Capturing changes in protein dynamics at the
level of individual amino acids could shed light on the mechanisms underlying human sequence
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

variations associated with disease.

It is known that disease-associated variants alter the stability of a protein (3, 27, 111). Con-
versely, a recent study based on high-throughput functional assays of over 2,000 variants revealed
that only one-third of mutations led to a decrease in protein stability (90). Rather than affect-
ing stability, a significant fraction of disease-associated variants impairs protein-ligand function or
enzymatic activity (13, 46, 104). Additionally, disease-associated variants are not always located at
highly conserved positions and/or at the positions close to or at functionally critical domains. To
further complicate the problem, studies that combine evolutionary approaches with biochemistry
for protein design have also revealed that disease-causing mutations at non-conserved sites can
involve very complex and poorly understood mechanisms.
The basic evolutionary principle that biochemically similar substitutions at non-conserved sites
do not alter function does not necessarily hold. On the contrary, regardless of biochemical simi-
larity, amino acid substitutions at non-conserved sites can lead to a wide range of outcomes, with
changes of increasing or decreasing functional activity at up to three orders of magnitude (i.e.,
rheostatic pattern of change) (94). Indeed, recent evolutionary analysis has revealed that approxi-
mately one-third of the residues in human proteins are relatively fast evolving (93), which means
that these positions vary considerably among species with many different amino acid types allowed
by natural selection. Thus, many disease-associated variants implicated in Mendelian and complex
diseases are also present in one or more species. A majority of the computational tools using se-
quence alignments with or without structural information to diagnose such variants are likely to
produce a wrong diagnosis for known disease-associated variants, which is indeed the case, as the
true positive rates have been reported to be rather low (49, 64). Incorporation of conformational
dynamics is therefore fundamental to more accurately identify human variants that impact bio-
logical function (e.g., non-neutral, disease-associated) and those that do not (neutral nSNVs) (14,
23, 46, 81).
Furthermore, from a biophysics perspective, variations in a human exome, which ultimately
determine the constellation of proteins expressed by an individual, are already associated with
more than a thousand major diseases. Because the disease-associated variants in proteins remain
the part of our genome that provides the best potential for understanding how sequence relates
to function via known phenotypic impact, they represent our best chance to bridge genomics and
evolution with biophysical aspects of proteins and, particularly, conformational dynamics.
The first question is whether flexibility per position could be useful to identify the sites that
are more susceptible to damaging, disease-associated mutation. To this end, a human proteome-
wide DFI analysis of 792 disease-associated variants and 788 neutral-associated variants has shown
that mutations associated with disease generally occurred at positions with low %DFI, which are

280 Campitelli et al.

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

structurally important rigid regions or hinge sites, whereas neutral-associated variants exhibited
opposite behavior, more often occurring at sites of intermediate or high flexibility (23). These
contrasting patterns establish that structural dynamics continuously shape the protein variation
present in the human population. They also suggest that metrics that measure protein dynamics,
such as the DFI, have the potential to provide information that is independent of multispecies
sequence alignment.
As discussed above, rigid hinge sites are often mechanistically essential and evolutionarily con-
served. However, not all disease-causing mutations occur at hinge sites, or in more complex sit-
uations, at non-hinge sites located distally from other important binding or catalytic regions. A
compelling test case is the human ferritin protein; disease variants are linked to a broad range
of conditions including neurodegenerative diseases such as Parkinson’s disease and Huntington’s
disease and early developed cataract syndrome. In the study of the wild-type human ferritin pro-
tein (47), which contains neutral and disease-associated variants, it was observed that the mutated
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

sites are neither located at nor in the vicinity of the experimentally identified functional regions
that act as hinges in controlling the overall motion. While neutral variants exhibit similar flexibil-
ity (i.e., DFI) profiles to those of the wild-type protein, disease mutations soften these distal and
functionally critical regions of human ferritin by increasing their DFI values (Figure 7a). No-
tably, the disease-associated variants allosterically induce changes in flexibility at two particular
functionally critical regions: the C-terminal end and a regulatory loop denoted L1 (46). Thus,
disease mutations may loosen hinges, in this case, impairing the allosterically regulated structural
dynamics [e.g., GFPs (41)]. Indeed, DCI analysis shows that the dynamic coupling among loop
L1, the C terminus, and the rest of the structure varied dramatically between the wild-type and
disease variants. This suggests that cataract syndrome–associated mutations disrupt necessary al-
losteric regulation (Figure 7b). Overall, DCI and DFI analyses suggest that disease-associated
mutations soften the functionally critical regions, leading to a floppy protein with the loss of al-
losterically regulated conformational dynamics. This is similar to the hinge of a door; if the hinge
is loosened, then the motion will not be adequately transmitted, and the door cannot function
properly.
Proteins are not isolated within a cell, and they interact with one another to engage in essential
biological functions (46, 47, 74, 75). Considering the critical role that protein interactions play
in cellular functions, a recent study of large-scale characterization of disease variants indicates
that the majority of disease variants do not alter structure or folding stability, but rather impair
protein interactions (90, 104). Different variants in the same gene lead to different interaction
profiles, often resulting in distinct disease phenotypes. Based on experimental analysis, one could
expect that positions with the most significant impact on binding dynamics contribute the most to
binding interactions, and that mutations at these positions may impair binding and, thus, function
(i.e., disease causing). Indeed, in a study over the full human proteome, DFI profiles of over a
thousand positions harboring neutral and disease variants revealed that interface residues have a
lower average %DFI (31%) than those present at non-interfaces (50%) when complex forms are
used, rather than single monomeric units alone, indicating the critical role in protein interaction
played by the interface residues between the monomeric units. Interestingly, interface sites with
disease-associated variants have significantly lower average %DFI (23%) compared to those of
neutral nSNVs (42%), a result that directly relates structural dynamics to functional importance
(13).
While mutations occur directly at important regions such as the positions that contribute most
to binding free energies (protein interface hotspots) (22, 101), distal mutations far from binding
sites could impact function through allosteric regulation. Some mutations allosterically impair
posttranslational modification, as observed in driver mutations in cancer (13). Disease-associated

www.annualreviews.org • Protein Dynamics and Evolution 281

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

0.030
a WT
L23M L23M
0.025 T30I (neutral)
C terminus

Site flexibility by DFI

0.020

0.015 Loop L1
T30I
(disease)
0.010

0.005
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

0.000
0 20 40 60 80 100 120 140 160
Residue number

b WT
1.4 Neutral
Coupling strength to loop L1 by DCI

Disease
1.2

1.0

0.8

0.6

0.4

0.2
0 20 40 60 80 100 120 140 160
Residue number
Figure 7
Disease-associated mutations can change flexibility at distant sites, as well as the overall dynamic coupling
between residues within a protein. (a) Dynamic flexibility index (DFI) values of the light chain subunit of
human ferritin protein (the full complex is shown in the inset) for wild-type (WT) variants (blue), average
DFI of neutral variants (green), and average DFI of disease variants (red). Flexibilities for disease variants
differ from the WT at functionally critical regions (loop L1, C terminus) while maintaining relatively similar
flexibilities in neutral variants. Note that these significant flexibility changes occur at regions far from the
mutational sites themselves. (b) Dynamic coupling profiles for loop L1 of WT variants (blue), average
dynamic coupling index (DCI) of neutral variants (green), and average DCI of disease variants (red).
Disease-associated variants exhibit overall stronger coupling to this region across the entire protein
structure, which disrupts necessary allosteric regulation.

variants can also change the functionally active (ON) versus inactive (OFF) populations by altering
the stability of particular conformations and/or conformational dynamics, as observed in cancer
driver mutations of kinases (45). Furthermore, they can lead to disease by shifting allosteric path-
ways (71). In these cases, positions not exactly identified as hotspots could be DARC spots that
may act as important nodes in the interaction network by remotely modulating the dynamics of

282 Campitelli et al.

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

the binding interface; mutations at these sites would alter the interface interactions, leading to
disease (46). Thus, distal allosteric mutations that modulate conformational dynamics may impact
function and cause disease.

CONCLUSIONS AND FUTURE PERSPECTIVES

Proteins are the machines of living systems that carry out a diverse set of essential biochemical
functions. Furthermore, the diversity of their functions has grown over time via molecular evo-
lution. Analyses of protein families indicate that proteins evolve for different functions through
sequence variation while conserving their 3D structures. This becomes even more apparent when
the 3D native folds of resurrected ancestral proteins are obtained. Since ancestral proteins were
adapted to intracellular and extracellular environments that are likely different from the environ-
ments hosting modern proteins, resurrected ancestral proteins exhibit high stability, substrate and
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

catalytic promiscuity, and altered patterns of interaction with other subcellular components (46).
Detailed computational conformational analyses support the hypothesis that proteins may have
evolved to new or more specific modern functions by altering their ensemble of conformational
states in their native (functional) state. Comparison of the distribution of flexibility of residues be-
tween ancestral proteins and their extant homologs reveals that the population density of highly
flexible and rigid sites increased as they evolved. This common feature of changing the flexibil-
ity of specific positions observed in evolution suggests a fine-tuning of their native ensemble. In
addition, functionally critical positions such as catalytic pockets or critical binding hotspots are
sequentially conserved. To adapt or to create a new function, nature uses substitutions of distal
positions that are dynamically coupled to functionally critical sites (DARC spot positions) rather
than substitutions at functional sites. Mutations at DARC spots allow proteins to evolve toward
a specific network of interactions that enables communication between the active sites and the
rest of the protein through conformational dynamics. The conformational dynamics analysis of
disease-associated variants in the human exome also suggests that disease-associated mutants hi-
jack the same physical principles of modulation of protein dynamics that lead to loss or gain in
function.
While natural selection involves beneficial mutations, the evolution of proteins involves many
other critical stochastic forces (58, 72). Indeed, neutral theory has become central to the study of
evolution at the molecular level. As stated by Kimura, “. . .the overwhelming majority of evolution-
ary changes at the molecular level are not caused by selection acting on advantageous mutants, but
by random fixation of selectively neutral or very nearly neutral mutants through the cumulative ef-
fect of sampling drift (due to finite population number) under continued input of new mutations”
(42, p. 381). Although this is still under debate, a bulk of conformational dynamics studies obtained
by directed evolution experiments and resurrection of ancestral proteins also support the presence
of stochastic forces that fine-tune function incrementally and even work to maintain function in
the face of the accumulation of deleterious variations due to population-level processes (77). By
bridging the fields of biophysics and evolutionary biology, we can explore how these stochastic
forces shape the biophysical landscape of proteins and address emerging questions about complex
nonadditive (epistatic) relationships among mutations that lead to interactions and dependence
among positions and proteins in evolution.

DISCLOSURE STATEMENT
The authors are not aware of any affiliations, memberships, funding, or financial holdings that
might be perceived as affecting the objectivity of this review.

www.annualreviews.org • Protein Dynamics and Evolution 283

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

ACKNOWLEDGMENTS
This work is supported in part by grants from the National Science Foundation (NSF-MCB
1715591) and the Moore Foundation to S.B.O. and from the National Science Foundation (Grow-
ing Convergence Research Award 1934848) to S.K. We thank I. Can Kazan and Nicholas Ose for
a careful review of the manuscript.

LITERATURE CITED
1. Abrusán G, Marsh JA. 2019. Ligand-binding-site structure shapes allosteric signal transduction and the
evolution of allostery in protein complexes. Mol. Biol. Evol. 36(8):1711–27
2. Agozzino L, Dill KA. 2018. Protein evolution speed depends on its stability and abundance and on
chaperone concentrations. PNAS 115(37):9092–97
3. Alber T. 1989. Mutational effects on protein stability. Annu. Rev. Biochem. 58:765–98
4. Atilgan C, Gerek ZN, Ozkan SB, Atilgan AR. 2010. Manipulation of conformational change in proteins
Access provided by Temple University Libraries on 06/09/20. For personal use only.
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

by single-residue perturbations. Biophys. J. 99(3):933–43

5. Bahar I, Atilgan AR, Demirel MC, Erman B. 1998. Vibrational dynamics of folded proteins: significance
of slow and fast motions in relation to function and stability. Phys. Rev. Lett. 80(12):2733–36
6. Bahar I, Cheng MH, Lee JY, Kaya C, Zhang S. 2015. Structure-encoded global motions and their role
in mediating protein-substrate interactions. Biophys. J. 109(6):1101–9
7. Baier F, Hong N, Yang G, Pabis A, Miton CM, et al. 2019. Cryptic genetic variation shapes the adaptive
evolutionary potential of enzymes. eLife 8:e40789
8. Ben-David M, Huang H, Sun MGF, Corbi-Verge C, Petsalaki E, et al. 2019. Allosteric modulation of
binding specificity by alternative packing of protein cores. J. Mol. Biol. 431(2):336–50
9. Bhabha G, Ekiert DC, Jennewein M, Zmasek CM, Tuttle LM, et al. 2013. Divergent evolution of protein
conformational dynamics in dihydrofolate reductase. Nat. Struct. Mol. Biol. 20(11):1243–49
10. Bloom JD, Labthavikul ST, Otey CR, Arnold FH. 2006. Protein stability promotes evolvability. PNAS
103(15):5869–74
11. Bolia A, Gerek ZN, Keskin O, Ozkan SB, Dev KK. 2012. The binding affinities of proteins interacting
with the PDZ domain of PICK1. Proteins 80(5):1393–408
12. Bolia A, Ozkan SB. 2016. Adaptive BP-Dock: an induced fit docking approach for full receptor flexibility.
J. Chem. Inf. Model. 56(4):734–46
13. Butler BM, Gerek ZN, Kumar S, Ozkan SB. 2015. Conformational dynamics of nonsynonymous variants
at protein interfaces reveals disease association: the role of dynamics in neutral and damaging nsSNVs.
Proteins Struct. Funct. Bioinform. 83(3):428–35
14. Butler BM, Kazan IC, Kumar A, Ozkan SB. 2018. Coevolving residues inform protein dynamics profiles
and disease susceptibility of nSNVs. PLOS Comput. Biol. 14(11):e1006626
15. Campbell E, Kaltenbach M, Correy GJ, Carr PD, Porebski BT, et al. 2016. The role of protein dynamics
in the evolution of new enzyme function. Nat. Chem. Biol. 12(11):944–50
16. Cooper A, Dryden DT. 1984. Allostery without conformational change: a plausible model. Eur. Biophys.
J. 11(2):103–9
17. Copp JN, Anderson DW, Akiva E, Babbitt PC, Tokuriki N. 2019. Exploring the sequence, function,
and evolutionary space of protein superfamilies using sequence similarity networks and phylogenetic
reconstructions. Methods Enzymol. 620:315–47
18. Coyle SM, Flores J, Lim WA. 2013. Exploitation of latent allostery enables the evolution of new modes
of MAP kinase regulation. Cell 154(4):875–87
19. Curado-Carballada C, Feixas F, Osuna S. 2019. Molecular dynamics simulations on Aspergillus niger
monoamine oxidase: conformational dynamics and inter-monomer communication essential for its ef-
ficient catalysis. Adv. Synth. Catal. 361(11):2718–26
20. Finnigan GC, Hanson-Smith V, Stevens TH, Thornton JW. 2012. Evolution of increased complexity
in a molecular machine. Nature 481(7381):360–64
21. Fuglebakk E, Echave J, Reuter N. 2012. Measuring and comparing structural fluctuation patterns in
large protein datasets. Bioinformatics 28(19):2431–40

284 Campitelli et al.

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

22. Geng C, Xue LC, Roel-Touris J, Bonvin AMJJ. 2019. Finding the G spot: Are predictors of bind-
ing affinity changes upon mutations in protein-protein interactions ready for it? Wiley Interdiscip. Rev.
Comput. Mol. Sci. 9(5):e1410
23. Gerek ZN, Kumar S, Ozkan SB. 2013. Structural dynamics flexibility informs function and evolution at
a proteome scale. Evol. Appl. 6(3):423–33
24. Gerek ZN, Ozkan SB. 2010. A flexible docking scheme to explore the binding selectivity of PDZ do-
mains. Protein Sci. 19(5):914–28
25. Gerek ZN, Ozkan SB. 2011. Change in allosteric network affects binding affinities of PDZ domains:
analysis through perturbation response scanning. PLOS Comput. Biol. 7(10):e1002154
26. Gobeil SMC, Ebert MCCJC, Park J, Gagné D, Doucet N, et al. 2019. The structural dynamics of engi-
neered β-lactamases vary broadly on three timescales yet sustain native function. Sci. Rep. 9:6656
27. Guerois R, Nielsen JE, Serrano L. 2002. Predicting changes in the stability of proteins and protein
complexes: a study of more than 1000 mutations. J. Mol. Biol. 320(2):369–87
28. Guo J, Zhou H-X. 2016. Protein allostery and conformational dynamics. Chem. Rev. 116:6503–15
Access provided by Temple University Libraries on 06/09/20. For personal use only.

29. Halabi N, Rivoire O, Leibler S, Ranganathan R. 2009. Protein sectors: evolutionary units of three-
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

dimensional structure. Cell 138(4):774–86

30. Haliloglu T, Bahar I. 2015. Adaptability of protein structures to enable functional interactions and evo-
lutionary implications. Curr. Opin. Struct. Biol. 35:17–23
31. Hawkins RJ, McLeish TCB. 2004. Coarse-grained model of entropic allostery. Phys. Rev. Lett.
93(9):098104
32. Hilser VJ. 2010. Biochemistry: an ensemble view of allostery. Science 327(5966):653–54
33. Hobbs JK, Shepherd C, Saul DJ, Demetras NJ, Haaning S, et al. 2012. On the origin and evolution of
thermophily: reconstruction of functional Precambrian enzymes from ancestors of Bacillus. Mol. Biol.
Evol. 29(2):825–35
34. Ingles-Prieto A, Ibarra-Molero B, Delgado-Delgado A, Perez-Jimenez R, Fernandez JM, et al. 2013.
Conservation of protein structure over four billion years. Structure 21(9):1690–97
35. Jiménez-Osés G, Osuna S, Gao X, Sawaya MR, Gilson L, et al. 2014. The role of distant mutations and
allosteric regulation on LovD active site dynamics. Nat. Chem. Biol. 10(6):431–36
36. Kaltenbach M, Jackson CJ, Campbell EC, Hollfelder F, Tokuriki N. 2015. Reverse evolution leads to
genotypic incompatibility despite functional and active site convergence. eLife 4:e06492
37. Keskin O, Bahar I, Jernigan RL, Beutler JA, Shoemaker RH, et al. 2000. Characterization of anti-
cancer agents by their growth inhibitory activity and relationships to mechanism of action and structure.
Anticancer Drug Des. 15(2):79–98
38. Khersonsky O, Roodveldt C, Tawfik DS. 2006. Enzyme promiscuity: evolutionary and mechanistic as-
pects. Curr. Opin. Chem. Biol. 10(5):498–508
39. Khersonsky O, Röthlisberger D, Dym O, Albeck S, Jackson CJ, et al. 2010. Evolutionary optimization
of computationally designed enzymes: Kemp eliminases of the KE07 series. J. Mol. Biol. 396(4):1025–42
40. Khersonsky O, Tawfik DS. 2010. Enzyme promiscuity: a mechanistic and evolutionary perspective.
Annu. Rev. Biochem. 79:471–505
41. Kim H, Zou T, Modi C, Dörner K, Grunkemeyer TJ, et al. 2015. A hinge migration mechanism unlocks
the evolution of green-to-red photoconversion in GFP-like proteins. Structure 23(1):34–43
42. Kimura M. 1991. The neutral theory of molecular evolution: a review of recent evidence. Idengaku Zasshi
66(4):367–86
43. Knies JL, Cai F, Weinreich DM. 2017. Enzyme efficiency but not thermostability drives cefotaxime
resistance evolution in TEM-1 β-lactamase. Mol. Biol. Evol. 34(5):1040–54
44. Knudsen M, Wiuf C. 2010. The CATH database. Hum. Genom. 4(3):207–12
45. Köhler S, Bauer S, Horn D, Robinson PN. 2008. Walking the interactome for prioritization of candidate
disease genes. Am. J. Hum. Genet. 82(4):949–58
46. Kumar A, Butler BM, Kumar S, Ozkan SB. 2015. Integration of structural dynamics and molecular evolu-
tion via protein interaction networks: a new era in genomic medicine. Curr. Opin. Struct. Biol. 35:135–42
47. Kumar A, Glembo TJ, Ozkan SB. 2015. The role of conformational dynamics and allostery in the disease
development of human ferritin. Biophys. J. 109(6):1273–81

www.annualreviews.org • Protein Dynamics and Evolution 285

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

48. Kumar S, Dudley JT, Filipski A, Liu L. 2011. Phylomedicine: an evolutionary telescope to explore and
diagnose the universe of disease mutations. Trends Genet. 27(9):377–86
49. Kumar S, Suleski MP, Markov GJ, Lawrence S, Marco A, Filipski AJ. 2009. Positional conservation and
amino acids shape the correct diagnosis and population frequencies of benign and damaging personal
amino acid mutations. Genome Res. 19(9):1562–69
50. Larrimore KE, Kazan IC, Kannan L, Kendle RP, Jamal T, et al. 2017. Plant-expressed cocaine hydrolase
variants of butyrylcholinesterase exhibit altered allosteric effects of cholinesterase activity and increased
inhibitor sensitivity. Sci. Rep. 7(1):10419
51. Li Z, Bolia A, Maxwell JD, Bobkov AA, Ghirlanda G, et al. 2015. A rigid hinge region is necessary for
high-affinity binding of dimannose to cyanovirin and associated constructs. Biochemistry 54(46):6951–60
52. Liang Z, Verkhivker GM, Hu G. 2019. Integration of network models and evolutionary analysis into
high-throughput modeling of protein dynamics and allosteric regulation: theory, tools and applications.
Brief. Bioinform. In press
53. Liberles DA, Teichmann SA, Bahar I, Bastolla U, Bloom J, et al. 2012. The interface of protein structure,
Access provided by Temple University Libraries on 06/09/20. For personal use only.

protein biophysics, and molecular evolution. Protein Sci. 21(6):769–85

Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

54. Lieberman RL, Wustman BA, Huertas P, Powe AC, Pine CW, et al. 2007. Structure of acid β-glucosidase
with pharmacological chaperone provides insight into Gaucher disease. Nat. Chem. Biol. 3(2):101–7
55. Liu X, Golden LC, Lopez JA, Shepherd TR, Yu L, Fuentes EJ. 2019. Conformational dynamics and
cooperativity drive the specificity of a protein-ligand interaction. Biophys. J. 116(12):2314–30
56. Liu Y, Bahar I. 2012. Sequence evolution correlates with structural dynamics. Mol. Biol. Evol. 29(9):2253–
63
57. Liu Y, Gierasch LM, Bahar I. 2010. Role of Hsp70 ATPase domain intrinsic dynamics and sequence
evolution in enabling its functional interactions with NEFs. PLOS Comput. Biol. 6(9):e1000931
58. Lynch M, Ackerman MS, Gout J-F, Long H, Sung W, et al. 2016. Genetic drift, selection and the
evolution of the mutation rate. Nat. Rev. Genet. 17(11):704–14
59. Maguid S, Fernandez-Alberti S, Echave J. 2008. Evolutionary conservation of protein vibrational dy-
namics. Gene 422(1–2):7–13
60. Maguid S, Fernández-Alberti S, Parisi G, Echave J. 2006. Evolutionary conservation of protein backbone
flexibility. J. Mol. Evol. 63(4):448–57
61. McLeish TCB, Rodgers TL, Wilson MR. 2013. Allostery without conformation change: modelling pro-
tein dynamics at multiple scales. Phys. Biol. 10(5):056004
62. McLeish TCB, Schaefer C, von der Heydt AC. 2018. The ‘allosteron’ model for entropic allostery of
self-assembly. Philos. Trans. R. Soc. B 373(1749):20170186
63. Mikulska-Ruminska K, Shrivastava I, Krieger J, Zhang S, Li H, et al. 2019. Characterization of differen-
tial dynamics, specificity, and allostery of lipoxygenase family members. J. Chem. Inf. Model. 59(5):2496–
508
64. Miller M, Bromberg Y, Swint-Kruse L. 2017. Computational predictors fail to identify amino acid sub-
stitution effects at rheostat positions. Sci. Rep. 7:41329
65. Mishra SK, Jernigan RL. 2018. Protein dynamic communities from elastic network models align closely
to the communities defined by molecular dynamics. PLOS ONE 13(6):e0199225
66. Mishra SK, Kandoi G, Jernigan RL. 2019. Coupling dynamics and evolutionary information with
structure to identify protein regulatory and functional binding sites. Proteins Struct. Funct. Bioinform.
87(10):850–68
67. Modi T, Huihui J, Ghosh K, Ozkan SB. 2018. Ancient thioredoxins evolved to modern-day stability-
function requirement by altering native state ensemble. Philos. Trans. R. Soc. B 373(1749):20170184
68. Modi T, Ozkan SB. 2018. Mutations utilize dynamic allostery to confer resistance in TEM-1 β-
lactamase. Int. J. Mol. Sci. 19(12):3808
69. Morley KL, Kazlauskas RJ. 2005. Improving enzyme properties: When are closer mutations better?
Trends Biotechnol. 23(5):231–37
70. Moult J. 2008. Comparative modeling in structural genomics. Structure 16(1):14–16
71. Navlakha S, Kingsford C. 2010. The power of protein interaction networks for associating genes with
diseases. Bioinformatics 26(8):1057–63

286 Campitelli et al.

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

72. Nei M, Kumar S. 2000. Molecular Evolution and Phylogenetics. Oxford, UK: Oxford Univ. Press
73. Neuwald AF. 2007. The CHAIN program: forging evolutionary links to underlying mechanisms. Trends
Biochem. Sci. 32(11):487–93
74. Nussinov R, Tsai C-J. 2013. Allostery in disease and in drug discovery. Cell 153(2):293–305
75. Nussinov R, Tsai C-J, Liu J. 2014. Principles of allosteric interactions in cell signaling. J. Am. Chem. Soc.
136(51):17692–701
76. Otten R, Liu L, Kenner LR, Clarkson MW, Mavor D, et al. 2018. Rescue of conformational dynamics
in enzyme catalysis by directed evolution. Nat. Commun. 9:1314
77. Patel R, Scheinfeldt LB, Sanderford MD, Lanham TR, Tamura K, et al. 2018. Adaptive landscape of
protein variation in human exomes. Mol. Biol. Evol. 35(8):2015–25
78. Pauling L, Zuckerkandl E, Henriksen T, Lövstad R. 1963. Chemical paleogenetics: molecular “restora-
tion studies” of extinct forms of life. Acta Chem. Scand. 17(Suppl.):9–16
79. Penkler D, Sensoy Ö, Atilgan C, Tastan Bishop Ö. 2017. Perturbation-response scanning reveals key
residues for allosteric control in Hsp70. J. Chem. Inf. Model. 57(6):1359–74
Access provided by Temple University Libraries on 06/09/20. For personal use only.

80. Perez-Jimenez R, Inglés-Prieto A, Zhao Z-M, Sanchez-Romero I, Alegre-Cebollada J, et al. 2011.

Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

Single-molecule paleoenzymology probes the chemistry of resurrected enzymes. Nat. Struct. Mol. Biol.
18(5):592–96
81. Ponzoni L, Bahar I. 2018. Structural dynamics is a determinant of the functional significance of missense
variants. PNAS 115(16):4164–69
82. Ponzoni L, Zhang S, Cheng MH, Bahar I. 2018. Shared dynamics of LeuT superfamily members
and allosteric differentiation by structural irregularities and multimerization. Philos. Trans. R. Soc. B
373(1749):20170177
83. Risso VA, Gavira JA, Mejia-Carmona DF, Gaucher EA, Sanchez-Ruiz JM. 2013. Hyperstability and
substrate promiscuity in laboratory resurrections of Precambrian β-lactamases. J. Am. Chem. Soc.
135(8):2899–902
84. Risso VA, Manssour-Triedo F, Delgado-Delgado A, Arco R, Barroso-delJesus A, et al. 2015. Mutational
studies on resurrected ancestral proteins reveal conservation of site-specific amino acid preferences
throughout evolutionary history. Mol. Biol. Evol. 32(2):440–55
85. Risso VA, Sanchez-Ruiz JM. 2017. Resurrected ancestral proteins as scaffolds for protein engineering.
In Directed Enzyme Evolution: Advances and Applications, ed. M Alcalde, pp. 229–55. Berlin: Springer
86. Risso VA, Sanchez-Ruiz JM, Ozkan SB. 2018. Biotechnological and protein-engineering implications of
ancestral protein resurrection. Curr. Opin. Struct. Biol. 51:106–15
87. Rocks JW, Ronellenfitsch H, Liu AJ, Nagel SR, Katifori E. 2019. Limits of multifunctionality in tunable
networks. PNAS 116(7):2506–11
88. Romero-Romero ML, Risso VA, Martinez-Rodriguez S, Ibarra-Molero B, Sanchez-Ruiz JM. 2016. En-
gineering ancestral protein hyperstability. Biochem. J. 473(20):3611–20
89. Saavedra HG, Wrabl JO, Anderson JA, Li J, Hilser VJ. 2018. Dynamic allostery can drive cold adaptation
in enzymes. Nature 558(7709):324–28
90. Sahni N, Yi S, Taipale M, Fuxman Bass JI, Coulombe-Huntington J, et al. 2015. Widespread macro-
molecular interaction perturbations in human genetic disorders. Cell 161(3):647–60
91. Sethi A, O’Donoghue P, Luthey-Schulten Z. 2005. Evolutionary profiles from the QR factorization of
multiple sequence alignments. PNAS 102(11):4045–50
92. Socolich M, Lockless SW, Russ WP, Lee H, Gardner KH, Ranganathan R. 2005. Evolutionary infor-
mation for specifying a protein fold. Nature 437(7058):512–18
93. Subramanian S, Kumar S. 2006. Evolutionary anatomies of positions and types of disease-associated and
neutral amino acid mutations in the human genome. BMC Genom. 7:306
94. Swint-Kruse L. 2016. Using evolution to guide protein engineering: The devil IS in the details. Biophys.
J. 111(1):10–18
95. Tan L, Serene S, Chao HX, Gore J. 2011. Hidden randomness between fitness landscapes limits reverse
evolution. Phys. Rev. Lett. 106(19):198102
96. Tawfik DS, Tokuriki N. 2009. Protein dynamism and evolvability. Science 324(5924):203–7
97. Taylor JL, Price JE, Toney MD. 2015. Directed evolution of the substrate specificity of dialkylglycine
decarboxylase. Biochim. Biophys. Acta Proteins Proteom. 1854(2):146–55

www.annualreviews.org • Protein Dynamics and Evolution 287

BB49CH13_Ozkan ARjats.cls April 28, 2020 19:22

98. Thirumalai D, Hyeon C, Zhuravlev PI, Lorimer GH. 2019. Symmetry, rigidity, and allosteric signaling:
from monomeric proteins to molecular machines. Chem. Rev. 119(12):6788–821
99. Tokuriki N, Tawfik DS. 2009. Stability effects of mutations and protein evolvability. Curr. Opin. Struct.
Biol. 19(5):596–604
100. Townsend PD, Rodgers TL, Glover LC, Korhonen HJ, Richards SA, et al. 2015. The role of protein-
ligand contacts in allosteric regulation of the Escherichia coli catabolite activator protein. J. Biol. Chem.
290(36):22225–35
101. Tuncbag N, Gursoy A, Keskin O. 2009. Identification of computational hot spots in protein inter-
faces: Combining solvent accessibility and inter-residue potentials improves the accuracy. Bioinformatics
25(12):1513–20
102. Turner JM, Graziano J, Spraggon G, Schultz PG. 2006. Structural plasticity of an aminoacyl-tRNA
synthetase active site. PNAS 103(17):6483–88
103. Wang S-W, Bitbol A-F, Wingreen NS. 2019. Revealing evolutionary constraints on proteins through
sequence analysis. PLOS Comput. Biol. 15(4):e1007010
Access provided by Temple University Libraries on 06/09/20. For personal use only.

104. Wang X, Wei X, Thijssen B, Das J, Lipkin SM, Yu H. 2012. Three-dimensional reconstruction of protein
Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

networks provides insight into human genetic disease. Nat. Biotechnol. 30(2):159–64
105. Wei G, Xi W, Nussinov R, Ma B. 2016. Protein ensembles: How does nature harness thermodynamic
fluctuations for life? The diverse functional roles of conformational ensembles in the cell. Chem. Rev.
116(11):6516–51
106. Weinreich DM, Delaney NF, Depristo MA, Hartl DL. 2006. Darwinian evolution can follow only very
few mutational paths to fitter proteins. Science 312(5770):111–14
107. Wilding M, Hong N, Spence M, Buckle AM, Jackson CJ. 2019. Protein engineering: the potential of
remote mutations. Biochem. Soc. Trans. 47(2):701–11
108. Wodak SJ, Paci E, Dokholyan NV, Berezovsky IN, Horovitz A, et al. 2019. Allostery in its many disguises:
from theory to applications. Structure 27(4):566–78
109. Woldeyes RA, Sivak DA, Fraser JS. 2014. E pluribus unum, no more: from one crystal, many conforma-
tions. Curr. Opin. Struct. Biol. 28(1):56–62
110. Yang G, Hong N, Baier F, Jackson CJ, Tokuriki N. 2016. Conformational tinkering drives evolution of
a promiscuous activity through indirect mutational effects. Biochemistry 55(32):4583–93
111. Yue P, Li Z, Moult J. 2005. Loss of protein structure stability as a major causative factor in monogenic
disease. J. Mol. Biol. 353(2):459–73
112. Zhang S, Li H, Krieger JM, Bahar I. 2019. Shared signature dynamics tempered by local fluctuations
enables fold adaptability and specificity. Mol. Biol. Evol. 36(9):2053–68
113. Zheng W, Brooks BR, Doniach S, Thirumalai D. 2005. Network of dynamically important residues in
the open/closed transition in polymerases is strongly conserved. Structure 13(4):565–77
114. Zheng W, Brooks BR, Thirumalai D. 2006. Low-frequency normal modes that describe allosteric tran-
sitions in biological nanomachines are robust to sequence variations. PNAS 103(20):7664–69
115. Zou T, Risso VA, Gavira JA, Sanchez-Ruiz JM, Ozkan SB. 2015. Evolution of conformational dynamics
determines the conversion of a promiscuous generalist into a specialist enzyme. Mol. Biol. Evol. 32(1):132–
43

288 Campitelli et al.

 BB49_FrontMatter ARI 9 April 2020 10:52

Annual Review of
Biophysics
Volume 49, 2020

Contents
Access provided by Temple University Libraries on 06/09/20. For personal use only.

The Physics of Cellular Decision Making During

Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

Epithelial–Mesenchymal Transition
Shubham Tripathi, Herbert Levine, and Mohit Kumar Jolly p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 1
Principles and Applications of Biological Membrane Organization
Wade F. Zeno, Kasey J. Day, Vernita D. Gordon, and Jeanne C. Stachowiak p p p p p p p p p p p p19
Mitochondria-Associated Proteostasis
Linhao Ruan, Yuhao Wang, Xi Zhang, Alexis Tomaszewski,
Joshua T. McNamara, and Rong Li p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p41
Milestoning: An Efficient Approach for Atomically Detailed
Simulations of Kinetics in Biophysics
Ron Elber p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p69
Enhanced Diffusion and Chemotaxis of Enzymes
Mudong Feng and Michael K. Gilson p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p87
Physical Principles Underlying the Complex Biology of Intracellular
Phase Transitions
Jeong-Mo Choi, Alex S. Holehouse, and Rohit V. Pappu p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 107
Multifunctional Chaperone and Quality Control Complexes in
Adaptive Immunity
Simon Trowitzsch and Robert Tampé p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 135
Temperature, Dynamics, and Enzyme-Catalyzed Reaction Rates
Vickery L. Arcus and Adrian J. Mulholland p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 163
Predicting Evolution Using Regulatory Architecture
Philippe Nghe, Marjon G.J. de Vos, Enzo Kingma, Manjunatha Kogenaru,
Frank J. Poelwijk, Liedewij Laan, and Sander J. Tans p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 181
Gene Regulation in and out of Equilibrium
Felix Wong and Jeremy Gunawardena p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 199

v
 BB49_FrontMatter ARI 9 April 2020 10:52

Nonequilibrium Thermodynamics in Cell Biology: Extending

Equilibrium Formalism to Cover Living Systems
Xiaona Fang and Jin Wang p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 227
RNA Droplets
Kevin Rhine, Velinda Vidaurre, and Sua Myong p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 247
The Role of Conformational Dynamics and Allostery in Modulating
Protein Evolution
Paul Campitelli, Tushar Modi, Sudhir Kumar, and S. Banu Ozkan p p p p p p p p p p p p p p p p p p p p 267
Light Microscopy of Mitochondria at the Nanoscale
Stefan Jakobs, Till Stephan, Peter Ilgen, and Christian Brüser p p p p p p p p p p p p p p p p p p p p p p p p p p 289
Access provided by Temple University Libraries on 06/09/20. For personal use only.

Insights into the Structure, Function, and Dynamics of the Bacterial

Annu. Rev. Biophys. 2020.49:267-288. Downloaded from www.annualreviews.org

Cytokinetic FtsZ-Ring
Ryan McQuillen and Jie Xiao p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 309

Indexes

Cumulative Index of Contributing Authors, Volumes 45–49 p p p p p p p p p p p p p p p p p p p p p p p p p p p 343

Errata

An online log of corrections to Annual Review of Biophysics articles may be found at

http://www.annualreviews.org/errata/biophys

vi Contents