Conformational Motions Regulate Phosphoryl Transfer in Related Protein Tyrosine Phosphatases

See allHide authors and affiliations

Science  23 Aug 2013:
Vol. 341, Issue 6148, pp. 899-903
DOI: 10.1126/science.1241735

Closing the Loop

Many studies have shown that protein dynamics are important to enzyme function. For example, enzyme protein movements have been shown to optimize the active site, enable binding of substrate and cofactor, and facilitate product release. Whittier et al. (p. 899) now show that in two tyrosine phosphatases, the rate of cleavage is coupled to motion of a loop. The two phosphatases have different catalytic rates; however, in both, a loop containing a catalytic residue switches between an inactive open and a catalytically competent closed state. The rates of closure are equivalent to the cleavage rates, suggesting that the leaving group tyrosine is protonated simultaneously with loop closure. Thus, tuning of the loop motion plays a regulatory role in the catalytic cycle.


Many studies have implicated a role for conformational motions during the catalytic cycle, acting to optimize the binding pocket or facilitate product release, but a more intimate role in the chemical reaction has not been described. We address this by monitoring active-site loop motion in two protein tyrosine phosphatases (PTPs) using nuclear magnetic resonance spectroscopy. The PTPs, YopH and PTP1B, have very different catalytic rates; however, we find in both that the active-site loop closes to its catalytically competent position at rates that mirror the phosphotyrosine cleavage kinetics. This loop contains the catalytic acid, suggesting that loop closure occurs concomitantly with the protonation of the leaving group tyrosine and explains the different kinetics of two otherwise chemically and mechanistically indistinguishable enzymes.

Molecular motions are crucial for the optimal functioning of enzymes. There has been much debate regarding what role motions play in the enzymatic conversion of substrates to products (1, 2), and recent studies, primarily solution nuclear magnetic resonance (NMR) relaxation experiments, have shown that enzyme motions are critical for optimizing the active site (36), enabling effective substrate or cofactor binding (7), and facilitating product dissociation (8). These motions often require collective movement of many amino acids over substantial molecular distances (9) and are indispensible for chemistry. Furthermore, motions appear to have been evolutionarily optimized to accommodate the extreme environments in which enzymes must function (10). Often, motions are rate-limiting to enzyme turnover; however, this is mainly via product release, whereas the chemical steps are usually much faster (11). As a result, the direct participation of molecular motions in the chemical steps of an enzyme reaction, with few exceptions (12, 13), remains largely untested experimentally.

We addressed this question through a comparison of active-site loop motions in two structurally similar protein tyrosine phosphatases (PTPs) (Fig. 1A and fig. S1A): YopH, a virulence factor from Yersinia (14), and PTP1B, a human phosphatase that is involved in regulating insulin, leptin, and epidermal growth factor (EGF) signaling (15). These enzymes catalyze the cleavage of the tyrosine phosphate monoester (pY) of their protein substrates followed by hydrolysis of the phosphoenzyme intermediate. YopH and PTP1B contain 10-residue WPD (16) active-site loops and conserved, eight-residue P-loops. The P-loops bind the phosphate moiety of the substrate and contain the catalytic nucleophiles C403 in YopH and C215 in PTP1B (Fig. 1, C and D). The WPD loop contains an aspartic acid D356 (D181 in PTP1B) that serves as the general acid to protonate the leaving group, such that Tyr leaves as the alcohol rather than the alkoxide (Fig. 1, B to D). The P-loop does not undergo a substantial conformational change during catalysis. However, motions of about 10 Å differentiate the open (apo) and the catalytically competent closed WPD loop (Fig. 1B). The closed WPD loop positions the general acid near the oxygen of the leaving group of the substrate. YopH and PTP1B reactions occur in two steps (Fig. 1D). First, after substrate binds, the WPD loop closes and catalyzes the cleavage (kcleavage, where k is the rate constant) of the phosphotyrosine moiety, generating a phosphoenzyme intermediate and a tyrosine peptide. Subsequently, the phosphoenzyme intermediate is hydrolyzed (khydrolysis), regenerating active enzyme and releasing inorganic phosphate. YopH and PTP1B share equivalent chemical mechanisms and transition states for P–O bond cleavage. In both, kinetic isotope effects (KIE) (17, 18) showed the leaving group oxygen to be fully protonated in the transition state for phosphotyrosine cleavage, a conclusion supported by the negligible dependence (β = –0.008) on leaving group basicity (19). Despite very similar substrate interactions and chemical mechanisms, their kcat values, which are partially limited by the rate of hydrolysis of the phosphoenzyme intermediate, are significantly different, with YopH about 20-fold more active than PTP1B for phosphopeptide substrate with temperature-adjusted catalytic rates of 700 to 1000 s–1 versus 15 to 30 s–1, respectively, at pH = 6.6 and 293 K (20). The cleavage of phosphopeptide is faster than the hydrolysis step, as evidenced by burst kinetics and an observed kinetic solvent deuterium isotope effect on the overall reaction (21). Overall, these kinetic data indicate that kcleavage values for YopH and PTP1B range from 1400 to 2000 s–1 and 25 to 80 s–1, respectively (methods). Here, we describe the WPD loop motion by solution-phase NMR relaxation dispersion studies and focus on the role of loop closure in the cleavage step.

Fig. 1 Comparison of YopH and PTP1B.

(A) YopH (cyan), PTP1B (gray), the phosphate analog, and vanadate (red spheres) localize the active site. (B) Open (light gray) and closed (blue) WPD loops of PTP1B. Vanadate (spheres) binds in the P-loop (orange), with C215 and R221 (16) in stick rendering. Movement of D181 is indicated by dashed lines. (C) Superposition of closed loops for PTP1B and YopH. The three catalytic residues are shown in stick rendering. (D) PTP catalytic reaction of cleavage and hydrolysis. PDB accession numbers for (A) are 2I42 and 3I80.

The apo (ligand-free) forms of PTP1B and YopH behave very differently. We applied 15N-Carr-Purcell-Meiboom-Gill (CPMG) relaxation dispersion experiments to PTP1B (22), in which the measured relaxation rate (R2) depends on the physical parameters for motion, [equilibrium populations pa/pb and chemical shift differences (Δω) between the two interconverting conformations and kinetics (kex = kf + kr), where ex indicates exchange; f and r indicate forward and reverse rates, respectively and τcp, which is the delay between 15N 180° pulses (eq. S1). For PTP1B, we have assigned amide resonances for W179, F182, G183, V184, S187, and A189 (16) in the WPD loop. Of these, resonances for G183 and S187 are too weak to quantitate, and V184 is partially overlapped precluding further analysis. The remaining WPD loop residues (W179, F182, and A189, fig. S1, B to D) show upward curving CPMG relaxation dispersion (Fig. 2, A to C). The exchange parameters are shown in Table 1. Their individual fits yield similar exchange rates with kex ~ 900 s–1, suggesting concerted loop motion. Therefore, F182 and A189 data were fit to a global model simultaneously at 600 and 800 MHz (Table 1). For these two residues, pa = 97.5 ± 1.4% (SE) and pb = 2.5%. Knowledge of pa and kex enables determination of kf = 22 ± 5 s–1 and kr = 890 ± 190 s–1. The larger uncertainties in measured R2 values for W179 precluded use of eq. S1; however, reliable kex values = 900 ± 200 s–1 and Rex values (= papbΔω2/kex) = 31.4 ± 4.6 s–1 are obtained for this residue by using eq. S2 (Fig. 1C). By using pa = 97.5% determined above, we calculated Δω for W179 to be 2.8 ± 0.9 parts per million (ppm) (Table 1).

Fig. 2 NMR-detected motions.

Open and closed PTP1B and YopH WPD loops (A). (B and C) The 15N-CPMG dispersion curves for A189, F182, and W179. (D) TROSY-detected off-resonance 1H-R dispersion curves for A359 (blue) and S361 (red). Error bars indicate standard errors determined from duplicate measurements.

Table 1

Summary of WPD loop motions.

Open/closed loop equilibrium values are derived from popen/pclosed. The 15N Δω values were determined from fits to relaxation dispersion data; 15N Δδcalc values were determined with ShiftX2 as described in the methods; and 15N Δδmeas values were determined from differences in measured chemical shifts between apo and peptide bound enzyme.

View this table:

On the basis of the many apo PTP1B crystal structures (23), we hypothesized that the major solution conformer is an open WPD loop with a minor population of closed WPD loop, supported by an apo closed crystal structure [Protein Data Bank (PDB) code 1SUG (24)] with a closed active site loop. This interconversion between open and closed conformations means that kf represents the rate for loop closure and kr the rate for loop opening. Support for this hypothesis comes from measurements of the chemical shift differences between the open and closed WPD loop, which were determined by direct measurement (Embedded Image) (materials and methods) of the differences in open and closed shifts for the apo and peptide bound (vide infra) enzyme. Moreover, the difference (Embedded Image) between calculated shifts for the closed conformation based on the crystal structure and the measured shifts for the open conformation was determined. These Δω, Embedded Image, and Embedded Image values are nearly identical for F182, A189, and similarly for W179, whose bound resonance disappears; Δω and Embedded Image are within error (Fig. 3, A and B, fig. S3A, and Table 1). The agreement between these shifts supports the hypothesis that the apo loop converts between a predominantly open conformation with a minor closed state and indicates that the WPD loop in apo PTP1B closes (kclose = kf = 22 ± 5 s–1) with a rate constant very similar to its kcleavage value of 25 to 80 s–1 (see methods) suggesting that, if the enzyme maintained these motions in the substrate-bound state, loop closure occurs simultaneously with protonation of the leaving group substrate.

Fig. 3 Chemical shift differences for open and closed WPD loops.

For PTP1B (A and B) and YopH (C and D), the NMR-measured (red symbols) and calculated (black open symbols) chemical shifts are shown for residues in the WPD loop.

Distinct from PTP1B, YopH is over an order of magnitude faster in its enzymatic reaction. WPD loop resonances for T358, A359, V360, and S361 were assigned (fig. S4A). In contrast to PTP1B, YopH 15N-CPMG and 15N-R dispersion curves are flat (fig. S4, B and C) indicating loop motions Embedded Image 30,000 s–1. To detect these faster motions, we modified a 1H-R experiment (25) for transverse relaxation-optimized spectroscopy (TROSY) detection (fig. S5) and applied it to YopH. In Fig. 2, D and E, dispersion is observed for loop residues A359 and S361 with global kex = 43,000 ± 6200 s–1. T358 is overlapped, and V360 had too low signal-to-noise for quantitation. Like PTP1B, our hypothesis is that the WPD loop moves between open and closed conformations, which is supported by the similarity in Embedded Image and Embedded Image 15N chemical shifts for A359 and S361 (Fig. 3, C and D) and for T358 and V360 (fig. S3, B and C).

The motion of the apo YopH WPD loop is in the fast exchange limit, and therefore other methods are needed to extract Δω and populations from these data because they appear as the product, papbΔω2 in Eq. 1

Embedded Image (1)

To circumvent this limitation, we measured the 15N-Rex values (Rex = papbΔω2/kex) from a 15N TROSY Hahn-echo experiment (26) (fig. S6) (for details, see supplementary materials) and used the measured kex and Embedded Image as a proxy for Δω to give pa = 97%. A loop-closing rate, kclose = 1240 ± 280 s–1 (Table 1), was calculated for YopH, which is about 1.7-fold greater than its kcat value and similar to kcleavage (1400 s–1 to 2000 s–1), suggesting that, like PTP1B, there is intimate involvement of WPD loop closure in the step of pY cleavage. Thus, for two enzymes with very different catalytic rates and identical chemical mechanisms, active-site loop motion in the apo enzyme is tuned to approximate the overall cleavage rate in each. Because the WPD loop must be closed for chemistry to occur, loop closing on the proper time scale will facilitate the chemical reaction.

We formed the complex between PTP1B and an octamer peptide, Ac-DADEXLIP-NH2 (16) (fig. S2A) derived from EGF receptor, where X is difluoromethylphosphono-phenylalanine, a noncleavable phosphotyrosine mimic. The inhibition constant value of the peptide (high nM, low μM) is similar to the substrate Michaelis constant, indicating that DADEXLIP behaves as a good substrate mimic (27). In PTP1B, F182 is significantly exchange-broadened in the peptide-bound form, and the resonance for W179 disappears, preventing further analysis. Both observations indicate a change in loop motions. The 15N-CPMG dispersion curves at 600 and 800 MHz for A189 in peptide-bound PTP1B are noticeably different from apo PTP1B and display pronounced oscillation at low CPMG pulsing frequencies, characteristic of slow conformational exchange (Fig. 4A and eq. S3). The A189 Δω = 1210 ± 130 s–1 (3.2 ± 0.3 ppm) is identical to that determined in ligand-free PTP1B (Table 1). The equivalence of Δω for apo and peptide-bound PTP1B is additional evidence that when substrate occupies the active site, the WPD loop interconverts between major closed and minor open conformations. Further support for this hypothesis comes from a PTP1B ligand-bound crystal structure [3EB1 (28)] in which the WPD loop is open, unlike the majority of other bound structures. From these data, we obtain a kclose = 30 ± 4 s−1 and kopen = 5 ± 1 s–1. Thus, loop closure is the same in apo and peptide-bound PTP1B; peptide binding only slows loop opening. The similarity of kclose and kcleavage along with KIE data (17, 18) indicating leaving group protonation in the transition state suggests that protonation of the tyrosine occurs concomitantly with loop closure with the proton being delivered by D181 on the WPD loop. Cleavage of phosphotyrosine is followed by hydrolysis of the phosphoenzyme intermediate, which occurs somewhat slower. These NMR data suggest that closing of the WPD loop occurs in close connection with the chemistry of phosphate cleavage and plays an important regulatory role in the overall catalytic cycle.

Fig. 4 NMR-detected motions.

The 15N-CPMG dispersion analyses for peptide-bound PTP1B (A) and YopH (B). (A) shows a global fit at 800 and 600 MHz for A189. (B) shows a global fit at 800 (red) and 600 (blue) MHz for A359 and S361 in YopH.

For YopH bound to peptide, loop motion slows substantially such that the 15N-CPMG dispersion experiment can now be used for characterization, in contrast to apo YopH (Fig. 4B). Global fits of eq. S1 to the dispersion data in Fig. 4B gives Δω values for A359 and S361 of 1350 ± 150 s–1 (3.5 ppm) and 1730 ± 290 s–1 (4.5 ppm), which are similar to the apo Δω, Embedded Image, and Embedded Imagevalues, as they are in PTP1B, indicating interconversion between open and closed conformations (Table 1). For the YopH loop, kex is 1790 ± 240 s–1 and pa is 99%, to give kclose = 1770 ± 240 s–1 and kopen = 18 ± 2 s–1. Like PTP1B, peptide binding to YopH only slows loop opening. Thus, in this faster enzyme kclose and kcleavage are similar, again suggesting that loop closure is concerted with protonation of the leaving group tyrosine as part of the cleavage reaction. Unlike the apo and peptide-bound enzymes, the product state mimic, tungstate, dissociates from PTB1B and YopH with identical rate constants (fig. S7 and methods). Thus, once cleavage and hydrolysis are complete, release of product is the same for both enzymes.

These NMR results for YopH are different than observed by other techniques. For apo YopH, Trp fluorescence anisotropy and temperature-jump (T-jump) fluorescent experiments suggest loop kinetics ranging from 3 ns to 3 μs, respectively (29, 30). The former value is consistent with molecular dynamics simulations (31) indicating conformational motion between open and partially closed states, unlike our data that show opening and closing times, similar to T-jump experiments, of 22 μs between open and fully closed conformations. Those fluorescence experiments, however, only report on the indole side chain of W179, whereas our NMR data pertain to the motion of the backbone and therefore are more closely connected to the motion of the entire loop. Comparison of YopH loop motion in the bound conformation to other studies is more difficult. We used a peptide, whereas the previous T-jump experiment exploits the fluorescence from p-nitrocatechol sulfate; the differences in ligand choice may explain the 50-fold difference in observed loop kinetics in the respective complexes. For PTP1B, there are no available data for comparison.

The data presented for this pair of phosphatases suggest that the tuning of loop motion is closely synchronized with the chemical reaction of phosphotryosine cleavage. During the femtosecond bond-making and -breaking steps of the enzyme reaction, the WPD loop is static on the time scale observable by NMR. However, our NMR experiments do suggest a contiguous energy landscape for loop motion and catalytic activity and that loop closure is closely coupled to the protonation of the tyrosine leaving group. Additionally, the nature of the occupancy of the active site modulates the WPD loop kinetics and is consistent with the dynamic energy landscape hypothesis of enzyme function (3).

A myriad of crucial biological pathways are regulated by posttranslational protein phosphorylation levels. Remarkably, our NMR data for these phophatases suggest that phosphorylation levels can be modulated through control of WPD loop kinetics. The additional Pro residues in the PTP1B loop are potentially responsible for the slower motions. Slower PTP1B loop motions likely reflect the physiological role of PTP1B as a tight regulator of cellular processes, in which its turnover rate must meet the strict growth requirements of the cell. In contrast, the faster loop motion and enzymatic activity in YopH would be beneficial for the rapid disruption of normal cellular pathways that enable this enzyme to participate in and facilitate Yersinia infection.

Supplementary Materials

Materials and Methods

Figs. S1 to S7

References (3252)

References and Notes

  1. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
  2. Acknowledgments: We thank J. Hoch, D. Korzhnev, M. Maciejewski, and G. Armstrong for access to their high-field NMR instruments and G. Khirich for running Monte Carlo simulations. S.K.W., A.C.H., and J.P.L. acknowledge support from grants NIH-T32GM008283, NIH-GM47297, and NSF-MCB1121372.
View Abstract

Stay Connected to Science

Navigate This Article