Proc Natl Acad Sci U S A 108(30): 12289-12294

PMC: PMC3145719

PMID: 21746900

Stereospecific gating of functional motions in Pin1

Results

Comparison of Ligand Binding Modes.

To define the ligand binding modes, we measured amide ^N-^{H chemical shift perturbations (CSPs) in Pin1 caused by adding saturating amounts of ligand (}Fig. 2 and SI Text). The substrate FFpSPR produced CSPs in both the PPIase and WW domains, with a strong bias toward the WW domain. This bias echoes other Pin1 substrates (4). By contrast, the CSPs of the locked inhibitors were heavily biased toward the PPIase domain, consistent with their competitive inhibition of PPIase activity (8). The cis inhibitor bound the isolated PPIase domain (40–163, vide infra) of Pin1 more tightly than the trans- (average An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq1.jpg , average ) (Table S1).

Fig. 2.

NH CSPs of the three ligands. (A) Substrate FFpSPR; (B) cis locked inhibitor; and (C) trans locked inhibitor. Significant CSPs were those > 0.05 ppm (horizontal dashed line). Except for H27, the significant cis CSPs (B) are just PPIase domain residues, whereas the trans CSPs (C) include residues of both domains.

The two inhibitors showed different domain binding preferences (Fig. 2). The trans inhibitor gave significant CSPs (> 0.05 ppm) in both the WW and PPIase domain, with the latter perturbations being more pronounced. The K_d values were similar for both domains (average An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq3.jpg , average ). By contrast, the cis inhibitor CSPs resided almost exclusively within the PPIase domain (average ); the only significant WW CSP was to H27 at the domain interface (Table S1). Thus, whereas trans inhibitor can bind either domain, the cis is a selective PPIase domain binder.

Within the PPIase domain, the cis and trans inhibitors gave distinct CSP signatures that we exploited in competition binding experiments. We initially saturated Pin1 with the trans inhibitor, and, as expected, we observed its CSP signature in both the PPIase and WW domains. Subsequent additions of the cis inhibitor gradually transformed the PPIase CSP signature from that of trans to cis, whereas the WW domain retained the trans signature (Fig. 3). The converse titration of trans into a preexisting solution of Pin1/cis produced the same results. These results show that Pin1 can bind simultaneously the cis conformer in the PPI domain, and the trans conformer in the WW domain.

Fig. 3.

Competition of the cis and trans locked inhibitors for binding to the WW (1–39), linker (40–49), and PPIase (50–163) domains illustrated by amide ^{H line-shapes from}^N-^{H cross-peaks. (}A rows): Pin1 saturated with the cis inhibitor (black traces and vertical blue solid lines) or the trans inhibitor (red traces and vertical dashed magenta lines). (B rows): Chemical shifts under simultaneous saturation of cis and trans: WW binds the trans inhibitor whereas the PPIase binds the cis inhibitor.

PPIase Activity Is Modulated by Remote Interactions at the Domain Interface.

Previous structural studies (9) showed that pS/T phospho-peptide substrates had slightly different binding affinities for the isolated PPIase domain versus full-length Pin1, suggesting the modulation of PPIase binding affinity by interdomain interactions. The locked inhibitors invited further investigation of this effect, with the advantage that we could now interpret the results stereospecifically.

We therefore generated a truncated Pin1 construct (Pin1-PPI, 40–163) lacking the WW domain (residues 1–39). The ^N-^{H spectra of Pin1-PPI and full-length Pin1 were nearly superimposable. Deviations were at the interdomain interface (F134, S138, F139, A140, L141, R142, S147, and I159), as with earlier investigations (}10). The WW domain thus acts as another ligand of the PPIase domain, with a binding surface defined, in part, by these interfacial residues. As in earlier studies, Pin1-PPI retained isomerase activity (5, 11, 12); however, it was higher than full-length Pin1 (13) by a factor of approximately 4.5. Similar behavior was observed with Pin1 substrates derived from the Tau protein (11).

We compared the inhibitor binding modes to Pin1-PPI versus full-length Pin1 by comparing the resulting ^N-^{H CSPs. The}cis inhibitor gave nearly identical CSPs for Pin1-PPI and full-length Pin1. By contrast, the trans CSPs differed between Pin1-PPI and full-length Pin1 at the domain interface (F139 and A140), the catalytic loop (H64, S71), and the substrate proline binding pocket (R127, G128, and Q131).

Unexpectedly, the CSPs of H59, S67, R119, and Q129 revealed slightly higher affinity (two- to fourfold smaller K_d values) of the cis inhibitor for the truncated Pin1-PPI over full-length Pin1 (Table S1). Except for R119, all of these residues are part of PPIase active site (14). This affinity mismatch cannot be due to direct binding to the WW domain because the cis inhibitor is a selective PPIase domain binder (vide supra). Instead, the K_ds suggest that interdomain interactions at one site (the WW domain interaction surface of the PPIase domain) can alter binding properties at a remote site—the substrate pocket on the opposite side of the PPIase domain.

Binding-Induced Changes of Side-Chain Flexibility.

We mapped the changes in Pin1 side-chain dynamics on the subnanosecond time scale caused by interactions with the substrate FFpSPR and the two locked inhibitors. This involved measuring deuterium (^{D) longitudinal (}R₁(^{D)) and transverse (}R_1ρ(^{D)) relaxation rate constants for the CH}₂D isotopomers of methyl-bearing side-chains in U-^N/^{C, 50%}^{D Pin1 at two static field strengths, 16.4 and 18.8 T (example spectra in}Fig. S1). Measurements of isolated (apo) Pin1 came from our previous studies (6).

The deuterium R₁(^{D) and}R_1ρ(^{D) values report on the subnanosecond (< 10}^{s) motions of the methyl}^C-^{D bond relative to the external field}B₀ (15). We used the Lipari–Szabo formalism (16 –18) to extract motional parameters from the R₁(^{D) and}R_1ρ(^{D) values. The Lipari–Szabo parameters describe}^C-^{D bond motions due to conformational dynamics via site-specific order parameters} An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq6.jpg , and effective correlation times, τ_e. is a pure number between 0 and 1 that measures the amplitude of the reorientational motions experienced by the methyl ^C-^{C symmetry axes.}decreases with increasing motional amplitude. The correlation time τ_e estimates the rapidity of the motions underlying An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq9.jpg .

Determining An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq10.jpg and τ_e requires prior characterization of overall Pin1 tumbling. Hence, we measured backbone amide ^{N relaxation parameters (}R₁(^N),R₂(^{N) and steady-state}^N-^{H NOE), followed by reduced spectral density mapping (}19 –21). This gave for each amide nitrogen-hydrogen (NH) bond two model-independent parameters of reorientational motion, J(0) and J(ω_N); both are highly sensitive to overall tumbling. The J(0) and J(ω_N) values consistently partitioned into two clusters, corresponding to the WW and PPIase domains (Fig. S2). This clustering suggested an approximate approach of using domain-specific, isotropic correlation times for overall tumbling (SI Text). Thus, after excluding NHs with outlying J(0) values due to substantial subnanosecond or μs-ms dynamics (Figs. S3 and S4) we determined τ_m,WW and τ_m,PPIase for each Pin1 state, using the corresponding R₂(^N)/R₁(^{N) ratios (}22).

We then fit the side chain deuterium R₁(^{D) and}R_1ρ(^{D) rates to the Lipari–Szabo spectral density function (}SI Text, Eq. S7) using An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq11.jpg and τ_e as adjustable parameters. After excluding methyls suffering resonance overlap or poor signal-to-noise, we could fit a common set of 53 methyl groups for apo Pin1 and the three complexes (Tables S2 and S3). The majority of these methyls (43) are in the PPIase domain. The An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq12.jpg values ranged from 0.08 to 0.98, and had an average value and uncertainty of 0.77 ± 0.03. The effective internal motion τ_e values ranged from 10 to 260 ps/rad, with an average value and uncertainty of 55 ± 4 ps/rad.

The formation of each Pin1-ligand complex caused changes in side-chain flexibility given by An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq13.jpg . Fig. 4 maps onto the X-ray crystal structure Protein Data Bank (PDB) code 1PIN (14) (corresponding bar-graphs are in Fig. S5). Positive indicates a loss of flexibility (red spheres), whereas negative indicates a gain of flexibility (blue spheres). The magnitudes of the An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq17.jpg were typically approximately 0.05–0.08, with the largest changes approximately ± 0.3.

Fig. 4.

Colored spheres showing An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq34.jpg for the methyl-bearing side chains of Pin1 caused by binding (A) FFpSPR; (B) cis locked inhibitor; and (C) trans locked inhibitor. (D) The for the truncated Pin1-PPI upon binding the cis locked inhibitor. The color code is An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq36.jpg (stiffening), (loosening), white (no significant change). The color intensity scales with the magnitude of .

To compare patterns of An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq18.jpg across the Pin1–ligand interactions, we wanted a metric sensitive to: (i) prominent changes; (ii) both positive and negative (flexibility losses and gains). Thus, we used a correlation coefficient

[1]

where the “i” and “j” refer to distinct Pin1 ligands. The summation index “k” runs over the methyls common to the two Pin1 interaction complexes. Methyls with similar An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq21.jpg , in both magnitude and sign, contribute positively to r(i,j), and r(i,j) values closer to 1.0 (the maximum value) indicate greater similarity. Uncertainties in r(i,j) were estimated using the standard error in the correlation coefficient (23) (SI Text).

Substrate Binding Reproduces the Hydrophobic Conduit.

Pin1 side-chain flexibility both increased ( An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq22.jpg ) and decreased () upon interaction with the substrate FFpSPR. Decreases occurred mainly for PPIase methyls within or adjacent to the substrate active site (L61δ₁, I78δ₁, L122δ₁, A124β, M130ϵ, and I156γ₂), and the domain interface (e.g., I28δ₁, I93δ₁, L141δ₁/δ₂, and V150γ₁) (Fig. 4). These stiffening side chains formed an internal conduit, similar to that seen for Cdc25 (6). This was supported by r(FFpSPR,Cdc25) = 0.66 ± 0.08. These similar changes in flexibility are reasonable, because both ligands are substrates. Yet, there were also local dissimilarities. Cdc25 caused greater stiffening of the conserved hydrophobic cluster, L60, L61, and V62 near the PPIase active site (FFpSPR showed significant stiffening only at L61), whereas FFpSPR showed greater stiffening in the substrate pocket (L122δ₁, A124β, and M130ϵ). Thus, the details of the conduit response may depend on whether the substrate is pT versus pS, and on the residues flanking the central pS/T-P core.

Inhibitor Binding Indicates Stereoselective changes in Flexibility.

The cis locked inhibitor caused the most extensive loss of side-chain flexibility. By contrast, the trans locked inhibitor caused a more heterogeneous pattern of An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq24.jpg , which included a greater number of flexibility increases (loosening). To better compare their dynamic effects, we correlated their effects with those of their parent substrate FFpSPR, and found r(FFpSPR,TRANS) = 0.19 ± 0.14, whereas r(FFpSPR,CIS) = 0.64 ± 0.08. These disproportionate r-values suggested that the cis inhibitor reproduced the substrate conduit response more prominently than the trans inhibitor (compare to Fig. 4). The Pin1-transr-value is lower in part because it lacks contributions from active site methyls L122δ₁/δ₂; they were nearly gone in the Pin1-trans deuterium relaxation spectra due to severe ^{C exchange broadening. Such broadening was absent for both Pin1-}cis and Pin1-FFpSPR. Such differential broadening provides another example of the different dynamic responses provoked by the cis versus trans inhibitors, and of the greater dynamical similarity of Pin1- cis (over Pin1-trans) to Pin1-FFpsSPR.

Further evidence for dynamical divergence between Pin1-cis and Pin1-trans came from the differences in their An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq25.jpg values (Fig. S6). Significant differences occurred at the proline binding pocket (L60δ₁, V62γ₂, M130ϵ, I158δ₁), the catalytic loop (I78δ₁, T79γ₂), and the domain interface (I89γ₂, I93δ₁, V150γ₁, T152γ₂, T162γ₂). Moreover, many of these same residues showed divergent cis/trans responses in other dynamical parameters including: (i) backbone NH J(0) values (ii) An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq26.jpg R₁ ∗ R₂ values for Pin1-cis versus Pin1-trans (Fig. S6). Thus, multiple parameters show that changes in Pin1 flexibility are sensitive to the cis versus trans configuration of the ligand.

Collectively, the FFpSPR substrate and corresponding cis- and trans-locked inhibitors demonstrated that: (i) the flexibility changes in the conduit—marked by stiffening of the more buried side chains—is an intrinsic response of Pin1 to substrate binding; (ii) the response is stereoselective—the cis conformer is the more powerful conduit stimulator.

WW Domain Binding Modulates the Conduit Response.

The cis inhibitor has negligible interactions with the WW domain (vide supra). Yet, deleting the WW domain (yielding Pin1-PPI) changed the binding affinity of the cis inhibitor at the remote PPIase active site. This provoked our interest in the conduit response in the absence of the WW domain. We therefore mapped An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq27.jpg for Pin1-PPI caused by binding of the cis locked inhibitor (Fig. 4). The values were similar to that of full-length Pin1, albeit less prominent, with an overall correlation of r(CIS : Pin1,CIS : Pin1-PPI) = 0.47 ± 0.11. Thus, whereas the conduit response is principally encoded within the PPIase domain, interdomain interactions can modulate its magnitude.

Dynamic Changes Conserved Across the Pin1-Ligand Complexes.

To define more precisely the flexibility changes important for binding, we pinpointed the recurrent changes across four interactions including the three ligands herein, and our previous Cdc25 study (6). For example, ligand binding caused consistent loosening ( An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq29.jpg ) at methyl axes of A53β, V55γ₁, T79γ₂, and A85β. The first two methyls are at the C-terminal end of the disordered linker connecting the WW and PPIase domains. The latter two are just after the flexible catalytic loop, at the N-terminal end of the long α helix -1. The increased mobility of these side chains may reflect loosening at a “flexure point” to facilitate rearrangements of the catalytic loop upon ligand binding. Such rearrangements are evident in X-ray crystal structures (9, 14), and implicated by our previous NMR exchange—broadening measurements (6).

Ligand binding caused consistent stiffening ( An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq30.jpg ) at methyl axes of L61δ₁, I93δ₁, A124β, L141δ₂, and I156γ₂. Methyl axes stiffening for at least three interactions included L60δ₂, L106δ₂, M130ϵ, V150γ₁, and I159δ₁. Critically, these methyls (except for A124β) are within or adjacent to the substrate binding pocket (L60, L61, M130, I156, and I159), or the interdomain interface (I93, L141, V150). Moreover, all of these methyls belong to residues of the proposed hydrophobic conduit, and thus, strengthen the notion that the conduit stiffening is endemic to the ligand binding process of Pin1.

Several dimethyl residues (ILV) gave anticorrelated changes in flexibility upon ligand binding. In particular, L122δ₂, V150γ₂, and I156δ₁ consistently loosened upon ligand binding, whereas their stereo-related partners L122δ₁, V150γ₁, and I156δ₂ consistently stiffened. These anticorrelated changes suggest an asymmetric reorganization of local steric interactions. The asymmetry could enhance contacts with one methyl, decreasing the mobility of its symmetry axis, while simultaneously enhancing the mobility of the other. L122δ₁ and δ₂ are candidates for this scenario, because they make nonequivalent van der Waals contacts with the substrate proline ring (14).

Residues Showing Dynamic Changes Are Highly Conserved.

If the above changes in side-chain flexibility are relevant for Pin1 substrate binding, then we expect conservation of the constituent residues among Pin1 homologs. We identified 43 highly conserved residues (≥90%) from the sequence alignment of Pin1 homologs (SI Text). We compared these residues with those showing consistent losses or gains of flexibility. All residues that showed a consistent loss or gain of flexibility were highly conserved (≥90%), with the exception of I93 and A124. Most of these were the conduit residues L60, L61, M130, L141, V150, I156, and I159. Thus, there is a strong overlap between residues that are evolutionarily conserved, and those with side chains showing consistent binding-related changes in flexibility (Fig. 5).

Fig. 5.

Side chains showing consistent changes in flexibility upon ligand binding belong to highly conserved residues. (A) Blue spheres are methyl symmetry axes that consistently loosen ( An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq39.jpg ), red spheres are methyl axes that consistently stiffen (). (B) Green spheres are methyl-bearing groups conserved at ≥90%.

Discussion

Comparing the dynamic consequences of binding cis versus trans conformers has been challenging for peptidyl-prolyl isomerases. We have answered this challenge for the human PPIase Pin1, by comparing how three related ligands—a substrate peptide FFpSPR and two inhibitor analogs locked in the cis and trans conformations—bind to Pin1 and alter its subnanosecond side-chain dynamics. The ligands differ at the core pS-P linkage. The substrate (FFpSPR) linkage undergoes cis-trans isomerization whereas the inhibitors cannot; otherwise, the composition and length of the three ligands are identical. We can therefore attribute the different dynamic responses to different conformational preferences of the pS-P linkage. The locked inhibitors are proxies of the substrate cis and trans ground states, and provide a unique opportunity to compare the effects of these two input conformers on Pin1 functional motions.

Chemical shifts showed different binding modes for the cis and trans inhibitors. The cis inhibitor is a nearly exclusive binder of the PPIase domain, whereas the trans inhibitor can bind both domains. The cis binds the PPIase active site more tightly; hence, saturating amounts of the cis inhibitor can knock off the trans inhibitor, which then binds just the WW domain. This suggests the design of fragment-linked inhibitors based on cis and trans binding cores targeting the PPIase and WW domains, respectively.

The methyl ^{D relaxation studies show that the substrate FFpSPR alters side-chain flexibility along a conduit of conserved hydrophobic residues linking the domain interface to the PPIase domain catalytic site. These alterations include both increases and decreases, with decreases (stiffening) dominating for the conduit residues that are less solvent accessible. This conduit overlaps with that observed in our previous study (}6) of Pin1 interacting with a different substrate. The emergence of this conduit response for two substrates with different sequences and lengths (FFpSPR versus EQPLpTPVTDL) suggests that it is intrinsic to Pin1 substrate recognition.

The locked inhibitors also induced conduit stiffening; thus, the stiffening reflects substrate binding rather than catalysis. However, the stiffening is more pronounced upon cis inhibitor binding; hence, the conduit response is stereoselective. Also, backbone NH J(0) and side-chain An external file that holds a picture, illustration, etc.
Object name is pnas.1019382108eq31.jpg relaxation corroborate this selectivity. Such selectivity may help route different input conformations to different signaling pathways.

Our K_d analyses showed that the PPIase domain of full-length Pin1 binds the cis locked inhibitor more tightly than the trans. Thus, tighter binding and conduit stiffening may be linked. Moreover, the truncated Pin1-PPI binds the cis inhibitor with slightly more affinity than the PPIase domain of full-length Pin1. This suggests that domain interactions can tune the breadth of active site conformations relevant for binding. Presumably, the PPIase active site in full-length Pin1 is optimized to bind the substrate transition state; this configuration may resemble cis more than trans (hence, the higher cis affinity), rather than being completely cis (hence, the higher affinity of Pin1-PPI versus full-length Pin1). This sparks interest in exploring transition-state analogues.

The Pin1 WW domain has been viewed as a docking module that enhances Pin1 specificity for pS/T-P motifs, and a modulator of PPIase activity due to the increase of local substrate concentration. Yet, mutations at the domain interface, in either the WW or PPIase domain, can reduce Pin1 activity as demonstrated by its inability to rescue yeast ESS1 knockouts (24). Together with the studies here, this suggests a parallel WW domain function: Its interaction with the PPIase domain can also tune the binding affinity of the remote (approximately 12 Å distance) PPIase domain catalytic site via an intraprotein signaling mechanism. Because the ^N-^{H spectra of}cis saturated PPIase domain were nearly identical in the presence and absence of the WW domain, this putative signaling mechanism does not invoke dramatic conformational change. New insight comes from our dynamics studies: WW domain binding can modulate the changes in side-chain flexibility, in particular, those along the conduit bridging the domain interface and the PPIase active site. Thus, whereas the intraprotein signaling may involve subtle structural changes, it may also involve propagated changes in flexibility (25). Precedent for such propagation includes millisecond motions for interdomain allostery (26), and subnanosecond motions for single domain allostery (27). This work expands the range of responses by suggesting that propagated changes in subnanosecond flexibility can assist interdomain communication within modular systems.

Given the above, conduit stiffening may serve to alter the breadth of conformational fluctuations at one functional site, upon ligand binding at another. This would be consistent with the disparity of time scales: The conduit motions are on the subnanosecond time scale, whereas PPIase activity occurs on the μs-ms time scale. The rapid conduit response would act as a high-speed communication link between the two functional sites of Pin1, in anticipation of the less frequent catalytic events.

Methyl-bearing residues are often highly conserved in proteins, because they provide a stabilizing hydrophobic core. Among Pin1 homologs, many of the highly conserved methyls also show consistent changes in mobility across four Pin1-ligand interactions (Fig. 5). Hence, their conservation likely reflects dual purposes: (i) maintaining overall protein stability; and (ii) providing a network of steric interactions, such as the conduit, that can enable intraprotein signaling via changes in flexibility. In fact, (ii) may be a general fortuitous consequence of ligand-induced stabilization of a protein hydrophobic core with spatially separate functional sites. Further testing of the conduit hypothesis will require mutating conduit residues followed by binding and dynamics studies.

In conclusion, our studies extend the function of conformational mobility in Pin1, a modular protein recruited by numerous signaling proteins. The dynamic responses of Pin1 are stereo-selective, and couple distinct functional sites. These dynamic responses entail flexibility changes that enable interactions between the modular domains to alter the interaction properties at the distal active site. Many proteins are modular and flexible, and thus, may exploit the dynamic gating strategies disclosed by Pin1.

Materials and Methods

Sample Preparation.

Full-length Pin1 was overexpressed and purified as described previously (6); the truncated Pin1-PPI (residues 40–163) followed a similar protocol (SI Text). The final NMR buffer was 30 mM imidazole-^{D—pH 6.6 with 30 mM NaCl, 0.03% NaN}₃, 5 mM DTT-D₁₀ and 90% H₂O/10% D₂O. Side-chain methyls and backbone nuclei were assigned at 295 K using standard triple-resonance experiments (6). The locked inhibitors and substrate FFpSPR were synthesized and purified as described previously (7, 8, 12). Further details regarding samples and ligand titrations are in SI Text.

NMR Relaxation Experiments and Analyses.

Deuterium relaxation rates R_1ρ(^{D) and}R₁(^{D) for CDH}₂ methyl isotopomers of full-length Pin1 and Pin1-PPI were measured with established pulse schemes (28, 29) at 16.4 and 18.8 T (700.13 and 800.13 MHz ^{H Larmor frequencies, respectively), 295 K. The}^{D hard-pulses were applied at 1.78 kHz whereas the spin-locks were applied at reduced strengths of 1.2 kHz. Backbone amide}R₁(^N),R_2-CPMG(^{N) and the steady-state}^HN-^{N NOE parameters were measured at 16.4 T using established pulse schemes (}30, 31). Further details for ^D,^{N, and}^{C relaxation and data analysis are in}SI Text.

^{Department of Chemistry and Biochemistry, University of Notre Dame, 251 Nieuwland Science Hall, Notre Dame, IN 46556;}

^{Department of Chemistry, Virginia Tech, Blacksburg, VA 24061}

^{To whom correspondence should be addressed. E-mail:}ude.dn@gnepj.

Edited by Peter E. Wright, The Scripps Research Institute, La Jolla, CA, and approved June 2, 2011 (received for review December 27, 2010)

Author contributions: F.A.E. and J.W.P. designed research; A.T.N., K.A.W., and J.W.P. performed research; A.T.N., X.J.W., B.X., A.Y.M.-C., and F.A.E. contributed new reagents/analytic tools; A.T.N., K.A.W., and J.W.P. analyzed data; and A.T.N., K.A.W., and J.W.P. wrote the paper.

Edited by Peter E. Wright, The Scripps Research Institute, La Jolla, CA, and approved June 2, 2011 (received for review December 27, 2010)

Freely available online through the PNAS open access option.

Abstract

Pin1 is a modular enzyme that accelerates the cis-trans isomerization of phosphorylated-Ser/Thr-Pro (pS/T-P) motifs found in numerous signaling proteins regulating cell growth and neuronal survival. We have used NMR to investigate the interaction of Pin1 with three related ligands that include a pS-P substrate peptide, and two pS-P substrate analogue inhibitors locked in the cis and trans conformations. Specifically, we compared the ligand binding modes and binding-induced changes in Pin1 side-chain flexibility. The cis and trans binding modes differ, and produce different mobility in Pin1. The cis-locked inhibitor and substrate produced a loss of side-chain flexibility along an internal conduit of conserved hydrophobic residues, connecting the domain interface with the isomerase active site. The trans-locked inhibitor produces a weaker conduit response. Thus, the conduit response is stereoselective. We further show interactions between the peptidyl-prolyl isomerase and Trp-Trp (WW) domains amplify the conduit response, and alter binding properties at the remote peptidyl-prolyl isomerase active site. These results suggest that specific input conformations can gate dynamic changes that support intraprotein communication. Such gating may help control the propagation of chemical signals by Pin1, and other modular signaling proteins.

Keywords: allostery, protein dynamics, ligand dynamics, protein evolution

Abstract

Phospho-serine/threonine-proline (pS/T-P) motifs are signaling motifs within intrinsically disordered loops of cell cycle proteins (1). The imide bond between the pS/T and P residues can adopt either the cis or trans conformation. These conformations differ in their susceptibility to kinases and phosphatases that propagate the chemical signals governing the cell cycle. Accordingly, the cell must regulate the cis/trans populations of these pS/T-P motifs to ensure proper signal routing.

In this context, the peptidyl-prolyl isomerase Pin1 has emerged as a critical regulator (2, 3). Pin1 is a reversible enzyme that catalyzes the cis-trans isomerization of the pS/T-P imide linkages (2, 3) of other signaling proteins, such as CDC25C, p53, c-Myc, NF-kB, cyclin D1, and tau (3). Pin1 engages when external events, such as S/T (de)-phosphorylation, change the cis-trans equilibrium. Pin1 then catalyzes the cis-trans isomerization, thereby accelerating the approach to the new equilibrium (1).

Pin1 is a modular protein of 163 residues consisting of a WW domain (1–39) and a larger peptidyl-prolyl isomerase (PPIase) domain (50–163) (Fig. 1). A flexible linker connects the two domains. Both domains are specific for pS/T-P motifs (1). The WW domain serves as a docking module, whereas catalysis is the sole province of the PPIase domain. Earlier structural studies of Pin1 revealed conformational changes upon substrate interaction, thus motivating flexibility-function studies of Pin1 (4 –6). Our previous NMR deuterium relaxation studies of Pin1 mapped the changes in flexibility of methyl-bearing side chains caused by interaction with an established phospho-peptide (pT) substrate (6). The interaction caused both gains and losses in side-chain flexibility. The flexibility losses occurred along a conduit of conserved hydrophobic residues within the PPIase domain, linking the domain interface with the PPIase catalytic site. These sites are on opposing sides of the PPIase domain, separated by distance of approximately 12 Å (Fig. 1).

Fig. 1.

The molecules of this study. (A) Surface representations of Pin1 [PDB ID code 1PIN (14)]; PPIase domain (aqua); WW domain (magenta); PPIase active site (yellow) with key residues labeled. (B) Pin1 methyls (spheres); methyls at functional regions (the domain interface, the PPIase active site, and the PPI catalytic loop) are colored according to domain (PPIase aqua; WW magenta). The gray spheres are methyls lacking a known function. (C) The three Pin1 ligands, left to right: substrate FFpSPR, cis locked inhibitor, and trans locked inhibitor.

Further interpretation of these results was hindered by the interconversion of the substrate pS/T-P imide bond between the cis and trans conformations. Pin1 recognizes both conformers. Thus, if and how the two conformers might impose different changes in dynamics remained indeterminate.

Etzkorn and coworkers designed peptidomimetic analogs of the Pin1 phospho-serine substrate (FFpSPR) (7, 8) (Fig. 1). These analogs replace the substrate pS-P core with alkene isosteres that lock the imide as cis (Ac–Phe–Phe–pSer–Ψ[(Z)CH = C]–Pro–Arg–NH₂), or trans (Ac–Phe–Phe–pSer–Ψ[(E)CH = C]–Pro–Arg–NH₂). These ligands are competitive inhibitors of Pin1 PPIase activity (K_i,cis = 1.7 ± 0.1 μM, and K_i,trans = 40 ± 2 μM) (8), and are proxies of the substrate cis and trans ground states, thus providing a unique opportunity to distinguish changes caused by the cis versus trans conformers.

Here we describe NMR studies of the interaction of these locked inhibitors and their parent substrate, FFpSPR, with Pin1 (Fig. 1C). The cis and trans inhibitors displayed different binding modes that imposed different changes in Pin1 side-chain flexibility. The cis inhibitor caused more prominent conduit rigidity than the trans inhibitor. Interactions between the PPIase and WW domains enhanced the conduit response and altered the ligand binding properties at the distal PPIase active site. These results suggest that specific input conformations gate changes in Pin1 dynamics that support intraprotein communication. Such stereoselective gating may be another means by which flexible proteins regulate the propagation of chemical signals.

Click here to view.

Acknowledgments.

We thank Dr. Thomas Nowak, John S. Zintsmaster, Jill Bouchard, and Dr. Jaroslav Zajicek for valuable suggestions. This work was supported by National Institutes of Health (NIH) Grants R01-GM083081 (J.W.P.) and Grant R01-CA110940 (F.A.E.).

Acknowledgments.

Footnotes

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1019382108/-/DCSupplemental.

Footnotes

References

1. Lu KP, Finn G, Lee TH, Nicholson LKProlyl cis-trans- isomerization as a molecular timer. Nat Chem Biol. 2007;3:619–629.[PubMed][Google Scholar]
2. Lu KP, Hanes SD, Hunter TA human peptidyl-prolyl isomerase essential for regulation of mitosis. Nature. 1996;380:544–547.[PubMed][Google Scholar]
3. Lu KP, Zhou XZThe prolyl isomerase PIN1: A pivotal new twist in phosphorylation signalling and disease. Nat Rev Mol Cell Biol. 2007;8:904–916.[PubMed][Google Scholar]
4. Jacobs DM, et al Peptide binding induces large scale changes in inter-domain mobility in human Pin1. J Biol Chem. 2003;278:26174–26182.[PubMed][Google Scholar]
5. Labeikovsky W, Eisenmesser EZ, Bosco DA, Kern DStructure and dynamics of pin1 during catalysis by NMR. J Mol Biol. 2007;367:1370–1381.[Google Scholar]
6. Namanja AT, et al Substrate recognition reduces side-chain flexibility for conserved hydrophobic residues in human Pin1. Structure. 2007;15:313–327.[PubMed][Google Scholar]
7. Wang XJ, et al Serine-cis-proline and serine-trans-proline isosteres: Stereoselective synthesis of (Z)- and (E)-alkene mimics by Still–Wittig and Ireland–Claisen rearrangements. J Org Chem. 2003;68:2343–2349.[PubMed][Google Scholar]
8. Wang XJ, Xu B, Mullins AB, Neiler FK, Etzkorn FAConformationally locked isostere of phosphoSer-cis-Pro inhibits Pin1 23-fold better than phosphoSer-trans-Pro isostere. J Am Chem Soc. 2004;126:15533–15542.[PubMed][Google Scholar]
9. Verdecia MA, Bowman ME, Lu KP, Hunter T, Noel JPStructural basis for phosphoserine-proline recognition by group IV WW domains. Nat Struct Biol. 2000;7:639–643.[PubMed][Google Scholar]
10. Bayer E, et al Structural analysis of the mitotic regulator hPin1 in solution: insights into domain architecture and substrate binding. J Biol Chem. 2003;278:26183–26193.[PubMed][Google Scholar]
11. Smet C, Wieruszeski JM, Buee L, Landrieu I, Lippens GRegulation of Pin1 peptidyl-prolyl cis-/trans- isomerase activity by its WW binding module on a multi-phosphorylated peptide of Tau protein. FEBS Lett. 2005;579:4159–4164.[PubMed][Google Scholar]
12. Namanja AT, et al Toward flexibility-activity relationships by NMR spectroscopy: Dynamics of Pin1 ligands. J Am Chem Soc. 2010;132:5607–5609.[Google Scholar]
13. Peng JW, Wilson BD, Namanja ATMapping the dynamics of ligand reorganization via and relaxation dispersion at natural abundance. J Biomol NMR. 2009;45:171–183.[Google Scholar]
14. Ranganathan R, Lu KP, Hunter T, Noel JPStructural and functional analysis of the mitotic rotamase Pin1 suggests substrate recognition is phosphorylation dependent. Cell. 1997;89:875–886.[PubMed][Google Scholar]
15. Abragam A Principles of Nuclear Magnetism. 1961. [PubMed][Google Scholar]
16. Lipari G, Szabo A. Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 1. Theory and range of validity. J Am Chem Soc. 1982;104:4546–4559.[PubMed]
17. Lipari G, Szabo A. Model-free approach to the interpretation of nuclear magnetic resonance relaxation of macromolecules. 2. Analysis of experimental results. J Am Chem Soc. 1982;104:4559–4570.[PubMed]
18. Nicholson LK, et al Dynamics of methyl groups in proteins as studied by proton-detected ^{C NMR spectroscopy. Application to the leucine residues of staphylococcal nuclease.}Biochemistry. 1992;31:5253–5263.[PubMed][Google Scholar]
19. Farrow NA, Zhang O, Szabo A, Torchia DA, Kay LESpectral density function mapping using ^{N relaxation data exclusively.}J Biomol NMR. 1995;6:153–162.[PubMed][Google Scholar]
20. Ishima R, Nagayama KProtein backbone dynamics revealed by quasi spectral density function analysis of amide N-15 nuclei. Biochemistry. 1995;34:3162–3171.[PubMed][Google Scholar]
21. Peng JW, Wagner GFrequency spectrum of NH bonds in eglin c from spectral density mapping at multiple fields. Biochemistry. 1995;34:16733–16752.[PubMed][Google Scholar]
22. Kay LE, Torchia DA, Bax ABackbone dynamics of proteins as studied by ^{N inverse detected heteronuclear NMR spectroscopy: Application to staphylococcal nuclease.}Biochemistry. 1989;28:8972–8979.[PubMed][Google Scholar]
23. Ghosh BAsymptotic expansions for the moments of the distribution of correlation coefficient. Biometrika. 1966;53:258–262.[PubMed][Google Scholar]
24. Behrsin CD, et al Functionally important residues in the peptidyl-prolyl isomerase Pin1 revealed by unigenic evolution. J Mol Biol. 2006;365:1143–1162.[PubMed][Google Scholar]
25. Cooper A, Dryden DTFAllostery without conformational change—a plausible model. Eur Biophys J Biophys Lett. 1984;11:103–109.[PubMed][Google Scholar]
26. Lipchock JM, Loria JPNanometer propagation of millisecond motions in V-type allostery . Structure. 2010;18:1596–1607.[Google Scholar]
27. Petit CM, Zhang J, Sapienza PJ, Fuentes EJ, Lee ALHidden dynamic allostery in a PDZ domain. Proc Natl Acad Sci USA. 2009;106:18249–18254.[Google Scholar]
28. Muhandiram DR, Yamazaki T, Sykes BD, Kay LEMeasurement of 2H T1 and T1r relaxation times in uniformly ^{C-labeled and fractionally 2H-labeled proteins in solution.}J Am Chem Soc. 1995;117:11536–11544.[PubMed][Google Scholar]
29. Millet O, Muhandiram DR, Skrynnikov NR, Kay LE. Deuterium spin probes of side-chain dynamics in proteins. 1. Measurement of five relaxation rates per deuteron in (13)C-labeled and fractionally (2)H-enriched proteins in solution. J Am Chem Soc. 2002;124:6439–6448.[PubMed]
30. Palmer AG., III NMR characterization of the dynamics of biomacromolecules. Chem Rev. 2004;104:3623–3640.[PubMed]
31. Kay LENMR studies of protein structure and dynamics. J Magn Reson. 2005;173:193–207.[PubMed][Google Scholar]