Dynamic and structural insights into allosteric regulation on MKP5 a dual-specificity phosphatase

Skeens, Erin; Maschietto, Federica; Manjula, Ramu; Shillingford, Shanelle; Murphy, James; Lolis, Elias J.; Batista, Victor S.; Bennett, Anton M.; Lisi, George P.

doi:10.1038/s41467-025-62150-w

Download PDF

Article
Open access
Published: 31 July 2025

Dynamic and structural insights into allosteric regulation on MKP5 a dual-specificity phosphatase

Nature Communications volume 16, Article number: 7011 (2025) Cite this article

2811 Accesses
1 Altmetric
Metrics details

Subjects

Abstract

Dual-specificity mitogen-activated protein kinase (MAPK) phosphatases (MKPs) directly dephosphorylate and inactivate the MAPKs. Although the catalytic mechanism of dephosphorylation of the MAPKs by the MKPs is established, a complete molecular picture of the regulatory interplay between the MAPKs and MKPs still remains to be fully explored. Here, we sought to define the molecular mechanism of MKP5 regulation through an allosteric site within its catalytic domain. We demonstrate using crystallographic and NMR spectroscopy approaches that residue Y435 is required to maintain the structural integrity of the allosteric pocket. Along with molecular dynamics simulations, these data provide insight into how changes in the allosteric pocket propagate conformational flexibility in the surrounding loops to reorganize catalytically crucial residues in the active site. Furthermore, Y435 is required for the interaction with p38 MAPK and JNK, thereby promoting dephosphorylation. Collectively, these results demonstrate critical roles for the allosteric site in coordinating both MKP5 catalysis and MAPK binding.

Evolutionary genomic relationships and coupling in MK-STYX and STYX pseudophosphatases

Article Open access 09 March 2022

Mechanism of activation and autophosphorylation of a histidine kinase

Article Open access 03 September 2024

Increased oxidative phosphorylation through pyruvate dehydrogenase kinase 2 deficiency ameliorates cartilage degradation in mice with surgically induced osteoarthritis

Article Open access 03 February 2025

Introduction

The mitogen-activated protein kinases (MAPKs) are a family of serine-threonine kinases that are responsible for a variety of critical cellular functions¹. The MAPKs are regulated in a well-controlled manner through phosphorylation by upstream kinases known as MAPK kinases (MKKs), which activate the MAPKs^2,3. In contrast, the MAPKs are negatively regulated via dephosphorylation by enzymes known as MAPK phosphatases (MKPs)^4,5. The coordinated activities between the MKKs and MKPs dictate the signaling output of the MAPKs to propagate responses such as gene transcription^4,6. Although there has been a substantial amount of work conducted on the actions of the MKKs, such as MEK1/2⁷ on MAPK activation, much less is known about the regulation of the MKPs on MAPK inactivation. Given the increasing realization that the MKPs represent novel therapeutic targets for the treatment of cancer, inflammatory and metabolic diseases, as well as rare diseases^8,9,10,11, a detailed understanding of MKP regulation has gained in its importance and significance.

The MKPs are a group of ten dual-specificity phosphatases (DUSPs) that directly inactivate the MAPKs through the dephosphorylation of the regulatory phosphotyrosine and phosphothreonine residues that reside within the activation loop of the MAPKs¹². The MKPs exhibit an exceptionally high degree of specificity for the MAPKs via direct interactions with the MAPKs through a conserved N-terminal MAPK docking site designated as the kinase interacting motif (KIM)¹³. Further selectivity to the MAPKs is conferred through a well-conserved catalytic pocket at the C-terminus comprising the essential cysteine residue. The structure of the catalytic pocket is sufficiently deep to accommodate the phosphotyrosyl residue of the MAPK to coordinate the first dephosphorylation event, followed by phosphothreonyl dephosphorylation^14,15. In addition to the combined contributions of the KIM and the active site in establishing MKP-MAPK dephosphorylation specificity, the MKPs typically reside in a low basal state of activity and often adopt an active conformation upon binding MAPK^15,16,17. Thus, the MKPs are highly regulated and direct exquisite MAPK substrate specificity through multiple mechanisms.

Despite these well-defined modes of MKP regulation, we identified an additional regulatory mechanism of the MKPs that indicates that phosphatase activity can be regulated through a novel allosteric site within the catalytic domain¹⁸. A high-throughput screen identified Compound 1 (Cmpd 1) as a small molecule inhibitor of MKP5 catalytic activity, and mutational analyses in combination with a solved X-ray structure of the MKP5 catalytic domain (MKP5-CD) bound to Cmpd 1 revealed that Cmpd 1 interacts with MKP5-CD at an allosteric pocket ~8 Å away from the catalytic C408¹⁸. The interaction of the inhibitor with MKP5-CD was mediated through hydrophobic interactions, most notably via Y435, and disruption of allosteric pocket residues attenuated Cmpd 1-MKP5-CD binding and MKP5 catalysis¹⁸. The discovery of the MKP5 allosteric site thereby uncovered a novel mode of regulation for MKP5 catalysis. Indeed, the importance of the MKP5 allosteric site for catalysis is supported by the observation that the critical Y435 residue is conserved amongst the active MKPs. We showed that MKP7, which is the nearest MKP in sequence homology to MKP5, contained an analogous allosteric site within its catalytic domain¹⁹. The MKP7 allosteric site is not only critical for MKP7 catalysis, but also contributes to binding with its substrates, p38 MAPK and c-Jun N-terminal kinase (JNK)¹⁹. These observations expanded the possible functions of the allosteric site, suggesting a role in not only regulating MKP catalysis but also MAPK engagement.

To further define the molecular characteristics and function of the MKP5 allosteric site, here we performed a combination of biophysical, computational, and biochemical experiments. We report that the MKP5 allosteric site confers profound conformational changes consistent with our previously reported deleterious effects of mutations in this region on MKP5 catalysis¹⁸. Molecular dynamics simulations infer that the allosteric site represents a conduit through which conformational dynamics are mediated, and together with NMR and X-ray crystallography, provide a plausible explanation for the impact of the allosteric site on MKP5 catalytic activity. Finally, we show that the MKP5 allosteric site represents an essential region through which the stress-activated MAPKs, p38 MAPK and JNK, interact to mediate dephosphorylation. Together, these data shed new light on the molecular and functional characteristics of the MKP5 allosteric site and provide further information on how selectively targeting this allosteric site within MKP5, and possibly other MKPs, leads to enzymatic inhibition.

Results

Inhibitor-bound MKP5-CD crystal structures exhibit conformational changes in both the allosteric and catalytic pockets

In our previous study using high-throughput screening, we identified compound 1 (Cmpd 1) as an inhibitor of MKP5 catalytic activity¹⁸. We determined the structure of the complex between MKP5-CD and Cmpd 1, characterized the kinetics of Cmpd 1-mediated inhibition of MKP5 activity, which led to its assignment as an allosteric MKP5 inhibitor¹⁸. The key allosteric residue that mediates Cmpd 1 binding, Y435, was identified to be essential for MKP5 catalysis¹⁸. To further define the molecular basis through which Y435 regulates MKP5 function, we reasoned that structures of mutations of this allosteric residue in the presence and absence of Cmpd 1 would reveal further insights into the subtleties of the allosteric pocket. Since the variant Y435W retains ~70% of WT MKP5-CD activity and binds Cmpd 1, whereas the Y435A and Y435S variants were catalytically impaired and failed to bind Cmpd 1, we focused on solving the structure of Y435W in the presence and absence of Cmpd 1. Crystals of the apo Y435W-CD structure and the Cmpd 1-bound complex were solved at 2.40 and 2.90 Å, respectively (Supplementary Table 1). These two new structures were compared to (a) each other, (b) to apo-WT MKP5-CD (PDB: 2OUD)^20,21, and (c) to Cmpd 1-WT MKP5-CD (PDB:6MC1)¹⁸ for the overall total root-mean-square deviation (RMSD) (Fig. 1). Prior to any comparisons, we confirmed MKP5 is a monomer by analytical ultracentrifugation (AUC) and oligomers observed in MKP5 crystal structures are due only to the high concentration necessary for crystallization (Supplementary Fig. 1). Structural alignment of the allosteric residue in the apo-form indicates the plane of the tryptophan in Y435W has tilted 42.8° away from the tyrosine in WT MKP5-CD (Fig. 1a). The electron density of Cmpd 1 is fully interpretable in Y435W (Supplementary Fig. 2). In the Y435W-Cmpd 1 complex, W435 is tilted 64.9° and shifted toward the α4 helix, creating space for Cmpd 1. No major differences were observed between the Cmpd 1-bound WT and Y435W complex structures (Fig. 1a). Additionally, multiple hydrophobic interactions are established between A416, I420, I445, N448, and P451 and the dimethyl butanone moiety of Cmpd 1. The backbone atoms of M452 and N448 also serve as electron donors for hydrogen bonding with Cmpd 1 (Fig. 1b). The Cmpd 1-WT MKP5-CD structure showcases alterations in the conformation of α4-α5 loop and the catalytic pocket that reduces the volume of the enzymatic site by ~18%¹⁸. There are some notable differences to account for the reduction, particularly the movement of loop residues (I445-N448) in the presence of Cmpd 1 that alters the hydrophilic interactions between the residues of α4-α5 loop (allosteric pocket) and α3 helix (catalytic pocket) (Fig. 1b). This observed change includes the side chain of S446 and P447 on the loop between α4-α5, and a slight rotation of helix α3, resulting in rearrangements distinct from apo MKP5-CD. Other structural modifications involving helix α3 are particularly important because its N-terminal region (residues G411-A416) separates the allosteric and enzymatic sites. A new hydrogen bond is formed between the backbone carbonyl of S446 and the hydroxyl group of S413 in the α3 helix by the rearrangement caused by Cmpd 1. These new interactions probably occur concurrently with the disruption of a hydrogen bond between the backbone atoms of S413 and N448. A new hydrogen bond between the backbone amide of S446 is made with the carboxyl of G411 after the disruption of the bond between S446 with V412. In the apo form, R442 interacts and stabilizes A410 and G411, but these interactions disappear and promote the conformational changes of the β5-α3 loop in the catalytic pocket. Significantly, this promotes a new interaction between catalytic residues C408 and Q409. Finally, the tilt between the Y435 and W435 in the Cmpd 1-bound structures contribute to changes in the allosteric and catalytic pockets. We explored whether the ~70% enzymatic activity of Y435W (relative to WT MKP5) was caused by a significant difference in the volume of the catalytic site, but it was not significant. Although the two other mutations, Y435A and Y435S, which have little enzymatic activity and do not bind Cmpd 1¹⁸, are well-suited for studying the structural and mechanistic source of catalysis, they were unstable for crystallization. Hence, we studied the effect of the Y435A and Y435S mutations on the apo proteins by NMR and molecular dynamics (MD) and focused on WT and Y435W for the Cmpd 1 bound proteins.

**Fig. 1: Tryptophan mutation at residue 435 confers structural and dynamic changes to the MKP5 allosteric pocket.**

Aromaticity of residue 435 is required for structural fidelity of the allosteric pocket

NMR studies were performed to characterize the chemical importance of the allosteric pocket residue Y435 by determining the structural effects of mutations at that allosteric site. Variants at Y435W, Y435S, and Y435A were generated, and all were found to be suitable for these NMR studies. Analysis of ¹H¹⁵N HSQC correlation spectra revealed chemical shift perturbations indicating localized structural changes, and line broadening suggesting heightened dynamics, at residues surrounding the site of mutation (Fig. 2a, b and Supplementary Figs. 3, 4). Though such structural and dynamic effects are often observed in regions adjacent to a mutation, the Y435S and Y435A variants exhibit considerably more chemical shift and dynamic perturbations than Y435W. These data suggest that substitution of the aromatic side chain at residue 435 disrupts the integrity of the allosteric pocket, which is consistent with previous work showing reduced binding of Cmpd 1 to MKP5 when Y435 is mutated to either a serine or alanine, while binding and activity are maintained by a tryptophan residue at the same site¹⁸. Further, resonances exhibiting chemical shift perturbations in Y435S and Y435A often shift on a similar trajectory relative to WT MKP5-CD. These observations suggest that the structural changes induced by Y435S and Y435A populate a similar conformation, which is distinct from WT MKP5-CD (Fig. 2c). In contrast, Y435W induces chemical shift perturbations at those same resonances, but on a divergent trajectory to that of the serine and alanine variants or remaining similar to WT MKP5-CD. Notably, all three Y435 mutations result in a chemical shift perturbation at the catalytic residue D377, demonstrating that disruption of the allosteric site through mutation is propagated to the catalytic site, as expected for an allosteric system. These data support previous findings that Y435 mutations decrease MKP5 activity, with the degree of NMR-detected perturbation directly correlating with the degree of catalytic attenuation¹⁸. Further, in all three mutants, we observe significant structural changes and/or line broadening in the α4-α5 allosteric site loop. Given that Y435W is partially active, and Y435S and Y435A are inactive, these data are consistent with analysis of the crystal structures in suggesting that the integrity of this loop is critical for catalysis, with the crystal structures providing a direct observation of interactions between the α4-α5 loop and the active site pocket residues in the α3 helix (Fig. 1). Similarly, thermal unfolding experiments revealed that the Y435S and Y435A mutations cause significant destabilization of MKP5-CD, with changes in the melting temperature (ΔT_m) of −11.9 °C and −14 °C, respectively, relative to WT (Supplementary Fig. 5). Conversely, Y435W only exhibits a ΔT_m of −4 °C. The rank order of changes in the melting temperature also correlates with the variant catalysis¹⁸. Thus, in addition to the integrity of the allosteric pocket, the aromaticity of the side chain at residue 435 contributes to the global stability of the MKP5-CD and further implicates the allosteric pocket as critical for MKP5 structural integrity.

**Fig. 2: Loss of aromaticity at residue 435 elicits structural and dynamic perturbations in the allosteric pocket.**

To further understand how changes in the allosteric pocket affect conformational flexibility in surrounding loops and reorganize crucial catalytic residues in the active site, we computed a traffic map of residue-to-residue shortest- and highest-correlation paths from the sampled MD trajectories of each mutant (Fig. 2d). This traffic map includes paths for each residue pair, resulting in N² routes (where N is the number of residues), allowing us to estimate the “backroads” and “highways” of information transfer through the polypeptide (see “Methods” for further details). We find similar traffic patterns in WT and Y435W, as opposed to remarkably altered communication pathways in Y435A and Y435S, each of which differs from WT in unique aspects (Fig. 2d). WT and Y435W show a high-traffic route directly connecting the allosteric region (via α5) with the active site, suggesting a preferred route of communication between allosteric and active sites (Fig. 2d, e). In contrast, the same pathway is altered in Y435A, missing the S413-S415 connection (Fig. 2d, e). This suggests an inefficient coupling between the two functional sites. Moreover, the main route deviates with respect to WT and Y435W and connects the α2 region to S415 rather than β5. The residues highlighted by the traffic map as relevant for communication follow the trend of NMR spectral perturbations (Fig. 2a, b). The extent of communication, as reflected by the map’s breadth, increases progressively from WT and Y435W to Y435S, where it spreads over a larger number of residues (Fig. 2e). The residues also match the dynamic perturbations in the allosteric pocket revealed by NMR (Fig. 2a, b). Y435S completely rewires the communication routes, consistent with the experimentally observed trends of strong NMR chemical shift perturbations (Fig. 2b) and poor catalytic activity¹⁸, with a substantially altered profile extending to the α2-β3 loop. In addition, a clear correlation is observed between the node and betweenness centrality distributions in WT and the mutants, quantitatively demonstrating a decreasing trend of similarity between Y435W to Y435A to Y435S (Supplementary Fig. 6). Moreover, the trend is not affected when the graph is constructed from backbone-only dihedral fluctuations or by including side-chain correlations, further validating our results (Supplementary Fig. 7). These data suggest that the structural changes induced by the Y435A and Y435S mutations contribute to the altered dynamics of neighboring residues, causing an attenuation of communication via the α5-β5 route to ultimately impact catalytic behavior.

Molecular dynamics identifies MKP5 allosteric site-to-active site communication channel

To gain further insight into the impact of the allosteric site mutations, we analyzed differences in root mean squared fluctuations (ΔRMSF) from MD simulations. Significant mutation-induced changes were noted in the vicinity of the MKP5 allosteric site, consistent with the NMR data (Fig. 2a), and the flexible regions encompassing residues 325–350 and 353–379 (Fig. 3a, b), the latter of which is unassigned in the NMR spectrum. The RMSF patterns by MD mirror the trends seen in the chemical shift and dynamic perturbations, with the greatest fluctuations detected in the serine variant (Fig. 3a, b).

**Fig. 3: Loss of aromaticity at the allosteric residue 435 reorganizes non-covalent interactions.**

We quantified changes in non-covalent interactions (ΔNCI) observed in MD simulations of the MKP5-CD variants relative to the WT MKP5-CD trajectory. Residues showing persistent polar and hydrophobic interactions are highlighted in Supplementary Fig. 8. We focused on the reorganization of interactions during microsecond-scale MD, which were performed in triplicate to ensure sufficient sampling and to enable comparison with experimental data. Figure 3c, d depict the cumulative changes in non-covalent interactions per residue, denoted as Σ(ΔNCI), and the absolute changes, denoted as Σ|ΔNCI|, arising from polar and apolar interactions. The Σ(ΔNCI) metric offers an assessment of the overall alteration to the chemical environment at each residue. The absolute value, Σ|ΔNCI|, highlights the residues experiencing the most significant alterations in the NCI network, irrespective of whether the net change is zero (i.e., in cases where disrupted interactions are compensated by the formation of new ones). Significant deviations from the WT MKP5-CD profile are observed, regardless of the mutation at residue 435, with the most notable changes concentrated in the active and allosteric sites (Fig. 3c, d). The Y435S variant exhibits a distinct Σ|ΔNCI| profile, especially in the allosteric site and at residues 334–344 (Fig. 3d). The profile of Σ|ΔNCI| indicates similarly reshuffled interactions caused by the Y435W and Y435A mutations, although there are significant differences between their relative intensities (Fig. 3d–f). The net change induced by each variant follows the trend Y435W > Y435A > Y435S, suggesting that the increasing disruption of NMR chemical shifts caused by non-aromatic mutations corresponds to a greater loss of non-covalent interactions induced by the variants. This loss is accounted for largely by polar interactions in Y435S, while the increase of ΔNCI in Y435W is mostly derived from an increase in hydrophobic contacts (Fig. 3e). Such changes include the disruption of the catalytic residue C408:SG–G411:O hydrogen bond, which decreases by over 35% with Y435W, Y435A, and Y435S mutations. Conversely, the N448:N–L449:N backbone hydrogen bond shows a 30% increase with Y435W, Y435A, and Y435S mutations. The changes in ΔNCI values occurring in more than 15% of the analyzed trajectories for each variant are detailed in Supplementary Table 2. The Y435S mutation causes over 40% of the examined scenarios, including a rearrangement of the hydrogen-bonding network in the allosteric site (residues T427-N--R428-N, T430-OG1--D433-N, T430-OG1--M431-N), which is not observed in Y435W (Supplementary Table 2).

The cumulative ΔNCI contributions for residues in the allosteric and active sites suggest that the major perturbations occur in the allosteric site (Fig. 3g). However, the variants differ in the impact they have on the active site, as Y435S leads to a reinforcement of the hydrophobic contact C408-N333, which is not observed in either Y435W or Y435A (Table 1). Figure 3h shows selected interactions with an overall persistency greater than 15% that are unique to specific variants, highlighting the regions that are most susceptible to the introduction of each mutation. The largest structural difference between Y435W and Y435A is the disappearance of the T355:Oγ1–E409:Nε2 hydrogen bond, which is absent in the latter. Additionally, both Y435A and Y435W exhibit the loss of the K439-Nζ–P433-O hydrogen bond, a disruption not observed in Y435S. These results suggest that, in comparison to WT MKP5-CD, the Y435A and Y435S mutations enhance the flexibility of the allosteric site and reorganize the active site, whereas the Y435W mutation leads to only minor changes, as shown by the distributions of the active site and allosteric site center-of-mass distances to β5-α3 loop and α4-α5 loop, respectively (Fig. 3i). These distances reflect the internal reorganization of these two functional sites, showing a shift of allosteric site distances to larger values in Y435A and Y435S (coupled to large changes in the active site in Y435S), which are not observed in either WT or Y435W. This distinction is further supported by pairwise distance analyses (Supplementary Fig. 9) and flexibility metrics (Fig. 3i).

Table 1 Non-covalent interaction changes estimated through molecular dynamics simulations

Full size table

Allosteric site ligand engagement alters dynamics and non-covalent interactions in the MKP5 catalytic domain

Biochemical studies of MKP5 in the presence of Cmpd 1 revealed that mutation of Y435 to serine or alanine dramatically reduced its ability to bind¹⁸. In contrast, the Y435W variant showed slightly reduced, but near-WT affinity¹⁸. We investigated the interaction of Cmpd 1 with MKP5-CD by NMR and found that when Cmpd 1 is bound to either WT MKP5-CD or Y435W, the associated ¹H¹⁵N HSQC NMR spectra exhibit only small structural changes via chemical shift perturbations (Supplementary Fig. 10), but significant line broadening (Fig. 4a, b and Supplementary Figs. 10–12). This effect, seen across the MKP5-CD, could be caused by a few different biophysical phenomena. Such extensive line broadening is surprising for small molecule binding and could occur due to the formation of large oligomers. However, AUC data showed that WT MKP5 is monomeric in both its apo and Cmpd 1-bound states (Supplementary Fig. 1). Line broadening is often a signature of heightened protein dynamics, and may suggest that Cmpd 1, when bound to the allosteric site, exerts its inhibitory effect by disrupting active-to-allosteric site communication by enhancing micro-millisecond fluctuations in MKP5. The observed line broadening may also be due to the on/off kinetics of Cmpd 1 causing the MKP5 backbone resonances to be in intermediate exchange. Though Cmpd 1 has limited solubility in an aqueous solution, we added Cmpd 1 to MKP5 at a molar ratio beyond its solubility limit, presuming that the highest amount of Cmpd 1 possible would be solubilized and interact with MKP5, and then centrifuged away any precipitated Cmpd 1. At the highest concentration of Cmpd 1 that can be bound to MKP5 under these experimental conditions, the spectrum remained significantly line broadened (Supplementary Fig. 13). Thus, intermediate exchange caused by the on/off kinetics of Cmpd 1 remains a possible source of the line broadening. To obtain a clearer structural picture of the entire MKP5 backbone, we also titrated Cmpd 1 into WT MKP5 while monitoring ¹H-¹³C ct-HSQC spectra and found that, like the amide resonances, many of the carbon resonances exhibited significant line broadening with only minor chemical shift perturbations (Supplementary Fig. 14). In contrast to WT and Y435W MKP5, titration of Cmpd 1 into Y435S produced minor NMR spectral changes (Supplementary Fig. 15), consistent with negligible binding and prior biochemical findings that can be attributed to structural and dynamic disruption of the allosteric pocket (Figs. 2 and 3) and loss of π-stacking interactions between Cmpd 1 with an aromatic side chain at residue 435¹⁸. Based on these experimental findings, we simulated the Cmpd 1-bound dynamics for WT and Y435W MKP5-CD. The stability of Cmpd 1 within the Y435W binding site resembles that of the WT. In both complexes, Cmpd 1 adopts a specific configuration supported by a strong π-stacking interaction between the aromatic portion of the ligand and tyrosine or tryptophan residues (Supplementary Fig. 16). The modified aromatic ring portion leads to a slightly increased flexibility of the Y435W binding site (Supplementary Fig. 17).

**Fig. 4: Allosteric site engagement alters the dynamic profile of the MKP5-CD.**

A comparison of the MD-derived conformations of the α4-α5 loop of the allosteric and active sites showed remarkable alignment (Fig. 4c–e). To characterize the difference in the conformational dynamics among apo and WT or Y435W Cmpd 1-bound structures, we measured the similarity of the MD configurations and crystal structure (6MC1) using the Kullback-Leibler divergence (KLD) metric computed at each frame of the trajectory. The probability distribution computed from the resulting values shows higher similarity between the WT and Y435W holo distributions (narrower Gaussian centered on smaller KLD values) when compared to the apo WT (Fig. 4d, e). These results agree with those derived from the crystal structure wherein a comparison of the holo WT and holo Y435W indicates little difference between the structures (Fig. 1). Moreover, the distance distributions between active site, allosteric site, and relevant loops (a4-a5 loop, β5-α3 loop) are conserved in Y435W with the allosteric site center-of-mass (COM) to a4-a5 loop distances showing a shift to larger distances, and active site COM to β5-α3 loop distances distribution becoming narrower in response to Cmpd 1 binding (Fig. 4f, g), consistent with an active site volume reduction¹⁸.

We evaluated the impact of Cmpd 1 on WT and Y435W MKP5-CD by assessing the cumulative response based on NCI alterations that occur upon binding. These changes represent a range of persistency levels sampled at various frequencies throughout the trajectory. In both systems, the overall change in NCIs (ΔNCI (holo–apo)) in WT and Y435W results in an apparent “loss” of interactions (Supplementary Figs. 18, 19). This suggests a higher number of non-covalent interactions are being strengthened rather than weakened, indicating a propensity for continuous partner exchanges. This behavior is illustrated by the specific changes in ΔNCI (holo–apo), such as polar, hydrophobic, and π-stacking interactions on the WT protein structure (Supplementary Fig. 18). In the WT MKP5-CD, Cmpd 1 binding induces widespread modifications across the entire structure, particularly affecting the allosteric site (α4, α5), the active site (β5-α3 loop), and neighboring secondary structures (β5-η2 loop, α1, η2, β4—Supplementary Fig. 19). A similar pattern is observed in Y435W, albeit with a noticeable decrease in the secondary structures at β3 and β5 accompanied by more pronounced changes in β5-η2 loop, α1, η2, and β4 (Supplementary Fig. 19). The reorganization of these interaction networks (with respect to the apo states) is consistent with the significant line broadening observed in NMR experiments (Fig. 4a, b) and support the notion that Y435 is required to maintain the structural integrity of the allosteric pocket. The dynamic effects introduced by Cmpd 1 are consistent with the altered catalytic output, though the differences in overall active-allosteric site atomic displacements and non-covalent interactions in WT and Y435W mirrors the Cmpd 1-induced functional effect in the two systems.

MKP5 allosteric site residue Y435 mediates MAPK binding

Structural insights from our NMR, crystallographic, and MD studies demonstrated that disruption of the allosteric site, either by mutation or Cmpd 1 binding caused disruption of loops, which appeared important for the transmission of structural changes to the active site. Furthermore, molecular modeling of Cmpd 1 bound to MKP5-CD with JNK1 predicts that an interaction interface is disrupted by Cmpd 1¹⁸. These observations suggest that the allosteric pocket could serve to mediate MAPK interactions with MKP5-CD leading to changes in active site conformation. Based on these observations, we tested whether the functional importance of these structural changes is evoked by the engagement of MKP5 with its substrates, p38 MAPK and JNK. To test this, we transiently transfected COS-7 cells with GST-tagged full-length WT MKP5 or MKP5 allosteric pocket variants, Y435W, Y435A, and Y435S. As a control, COS-7 cells were also transfected with the catalytically inactive and substrate-trapping variant of MKP5 containing an active site C408-to-S408 mutation (C408S). To test for p38 MAPK or JNK binding to these MKP5 variants, either HA-tagged p38 MAPK or GFP-JNK expression vectors were co-transfected, respectively. Following affinity precipitation of MKP5 from COS-7 cells, we found that Y435A, Y435S, and Y435W MKP5 were significantly impaired in their ability to complex with p38 MAPK as compared with WT MKP5 (Fig. 5a). Similarly, JNK was also reduced in its amount complexed with the MKP5 allosteric pocket variants, as compared with WT MKP5 (Fig. 5b). In contrast, the catalytically inactive and substrate-trapping variant C408S bound to both p38 MAPK and JNK at substantially higher levels, consistent with the establishment of a dead-end MKP5-JNK and MKP5-p38 MAPK complex. Consistent with the lack of p38 MAPK and JNK binding to MKP5 by the allosteric mutant variants and reduced catalytic activity we found that these mutants when overexpressed were diminished in their capacity to dephosphorylate endogenous p38 MAPK and JNK in COS-7 cells stimulated with sorbitol (Supplementary Fig. 20). These results indicate that the MKP5 allosteric pocket not only mediates catalysis, as demonstrated previously¹⁸, but also mediates both p38 MAPK and JNK binding which correlates with the loss of MKP5 activity in cells. The implications of MAPK docking to the allosteric site based collectively on our structural findings described herein, infer that this interaction induces similar conformational changes that propagate to the structural reorganization of the active site.

**Fig. 5: MKP5 residue Y435 mediates MAPK binding.**

MKP5 allosteric site differentially orders p38 MAPK and JNK active site engagement

To further understand the relationship between MAPK-MKP5 allosteric and active site engagement, we tested the effects of the catalytically inactive substrate-trapping variant, C408S, to bind either p38 MAPK or JNK in the context of the allosteric Y435S variant. As shown when Y435S was expressed in COS-7 cells, both p38 MAPK and JNK failed to interact (Fig. 5). Consistent with the substrate-trapping properties of C408S, both p38 MAPK and JNK bound with significantly increased levels to this MKP5 variant as compared with WT MKP5 (Fig. 6). However, when both the allosteric site (Y435A) and catalytically inactive substrate-trapping (C408S) variants were combined to generate the double Y435A/C408S MKP5 mutant, we found that p38 MAPK was still capable of complex formation with MKP5 (Fig. 6a–c). In contrast, JNK binding to MKP5 was markedly inhibited in the Y435A/C408S variant (Fig. 6d–f). Based upon our NMR findings that mutation of Y435 to A435 causes structural and dynamic perturbations in α4 and α5 regions of the allosteric pocket, these results suggest that JNK utilizes these regions to bind MKP5 more so than p38 MAPK. The levels of phosphorylated p38 MAPK and JNK present in the complex of Y435A/C408S MKP5 were reduced slightly for p38 MAPK and JNK as compared with C408S MKP5 alone (Fig. 6). We interpret these results to suggest that the active site conformational changes induced by Y435A at C408 are more deleterious to the formation of the phospho-thioate intermediate of phosphorylated p38 MAPK and JNK. Collectively, these data suggest that p38 MAPK and JNK exhibit distinct differences in the constraints required for active site engagement when bound to the allosteric pocket.

**Fig. 6: Effects of MKP5 residue Y435 on MAPK engagement at the active site.**

Discussion

The MKPs play fundamental and non-redundant signaling roles through their specific ability to dephosphorylate and inactivate the MAPKs in a spatio-temporal manner^4,5. Serving as nodal regulators of the MAPKs, the MKPs integrate multiple MAPK responses from upstream stimuli. As such, MKPs have been shown to mediate both positive and negative signaling outcomes and have become attractive therapeutic targets for certain human diseases⁸. Although the molecular basis of the MKP’s to directly dephosphorylate the MAPKs is appreciated, how the MKPs derive signaling selectivity is not fully understood, suggesting that the full complexity of their regulation has yet to be uncovered. The identification of a novel allosteric pocket that regulates MKP5 activity revealed an unanticipated mode of regulation that raised questions about the site-specific conformational changes at the molecular level that influence MKP5 catalysis. Here, we have integrated structural, computational, and biological approaches to demonstrate that the integrity of the allosteric site is strictly maintained through side-chain aromaticity at residue 435 located within the phosphatase domain. This interpretation is based on the conformation populated by the catalytically competent Y435W variant, visualized by crystallography and NMR, which is distinct from the dynamically unstable conformation of Y435S and Y435A. The Y435S and Y435A variants do not crystallize in the presence or absence of Cmpd 1. There are no major conformational changes between WT and Y435W structures in the presence or absence of Cmpd 1. The only minor change is the tilt of the bound Cmpd 1 in WT and Y435W, but this modification does not alter the volume of the active site, which is purported to be partly responsible for the decrease in enzymatic activity¹⁸. NMR and MD simulations highlight fluctuations near the allosteric site that are similar among the three variants, but also reveal that the distinct catalytic behaviors stem from a network of disrupted non-covalent interactions that follows a rank order of potency Y435W > Y435A > Y435S. Molecular conduits of communication between the allosteric and active sites interpreted by MD simulations provide insight into how ligand engagement (e.g., MAPK binding) propagates conformational changes to the active site. Finally, we show that both p38 MAPK and JNK represent physiological ligands of the allosteric site, suggesting that MKP5 undergoes dynamic conformational reorganization that likely facilitates MAPK dephosphorylation.

Although there have been other protein tyrosine phosphatases (PTPs) that have been reported to be modulated by allosteric small molecule binding, such as SHP2²² and PTP1B²³, MKP5 represents the first MKP targeted in this precise manner²⁴. MKP3/DUSP6 has been shown to be inhibited by the small molecule 3-(Cyclohexylamino)-2,3-dihydro-2-(phenylmethylene)-1H-inden-1-one (BCI) through a mechanism suggestive of allostery based upon in silico docking^25,26. However, how BCI, through this putative allosteric site, transmits its inhibitory effect to the active site remains unknown. Here, we have expanded upon our initial observation that Cmpd 1 collapses the active site volume to inhibit catalysis, by defining the communication channels from the allosteric site to the active site. Mutations at the allosteric site that deform hydrophobicity exert significant destabilization of the MKP5 catalytic domain. The Y435W and Y435A mutations impact the β3-η2 loop, resulting in a restructuring of hydrogen bonds, more notably disturbed in Y435S (see Supplementary Table 2). Notably, within the β4-η3 loop, the catalytic residue D377 displays significant positional variation across all allosteric mutants, consistent with NMR chemical shift perturbations, with the interaction networks showing more pronounced changes in Y435A and Y435S. Thus, specific interactions affected by each variant in silico align with all the observed experimental trends, supporting the interpretation that aromatic residues prime the MKP5-CD to facilitate its allosteric crosstalk. Thus, binding of Cmpd 1, and by inference either p38 MAPK or JNK, to its allosteric site rewires non-covalent interactions (especially hydrogen bonds) and increases flexibility surrounding the pocket in a mutation-specific fashion. The distinct interaction networks observed from MD simulations with Cmpd 1 bound to MKP5 variants are also consistent with experimental findings.

Our biophysical findings, together with MAPK interaction assays, show that the allosteric pocket of MKP5 represents a multifunctional region within the catalytic domain to control MAPK docking and enzymatic activity. Previous studies demonstrated that MKP5 exhibited a structure that is consistent with evidence that it has constitutive phosphatase activity^21,27. In other MKPs that have been reported to exhibit inducible activity, the catalytic aspartic acid residue is held away from the other catalytic residues by a hydrogen bond with a nearby asparagine, which is absent in MKP5^20,21,28. The experiments reported here, which indicate the establishment of non-covalent networks from the allosteric site to the active site suggest that additional regulatory features of MKP5 are operative upon MAPK binding that could further influence substrate selectivity, engagement and/or enhancement of phosphatase activity. Although both p38 MAPK and JNK interact with MKP5, they exhibit distinct modes of binding^29,30,31. p38 MAPK binds to the KBD of MKP5 whereas JNK utilizes a F/L-x-F motif within the catalytic domain of MKP5³¹. We now show that p38 MAPK and JNK differentially engage the active site cysteine since JNK fails to bind in context of the allosteric site/substrate-trapping mutant whereas p38 MAPK binding is retained. These results further support the observations of the distinct binding mechanisms of p38 MAPK and JNK to MKP5^20,21,29. These results demonstrate that the allosteric site represents a mechanism for establishing additional features of substrate selectivity between p38 MAPK and JNK for dephosphorylation which could influence downstream signaling. Indeed, we have recently demonstrated that MKP5 regulates TGFβ signaling through a JNK-, but not p38 MAPK-dependent, signaling mechanism³². Whether differential allosteric site engagement underlies these selectivity effects remains to be determined. Regardless, given that both p38 MAPK and JNK are predicted to alter catalytic site conformation, the extent to which allosteric site engagement regulates MKP5 activity will need to be determined. Furthermore, how p38 MAPK and JNK binding to this allosteric site is ordered in context of the KBD and Y/F-x-F motifs also requires further investigation. It is noteworthy that the allosteric site is conserved amongst all the active MKPs, including MKP7 where we have shown that the analogous allosteric site also contributes to both activity, p38 MAPK, and JNK binding¹⁹. Interestingly, MKP7 does not exhibit the observed differential binding with p38 MAPK and JNK via its active and allosteric site as seen for MKP5¹⁹. Thus, this allosteric site is likely to be a common site of MAPK binding amongst the MKPs but with differing contributions for each particular MAPK relative to active site engagement.

MKP5 has been implicated as a potential therapeutic target for the treatment of fibrosis in skeletal muscle, heart, and the lung^33,34,35. Given that there are limited treatments available for fibrotic tissue disease, the importance of understanding the molecular mechanisms of MKP5 allosteric inhibition is of significance. Collectively, this study provides molecular insight into the dynamic structural changes initiated upon ligand engagement of the MKP5 allosteric site. NMR and MD analyses reveal that mutations at the allosteric Y435 residue all result in disruption of the catalytic site at C408 and the general acid at D377 to varying degrees, providing direct evidence for allostery. MD simulations illuminate channels of communication between Y435 and the active site by critical non-covalent interactions. Finally, our data indicate that an additional feature of the MKP5 allosteric site is to bind MAPK. Collectively, these data uncover previously unappreciated modes of MKP5 regulation, which can be harnessed to inform allosteric inhibitor design.

Methods

Protein expression and purification

The WT and single-point mutants of the MKP5-CD were cloned into the pET28a vector with a TEV cleavage site and subsequently transformed into BL21 Gold (DE3) cells. Variant constructs were generated using Pfu Turbo DNA polymerase (Agilent). Protein expression and purification followed our previously reported protocol¹⁸, with minor modifications. Bacterial cells were induced using 1 mM IPTG and allowed to express the protein overnight at 20 °C. The cells were suspended in lysis buffer (20 mM HEPES, pH 7.4, 500 mM NaCl, 10% glycerol, 5 mM imidazole, 2 mM β-mercaptoethanol, DNase1, and complete EDTA-free protease inhibitor). After disruption, the soluble fraction was separated by centrifugation and loaded onto a prepacked Ni-NTA column, equilibrated in the lysis buffer. The protein was eluted with an imidazole gradient. Prior to TEV protease cleavage, eluent fractions were dialyzed against cleavage buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 2.5 mM CaCl₂). After complete digestion, Ni-NTA affinity chromatography was performed again to remove undigested protein fractions. Size exclusion chromatography was then carried out using a Hi-Load 16/600 Superdex75 GL column to obtain a pure, monodispersed sample. After concentration, the protein was stored at −80 °C in storage buffer (20 mM Tris, pH 8.0, 150 mM NaCl, 5% glycerol, 5 mM dithiothreitol).

Crystallization and structure determination

The MKP5-CD variant Y435W, at a concentration of 6–8 mg/ml, underwent co-crystallization with the Cmpd 1¹⁸ using the sitting-drop method. The protein was incubated overnight with Cmpd 1 (5 mM). Drops of 300 nl protein mixed with 200 nl reservoir solution were set using the Formulatrix NT8 drop setter in a 96-well plate. Crystals of the Y435W in complex with Cmpd 1 were grown in a buffer of 0.1 mM Tris, 300 mM NaCl, pH 8.5. X-ray crystallographic data sets were collected at Brookhaven Lab National Synchrotron Light Source II, at the AMX and FMX beamlines. Molecular replacement, employing either Molrep in the CCP4 software package or phaser³⁶, was carried out using the MKP5-CD structure in complex with Cmpd 1 (PDB: 6MC1) as the search model. The 2|Fo|- |Fc| and |Fo|- |Fc| Fourier maps were analyzed, revealing no electron density for any compounds other than Cmpd 1. Energy-minimized coordinates and Crystal Information Files (CIF) for Cmpd 1 were generated using Jligand³⁷. Structure refinement utilized Phenix.refine³⁸ and REFMAC5³⁹, with several rounds of refinement cycles and model building in Coot⁴⁰ to achieve the final structure. Stereochemistry assessment with MOLPROBITY⁴¹ confirmed good quality in the crystal structures. Data collection, scaling, and refinement statistics are summarized in (Supplementary Table 1). The structural coordinates of the complex have been deposited in the RCSB Protein Data Bank PDB (9BPN: apo-Y435W, 9BU4: Cmpd 1-Y435W).

Analytical ultracentrifugation

All MKP5-CD crystallized as dimers, but we used AUC to determine the oligomeric state in solution for interpreting biological studies at lower concentrations. MKP5 protein samples were prepared at three concentrations (20 μM, 10 μM, and 5 μM) with and without 100 μM Cmpd 1 and centrifuged at three speeds (6427 × g, 15,564 × g, and 25,722 × g) at 4 °C until equilibrium was reached using an AnTi-60 rotor in an XL-I analytical ultracentrifuge (Beckman), controlled using ProteomeLab XL-A/XL-I software (v. 6.2). In all experiments, radial measurements of absorbance at 280 nm (A₂₈₀), 250 nm (A₂₅₀) and 290 nm (A₂₉₀) were taken using step scan(s) from 5.8–7.3 cm at 0.003 cm intervals, with 5 replicates. Reference samples contained sample buffer and compound when appropriate. Data points in Supplementary Fig. 1 are raw AUC centrifugation data plotted as the natural logarithm (ln) A₂₈₀ transposed to the origin against a function of radius squared (r² − r_o²)/2, where r is the radial position in the sample and r_o is the radial position of the meniscus. For a single ideal species, this representation gives a straight line with slope proportional to its molecular mass. Dashed lines plotted in Supplementary Fig. 1 correspond to expected results for monomer, dimer and monomer bound to compound1. A global fit of all data for both with and without compound1 was done to a model of an ideal species of monomer. Global fits to a monomer-dimer model did not converge. Data was analyzed HeteroAnalysis v 1.1.60⁴² and GraphPad Prism v. 9.5.1.

NMR spectroscopy

MKP5-CD samples for NMR were expressed in M9 minimal media with ¹⁵NH₄Cl and ¹³C₆H₁₂O₆ (Cambridge Isotope Laboratories) as the sole nitrogen and carbon sources. The cells were cultured at 37 °C to an OD₆₀₀ of 0.8–1.0, induced with 1 mM IPTG, and then incubated at 20 °C for an additional 16–18 h. Isotopically labeled MKP5-CD samples were purified by the protocol described above. The samples were dialyzed into a buffer containing 40 mM sodium phosphate, 2 mM DTT, and 1 mM EDTA at pH 7.4. Samples were concentrated and 5% D₂O was added prior to NMR experiments. Backbone resonance assignment experiments were collected on an 800 μM MKP5-CD sample. For studies of Cmpd 1 binding, 0.1 mM MKP5-CD was incubated with Cmpd 1 (dissolved in 100% DMSO) at the molar amounts indicated (relative to 0.1 mM MKP5-CD) for 1 h at room temperature prior to data collection and compared to spectra of 0.1 mM MKP5-CD (apo).

All NMR experiments were collected on a Bruker Avance NEO 600 MHz spectrometer at 298 K, equipped with pulsed field gradients and a triple resonance cryoprobe, and using TopSpin (v. 4.0.9). NMR spectra were processed with NMRPipe (v. 11.3)⁴³ and analyzed in NMRFAM-SPARKY (v. 1.47 powered by Sparky v. 3.19)⁴⁴. The backbone resonance assignments were completed using a TROSY HSQC (trosyf3gpphsi19.2)^45,46,47 and the following TROSY triple resonance experiments: HNCA (trhncagp3d), HN(CO)CA (trhncocagp3d), HN(CA)CB (trhncacbgp3d), HN(COCA)CB (trhncocacbgp3d), HN(CA)CO (trhncacogp3d), and HNCO (trhncogp3d)^48,49. Three-dimensional correlations and resonance assignments were determined in CARA (version 1.9.1.7)⁵⁰. 37.3% of the MKP5-CD sequences remain unassigned due to line broadening that precludes reliable visualization of these resonances. Backbone assignments for the MKP5-CD variants were transferred from the WT MKP5-CD spectrum under buffer matched conditions. ¹H¹⁵N combined chemical shift perturbations (Δδ) were determined from ¹H¹⁵N TROSY HSQC spectra by Δ\(\delta=\sqrt{\frac{1}{2}(({\delta }_{\mathrm{HN}}^{2})+({\delta }_{N}^{2}/25))}\), where δ_HN and δ_N \(=|{\delta }_{\mathrm{WT}}-{\delta }_{\mathrm{variant}}|\). Significant chemical shift perturbations were determined as values 1.5σ above the 10% trimmed mean of all data sets. ¹H¹³C spectra were collected using a constant-time HSQC (ct-HSQC) pulse sequence (hsqcctetgpsp.2)⁵¹.

CD spectroscopy

MKP5-CD samples (10 μM) were dialyzed into a buffer containing 40 mM sodium phosphate and 1 mM EDTA at pH 7.4. Thermal unfolding circular dichroism (CD) experiments were collected on a JASCO J-815 spectropolarimeter equipped with a variable temperature Peltier device using a 2 mm quartz cuvette. Denaturation curves were recorded at 223 nm over a temperature range of 20–90 °C. T_m values were determined via nonlinear curve fitting in GraphPad Prism.

Molecular dynamics simulations

The structural models for the apo and Cmpd 1-bound protein of MKP5 were based on the crystal structure of MKP5 with (PDB: 6MC1)¹⁸ and without Cmpd 1 (PDB: 1ZZW)²¹. MD simulations were performed using the AMBER-ff19SB, and General Amber (GAFF) force field for the protein and ligand, respectively, as included in the Amber22 software package^52,53. We simulated the dynamics of eight distinct states of MKP5, including WT and three variants (Y435W, Y435A, Y435S). Additionally, WT and Y435W were simulated in the presence of Cmpd 1. MD simulations over 1 µs were carried out for each system. AmberTools22 were employed to define standard protonation states and complete hydration of the models using TIP3 water solvent molecules to ensure density values ≥0.9 mol Å⁻³⁵³.

MD simulations were prepared as follows: first, the solvent was minimized by restraining all atoms but water and ions at the positions defined by the PDB models. Structural constraints were then released, and the solvated structures were then gradually heated to 303 K, performing MD simulations (of at least 10 ns) in the canonical NVT ensemble using Langevin dynamics. Unconstrained MD simulations were carried out for 90 ns, resulting in a total pre-equilibration simulation time of about 100 ns. The pre-equilibrated model systems were then simulated in the NPT ensemble at 300 K and 1 atm, using Langevin dynamics for 1µs. All simulations were performed using periodic boundary conditions with a switching cutoff distance of 12 Å. Three independent replicas were run for WT and each variant in the unbound forms and WT and Y435W bound forms for a total of 18 µs simulations analyzed in this study.

The MD trajectories were analyzed using the MDiGest⁵⁴ and MDAnalysis⁵⁵ Python packages after sampling one snapshot per 0.1 ns (i.e., a total of 10,000 frames for each model system). The RMSD analysis and non-covalent interaction analyses were performed as implemented in MDiGest⁵⁴. For the latter, 1 frame per ns and the default angular and distance settings were used to estimate the percentage of frames in which each interaction was present⁵⁴.

Allosteric traffic maps

Traffic plots were generated using a novel implementation of Dijkstra’s algorithm⁴⁹ to identify the shortest pathways between every pair of residues in the protein. In this approach, the protein was modeled as a graph where each amino acid is a node, and the edges connecting them were weighted according to generalized correlation values derived from backbone and side-chain dihedral fluctuations from MD trajectories. The mutual information matrices were calculated using sine and cosine projections of the dihedral angles at each trajectory step, which were subsequently used to compute a generalized correlation matrix, as implemented in MDigest⁴⁷.

This matrix was made sparse to create a network for the traffic analysis: only those edges between residue pairs with Cα distances below 6 Å in over 75% of the simulation frames were retained. A −log transformation was applied to the matrix so that highly correlated residue pairs had shorter edge distances, while less correlated pairs (values closer to 0) had longer edge distances. This graph, which combines proximity and correlation information, was then analyzed by computing the shortest pathways between every pair of residues using Dijkstra’s algorithm⁴⁹. The resulting traffic was quantified using normalized edge and node betweenness centralities, scaled between 0 and 1 for consistent comparison across graphs. For visualization purposes, a threshold of 0.3 was set to highlight the connections with the most significant influence on the protein conformational dynamics, and these values were mapped onto the protein structure.

Generation of MKP5 mutant expression vectors and cell transfections

Generation of MKP5 mutant variants, MKP5-Y535S, MKP5-Y435A, and MKP5-Y435W have all been described previously¹⁸. Cell-based assays for MAPK binding experiments utilized COS-7 cells grown at 37 °C at 5% CO₂. Cells were maintained in Dulbecco’s modified eagle’s medium (DMEM) (cat# 11965092) supplemented with 10% fetal bovine serum (FBS), 1% sodium pyruvate (Gibco cat# 11360070), and 1% penicillin-streptomycin (Gibco cat# 15140122). COS-7 cells were seeded at a density of 8.5 × 10⁵ in 100 mm dishes and maintained in DMEM supplemented with 10% FBS for 24 h before transfection. Cells were co-transfected with 6.8 µg DNA of either pEBG vector (expressing GST alone) or full-length pEBG-GST-MKP5 (WT or variants), and 3.4 µg of either pcDNA3-HA-p38α MAPK or pEGFP-JNK1β using FuGENE transfection reagent as per manufacturer’s instructions. Cells were transfected for 48 h in 10% FBS DMEM before harvest.

Immunoblotting and immunoprecipitation

Cells were lysed on ice using Pierce^TM IP lysis buffer (cat# 87787). Lysates were affinity purified using GSH-Sepharose beads. For sorbitol stimulation, cells were transfected with 8.5 μg DNA of either pcDNA3.1 or GST-MKP5 (WT or mutants) using FuGENE transection reagent as per manufacturer’s instructions. Twenty four hours after transfection, cells were washed once with PBS and starved overnight in 0.1% FBS in DMEM. Next day, cells were washed with PBS followed by incubation with 500 mM sorbitol in 0.1% FBS in DMEM for 30 min before lysis followed by immunoblot analysis for detection of endogenous p38 MAPK, JNK1/2, and ERK1/2. Affinity purified complexes containing full-length MKP5 (WT or variants) were separated by boiling and complexed HA-p38α MAPK or pEGFP-JNK1β proteins detected by immunoblotting using either anti-HA (3F10, Roche cat# 121518167001) or anti-GFP (Cell Signaling Technologies cat# 2955) antibodies. GST input controls were detected using anti-GST antibodies. For the assessment of MAPK phosphorylation, cell lysates were clarified by centrifugation, and proteins were resolved by SDS-PAGE and transferred to nitrocellulose membranes by semi-dry transfer for 1 h followed by immunoblot analysis for detection of endogenous phosphorylated p38 MAPK, JNK1/2, and ERK1/2 using anti-phospho-p38 MAPK (Cell Signaling Cat# 9215), anti-phospho-JNK1/2 (Cell Signaling cat#4668) and anti-phospho-ERK1/2 (Cell Signaling cat# 9101) antibodies. Total levels of MAPKs were assessed by immunoblotting of re-probed membranes using anti-p38α MAPK (Santa Cruz cat# 81621), anti-JNK1/2 (Santa Cruz cat# 3708), and anti-ERK1/2 (Cell Signal cat# 9107) antibodies.

Statistical analyses

No statistical methods were used to predetermine sample size. The number of samples used in each experiment is shown. All cell-based experiments were performed at least three times independently. Statistical analysis and graphing were performed using GraphPad Prism 9.4.1 software. We did not estimate variations in the data. The variances are similar between the groups that are being statistically compared. All data represent the mean ± standard errors of the means (SEM). For p value determinations, we used one-way or two-way ANOVA with multiple comparisons.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Unless otherwise stated, all data supporting the results of this study can be found in the article, supplementary, and source data files. The MKP5 structures have been deposited to PDB under accession numbers 9BPN and 9BU4 (with inhibitor) [https://www.rcsb.org/structure/9BU4]. The raw diffraction data for the MKP5 variant has been deposited under accession number 9BPN and 9BU4 (with inhibitor) [https://proteindiffraction.org/project/MKP5-Y435W-allosteric-inhibitor_9bu4/]. NMR backbone resonance assignments of MKP5-CD generated in this study have been deposited in the BMRB database under accession code #53231 [https://doi.org/10.13018/BMR53231]. NMR chemical shift and peak height data generated in this study are provided in the Supplementary Information/Source Data file. The raw thermal unfolding data generated in this study via CD spectroscopy are provided in the Supplementary Information/Source Data file. Source data are provided with this paper.

Code availability

All codes used in the paper are publicly available at Github [https://github.com/fmaschietto/mdigest].

References

Hotamisligil, G. S. & Davis, R. J. Cell signaling and stress responses. Cold Spring Harb. Perspect. Biol. 8, a006072 (2016).
Article PubMed PubMed Central Google Scholar
Cargnello, M. & Roux, P. P. Activation and function of the MAPKs and their substrates, the MAPK-activated protein kinases. Microbiol. Mol. Biol. Rev. 75, 50–83 (2011).
Article CAS PubMed PubMed Central Google Scholar
Pearson, G. et al. Mitogen-activated protein (MAP) kinase pathways: regulation and physiological functions. Endocr. Rev. 22, 153–183 (2001).
CAS PubMed Google Scholar
Caunt, C. J. & Keyse, S. M. Dual-specificity MAP kinase phosphatases (MKPs): shaping the outcome of MAP kinase signalling. FEBS J. 280, 489–504 (2013).
Article CAS PubMed PubMed Central Google Scholar
Dickinson, R. J. & Keyse, S. M. Diverse physiological functions for dual-specificity MAP kinase phosphatases. J. Cell Sci. 119, 4607–4615 (2006).
Article CAS PubMed Google Scholar
Murphy, L. O., MacKeigan, J. P. & Blenis, J. A network of immediate early gene products propagates subtle differences in mitogen-activated protein kinase signal amplitude and duration. Mol. Cell Biol. 24, 144–153 (2004).
Article CAS PubMed PubMed Central Google Scholar
Caunt, C. J., Sale, M. J., Smith, P. D. & Cook, S. J. MEK1 and MEK2 inhibitors and cancer therapy: the long and winding road. Nat. Rev. Cancer 15, 577–592 (2015).
Article CAS PubMed Google Scholar
Shillingford, S. R. & Bennett, A. M. Mitogen-activated protein kinase phosphatases: no longer undruggable? Annu. Rev. Pharm. Toxicol. 63, 617–636 (2023).
Article CAS Google Scholar
Jeffrey, K. L., Camps, M., Rommel, C. & Mackay, C. R. Targeting dual-specificity phosphatases: manipulating MAP kinase signalling and immune responses. Nat. Rev. Drug Discov. 6, 391–403 (2007).
Article CAS PubMed Google Scholar
Zandi, Z. et al. Dual-specificity phosphatases: therapeutic targets in cancer therapy resistance. J. Cancer Res Clin. Oncol. 148, 57–70 (2022).
Article PubMed PubMed Central Google Scholar
Hota, A. & Bennett, A. M. MAP kinase phosphatases in metabolic diseases. Trends Endocrinol. Metab. 40555575 https://doi.org/10.1016/j.tem.2025.05.002 (2025).
Raman, M., Chen, W. & Cobb, M. H. Differential regulation and properties of MAPKs. Oncogene 26, 3100–3112 (2007).
Article CAS PubMed Google Scholar
Slack, D. N., Seternes, O. M., Gabrielsen, M. & Keyse, S. M. Distinct binding determinants for ERK2/p38α and JNK MAP kinases mediate catalytic activation and substrate selectivity of MAP kinase phosphatase-1. J. Biol. Chem. 276, 16491–16500 (2001).
Article CAS PubMed Google Scholar
Yuvaniyama, J., Denu, J. M., Dixon, J. E. & Saper, M. A. Crystal structure of the dual specificity protein phosphatase VHR. Science 272, 1328–1331 (1996).
Article ADS CAS PubMed Google Scholar
Zhao, Y. & Zhang, Z.-Y. The mechanism of dephosphorylation of extracellular signal-regulated kinase 2 by mitogen-activated protein kinase phosphatase 3. J. Biol. Chem. 276, 32382–32391 (2001).
Article CAS PubMed Google Scholar
Camps, M. et al. Catalytic activation of the phosphatase MKP-3 by ERK2 mitogen-activated protein kinase. Science 280, 1262–1265 (1998).
Article ADS CAS PubMed Google Scholar
Muda, M. et al. The mitogen-activated protein kinase phosphatase-3 N-terminal noncatalytic region is responsible for tight substrate binding and enzymatic specificity. J. Biol. Chem. 273, 9323–9329 (1998).
Article CAS PubMed Google Scholar
Gannam, Z. T. K. et al. An allosteric site on MKP5 reveals a strategy for small-molecule inhibition. Sci. Signal 13, eaba3043 (2020).
Article CAS PubMed PubMed Central Google Scholar
Shillingford, S. et al. A novel site on dual-specificity phosphatase MKP7/DUSP16 is required for catalysis and MAPK binding. J. Biol. Chem. 298, 102617 (2022).
Article CAS PubMed PubMed Central Google Scholar
Jeong, D. G. et al. Crystal structure of the catalytic domain of human MAP kinase phosphatase 5: structural insight into constitutively active phosphatase. J. Mol. Biol. 360, 946–955 (2006).
Article CAS PubMed Google Scholar
Tao, X. & Tong, L. Crystal structure of the MAP kinase binding domain and the catalytic domain of human MKP5. Protein Sci. 16, 880–886 (2007).
Article CAS PubMed PubMed Central Google Scholar
Chen, Y. P. et al. Allosteric inhibition of SHP2 phosphatase inhibits cancers driven by receptor tyrosine kinases. Nature 535, 148–152 (2016).
Article ADS MathSciNet CAS PubMed Google Scholar
Krishnan, N. et al. Targeting the disordered C terminus of PTP1B with an allosteric inhibitor. Nat. Chem. Biol. 10, 558–566 (2014).
Article CAS PubMed PubMed Central Google Scholar
Gannam, Z. T. K. et al. Defining the structure-activity relationship for a novel class of allosteric MKP5 inhibitors. Eur. J. Med. Chem. 243, 114712 (2022).
Article CAS PubMed PubMed Central Google Scholar
Korotchenko, V. N. et al. In vivo structure-activity relationship studies support allosteric targeting of a dual specificity phosphatase. ChemBioChem 15, 1436–1445 (2014).
Article CAS PubMed PubMed Central Google Scholar
Molina, G. et al. Zebrafish chemical screening reveals an inhibitor of Dusp6 that expands cardiac cell lineages. Nat. Chem. Biol. 5, 680–687 (2009).
Article CAS PubMed PubMed Central Google Scholar
Tanoue, T., Moriguchi, T. & Nishida, E. Molecular cloning and characterization of a novel dual specificity phosphatase, MKP-5. J. Biol. Chem. 274, 19949–19956 (1999).
Article CAS PubMed Google Scholar
Lu, C. et al. Structural and dynamic insights into the mechanism of allosteric signal transmission in ERK2-mediated MKP3 activation. Biochemistry 56, 6165–6175 (2017).
Article CAS PubMed Google Scholar
Zhang, Y. Y., Wu, J. W. & Wang, Z. X. A distinct interaction mode revealed by the crystal structure of the kinase p38alpha with the MAPK binding domain of the phosphatase MKP5. Sci. Signal 4, ra88 (2011).
Article PubMed Google Scholar
Theodosiou, A., Smith, A., Gillieron, C., Arkinstall, S. & Ashworth, A. MKP5, a new member of the MAP kinase phosphatase family, which selectively dephosphorylates stress-activated kinases. Oncogene 18, 6981–6988 (1999).
Article CAS PubMed Google Scholar
Liu, X. et al. A conserved motif in JNK/p38-specific MAPK phosphatases as a determinant for JNK1 recognition and inactivation. Nat. Commun. 7, 10879 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Dorry, S., Perla, S. & Bennett, A. M. Mitogen-activated protein kinase phosphatase-5 is required for TGF-beta signaling through a JNK-dependent pathway. Mol. Cell. Biol. 45, 17–31 (2025).
Shi, H. et al. Improved regenerative myogenesis and muscular dystrophy in mice lacking Mkp5. J. Clin. Investig. 123, 2064–2077 (2013).
Article CAS PubMed PubMed Central Google Scholar
Xylourgidis, N. et al. Role of dual-specificity protein phosphatase DUSP10/MKP-5 in pulmonary fibrosis. Am. J. Physiol. Lung Cell. Mol. Physiol. 317, L678–L689 (2019).
Article CAS PubMed PubMed Central Google Scholar
Zhong, C. et al. MAP kinase phosphatase-5 deficiency protects against pressure overload-induced cardiac fibrosis. Front. Immunol. 12, 790511 (2021).
Article CAS PubMed PubMed Central Google Scholar
Vagin, A. & Teplyakov, A. Molecular replacement with MOLREP. Acta Crystallogr. D. Biol. Crystallogr. 66, 22–25 (2010).
Article ADS CAS PubMed Google Scholar
Lebedev, A. A. et al. JLigand: a graphical tool for the CCP4 template-restraint library. Acta Crystallogr. D. Biol. Crystallogr. 68, 431–440 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Terwilliger, T. C. et al. Iterative model building, structure refinement and density modification with the PHENIX AutoBuild wizard. Acta Crystallogr. D. Biol. Crystallogr. 64, 61–69 (2008).
Article ADS CAS PubMed Google Scholar
Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D. Biol. Crystallogr. 53, 240–255 (1997).
Article ADS CAS PubMed Google Scholar
Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. D. Biol. Crystallogr. 60, 2126–2132 (2004).
Article ADS PubMed Google Scholar
Chen, V. B. et al. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D. Biol. Crystallogr. 66, 12–21 (2010).
Article ADS CAS PubMed Google Scholar
Cole, J. L. Analysis of heterogeneous interactions. in Methods in Enzymology (Academic Press, 2004).
Delaglio, F. et al. NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293 (1995).
Article CAS PubMed Google Scholar
Lee, W., Tonelli, M. & Markley, J. L. NMRFAM-SPARKY: enhanced software for biomolecular NMR spectroscopy. Bioinformatics 31, 1325–1327 (2015).
Article PubMed Google Scholar
Rance, M., Loria, J. P. & Palmer, A. G. Sensitivity improvement of transverse relaxation-optimized spectroscopy. J. Magn. Reson. 136, 92–101 (1999).
Article ADS CAS PubMed Google Scholar
Pervushin, K. V., Wider, G. & Wüthrich, K. Single transition-to-single transition polarization transfer (ST2-PT) in [15N,1H]-TROSY. J. Biomol. NMR 12, 345–348 (1998).
Article CAS PubMed Google Scholar
Czisch, M. & Boelens, R. Sensitivity enhancement in the TROSY experiment. J. Magn. Reson. 134, 158–160 (1998).
Article ADS CAS PubMed Google Scholar
Salzmann, M., Pervushin, K., Wider, G., Senn, H. & Wuthrich, K. TROSY in triple-resonance experiments: new perspectives for sequential NMR assignment of large proteins. Proc. Natl. Acad. Sci. USA 95, 13585–13590 (1998).
Article ADS CAS PubMed PubMed Central Google Scholar
Pervushin, K., Riek, R., Wider, G. & Wuthrich, K. Attenuated T2 relaxation by mutual cancellation of dipole-dipole coupling and chemical shift anisotropy indicates an avenue to NMR structures of very large biological macromolecules in solution. Proc. Natl. Acad. Sci. USA 94, 12366–12371 (1997).
Article ADS CAS PubMed PubMed Central Google Scholar
Keller, R. L. The Computer Aided Resonance Assignment Tutorial (Cantina Verlag, 2004).
Vuister, G. W. & Bax, A. Measurement of two-bond JCOHα coupling constants in proteins uniformly enriched with13C. J. Biomol. NMR 2, 401–405 (1992).
Article CAS PubMed Google Scholar
Tian, C. et al. ff19SB: amino-acid-specific protein backbone parameters trained against quantum mechanics energy surfaces in solution. J. Chem. Theory Comput. 16, 528–552 (2020).
Article PubMed Google Scholar
Case, D. et al. AMBER2020, University of California, San Fransisco. J. Am. Chem. Soc. 142, 3823–3835 (2020).
Google Scholar
Maschietto, F., Allen, B., Kyro, G. W. & Batista, V.S. MDiGest: a Python package for describing allostery from molecular dynamics simulations. J. Chem. Phys. 158, 215103 (2023).
Gowers, R. J. et al. MDAnalysis: a Python package for the rapid analysis of molecular dynamics simulations. In Conference: Proc. 15th Python in Science Conference(SCIPY 2016), 2016-07-11–2016-07-11 (2019).

Download references

Acknowledgements

This work was partially supported by NIH grant R01 GM144451 to G.P.L. A.M.B. and E.L. were supported by R01 HL158876, and A.M.B. was supported by R01 AR080152. Simulations were partially executed on resources supported by the Simons Center for Computational Physical Chemistry (SCCPC) at NYU (SF Grant No. 839534). F.M. acknowledges support from a postdoctoral fellowship awarded by the SCCPC at NYU. Part of the simulations were executed on the Yale IT High-Performance Commuting resources, which are additionally thanked for services and expertise.

Author information

Shanelle Shillingford
Present address: Boston Consulting Group, Summit, NJ, USA
These authors contributed equally: Erin Skeens, Federica Maschietto, Ramu Manjula, Shanelle Shillingford.

Authors and Affiliations

Department of Molecular Biology, Cell Biology and Biochemistry, Brown University, Providence, RI, USA
Erin Skeens & George P. Lisi
Department of Chemistry, New York University, New York, NY, USA
Federica Maschietto
Simons Center for Computational Physical Chemistry at New York University, New York, NY, USA
Federica Maschietto
Department of Chemistry, Yale University, New Haven, CT, USA
Federica Maschietto, Shanelle Shillingford & Victor S. Batista
Department of Pharmacology, Yale School of Medicine, Yale University School of Medicine, New Haven, CT, USA
Ramu Manjula, Shanelle Shillingford, James Murphy, Elias J. Lolis & Anton M. Bennett
Yale Center for Molecular and Systems Metabolism, Yale University School of Medicine, New Haven, CT, USA
Anton M. Bennett

Authors

Erin Skeens
View author publications
Search author on:PubMed Google Scholar
Federica Maschietto
View author publications
Search author on:PubMed Google Scholar
Ramu Manjula
View author publications
Search author on:PubMed Google Scholar
Shanelle Shillingford
View author publications
Search author on:PubMed Google Scholar
James Murphy
View author publications
Search author on:PubMed Google Scholar
Elias J. Lolis
View author publications
Search author on:PubMed Google Scholar
Victor S. Batista
View author publications
Search author on:PubMed Google Scholar
Anton M. Bennett
View author publications
Search author on:PubMed Google Scholar
George P. Lisi
View author publications
Search author on:PubMed Google Scholar

Contributions

E.S.: expressed and purified MKP5-CD proteins, performed NMR and CD experiments and associated data analysis; F.M.: performed MD simulations and associated data analysis; R.M.: expressed and purified MKP5-CD proteins, performed X-ray crystallography and associated data analysis; S.S.: performed MKP5 cellular experiments and associated data analysis; J.M. performed analytical ultracentrifugation experiments; E.J.L.: conceived the study, supervised X-ray crystallography; V.S.B.: conceived the study, supervised MD simulations; A.M.B.: conceived the study, supervised cellular experiments; G.P.L.: conceived the study, supervised NMR spectroscopy. The manuscript was written through contributions of all authors.

Corresponding authors

Correspondence to Elias J. Lolis, Victor S. Batista, Anton M. Bennett or George P. Lisi.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Christopher McClendon, Hongwei Yao, and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Reporting Summary

Transparent Peer Review file

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Skeens, E., Maschietto, F., Manjula, R. et al. Dynamic and structural insights into allosteric regulation on MKP5 a dual-specificity phosphatase. Nat Commun 16, 7011 (2025). https://doi.org/10.1038/s41467-025-62150-w

Download citation

Received: 31 August 2024
Accepted: 11 July 2025
Published: 31 July 2025
DOI: https://doi.org/10.1038/s41467-025-62150-w