Subscriber access provided by READING UNIV
Article
Unraveling Allosteric Mechanisms of Enzymatic Catalysis with an Evolutionary Analysis of Residue-residue Contact Dynamical Changes Phuoc J. Vu, Xin-Qiu Yao, Mohamed Faizan Momin, and Donald Hamelberg ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.7b04263 • Publication Date (Web): 01 Feb 2018 Downloaded from http://pubs.acs.org on February 2, 2018
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
ACS Catalysis is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 36 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
ACS Catalysis
1 2
Unraveling Allosteric Mechanisms of Enzymatic Catalysis with an Evolutionary Analysis of Residue-residue Contact Dynamical Changes
3 4
Phuoc Jake Vu1, Xin-Qiu Yao1, Mohamed Momin, Donald Hamelberg*
5
Department of Chemistry, Georgia State University, Atlanta, Georgia 30303-2515, USA.
6 7 8 9 10
1
Equal contributions
*Corresponding to: Prof. Donald Hamelberg; Department of Chemistry, Georgia State University, 29 Peachtree Center Ave NE, Atlanta, Georgia 30303-2515, USA. Telephone: (404) 413-5564; E-mail:
[email protected].
11 12
1 ACS Paragon Plus Environment
ACS Catalysis 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 2 of 36
13
Abstract
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
The evolution of protein conformational dynamics contains important information about protein function and regulation. Here, we describe an approach to dynamical-evolution analysis based on multiple microsecond molecular dynamics simulations and residueresidue contact analysis. We illustrate our approach by comparing three human cyclophilin isoforms, cyclophilin A, D, and E, which belong to a family of enzymes catalyzing peptidyl-prolyl cis-trans isomerization. Our results reveal that despite distinct overall equilibrium conformations between cyclophilins under substrate-free conditions, functional dynamical changes resembling substrate-binding and catalytic processes tend to be conserved. Key residues displaying either concerted or specific dynamical changes among isoforms during the reactions are identified, which delineate two distinct allosteric pathways for cyclophilin function consistent with recent nuclear magnetic resonance experiments. A sequence-based coevolution analysis is also employed for further understanding dynamical consequences. Our results collectively provide a framework where both common and specific functional mechanisms of a protein family can be elucidated.
29 30 31
Keywords: evolution, allosteric regulation, enzyme dynamics, residue-residue contact, cyclophilin, molecular dynamics
2 ACS Paragon Plus Environment
Page 3 of 36 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
ACS Catalysis
32
Introduction
33
Protein internal motions or dynamics have been increasingly recognized to play a crucial
34
role in protein function and regulation.1-3 Typical protein dynamics span a broad range of
35
spatiotemporal scales, from subtle backbone and sidechain fluctuations (~10-9s) and domain
36
motions (~10-6s) to large-amplitude conformational changes normally observed in
37
biological molecular machines (10-3-102s). Notable examples in which dynamics determine
38
function include the dynamical rearrangements in enzymes to facilitate catalytic turnover,4
39
the conformational changes in transporters to pump small molecules in and out of cell
40
membrane,5 the force-producing structural changes in molecular motors,6 and the prevailing
41
roles of dynamics in the allosteric regulation during signal transduction.7-9 Despite
42
numerous efforts in inspecting protein dynamics with both experimental and computational
43
methods, such as nuclear magnetic resonance (NMR)10 and molecular dynamics (MD)
44
simulation,11 how protein molecules harness thermal fluctuations for function remains
45
elusive.
46
47
Protein dynamics reflect the underlying energy landscape, which is predominantly
48
determined by protein sequence. In this perspective, evolution modified protein dynamics
49
and function by altering the energy landscape. During evolution, certain patterns of protein
50
dynamics must be conserved to retain the core protein function that arose early. Similar to
51
how multiple sequence alignment screens out variable residues to detect structurally or
52
functionally critical sites, comparing protein dynamics at aligned residues across a protein
3 ACS Paragon Plus Environment
ACS Catalysis 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 4 of 36
53
family identifies dynamically conserved sites that are intimately related to the core function.
54
Meanwhile, variable dynamics in synergy with variable sequences across family members
55
underlie the subtle functional diversity, and hence, comparative analysis of dynamics also
56
helps to foster our understanding of protein functional specificity (e.g., selective substrate
57
binding or distinct kinetics). Indeed, this new paradigm of evolution-based approach to
58
delineating the complex sequence-structure-dynamics-function relationship has gained
59
increasing interests. By comparing the crystallographic structures of several homologous
60
proteins under distinct functional states, Babu and colleagues recently derived the universal
61
mechanisms governing the activation of heterotrimeric guanine nucleotide-binding proteins
62
(G proteins) and G protein-coupled receptors.12-13 With X-ray crystallography and NMR
63
spectroscopy, Wright and colleagues compared the dynamics of human and Escherichia
64
coli dihydrofolate reductase (DHFR), an enzyme that catalyzes the NADPH-dependent
65
reduction of dihydrofolate to tetrahydrofolate, and identified the residue determinants
66
underlying the distinct catalytic efficiency and robustness between the human and the
67
bacterial enzymes.14 In addition to the comparisons between extant proteins, the structures
68
and dynamics of ancestral proteins have been modeled with a computational ancestor
69
reconstruction method, which empowers a direct tracking of the evolution of protein
70
dynamics.15-17 These recent advances demonstrate how a study of protein dynamics in the
71
context of evolution provides unprecedented insights into functional mechanisms. The
72
knowledge obtained complements that derived from a large-scale evolutionary analysis
73
aimed at inferring sequence and structural traits underlying conserved and divergent
74
functions of enzyme superfamilies18-19 and can be further leveraged to develop new
4 ACS Paragon Plus Environment
Page 5 of 36 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
ACS Catalysis
75
algorithms for rational protein design and drug discovery. However, a general approach to
76
the evolutionary analysis of conformational dynamics for most biological systems is still
77
lacking.
78
79
Cyclophilin A (CypA) is a ubiquitous protein where function is strongly coupled with
80
dynamics.4, 20 As a peptidyl-prolyl cis-trans isomerase (PPIase), cyclophilin A catalyzes the
81
interconversion between the cis (with the peptidyl-prolyl torsion angle ω=0º) and the trans
82
(ω=±180º) states of the prolyl peptide bond in Xaa-Pro motifs, where Xaa represents any
83
amino acid (Figure 1A & B). Humans have 17 cyclophilin isoforms (Figure S1), among
84
which CypA is the best characterized, both experimentally and computationally.
85
Cyclophilins are known targets for immunosuppressant, and their function is critical to
86
many important cellular processes, including protein folding and signal transduction.21
87
Cyclophilin D (CypD) and cyclophilin E (CypE) are the family members closest to CypA,
88
with high sequence identity (68-75%) and almost identical structures to CypA (the
89
backbone root mean square deviation, or RMSD, is within 0.50-0.65 Å; see Figure 1A).
90
The three isoforms of cyclophilin have the same PPIase activity; however, they function in
91
different subcellular locations: CypA is generally found in the cytosol, CypD in the
92
mitochondria matrix, and CypE in the nucleus.22 We recently examined the dynamical
93
properties, represented by the breaking and formation of residue-residue contacts, of CypA
94
under various substrate-binding and mutational conditions using a combined approach of
95
MD simulations and NMR experiments.23 In particular, we found an interesting ‘dynamic
5 ACS Paragon Plus Environment
ACS Catalysis 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 6 of 36
96
cluster’ containing residues showing substantial dynamical changes upon substrate binding
97
located ~15 Å away from the active site. This finding reveals a new allosteric mechanism in
98
CypA, but its generality across the cyclophilin family is not known.
99
100
In this work, we develop a new approach for general evolutionary analysis of protein
101
conformational dynamics based on multiple microsecond-long MD simulations and the
102
similar contact analysis method previously described.23 We illustrate our approach by
103
comparing the dynamics between human CypA, CypD, and CypE derived from MD
104
simulations (totaling 26 µs). For each cyclophilin isoform, three functional states
105
collectively representing substrate-binding and catalytic isomerization processes are
106
examined. This allows us to evaluate the conservation of dynamics across isoforms with
107
respect to distinct enzymatic processes. Dissecting the contact dynamics further enables us
108
to identify the key residues determining the common and isoform-specific dynamical
109
changes. We also perform a sequence coevolution analysis to identify correlated amino acid
110
substitutions between residues in the cyclophilin family and find that dynamically
111
conserved contacts do not always represent the most highly coevolving residue pairs,
112
suggesting that dynamics is the missing link and sequence and structure alone cannot fully
113
describe function. The consensus groups of residues derived from the dynamical
114
conservation analysis are consistent with recent NMR experiments.24
115
116
6 ACS Paragon Plus Environment
Page 7 of 36 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
ACS Catalysis
117 118 119 120 121 122 123 124 125 126
Figure 1. Backbone and active site structures are highly similar across CypA, CypD, and CypE. (A) The substrate-free crystallographic structures of CypA (white; PDB: 3K0M), CypD (red; PDB: 4O8H), and CypE (green; PDB: 3UCH) are superimposed and are represented as cartoon. The modeled substrate (See Methods) is displayed as cartoon (yellow) with ‘Gly-Pro’ motif shown as licorice and colored by atom types. The enlarged view shows the substrate-binding pocket represented as white transparent surface. Side chains of active site residues are displayed as sticks and are color-coded the same as backbone. (B) Schematic cis-trans isomerization of the ‘Gly-Pro’ motif. Ts, transition state.
127
Results and Discussion
128
Protein backbone and residue side chain dynamics under the substrate-free state are
129
distinct among CypA, CypD, and CypE.
130
Fluctuation analysis and principal component analysis (PCA) of MD simulation trajectories
131
reveal that backbone dynamics are different between CypA, CypD, and CypE. Multiple
132
long-time (2.2-2.7 µs) MD simulations were performed under substrate-free conditions for
7 ACS Paragon Plus Environment
ACS Catalysis 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 8 of 36
133
each cyclophilin isoform. Snapshots of the latter 2-µs of each simulation trajectory were
134
analyzed. Residue-wise averaged root mean square fluctuation (RMSF) of backbone atoms
135
derived from the simulations shows that, whilst the atomic fluctuations at most residues
136
look similar across isoforms, the fluctuations located at the β4-β5 loop (the loop between
137
the fourth and the fifth β-strands), the β5-β6 loop, the β6-β7 loop, and the α2-β8 loop (the
138
loop between the second α-helix and the eighth β-strand) are apparently different (Figure
139
2A). In general, CypA and CypE are more flexible than CypD. PCA was then performed on
140
the Cartesian coordinates of backbone atoms from the simulations to examine backbone
141
conformational distributions at equilibrium (Figure 2B). It reveals that CypE has the
142
broadest distribution and samples multiple conformational states in the subspace spanned
143
by the top two principal components (i.e., PC 1 and PC 2, which collectively capture nearly
144
50% of total atomic mean displacements or variance), indicating the overall highest
145
flexibility of CypE among the isoforms. CypA has a distribution with modest broadness but
146
it still samples at least two conformational states, and it samples the space largely
147
overlapped with that populated by CypE. In contrast, CypD has the narrowest distribution,
148
which is clearly separated from those of CypA and CypE, and samples only one
149
conformational state, indicating that the backbone of CypD is rigid and has distinct
150
equilibrium properties from CypA and CypE.
151
152
Inspection of residue-residue contacts reveals distinct side chain dynamics among CypA,
153
CypD, and CypE. A pair of residues is in contact whenever their minimal non-hydrogen
8 ACS Paragon Plus Environment
Page 9 of 36 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
ACS Catalysis
154
atomic distance is at most 4.5 Å and they are separated by at least three residues in
155
sequence (i to i+n, n≥3). The 4.5 Å-threshold employed here represents a well defined
156
interaction range for most amino acids and has been applied to contact analysis in previous
157
simulation studies.23, 25-26 Other parameters were also used to test the robustness of our
158
results (See discussion all over Results and Discussion). For each residue pair, the
159
probability of contact formation during the simulation was calculated and compared
160
between cyclophilin isoforms. It shows that the residue pairs with substantial difference in
161
contact formation probability (i.e., |df=fCypY-fCypX|≥0.1, where fCypX and fCypY are contact
162
probabilities calculated for the two compared isoforms and the threshold 0.1 represents an
163
error estimate of f23) distribute all over the cyclophilin molecule for all the comparisons
164
(Figure 2C-E), indicating that side chain dynamics are distinct among the isoforms. In
165
summary, although in both sequence and structure cyclophilin isoforms are very similar,
166
they possess different conformational ensembles at equilibrium at least under the substrate-
167
free state.
168
9 ACS Paragon Plus Environment
ACS Catalysis 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 10 of 36
169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187
Figure 2. Backbone and sidechain dynamics under the substrate-free state are different between CypA, CypD, and CypE. (A) Residue-wise averaged root mean square fluctuation (RMSF) of backbone atoms derived from the MD simulations under substratefree conditions. Residue numbers are based on CypA. Secondary structure elements are annotated as black (α helices) and grey (β strands) rectangles at the top and bottom of the plot. (B) PCA performed on the Cartesian coordinates of backbone atoms from the simulations for CypA (grey), CypD (red), and CypE (green). Simulation-generated conformational snapshots are projected in the subspace spanned by the two principal components capturing the largest structural variance (PC1 and PC2; the number in the axis label indicates the percentage of variance captured by the corresponding PC). Probability density distributions of the conformational samples are represented as contour lines. The sampled space of the CypA and CypD simulations are also outlined. (C-E) Contact probability difference under the substrate-free state between isoforms (df=fCypY-fCypX, where f is the probability of contact formation and CypX and CypY are the corresponding cyclophilins under the comparison CypX/CypY) mapped to the crystal structure of CypA (PDB: 1M9F). Blue and red cylinders represent contacts with df≥0.1 and df≤-0.1, respectively, where cylinder radius is proportional to |df|. The yellow star indicates the active site.
10 ACS Paragon Plus Environment
Page 11 of 36 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
ACS Catalysis
188
189
Dynamical changes of residue-residue contacts during substrate binding are conserved
190
between CypA and CypE but are not conserved between CypA/E and CypD.
191
Although the conformational dynamics under an individual state (i.e., the substrate-free
192
state) are very different between the cyclophilin isoforms, dynamical changes from the
193
substrate-free to the substrate-bound state display a certain extent of similarity. Multiple
194
2.4-µs additional simulations were performed under substrate-bound conditions where the
195
peptidyl-prolyl torsion angle (ω) was in the cis-conformation (termed ‘cis-bound state’).
196
The probability difference of contact formation during simulations between the substrate-
197
free and the cis-bound states was used to characterize the dynamical changes during
198
substrate binding. Analysis of the CypA simulations reveals a site ~15 Å away from the
199
active site showing substantial dynamical changes (|df=fY-fX|≥0.1, where fX and fY are
200
contact probabilities calculated for the two compared states) of residue contacts (Figure
201
3A). This observation resembles the dynamic cluster identified in the previous simulation
202
study of CypA,23 although the substrate employed in the present work is five-residue longer
203
than that used in the previous study.23 This consistency indicates that the revealed
204
dynamical changes are an intrinsic dynamical characteristic of CypA independent from the
205
identity of the bound substrate. Intriguingly, overall similar patterns of dynamical changes
206
are observed between CypA and CypE (Figure 3A & C). A large portion of contacts that
207
are either more often formed (with an increase of contact probability) or more often broken
208
(decrease of contact probability) upon substrate binding in CypA are shown to have the
11 ACS Paragon Plus Environment
ACS Catalysis 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 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 12 of 36
209
same direction of changes in CypE (Figure 3E). These contacts with the same trends of
210
changes from one state to the other between isoforms are defined as ‘dynamically
211
conserved contacts.’
212
213
To quantitatively measure the overall similarity of dynamical changes between isoforms,
214
we developed a dynamical conservation index (DCI), which is defined by the percentage of
215
dynamically conserved contacts. In the calculation of DCI, contacts showing small
216
dynamical changes (absolute contact probability difference |df|