Compound Molecular Logic in Accessing the ... - ACS Publications

Subscriber access provided by Kaohsiung Medical University

Article

Compound Molecular Logic in Accessing the Active Site of M. Tuberculosis Protein Tyrosine Phosphatase B Thomas E. Morrell, Ilona Urszula Rafalska-Metcalf, Haw Yang, and Jhih-Wei Chu J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b08070 • Publication Date (Web): 10 Oct 2018 Downloaded from http://pubs.acs.org on October 10, 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 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 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.

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 21 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

Journal of the American Chemical Society

Compound Molecular Logic in Accessing the Active Site of M. Tuberculosis Protein Tyrosine Phosphatase B Thomas E. Morrell,† Ilona U. Rafalska-Metcalf,† Haw Yang,∗,† and Jhih-Wei Chu∗,‡ †Department of Chemistry, Princeton University, Princeton, NJ 08544, USA ‡Institute of Bioinformatics and Systems Biology, Department of Biological Science and Technology, and Institute of Molecular Medicine and Bioengineering, National Chiao Tung University, Hsinchu, Taiwan 30068, ROC E-mail: [email protected]; [email protected]

Abstract Protein tyrosine phosphatase B (PtpB) from Mycobacterium tuberculosis (Mtb) extends the bacteria’s survival in hosts and hence is a potential target for Mtb-specific drugs. To study how Mtb-specific sequence insertions in PtpB may regulate access to its active site through large-amplitude conformational changes, we performed free-energy calculations using an all-atom explicit solvent model. Corroborated by biochemical assays, the results show that PtpB’s active site is controlled via an “either/or” compound conformational gating mechanism—an unexpected discovery that Mtb has evolved to bestow a single enzyme with such intricate logical operations. In addition to providing unprecedented insights for its active-site surroundings, the findings also suggest new ways of inactivating PtpB.

1

ACS Paragon Plus Environment

Journal of the American Chemical Society 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

Introduction Protein tyrosine phosphatases (PTPases) remove the phosphate group from a phosphorylated tyrosine residue on a protein. They are essential in cellular signal transduction as well as during cell growth, 1 and are regulated post-translationally through reversible oxidation-inactivation of the cysteine moiety at the active site, 2 likely forming a cyclic sulfenamide bond with the main chain nitrogen of the neighboring serine. 3 Many pathogenic organisms secrete PTPases into the host to promote their survival. 4 The 30-kDa Mtb PtpB, for example, is considered a key virulence factor secreted into macrophages, presumably manipulating the host cell’s signaling pathways. 5 The Mtb-specfic structural features make PtpB an attractive target for new drugs 6 —the need for which is becoming more urgent with increasing cases of multi-drug resistant strains. 7 Even though its cognate substrates remain elusive, 8 PtpB has been suggested to be a multi-specificity enzyme. 9 As shown in Figure 1, the PtpB structure contains a catalytic core common to PTPases, 10 and additional structural insertions surrounding the cysteine active site. These extra components include α7 (residues 197-212) and α8 (residues 217-226) that form a two-helix lid covering the entrance to the active site of PtpB, and α3a (residues 88-105) that flanks the active-site opening. This is in sharp contrast to a typical PTPase where the catalytically conserved residues are exposed. In the PO4 -cocrystalized structure of PtpB (Figure 1A), α8 is in a closed conformation with α3a unresolved, presumably partially unstructured. 11 In the structure of PtpB complexed with an inhibitor, (oxalylaminomethylene)thiophene sulfonamide (OMTS), α8 adopts an open form with some amino acids shifting as large as ˚ comparing to positions in the PtpB/PO4 structure (Figure 1B). 12 In this structure, α7 27 A remains in the closed conformation, and α3a is seen to take shape as part of the active-site enclosure. A recent single-molecule study showed that both α7 and α8 may dynamically open up in solution to allow access to the active site even in the absence of an inhibitor. 13 For apo PtpB, the open-closed transition rates of α8 are slower than those of α7 by a factor of ∼2 even though the two helices are next to each other with a connecting turn. The 2


Page 2 of 21

Page 3 of 21 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


A α7 α8 active site

PDB ID: 1YWF

B

α8

active site (Cys160)

α7 α3a Cys160

Asp165

Arg166 PDB ID: 2OZ5

Figure 1: Ribbon diagrams for the structure of protein tyrosine phosphatase B (PtpB) with the α8 helix in its (A) closed form (1YWF), 11 and (B) open form (2OZ5). 12 In each panel, the left shows the overall conformation and the right shows a zoom-in view of the active site. The α3a helix is not resolved in the 1YWF structure, and is represented here by an orange dashed trace. For clarity, the co-crystallizing small molecules, PO4 and OMTS, have been removed. slower, rate-limiting transition rate in α8 was attributed to its interactions with the nearby α3a helix. These prior experiments therefore suggest that α8 and α3a play a central role in controlling the dynamic reconfiguration around PtpB active site; as such, the current study will focus on the interplay between the α8 and α3a conformational changes.

Methods Development of all-atom models of PtpB The PtpB/PO4 (PDB code: 1YWF) and PtpB/OMTC (PDB code: 2OZ5) (cf. Figure 1) X-ray structures were used as templates. In the 2OZ5 structure, α8 was open and α3a was attached, 12 and the coordinates of protein atoms in 2OZ5 thus served to develop an 3



all-atom model for the α8 open and α3a attached configuration, (α8o -α3aa ). Hereafter, the superscript ‘o’ or ‘c’ denotes the open or the closed conformation for the α8 helix, and ‘d’ or ‘a’ the detached or the attached conformation for the α3a helix, respectively. In the 1YWF structure, α8 was closed and α3a was not resolved. 11 We thus grafted the α3a coordinates in 2OZ5 onto the 1YWF structure using the NEST homology modeling program 14 for an all-atom model of the α8 closed and α3a attached (α8c -α3aa ) configuration. In the development of all-atom models, after adding coordinates of hydrogen atoms and missing residues via the internal coordinate facility in the CHARMM program 15 and the LOOPY software, 14 the structure was energy minimized to remove local clashes that may occur during model building. To attain configurations in which α3a is detached, i.e., (α8o -α3ad ) and (α8c -α3ad ), we conducted self-guided Langevin dynamics (SGLD) simulations as described in the following starting from the α3a attached configurations, (α8o -α3aa ) and (α8c -α3aa ). As detachment of α3a was indeed observed in several SGLD trajectories started from the (α8c -α3aa ) configuration, the coordinates of detached α3a in a selected structure were grafted onto the α8-closed and α8-open crystal structures as models for (α8c -α3ad ) and (α8o -α3ad ) configurations.

Self-guided Langevin dynamics We employed the self-guided Langevin dynamics (SGLD) method 16 as implemented in CHARMM 35 15 to more rapidly sample the configurational space of PtpB. For each trajectory run, the friction constant in SGLD was first set to 10 ps−1 for 900 ps with a 1.5 fs time step. Following the 900 ps period, the friction coefficient was adjusted to 2 ps−1 and at least 4 ns of trajectory was collected in a particular run. In all SGLD simulations, the local average time was 0.5 ps, the guiding factor was 1, and the system temperature was 300 K. The GBMV2 implicit solvent model 17 was employed in the SGLD simulations using ˚ for non-bound interactions with a switching function activated at 16 A. ˚ a cutoff of 18 A For the (α8c -α3ad )-(α8c -α3aa ) transition, we started from the α8-closed conformation, (α8c 4


Page 4 of 21

Page 5 of 21 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


α3ad ), and computed a total of 32 independent trajectories with different initial velocities. The accumulated total trajectory length was 174 ns. Starting from the α8 open configuration of (α8c -α3aa ), we performed 10 independent runs with the accumulated trajectory length of 74 ns.

Reaction path optimization The objective of reaction path optimization was to generate a smooth transition of structures that will be used as initial configurations for free energy calculations in an explicit solvent. To this end, we followed the procedure developed by Brokaw et al.. 18 Briefly, a path was represented as a chain of duplicated structures of PtpB, and each structure was referred to as a replica. The initial and final replicas were pre-optimized and were fixed during path optimization. In optimizing a path for α3a detachment under the α8-closed condition, for example, the reactant state and the product state were (α8c -α3aa ) and (α8c -α3ad ), respectively. An order parameter was used to define the distance between neighboring replicas that represented the progression of the conformational change along the chain. Since the salt bridges formed by residues Glu95 and Asp105 were prominent features that broke during α3a detachment in the exploratory SGLD simulations, the mass weighed root-mean-square deviation (RMSD) of the side-chain heavy atoms of Glu95 and Asp105 was used to define the distance between replicas, and was found to generate a smooth path consistent with the detachment events observed in SGLD simulations. Reaction path optimization proceeded via minimizing the sum of potential energies of all replicas subject to the holonomic constraints of equal distances between neighboring replicas. 18 The steepest descent (SD) approach was used in the initial stage of path optimization, followed by the adopted basis Newton-Raphson (ABNR) method to satisfy ˚ for path optimization. The FACTS 19 implicit the gradient tolerance of 0.001 kcal/mol/A solvent model was employed for all calculations of path optimization with a non-bound ˚ and a switching function activated at 10 A. ˚ interaction cutoff of 12 A 5



We started a path optimization with a single intermediate replica for quicker convergence. In the second round, a new replica was added in the middle of two neighbors to double the resolution. 18 This procedure was repeated until a smooth path was optimized. For path optimization with more then nine replicas, a kinetic energy potential 18 of 0.5 kcal/mol was added to prevent developing kinks along the path. For α3a detachment, ˚ of RMSD between replicas, and the optimized a total of 17 replicas resulted in less than 1 A geometries along the path were used as initial configurations for computing the free energy profile of α3a detachment under the α8-closed as well as the α8-open conditions. The optimized path was also employed to define the order parameter for computing the free energy profile of a conformational change. For α3a detachment, a combination of two inter-residue distances (IRD) was deduced. IRD1 was between Arg59 in the α3 helix surrounding the active site cleft and Glu95 in α3a, defined as the distance between the mass weighted centers of Cα and HA of the two residues. IRD2 was the distance between the weighted centers of Cα and HA of Arg226 in α8 and Asp105 in α3a. IRD1 and IRD2 were combined as the order parameter ξ = IRD1 + 0.4 · IRD2 for computing the free energy profile of α3a detachment, and the 0.4 coefficient was deduced from the replica structures along the optimized reaction path to determine a single coordinate for representing the conformational change.

Potential of mean force calculations ˚ by 70 A ˚ The structure of each replica was solvated in an orthorhombic water box of 90 A ˚ with TIP3P water for which the boundaries were at least 10 angstroms away from by 65 A any protein atom. Sodium chloride ions were added, amounting to 0.1 M in concentration. The particle mesh Ewald method 20 was used for calculating the electrostatic interactions, ˚ with and the cutoff for short-range electrostatics and non-bound interactions was 12 A ˚ Each system was energy minimized via 200 steps a switching function activated at 8 A. of SD and 1000 steps of ABNR, and then heated to 300 K via 500 ps of velocity assigning 6


Page 6 of 21

Page 7 of 21 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


and equilibrated for 3 ns at 300 K and 1 atm via the Langevin thermostat and piston. 21 The bond lengths involving a hydrogen atom were constrained at their equilibrium values via the SETTLE algorithm, 22 and a 2 fs time step was used for time integration. After equilibration, the Langevin thermostat and piston were also used for production run at 300 K and 1 atm. The system corresponding to each replica was restrained at a value of ξ at 300 K and 2

˚ to compute the 1.0 atm via a harmonic potential with the force constant of 50 kcal/mol/A mean force. 23 The restrained values of ξ were equally spaced along the optimized path of 17 replicas, and the definition of the order parameters ξ was described in the previous section. All of the molecular dynamics simulations in this part were conducted using the NAMD 2.9 software. 24 The 14 ns restrained MD simulations of all 17 replicas were repeated three times and the results averaging over the three profiles of potential of mean force (PMF) are shown later. The statistical uncertainty of the mean force of each replica was estimated via the standard deviation of block-averaged values for the window size of 2 ns. The mean force variances were then summed along the path and the square root of the sum was used to estimate the statistical uncertainties of the PMF values along the profile.

Limited proteolysis assay of PtpB A primary trypsin cleavage site of PtpB, Arg100, was located in the α3a helix. Since the site becomes significantly more exposed when the α3a helix is detached, the trypsin digestion assay could thus provide a means to compare the simulation prediction of coupled conformational changes of α3a and α8 with experimental measurement. Proteolysis at Arg100 produces an N-terminal 11-kDa fragment and a C-terminal 19-kDa fragment, and the intensity of gel bands resolved by SDS-PAGE could be quantified and related back to the concentration of α3a in the detached state. Since the α8 helix conformation could be shifted by the concentration of glycerol in the protein environment, the relative amount of 7



α3a that was detached under different α8 helix conformations was determined using a limited proteolysis assay under different glycerol concentrations. Other details for this series of experiments, including protein protein expression, purification, and labeling, kinetic modeling, and calibration of the dependence of trypsin activity on glycerol concentration were reported in SI.

Results and Discussion α3a tends to detach as α8 was closed. We begin with the α8-closed state, for which the PtpB/PO4 X-ray structure (cf. Figure 1A) was used for model building as described in Methods. The atomic coordinates of α3a in the PtpB/OMTS structure (cf. Figure 1B) were employed to supplement the missing information followed by energy minimization to eliminate local clashes (see Methods for details). For the α8-open state, the PtpB/OMTS Xray structure was used as the initial configuration for molecular dynamics (MD) simulation. Since the α3a coordinates in the PtpB/OMTS structure were used for model building, α3a was attached to the PtpB core in the initial structures for both the α8-closed and α8-open states. A number of independent trajectories were then started from the (α8c -α3aa ) and the (α8o -α3aa ) configurations. These simulations used an implicit solvent model to explore the landscape of protein structure, where the self-guided Langevin dynamics (SGLD) scheme was employed to expand the scope of sampling. 25–27 In ∼60% of 32 independent SGLD simulations starting from the (α8c -α3aa ) configuration (174 ns accumulated length), α3a retained its helical structure and simply underwent detachment from the PtpB core as shown in Figure 2A, as opposed to becoming an unstructured random coil. 11 In the rest of α8-closed simulations, α3a remained attached to ˚ as measured the PtpB core. The α3a detachment was observed to displace as much as 20 A, by the ξ order parameter defined earlier in the Methods section. The histograms of ξ values sampled in SGLD simulations starting from the α8-closed and the α8-open states

8


Page 8 of 21

Page 9 of 21

A

Arg226 α7 α8 α3a Asp105 Glu95

Arg59

8

B

6 4

counts / 1000

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


2 0

C

3 2 1 0 10

15

20

25

30

35

40

order parameter ξ in α3a detachment (Å)

Figure 2: Histograms of the ξ order parameter for describing α3a detachment in SGLD simulations. (A) A ribbon representation of an α3a-detached, α8-closed structure of PtpB. (B) The ξ histogram for trajectories starting from the α8-closed state. (C) The ξ histogram for trajectories starting from the α8-open state. are shown Figure 2B and Figure 2C, respectively. Over the 10 independent simulations (accumulated length 74 ns) starting from the α8-open state (cf. Figure 1A, the (α8o -α3aa ) configuration), α3a stayed attached (cf. Figure 2C). Therefore, the α8o -α3aa configuration remained stable without the inhibitor. The SGLD simulations thus indicated that the α8 and α3a helices interact in such a way that when α8 is open, α3a tends to stay attached; whereas when α8 is closed, the α8c -α3ad configuration is more populated than the α8c -α3aa configuration. PMF of α3a detachment depends on α8 conformation. To characterize the energetics of this coupling, free-energy calculations were carried out using an all-atom explicit solvent model. Briefly, the X-ray structures and configurations sampled in the exploratory SGLD 9



simulations discussed earlier were employed to define, and link, the end states with reaction-path optimization. 18 A path here is a chain of replicated PtpB structures separated in equal distances along an empirically determined order parameter to connect the initial and the final state. Each protein structure along the chain, denoted as a ‘replica,’ was energy minimized subjected to the equal-distance constraint. 18 The atomic structures along each path were first energy minimized in the implicit solvent as in our SGLD simulations, and then were solvated in a box of explicit TIP3P water molecules. The energy-minimized structures were employed as initial configurations for computing the mean forces along the ξ order parameter defined in Methods via restrained MD simulations. These calculations were conducted under the periodic boundary condition at 300 K and 1 atm. In restraining a system at a given value of ξ, 3 ns of equilibration was conducted before the 14 ns production run was collected to compute the mean force. The total energy as a function of time for a particular run is shown in Figure S1 to showcase that the system is well equilibrated in the restrained MD simulations. The procedure of 3 ns equilibration followed by 14 ns production run under equilibrium conditions was repeated three times for each of the 17 replicas. As can be seen from top-left panles in Figures S2–S4 and Figures S5–S7, the free-energy features were very well reproduced in repeated runs, indicating convergence in the free-energy profiles. The statistical significance can be observed in the resulting free energy differences as shown in Figure 3. Other computational details are reported in Methods. We focus on two paths—the (α8o -α3aa )–(α8o -α3ad ) path and the (α8c -α3aa )–(α8c -α3ad ) path—because they are sufficient to generate experimentally testable predictions. The computed free-energy profiles displayed in Figure 3 show that when α8 is closed, α3a detachment leads to free energy reduction (Figure 3B). On the other hand, α3a detachment becomes an activated process when α8 adopts an open open conformation, with a freeenergy penalty of ∼3.15 kcal/mol (Figure 3A)—an energy penalty consistent with the observation in SGLD simulations that α3a remained attacted in the trajectories starting

10


Page 10 of 21

Page 11 of 21

3

A

2 potential of mean force (kcal/mol)

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


α8o-α3ad

1 0

ΔGo,a → d = 3.15 ± 0.33 kcal/mol

-1

14 1 0

α8 helix open

α8o-α3aa

16 18 20 22 24 26 order parameter ξ in α3a detachment (Å)

B

28

α8 helix closed c

α8 -α3a

a

ΔGc,a → d = -2.51 ± 0.14 kcal/mol α8c-α3ad

-1 -2 -3 16

18 20 22 24 26 28 order parameter ξ in α3a detachment (Å)

30

Figure 3: The potential of mean force along the pathways of α3a detachment. Top panel: The profile for the α8 helix being the open state. ∆Go,a→d was estimated from averaging over the plateau PMF values for the detached state of α3a as indicated by the black lines. Bottom panel: The profile for the α8 helix being the closed state. ∆Gc,a→d was estimated from the free energy well when α3a is partially detached, as indicated by the black line. from the α8o -α3aa configuration. The results clearly illustrate, and quantify, the coupling between the conformational transitions for α8 and α3a helices. The free-energy profile for the closed-to-open (or attached-to-detached) transition of one of the helices thus depends on the conformational state of the other. Structurally, as one of the helices closes (attaches to the protein core), the resulting molecular environment hinders the other from fully interacting with the protein core, hence leading to the observed changes in the free-energy landscape. In other words, the active site is at least partially occluded thermodynamically either via α8 closure or α3a attachment. This is reminiscent of logical either/or, but here the logical operation should be understood as a “fuzzy logic” because both helices dynamically and stochastically vary their conformational states. Such compound logical operations, which are beyond the simple and widespread on/off gate, point to a potentially powerful new way of using logic-circuit concepts to analyze protein conformational changes. Relating PMF calculations to an experimental observable. To investigate possible experimental manifestations, we note that the equilibrium fraction of PtpB in its α8-

11



Page 12 of 21

open state (x) is expected to correlate with that in the α3a-detached state (u) because the free-energy difference between the open and and closed states of α8 depends on the attachment/detachment of α3a. Using a simple 4-configuration model, we show that the relative populations of α8 and α3a conformational states are linearly correlated, uˆ = sˆxˆ + c, with (see SI for derivation), h i h i ∆Gc,a→d ∆Go,a→d − x exp x0 exp 0 kB T kB T h i h i. sˆ = ∆Gc,a→d ∆Go,a→d 1 + x0 exp + 1 − x exp ( ) 0 k T k T B

(1)

B

In eq 1, the normalized relative population, xˆ ≡ x/x0 (or uˆ ≡ u/u0 ), is used with x0 (or u0 ) being the reference-state population because it is generally difficult to experimentally determine the absolute population. Based on our computational results (cf. Figure 3), the free-energy differences are ∆Go,a→d = 3.15 ± 0.33 kcal/mol and ∆Gc,a→d = −2.51 ± 0.14 kcal/mol for the α3aa -to-α3ad transition while α8 being in the open or closed states, respectively. Using the experimental x0 = 0.37 ± 0.07 in buffer, 13 the normalized slope in eq 1 gives sˆ = −0.58 ± 0.17. Measurement of conformational coupling between α3a and α8. The above theoretical prediction for the thermodynamics of conformational changes can in principle be examined by ensemble-averaged experiments. In a previous study, it was found that the extent of α8 opening can be tuned by the amount of glycerol present in solution. 13 The adjustability of α8 conformational state provides a unique opportunity to experimentally examine the conformation coupling in PtpB because the computation-predicted interaction between α8 and α3a implies that adding glycerol can also modulate the opening of α3a through the conformational state of α8. Following the experimental protocol published previously, 13 we systematically varied the glycerol content in the reaction buffer and measured the level of α3a exposure (detachment) using limited trypsin digestion (Figure 4A). Further, the α8-open population in various glycerol concentrations can be obtained using the available single-molecule data (Figure 4B). 13 Taken together, we were able to compare the theoretical

12


20% Glycerol kDa M 1 2 3 4 5 6 7 75 50 37

A

25 20

B

0.7 0.6 0.5

19 kDa

15

0.4 0.3

10

0

C relative fraction of α 3a detached

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


percent α8 lid open

Page 13 of 21

5 10 15 20 percent glycerol

25

theoretical prediction ^ stheory = -0.58 ± 0.17

0.7 0.6

experimental test sexpt = −0.63 ± 0.33

0.5

^

0.4 0.3 0.2 Arg100 (proteolysis site)

0.1 0 0

0.2 0.4 0.6 0.8 relative fraction of α8 lid open

1

Figure 4: Experimental test for the computationally predicted α8-α3a coupling. (A) A representative gel image for the limited trypsin digestion assay at 20% glycerol, which allows the evaluation for the relative fraction of PtpB with detached α3a. The arrow indicates the 19-kDa C-terminal fragment after trypsin cleavage at the Arg100 site (see inset in panel C). Numbers above the gels indicate times when samples were collected: (1) 0, (2) 5, (3) 15, (4) 30, (5) 60, (6) 120, (7) 180 minutes. The molecular ladder is shown in the ‘M’ lane. (B) The percentage of α8-helix being open as a function of glycerol concentration calculated using the published data from Ref. 13. The dashed line is uncertainty-weighted linear fit as an eye guide. (C) Comparison of computational prediction (red line) and experimental data (black circles) for the α8-α3a coupling. The black dashed line is the linear fit to the experimental data. The vertical offset of the computational prediction has been adjusted to best fit the experimental data points. prediction in eq 1 with experiments. As shown in Figure 4C, there is a quantitative agreement within statistical uncertainties between molecular simulation and experiment. A molecular picture for the conformational coupling. With the results of PMF calculations corroborated by experiments, the molecular interactions that underlie PtpB’s active-site control can be understood as follows. In general, the interplay between α8 and α3a is seen to arise primarily from mutual occlusion. The α8 helix—consisted of two hydrophobic segments (residues 216–225 and 230–233) separated by a short hydrophilic stretch—exhibits a tendency to form a helix-turn-helix arrangement. This attribute also 13



A

B

α8 open Arg226

Page 14 of 21

α8 closed Arg226 Asp105

Asp105

Glu95 Arg59

Glu95 Arg59

Figure 5: Molecular interactions between the α8 and the α3a helices when (A) α8 is open or (B) closed. Salt bridges are shown as stick with residues from the α3a helix colored in gold and those from the α8 colored in red (open) or blue (closed). explains why the α8-open conformation, which is still populated in the buffer without any binding partner, would become more favorable in less-polar environments (cf. Figure 4B; N polarity index: 28 EN T (water) = 1.000, ET (glycerol) = 0.822). In the closed conformation,

Phe222 in α8 can insert into the active-site cleft to block the exposure of Cys160; 11 as such, α8 serves as the main control of active-site access. As for α3a, the Glu95-Arg59 salt bridge (∆G ∼ −0.46 kcal/mol 29 ) allows coupling to the protein core, and this interaction can stay intact whether α8 takes an open or a closed conformation as shown in Figure 5. The α8-open conformation, on the other hand, allows formation of an additional stronger salt bridge (Asp105-Arg226, ∆G ∼ −1.87 kcal/mol, 29 Figure 5A) between α3a and the protein core, but this interaction cannot form when α8 is closed due to volume exclusion (Figure 5B). These residue-level couplings between α8 and α3a also explain how α8 open-close transition rate is 2x slower than that of α7. The free-energy profiles connecting the α8-closed and the α8-open states, which were separately calculated (see SI for details), indeed illustrated the impact of α3a attachment conformational state on the conformational change of α8. When α3a was in the detached state, the closed state of α8 was at the minimum of the PMF and opening of α8 led to free energy increase (Figure S8). On the other hand, if α3a was in the attached state, the α8 open-closed transition exhibited a free-energy barrier (Figure S9). These observations demonstrated once again a thermodynamic compound logic through coupling in confor-

14


Page 15 of 21 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


mational changes. The PMF profiles also predicted that both the α8-closed and the α8-open states should coexist in room-temperature solution and that the α8-closed state would be more populated—in agreement with the single-molecule results. 13 In principle, it is possible to infer the structural 30 and kinetic 31–34 elements by deep analyses of single-molecule traces and compare the results with the computational PMF obtained here; 35 work along this line is currently underway.

Conclusions A crucial function for the Mtb-specific motif has been identified as active-site protection against reactive oxygen species in order to prolong Mtb potency. 13 Mtb has evolved to effectuate a series of PtpB structural insertions that protect its active site from oxidative inactivation while dynamically allowing remodeling of the active-site surrounding to afford substrates. Thermodynamically, the active site is either blocked by α8 or occluded by α3a. Although the two Mtb specific helices cover the active site from different directions, they occupy some common space if both are closed at the same time. Such a conformationchange coupling is evocative of an exclusive logic gate (XOR) in digital circuits. It should be illuminating to see what roles does this compound conformational gating (molecular logic) play in interfering with the signaling circuitry of the host once the exact mode of PtpB action is known. This work also suggests that disrupting the molecular logical operation, e.g., locking α3a to one conformational state, could be an effective way of inactivating PtpB.

Acknowledgement The authors wish to dedicate this work to the late Prof. Tom Alber. This work was supported by the Princeton University, the Global Networking Talent 3.0 Plan of the

15



National Chiao Tung University, the Ministry of Science and Technology of Taiwan (No. 1032628-M-009-003-MY3, 106-2113-M-009-020-, and 107-2113-M-009-003-), and the Center for Intelligent Drug Systems and Smart Bio-devices (IDS2 B) from the Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan. The US National Science Foundation is acknowledged for a Graduate Research Fellowship (No. DGE-0646086, to TEM), and for part of the computational resources, XSEDE (No. TG-CHE110084).

Supporting Information Available Details of all-atom MD PMF calculations, computational structures along the pathways of conformational changes, and biochemical assay. This material is available free of charge via the Internet at http://pubs.acs.org/.

References (1) Bohmer, F.; Szedlacsek, S.; Tabernero, L.; Ostman, A.; den Hertog, J. Protein tyrosine phosphatase structure-function relationships in regulation and pathogenesis. FEBS J. 2013, 280, 413–431. (2) Ostman, A.; Frijhoff, J.; Sandin, A.; Bohmer, F. D. Regulation of protein tyrosine phosphatases by reversible oxidation. J. Biochem. 2011, 150, 345–356. (3) Groen, A.; Lemeer, S.; van der Wijk, T.; Overvoorde, J.; Heck, A. J. R.; Ostman, A.; Barford, D.; Slijper, M.; den Hertog, J. Differential Oxidation of Protein-tyrosine Phosphatases. J. Biol. Chem. 2005, 280, 10298–10304. (4) Sajid, A.; Arora, G.; Singhal, A.; Kalia, V. C.; Singh, Y. Protein Phosphates of Pah-

16


Page 16 of 21

Page 17 of 21 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


togenic Bacteria: Role in Physiology and Virulence. Ann. Rev. Microbiol. 2015, 69, 527–547. (5) Zhou, B.; He, Y. T.; Zhang, X.; Xu, J.; Luo, Y.; Wang, Y. H.; Franzblau, S. G.; Yang, Z. Y.; Chan, R. J.; Liu, Y.; Zheng, J. Y.; Zhang, Z. Y. Targeting mycobacterium protein tyrosine phosphatase B for antituberculosis agents. Proc. Natl. Acad. Sci. U. S. A. 2010, 107, 4573–4578. (6) Zhang, Z. Y. Drugging the Undruggable: Therapeutic Potential of Targeting Protein Tyrosine Phosphatases. Acc. Chem. Res. 2017, 50, 122–129. (7) WHO, Global tuberculosis report 2016. http://www.who.int/tb/publications/global report/. (8) Sherman, D. R.; Grundner, C. Agents of change - concepts in Mycobacterium tuberculosis Ser/Thr/Tyr phosphosignalling. Mol. Microbiol. 2014, 94, 231–241. (9) Beresford, N.; Patel, S.; Armstrong, J.; Szoor, B.; Fordham-Skelton, A. P.; Tabernero, L. MptpB, a virulence factor from Mycobacterium tublerculosis, exhibits triple-specificity phosphatase activity. Biochem. J. 2007, 406, 13–18. (10) Beresford, N. J.; Saville, C.; Bennett, H. J.; Roberts, I. S.; Tabernero, L. A new family of phosphoinositide phosphatases in microorganisms: identification and biochemical analysis. BMC Genomics 2010, 11, 457. (11) Grundner, C.; Ng, H.-L.; Alber, T. Mycobacterium tuberculosis protein tyrosine phosphatase PtpB structure reveals a diverged fold and a buried active site. Structure 2005, 13, 1625–34. (12) Grundner, C.; Perrin, D.; Hooft van Huijsduijnen, R.; Swinnen, D.; Gonzalez, J.; Gee, C. L.; Wells, T. N.; Alber, T. Structural basis for selective inhibition of Mycobacterium tuberculosis protein tyrosine phosphatase PtpB. Structure 2007, 15, 499–509.

17



(13) Flynn, M. E.; Hanson, J. A.; Alber, T.; Yang, H. Dynamic active-site protection by the M. tuberculosis protein tyrosine phosphatase PtpB lid domain. J. Am. Chem. Soc. 2010, 132, 4772–4780. (14) Petrey, D.; Xiang, Z.; Tang, C. L.; Xie, L.; Gimpelev, M.; Mitros, T.; Soto, C. S.; Goldsmith-Fischman, S.; Kernytsky, A.; Schlessinger, A.; Koh, I. Y. Y.; Alexov, E.; Honig, B. Using multiple structure alignments, fast model building, and energetic analysis in fold recognition and homology modeling. Proteins: Struct., Funct., Bioinf. 2003, 53, 430–435. (15) Brooks, B.; Brooks III, C.; Mackerell Jr, A.; Nilsson, L.; Petrella, R.; Roux, B.; Won, Y.; Archontis, G.; Bartels, C.; Boresch, S.; Others, CHARMM: the biomolecular simulation program. J. Comput. Chem. 2009, 30, 1545–1614. (16) Wu, X.; Brooks, B. R. Self-guided Langevin dynamics simulation method. Chem. Phys. Lett. 2003, 381, 512–518. (17) Mongan, J.; Simmerling, C.; McCammon, J. A.; Case, D. A.; Onufriev, A. Generalized Born model with a simple, robust molecular volume correction. J. Chem. Theory Comput. 2007, 3, 156–169. (18) Brokaw, J. B.; Haas, K. R.; Chu, J.-W. Reaction Path Optimization with Holonomic Constraints and Kinetic Energy Potentials. J. Chem. Theory Comput. 2009, 5, 2050–2061. ¨ U.; Caflisch, A. FACTS: fast analytical continuum treatment of solvation. (19) Haberthur, J. Comput. Chem. 2008, 29, 701–715. (20) Darden, T.; York, D.; Pedersen, L. Particle mesh Ewald: An N log(N) method for Ewald sums in large systems. J. Chem. Phys. 1993, 98, 10089–10092. (21) Feller, S. E.; Zhang, Y.; Pastor, R. W.; Brooks, B. R. Constant pressure molecular

18


Page 18 of 21

Page 19 of 21 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


dynamics simulation: The Langevin piston method. J. Chem. Phys. 1995, 103, 4613– 4621. (22) Miyamoto, S.; Kollman, P. SETTLE: an analytical version of the SHAKE and RATTLE algorithm for rigid water models. J. Comput. Chem. 1992, 13, 952–962. (23) Maragliano, L.; Fischer, A.; Vanden-Eijnden, E.; Ciccotti, G. String method in collective variables: Minimum free energy paths and isocommittor surfaces. J. Chem. Phys. 2006, 125, 024106. (24) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid, E.; Villa, E.; Chipot, C.; Skeel, R. D.; Kal´e, L.; Schulten, K. Scalable molecular dynamics with NAMD. J. Comput. Chem. 2005, 26, 1781–802. (25) Lee, C.-I.; Chang, N.-y. Characterizing the denatured state of human prion 121-230. Biophys. Chem. 2010, 151, 86–90. (26) Olson, M. A.; Chaudhury, S.; Lee, M. S. Comparison between self-guided Langevin dynamics and molecular dynamics simulations for structure refinement of protein loop conformations. J. Comput. Chem. 2011, 32, 3014–3022. (27) Damjanovi´c, A.; Garc´ıa-Moreno E, B.; Brooks, B. R. Self-guided Langevin dynamics study of regulatory interactions in NtrC. Proteins: Struct., Funct., Bioinf. 2009, 76, 1007–1019. (28) Reichardt, C. Solvents and solvent effects in organic chemistry, 4th ed.; Wiley-VCH: Weinheim, Germany, 2011. (29) White, A. D.; Keefe, A. J.; Ella-Menye, J.-R.; Nowinski, A. K.; Shao, Q.; Pfaendtner, J.; Jiang, S. Free Energy of Solvated Salt Bridges: A Simulation and Experimental Study. J. Phys. Chem. B 2013, 117, 7254–7259.

19



(30) Sun, X.; Morrell, T. E.; Yang, H. Extraction of Protein Conformational Modes from Distance Distributions Using Structurally Imputed Bayesian Data Augmentation. J. Phys. Chem. B 2016, 120, 10469–10482. (31) Haas, K. R.; Yang, H.; Chu, J. W. Expectation-Maximization of the Potential of Mean Force and Diffusion Coefficient in Langevin Dynamics from Single Molecule FRET Data Photon by Photon. J Phys Chem B 2013, 117, 15591–15605. (32) Haas, K. R.; Yang, H.; Chu, J. W. Fisher information metric for the Langevin equation and least informative models of continuous stochastic dynamics. J Chem Phys 2013, 139. (33) Haas, K. R.; Yang, H.; Chu, J. W. Analysis of Trajectory Entropy for Continuous Stochastic Processes at Equilibrium. J Phys Chem B 2014, 118, 8099–8107. (34) Haas, K. R.; Yang, H.; Chu, J. W. Trajectory Entropy of Continuous Stochastic Processes at Equilibrium. J Phys Chem Lett 2014, 5, 999–1003. (35) Chu, J.-W.; Yang, H. Identifying the structural and kinetic elements in protein largeamplitude conformational motions. Int. Rev. Phys. Chem. 2017, 36, 185–227.

20


Page 20 of 21

Page 21 of 21

Graphical TOC Entry

closed

α8 helix conformation

open d

21


thermodynami cally preferred?

α8 open? 1 1 e tat hed s c 0 elix atta h a α3 d 0 e ch eta

M. tuberc tuberculosis PtpB

α3a detached?

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


1 0 1 0

0 1 1 0

Compound Molecular Logic in Accessing the ... - ACS Publications

Recommend Documents