Evaluation of Energetic and Dynamic Coupling Networks in a PDZ Domain Protein

J. Mol. Biol. (2006) 364, 337–351

doi:10.1016/j.jmb.2006.08.076

Evaluation of Energetic and Dynamic Coupling Networks in a PDZ Domain Protein Ernesto J. Fuentes 1 , Steven A. Gilmore 1 , Randall V. Mauldin 2 and Andrew L. Lee 1,2 ⁎ 1

Division of Medicinal Chemistry and Natural Products, School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA 2

Department of Biochemistry & Biophysics, School of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA

A number of computational and experimental studies have identified intramolecular communication “pathways” or “networks” important for transmitting allostery. Here, we have used mutagenesis and NMR relaxation methods to investigate the scope and nature of the communication networks found in the second post-synaptic density-95/discs large/ zonula occludens-1 (PDZ) domain of the human protein tyrosine phosphatase 1E protein (hPTP1E) (PDZ2). It was found that most mutations do not have a significant energetic contribution to peptide ligand binding. Three mutants that showed significant changes in binding also displayed context-dependent dynamic effects. Both a mutation at a partially exposed site (H71Y) and a buried core position (I35V) had a limited response in sidechain 2H-based dynamics when compared to wild-type PDZ2. In contrast, a change at a second core position (I20F) that had previously been shown to be part of an energetic and dynamic network, resulted in extensive changes in side-chain dynamics. This response is reminiscent to that seen previously upon peptide ligand binding. These results shed light on the nature of the PDZ2 dynamic network and suggest that position 20 in PDZ2 acts as a “hub” that is energetically and dynamically critical for transmitting changes in dynamics throughout the PDZ domain. © 2006 Elsevier Ltd. All rights reserved.

*Corresponding author

Keywords: protein dynamics; allostery; PDZ domain; dynamic coupling; protein–protein interaction

Introduction Allostery, the transmission of energy within or between proteins, can result from structural rearrangements or dynamical changes (or both).1,2 Mounting experimental and theoretical evidence suggests that allosteric behavior may be a fundamental property of all proteins, not necessarily limited to multi-domain proteins or complexes.3

Present address: E. J. Fuentes, University of Iowa, Department of Biochemistry, Roy J. and Lucille A. Carver College of Medicine, Iowa, USA. Abbreviations used: PDZ, post-synaptic density-95/ discs large/zonula occludens-1; hPTP1E, human protein tyrosine phosphatase 1E; PDZ2, 2nd PDZ domain from hPTP1E; PSD-95, post-synaptic density-95; WT, wild-type; HSQC, heteronuclear single quantum coherence; RDC, residual dipolar coupling; VDW, van der Waals. E-mail address of the corresponding author: [email protected]

Thus, allosteric regulation of function might occur within a single domain protein via disparate sites that are energetically connected through structural rearrangements and/or motional dynamics. Although it is evident that such intramolecular communication exists, the nature and mechanism by which it occurs has not been clearly delineated. Recent studies using deuterium spin dynamics of methyl and methylene side-chain moieties indicate that proteins posses a rich complement of dynamics heterogeneously distributed throughout the protein. We and others have shown that 2H-methyl relaxation methods can be used to identify sites having altered side-chain dynamical properties as a function of effector ligand,4–10 solution conditions,11 or mutation.12–17 These methods offer a possible means for detecting changes in dynamics resulting from (or leading to) the propagation of energetic perturbations from remote sites within a protein to a catalytic or ligand binding site, i.e. allostery. Post-synaptic density-95/discs large/zonula occludens-1 (PDZ) domains are small, ∼100 amino acid protein domains, that generally bind the final

0022-2836/$ - see front matter © 2006 Elsevier Ltd. All rights reserved.

Methyl Dynamics of hPTP1E PDZ2 Mutants

338 four to five C-terminal residues of partner proteins.18 Solution and crystal structures for various members of this domain family indicate that they share a common topology comprised of a β-barrel structure flanked by a prominent α-helix (Figure 1). Residues found within helix α2, beta-strand β2, and the so-called “carboxylate-binding” loop connecting beta-strands β1 and β2 form a hydrophobic pocket that accommodates the C-terminal residues of target proteins. In the second PDZ domain from the human protein tyrosine phosphatase 1E (PDZ2), residues Leu18, Ile20, and Leu78 (numbering based on Fuentes et al.10) contribute to this ligand-binding pocket. Histidine 71, located in helix α2, defines PDZ2 as a class I PDZ domain and participates in hydrogen bonding with a Thr or Ser side-chain of target proteins.19 Our previous findings showed that upon ligand binding, changes in side-chain dynamics occurred at the binding site, namely at Leu18, Ile20, and Leu78. Furthermore, side-chains flanking these residues (located up to ∼7 Å away, e.g. Val61, Val64, and Val85) were also dynamically perturbed, suggesting a redistribution of binding energy that propagates throughout the protein upon ligand binding. The total response in side-chain dynamics on ligand binding defines the PDZ2 “dynamic network”. Interestingly, various residues at these same positions in the third PDZ domain of PSD-95 were found to be thermodynamically coupled to His71 in helix α2,20 suggesting a link between motional dynamics and energetic propagation within this domain. Based on these observations we rationalized that perturbations to the PDZ2 protein, such as a point mutation, might begin to reveal the scope and nature of the dynamically coupled network. Here, we created several point mutations within the PDZ2 dynamic network in order to investigate their relative energetic contribution to binding and effect

on fast timescale dynamics. Comparison of the 2H methyl-bearing side-chain dynamics (S2axis and τe) between mutant proteins and the wild-type (WT) showed that the effects of mutations are contextdependent. Both a mutation at a partially exposed site (H71Y) and a buried core position (I35V) had a limited response in dynamics. In contrast, a change at a second core position (I20F), that had previously been shown to be part of an energetic and dynamic “network”, resulted in extensive changes in sidechain dynamics. This response was reminiscent to that seen previously upon peptide ligand binding. These results shed light on the nature of the PDZ2 dynamic network and suggests that position 20 in PDZ2 may act as a “hub” that is energetically and dynamically crucial for transmitting changes in dynamics throughout the PDZ domain.

Results Amino acid substitutions were introduced at various positions in PDZ2 to examine their effect on the energetics of ligand binding and protein dynamics. The various point mutants used in this study (see Table 1) were chosen based on previous results that suggested these sites might be energetically20 or dynamically10 “coupled” to the binding of peptide ligand. As shown in Figure 1, these sites are distributed throughout the protein including several “distal” sites that are >7 Å from direct interaction with peptide ligand. Peptide binding The panel of mutants was surveyed for the ability to bind a C-terminal peptide derived from the RAGEF2 protein using either a fluorescence or NMR titration assay. Fluorescence titration data

Figure 1. The solution structure of the second PDZ domain of human PTP1E. Representative structures of PDZ2 in the absence (a) and presence (b) of peptide-ligand are shown with the elements of secondary structure labeled. (a) The free mouse NMR structure (PDB code 1GM145). (b) PDZ2 bound to RA-GEF2 peptide (PDB code 1D5G; human form) and a 90° rotation. Amino acid positions mutated in the work described here are highlighted. This Figure was produced using the program MOLMOL.52

Methyl Dynamics of hPTP1E PDZ2 Mutants

339

Table 1. Binding free energy for the interaction between PDZ2 mutants and the RA-GEF2 peptide ligand determined at 25 °C

WT I20Fa,b G24Ab I35A I35Vc V40Ia V40La A45Pb A46Ib A46Vb I52Vb V61Aa,b V64Aa T70Ab T70Sb H71Yb,c V84S V85Aa,b

ΔGb (kcal/mol)

ΔΔGb (kcal/mol)

Kd (μM)

Fold effect

− 6.74 (0.10) − 6.50 (0.03) − 6.78 (0.04) nd − 5.48 (0.16) − 6.78 (0.03) − 6.94 (0.07) − 6.76 (0.04) − 6.55 (0.03) − 6.77 (0.16) − 6.46 (0.10) − 6.69 (0.05) − 6.58 (0.03) − 6.73 (0.15) − 6.60 (0.01) − 3.75 (0.15) − 6.70 (0.04) − 6.67 (0.09)

− 0.23 (0.11) 0.04 (0.11) nd −1.25 (0.19) 0.04 (0.11) 0.21 (0.12) 0.02 (0.11) − 0.18 (0.11) 0.03 (0.19) −0.28 (0.14) − 0.04 (0.11) −0.15 (0.10) 0.00 (0.18) −0.13 (0.10) −2.99 (0.18) − 0.04 (0.11) −0.07 (0.13)

11.6 (2.2) 17.1 (0.8) 10.7 (0.6) ∼100 97.3 (25.6) 10.7 (0.6) 8.1 (1.0) 11.1 (0.8) 15.6 (0.8) 11.1 (2.9) 18.5 (3.0) 12.4 (1.0) 14.9 (0.7) 11.9 (3.2) 14.3 (0.2) 1834 (470) 12.2 (0.8) 12.9 (1.9)

1.0 1.5 0.9 ∼10 8.4 0.9 0.7 1.0 1.3 0.9 1.6 1.1 1.3 1.0 1.2 157.5 1.1 1.1

nd, not determined. a Found in dynamic network.10 b Found in energetic network.20 c Binding energetics determined by NMR titration data.

established that PDZ2 WT has an affinity (Kd) for peptide of ∼12 μM and the affinities for the mutant panel had a 0.7–160-fold effect in binding affinity (Table 1). Using this fluorescence assay, only very weak binding was noted for H71Y (Kd ≫ 100 μM). NMR chemical shifts monitored as a function of peptide concentration yielded Kd estimates of 1.8 ± 0.5 mM and 3.7 ± 1.3 mM for H71Y and H71Y/I20F, respectively. Based on fluorescence measurements, the I35V mutation had a decreased affinity for peptide (Kd ∼100 μM) which was confirmed by NMR titration data (Kd ∼101 ± 24 μM). Thermodynamic stability The thermodynamic stability for PDZ2 WT, H71Y, I20F, I35V and H71Y/I20F was assessed by guanidine hydrochloride chemical denaturation. The fitted stabilities (ΔGu) range from 1.8 to 4.0 kcal/ mol (Table 2). PDZ2 WT and H71Y displayed similar unfolding energetics (∼3.0 kcal/mol), while proteins containing the I20F substitution showed an increase in unfolding free energy up to 4 kcal/mol. The I35V mutant has a stability (ΔGu) of 1.80(±0.04) kcal/mol, which is destabilized by 1.3 kcal/mol relative to WT. Using these data, a thermodynamic mutant site cycle21 could be constructed for residues H71Y and I20F (see Discussion). Mutational effects on protein structure In general, 15 N- 1 H and 13 C- 1 H heteronuclear single quantum coherence (HSQC) spectra of PDZ2 mutants had well dispersed chemical shifts, indicating single, stable conformations. However, notable exceptions were proteins with mutations at amino

acid position 35. Substitution of the wild-type Ile for Ala at this position severely impaired the ability to bind peptide ligand (Table 1) and the 15N-1H HSQC spectrum suggested the presence of multiple conformations (data not shown). The I35V mutation had a milder effect with respect to both peptide binding (∼eightfold reduction in affinity) and conformational exchange. Detailed NMR analysis was possible using fresh samples, however, over time (∼five days) there was a tendency for the protein to unfold. Chemical shift changes ( 15 N- 1 H and 13 C- 1 H3 pairs) show evidence for local perturbations at sites bordering the mutation, but suggest no gross changes in structure. The average backbone (15N-1H) chemical shift change (normalized chemical shift change of (0.5*((ΔωH)2 + (γ*ΔωN) 2 ))0.5 , with γ = 0.2) for I20F, I35V and H71Y was 0.09 ± 0.13, 0.07 ± 0.05 and 0.06 ± 0.07, respectively. Similarly, the average chemical shift changes for methyl groups (13C-1H3, using γ = 0.3) for I20F, I35V and H71Y was 0.09 ± 0.10, 0.02 ± 0.02 and 0.03 ± 0.04, respectively. In comparison, the PDZ2/RA-GEF2 complex had relatively large chemical shift changes along the backbone (0.14 ± 0.14) and side-chain methyl groups (0.07 ± 0.06), yet only minor structural changes occur upon peptide binding.10 The larger change in amide chemical shifts upon peptide binding may reflect the ionic interactions made by the complex rather than structural changes. Overall, these results suggest that PDZ2 mutants have only small changes in backbone and side-chain structure. To further assess changes in structure, amide nitrogen residual dipolar coupling measurements (RDCs) were obtained for WT, I20F and H71Y proteins. A comparison between WT and mutant RDC measurements is shown in Figure 2. Panel (a) indicates that independent RDC measurements for PDZ2 WT were highly reproducible, r = 0.99. Comparison of the mutant PDZ2 RDCs with those obtained for the WT were in very good agreement for the vast majority of residues (Figure 2(b) and (c)), which supports the conclusion that no major Table 2. Pair-wise coupling free energies (ΔΔΔGi) for PDZ2 I20F and H71Y mutants determined at 25 °C

PDZ2 WT I20F H71Y H71Y/I20F

PDZ2 WT I20F H71Y H71Y/I20F

ΔGu (kcal/mol)

ΔΔGua

ΔΔΔGib

Σ singles

3.13 (0.15) 4.04 (0.06) 3.09 (0.14) 3.72 (0.09)

0.91 (0.16) –0.04 (0.20) 0.59 (0.17)

–0.28 (0.31)

0.87 (0.26)

ΔGb (kcal/mol)

ΔΔGba

ΔΔΔGi b

Σ singles

–6.74 (0.10) –6.50 (0.03) –3.75 (0.15) –3.35 (0.19)

0.24 (0.10) 2.99 (0.18) 3.39 (0.21)

0.16 (0.30)

3.23 (0.21)

u=unfolding, b=binding. a WT ΔΔGu/b = [ΔGmut u/b – ΔGu/b]. b H71Y ΔΔΔGi = [ΔGH71Y/I20F – (ΔGI20F u/b u/b + ΔGu/b )].

340

Methyl Dynamics of hPTP1E PDZ2 Mutants

Figure 2. One-bond 15N-1H residual dipolar couplings (RDCs) for the (a) WT, (b) H71Y, and (c) I20F plotted against PDZ WT. In (a), two independent experiments for WT are shown. The linear-correlation coefficient, r, for each fit is indicated.

conformational change occurred along the backbone upon mutation. Backbone and methyl side-chain order parameters Backbone 15N-derived order parameters (S2) were determined for I20F, I35V and H71Y mutants using the model-free approach.22,23 Comparison of these results with WT shows that only small changes in the derived backbone S2 are evident. The average change in S2 (mut – WT) was −0.022 ± 0.024, −0.029 ± 0.025 and −0.016 ± 0.026 for I20F, I35V and H71Y, respectively. These results indicate the rigidity of the backbone for these mutants is very similar to PDZ WT. Side-chain methyl dynamics parameters were determined for each mutant PDZ2 protein by the use of 2H spin-relaxation measurements at two field strengths, followed by fitting to model-free parameters S2axis and τe (Table 3).24,25 Figures 3 and 4 summarize the change in S2axis and τe for H71Y, I35V and I20F compared to PDZ2 WT.10 Changes were deemed significant if the difference observed in the fitted parameter was twice that of the derived error. Using this criterion, four residues in H71Y show significant dynamical changes: Ala12β, Val30γ2, and Ala74β show changes in S2axis, while Thr28γ2 and Ala74β show a change in τe (Figure 3(a)). Similarly, the I35V mutation had an effect on side-chain dynamics for six residues: Val9γ2, Val22γ2, Leu78δ1, Val85γ2, and Leu89δ2 had changes in S2axis and Val61γ1 had a change in τe (Figure 3(b)). In contrast, the I20F mutation had a more pronounced effect – 14 residues showed significant changes in dynamics (Figure 3(c)). With respect to changes in S 2 axis, Val9 γ1 , Ala12 β , Leu18δ1, Val22γ1, Val26γ1, Val30γ2, Ile41γ2, Val58γ1, Val64γ2, Ala69β, and Leu78δ1 showed significant changes; the methyl groups of Val22γ2, Val37γ2, Ala46β, and Val85γ1 had significant changes in τe. Coupling constants and χ1 rotamer populations Three bond 3JCγN and 3JCγCO coupling constants were measured for Ile, Val, and Thr amino acid

residues of PDZ2 WT, PDZ2/RA-GEF2, and I20F. In general, our results indicate that coupling constants for most residues do not change as a function of ligand (PDZ2/RA-GEF2) or mutation (I20F). Based on the measured coupling constants, rotameric averaging for many residues was inferred (Table 4). In addition, rotameric χ1 populations were estimated using a three-site jump model.9 In a few cases, there were significant changes in coupling constant values relative to PDZ2 WT. Specifically, the PDZ2/RA-GEF2 complex had significant changes in 3JCγCO for Val61γ1, Val64γ2, Val75γ1 and Val85γ1. In addition, Thr28γ2 had a change in 3 JCγN. Changes in coupling constants were also detected in I20F: Val9γ1 and Val61γ1 had changes in 3JCγCO, and Val61γ1 had a change in 3JCγN. The approximate timescale of rotameric averaging could be inferred by comparing the rotamer populations determined from coupling constants with those predicted from S2axis values by use of a semi-quantitative analytical expression.26 Residues whose rotameric populations agreed well (within ∼20%) indicate averaging on a fast timescale (ns–ps), while those that did not should average on a slower timescale, most likely μs–ms. Correlation between order parameter and chemical shift changes Because chemical shifts should in principle be sensitive to changes in dynamics, changes in chemical shifts were compared to changes in S2axis values (Figure 5). On a gross level, more extensive changes in S2 axis values upon mutation (e.g. I20F) were accompanied by more extensive changes in methyl proton and carbon chemical shifts. However, taken on an individual site basis, chemical shift changes are not reliable predictors of changes in ps–ns dynamics (Figure 5(b)). This is not entirely surprising given that the chemical shift has contributions from not only local side-chain structure but also

Methyl Dynamics of hPTP1E PDZ2 Mutants

341

Table 3. Fitted methyl side-chain dynamic parameters for PDZ2 WT, H71Y, I35V and I20F mutants PDZ2 WT Methyl δ1

Ile6 Ile6γ2 Val9γ1 Val9γ2 Leu11δ1 Leu11δ2 Ala12β Leu18δ1 Leu18δ2 Ile20δ1 Ile20γ2 Val22γ1 Val22γ2 Thr23γ2 Val26γ1 Val26γ2 Thr28γ2 Val30γ1 Val30γ2 Ile35δ1 Ile35γ2 Val37γ1 Val37γ2 Ala39β Val40γ1 Val40γ2 Ile41δ1 Ile41γ2 Ala45β Ala46β Ile52δ1 Ile52γ2 Val58γ1 Val58γ2 Leu59δ1 Leu59δ2 Ala60β Val61γ1 Val61γ2 Val64γ1 Val64γ2 Leu66δ1 Leu66δ2 Ala69β Thr70γ2 Ala74β Val75γ1 Val75γ2 Thr77γ2 Leu78δ1 Leu78δ2 Thr81γ2 Val84γ1 Val84γ2 Val85γ1 Val85γ2 Leu87δ1 Leu87δ2 Leu88δ1 Leu88δ2 Leu89δ1 Leu89δ2 Thr96γ2

H71Y

I35V

I20F

S2axis

τe (ps)

S2axis

τe (ps)

S2axis

τe (ps)

S2axis

τe (ps)

0.26 (0.01) 0.68 (0.02) 0.51 (0.02) 0.51 (0.03)

33.8 (1.0) 43.4 (1.9) 61.7 (2.2) 79.8 (3.4)

0.25 (0.01) 0.71 (0.01) 0.51 (0.01) 0.51 (0.02)

32.9 (0.5) 42.2 (1.1) 60.8 (1.2) 80.3 (2.0)

0.26 (0.01) 0.68 (0.02) 0.51 (0.02) 0.51 (0.03)

33.8 (1.0) 43.4 (1.9) 61.7 (2.2) 79.8 (3.4)

0.28 (0.01) 0.71 (0.01)

33.0 (0.6) 42.7 (1.1)

0.59 (0.02)

74.3 (2.2)

0.28 (0.05) 0.74 (0.02) 0.61 (0.07) 0.54 (0.04) 0.60 (0.02) 0.35 (0.01) 0.80 (0.04) 0.76 (0.03) 0.72 (0.05)

68.7 (6.7) 43.8 (1.6) 50.5 (7.4) 40.8 (3.5) 22.3 (1.6) 47.6 (1.4) 58.4 (3.4) 32.7 (2.3) 97.5 (6.0)

0.30 (0.03) 0.81 (0.01) 0.57 (0.04) 0.61 (0.02) 0.64 (0.01) 0.37 (0.01) 0.82 (0.02) 0.80 (0.02) 0.83 (0.03)

68.7 (3.6) 42.1 (1.0) 48.8 (4.3) 35.9 (1.8) 20.2 (0.9) 47.0 (0.8) 58.8 (1.9) 34.2 (1.3) 102.0 (3.9)

0.28 (0.05) 0.74 (0.02) 0.61 (0.07) 0.54 (0.04) 0.60 (0.02) 0.35 (0.01) 0.80 (0.04) 0.76 (0.03) 0.72 (0.05)

68.7 (6.7) 43.8 (1.6) 50.5 (7.4) 40.8 (3.5) 22.3 (1.6) 47.6 (1.4) 58.4 (3.4) 32.7 (2.3) 97.5 (6.0)

0.23 (0.02) 0.83 (0.01) 0.75 (0.07) 0.72 (0.03)

67.3 (3.1) 40.2 (1.0) 44.1 (5.1) 32.2 (1.9)

0.90 (0.03) 0.83 (0.02) 0.80 (0.03)

74.0 (2.8) 29.1 (1.3) 90.9 (3.2)

0.56 (0.02) 0.74 (0.04) 0.55 (0.02) 0.53 (0.03) 0.65 (0.03) 0.84 (0.04) 0.88 (0.04)

40.9 (1.7) 70.1 (3.5) 66.8 (2.4) 58.7 (2.9) 38.0 (3.1) 38.6 (3.1) 49.5 (3.6)

0.55 (0.01) 0.68 (0.02) 0.42 (0.01) 0.53 (0.01) 0.67 (0.02) 0.86 (0.02) 0.92 (0.03)

42.9 (1.0) 60.8 (1.6) 63.1 (1.1) 56.8 (1.6) 34.0 (1.9) 40.2 (1.8) 50.7 (2.1)

0.56 (0.02) 0.74 (0.04) 0.55 (0.02) 0.53 (0.03)

40.9 (1.7) 70.1 (3.5) 66.8 (2.4) 58.7 (2.9)

0.88 (0.04)

49.5 (3.6)

0.63 (0.01) 0.82 (0.02) 0.61 (0.01) 0.58 (0.02) 0.69 (0.02) 0.89 (0.03) 0.95 (0.03) 0.91 (0.02)

39.7 (1.1) 67.5 (2.1) 63.6 (1.3) 59.3 (1.8) 37.9 (1.6) 37.1 (1.8) 30.1 (1.7) 25.9 (1.5)

0.87 (0.02) 0.61 (0.02)

34.0 (1.7) 41.8 (2.0)

0.91 (0.01) 0.62 (0.01)

34.5 (1.0) 42.0 (1.2)

0.87 (0.02) 0.61 (0.02)

34.0 (1.7) 41.8 (2.0)

0.18 (0.01) 0.78 (0.03) 0.91 (0.04) 0.92 (0.05) 0.71 (0.03) 0.76 (0.03) 0.81 (0.04) 0.81 (0.03) 0.55 (0.04) 0.54 (0.03)

29.5 (1.2) 53.7 (2.3) 49.5 (3.2) 62.4 (4.7) 16.5 (2.0) 40.1 (2.7) 63.1 (4.0) 32.4 (2.4) 42.7 (4.1) 50.3 (2.8)

0.19 (0.01) 0.80 (0.02) 0.89 (0.02) 0.93 (0.03) 0.76 (0.02) 0.80 (0.02) 0.88 (0.03) 0.86 (0.02) 0.58 (0.03) 0.55 (0.02)

30.5 (0.5) 54.7 (1.4) 52.2 (1.9) 60.9 (2.8) 13.6 (1.2) 40.1 (1.5) 62.4 (2.3) 31.2 (1.4) 44.8 (2.6) 52.7 (1.7)

0.18 (0.01) 0.78 (0.03) 0.91 (0.04) 0.92 (0.05) 0.71 (0.03) 0.76 (0.03) 0.81 (0.04) 0.81 (0.03) 0.55 (0.04) 0.54 (0.03)

29.5 (1.2) 53.7 (2.3) 49.5 (3.2) 62.4 (4.7) 16.5 (2.0) 40.1 (2.7) 63.1 (4.0) 32.4 (2.4) 42.7 (4.1) 50.3 (2.8)

0.21 (0.01) 0.71 (0.01) 0.94 (0.02) 0.99 (0.03) 0.72 (0.02) 0.78 (0.02)

27.3 (0.4) 51.7 (1.2) 44.8 (1.7) 45.3 (2.2) 13.8 (1.0) 37.8 (1.5)

0.89 (0.02)

28.5 (1.4)

0.56 (0.02)

51.8 (1.7)

0.77 (0.03) 0.50 (0.01) 0.48 (0.01)

38.2 (2.3) 48.2 (1.3) 44.8 (0.7)

0.81 (0.02) 0.79 (0.02) 0.54 (0.01)

36.4 (1.2) 38.3 (1.4) 51.0 (1.1)

0.77 (0.03) 0.50 (0.01) 0.48 (0.01)

38.2 (2.3) 48.2 (1.3) 44.8 (0.7)

0.63 (0.01)

47.9 (1.1)

0.60 (0.07) 0.80 (0.04) 0.80 (0.04) 0.64 (0.05)

50.4 (6.5) 42.0 (2.5) 36.8 (2.8) 71.0 (4.7)

0.57 (0.03) 0.53 (0.01)

51.0 (3.3) 45.8 (1.3)

42.0 (1.2) 44.8 (1.7)

90.9 (3.9)

50.4 (6.5) 42.0 (2.5) 36.8 (2.8) 71.0 (4.7)

0.56 (0.01) 0.94 (0.02)

0.88 (0.04)

0.60 (0.07) 0.80 (0.04) 0.80 (0.04) 0.64 (0.05)

0.57 (0.02)

69.0 (2.4)

0.81 (0.06)

52.1 (5.6)

52.1 (5.6)

0.84 (0.04)

54.3 (3.1)

45.0 (1.9) 86.0 (6.2) 50.6 (1.9)

31.1 (1.0) 57.8 (3.2) 56.0 (1.4) 45.4 (1.1) 84.8 (3.6) 53.6 (1.2)

0.81 (0.06)

0.48 (0.02) 0.87 (0.06) 0.69 (0.02)

0.67 (0.01) 0.80 (0.03) 0.53 (0.01) 0.51 (0.01) 0.92 (0.04) 0.70 (0.01)

0.48 (0.02) 0.87 (0.06) 0.69 (0.02)

45.0 (1.9) 86.0 (6.2) 50.6 (1.9)

0.68 (0.02) 1.00 (0.04)

50.0 (1.6) 85.0 (3.9)

0.67 (0.03) 0.70 (0.03) 0.43 (0.06)

73.3 (3.6) 72.0 (3.6) 51.8 (6.3)

0.68 (0.02) 0.68 (0.02) 0.48 (0.03)

77.3 (2.2) 75.0 (2.2) 51.6 (3.7)

0.67 (0.03) 0.70 (0.03) 0.43 (0.06)

73.3 (3.6) 72.0 (3.6) 51.8 (6.3)

0.65 (0.02) 0.72 (0.02) 0.59 (0.17)

79.6 (2.3) 81.2 (2.6) 88.9 (17.4)

0.63 (0.02)

42.1 (2.1)

0.63 (0.01)

41.2 (1.2)

0.63 (0.02)

42.1 (2.1)

0.52 (0.08) 0.58 (0.13)

45.9 (8.2) 60.9 (15.3)

0.48 (0.04) 0.47 (0.07)

53.6 (4.8) 64.3 (8.3)

0.52 (0.08) 0.58 (0.13)

45.9 (8.2) 60.9 (15.3)

0.65 (0.01) 0.71 (0.03) 0.49 (0.04) 0.51 (0.08)

41.4 (1.2) 49.7 (2.2) 45.8 (4.2) 59.3 (8.6)

from surrounding structures and their fluctuations over a wide timescale range. In contrast, S2 axis are a purely local measure of dynamics and are only sensitive to motions on a sub-τm timescale (ps–ns).

Discussion Classic allostery is generally assumed to involve changes in structure between at least two states that are regulated at a distance. Recent experimental and

342

Methyl Dynamics of hPTP1E PDZ2 Mutants

Figure 3. The response of side-chain methyl dynamics to point mutations in PDZ2. The change in side-chain order parameter (ΔS2axis) and the associated timescale of motion (Δτe) relative to PDZ2 WT are shown for (a) H71Y, (b) I35V and (c) I20F. Methyl groups with significant changes are shaded black.

theoretical studies point towards a new, broader definition of allostery which includes single domain proteins, 3 protein dynamics, and shifts in the population of conformational sub-states.1,2,27,28 In this new view, it is of interest to explore how protein dynamics might contribute to allostery and/or facilitate communication within a protein. We and

others have used NMR relaxation methods to detect long-range communication within single-domain proteins. 7, 8, 10, 13, 15, 17 For example, 2 H-based methyl side-chain dynamics were used to detect changes in side-chain dynamics parameters in a PDZ domain protein upon peptide binding, several of which were long-range (>9 Å).10 These

Methyl Dynamics of hPTP1E PDZ2 Mutants

343

Figure 4. The I20F mutation recapitulates the side-chain dynamic response due to peptide ligand. The dynamic changes are mapped onto the PDZ2 structure for (a) peptide ligand10 (PDB code 1D5G19), (b) I35V, and (c) I20F mutation (PDB code 1GM145) are shown. The protein secondary structure is colored gray. Side-chain methyl residues with changes ΔS2axis and Δτe are displayed as red or yellow VDW surfaces, respectively. Methyl groups that do not have perturbed dynamics parameters are colored gray. Methyl groups not analyzed are colored black and rendered as sticks. The peptide ligand and mutation are colored green. (d) Changes in ΔS2axis and Δτe for PDZ2/RA-GEF2, I35V, and I20F are mapped onto the linear amino acid sequence.

results revealed a dynamic network within PDZ2. Interestingly, many of these sites were found to be statistically coupled based on the analysis of primary sequence alignments in the PDZ family,20 suggesting a role for dynamics in energetic transmission (i.e. allostery). Here, we have used mutagenesis coupled with peptide binding experiments to probe the energetic features of the PDZ2 dynamic network. Moreover, we have used 2Hbased side-chain relaxation experiments to characterize the dynamic response of various point mutant proteins. A major goal was to establish the energetic and dynamic consequences in response to perturbation of particular sites within the PDZ2 network. Network mutations identify binding “hot-spot” residues in PDZ2 Figure 1 shows the 13 sites where individual point mutations were incorporated into PDZ2. These sites were selected based on previous studies that had identified a network of residues that were energetically20 or dynamically10 linked to peptide binding. Lockless & Ranganathan20 identified about ten residues within the family of PDZ domains that defined an “energetic pathway” and our previous 2H-based methyl relaxation results revealed 14 residues that had perturbed dynamics upon peptide ligand binding.10 Six positions of the energetic pathway

were tested and two showed decreased binding affinity (I20 and I52). Of the 14 dynamically perturbed residues, five were tested and only I20F had a significant effect on binding (Table 1). Two additional sites were tested: a hydrophobic core site (I35; R. Ranganathan, personal communication) and a solvent exposed site (V84). The V84S mutation had no effect on binding, whereas mutations at position 35 had deleterious effects on peptide binding, stability, and displayed a tendency for conformational heterogeneity. Thus, of the 17 mutants tested only four sites showed significant changes in binding affinity (residues 20, 35, 52 and 71). Interestingly, all four of these sites are found in the energetic network while residues at position 20 and 71 are also common to the dynamic network. Together, these results indicate that most residues do not have a large energetic contribution to binding and suggest a “special” role for residues at position 20 and 71. Examination of the PDZ2/RA-GEF2 structure offers clues to rationalize the changes in binding affinity. The side-chain of residue I20 contributes to the hydrophobic cleft that accommodates a Cterminal hydrophobic residue (Val in RA-GEF2). The histidine in helix α2 (H71) makes a hydrogen bond with the penultimate RA-GEF2 residue (Thr side-chain) and this interaction is a key determinant of binding specificity. Thus, reduction in binding affinity at these two positions likely results

344

Table 4. Rotameric side-chain populations for Ile, Val, and Thr from 3JCγN and 3JCγCO coupling constant data for PDZ2, PDZ2/RA-GEF2, and I20F Methyl group

PDZ2 WT 3

JCγCO

3

PDZ2 WT/RAGEF-2

JCγN

P-60

P60

P180

3

JCγCO

3

JCγN

P-60

P60

P180

1.23 (0.13) 2.95 (0.10) 0.90 (0.30) 1.13 (0.16) 0.98 (0.24)

1.78 (0.10) 0.68 (0.09) 2.04 (0.07) 1.84 (0.05) 1.93 (0.06)

0.80 0.16 0.93 0.84 0.89

0.20 0.78 0.07 0.16 0.11

0.00 0.05 0.00 0.00 0.00

1.21 (0.12) 3.07 (0.10) 0.83 (0.30) 1.06 (0.15) 0.95 (0.24)

1.79 (0.05) 0.60 (0.14) 2.05 (0.08) 1.80 (0.06) 1.94 (0.08)

0.81 0.11 0.95 0.83 0.90

0.19 0.82 0.05 0.15 0.10

0.00 0.06 0.00 0.02 0.00

Val9γ1 Val9γ2 Val22γ1 Val22γ2 Val26γ1 Val26γ2 Val30γ1 Val30γ2 Val37γ1 Val37γ2 Val40γ1 Val40γ2 Val58γ1 Val58γ2 Val61γ1 Val61γ2 Val64γ1 Val64γ2 Val75γ1 Val75γ2 Val84γ1 Val84γ2 Val85γ1 Val85γ2

1.45 (0.10) 2.81 (0.06) 3.78 (0.06) 1.00 (0.18)

1.31 (0.06) 0.00 (0.10) 0.47 (0.23) 0.45 (0.21)

0.39

− 0.03

0.64avg

−0.05

0.64avg

− 0.04

0.09

0.90

0.02

0.08

0.49

0.34

0.17avg

0.67

0.24

0.09avg

1.10 (0.12) 2.42 (0.06) 1.81 (0.10) 1.01 (0.24)

0.98 (0.07) 1.03 (0.07) 0.69 (0.13) 2.04 (0.06)

1.32 (0.07) 0.00 (0.10) 0.44 (0.27) 0.59 (0.22) 0.51 (0.20) 0.99 (0.11)

0.41

0.95

1.53 (0.09) 2.80 (0.06) 3.75 (0.06) 1.01 (0.20) 3.00 (0.05) 0.92 (0.18) 1.92 (0.12)

0.63 (0.24)

1.07 (0.15) 0.79 (0.28) 3.31 (0.07)

0.00 (0.10) 3.03 (0.06)

1.58 (0.06) 0.56 (0.17)

− 0.05

0.30

0.75

Thr23γ2 Thr28γ2 Thr70γ2 Thr77γ2 Thr81γ2

1.63 (0.21) 2.43 (0.12) 3.02 (0.10) 1.21 (0.40) 3.35 (0.14)

0.52 (0.33) 0.87 (0.15) 0.86 (0.17) 1.87 (0.12) 0.57 (0.37)

0.19 0.36 0.26 0.85 0.12

0.41 0.64 0.74 0.15 0.88

0.40avg 0.00avg 0.00avg 0.00 0.00

a b c d

0.52

0.10

0.39

0.43

0.13

avg

0.44

0.14

− 0.10

0.97

1.83 (0.10)

0.16

0.00

0.84

1.21 (0.12)

1.78 (0.05)

0.20

−0.02

0.81

1.95 (0.06) 0.82 (0.14)

− 0.05

0.14

0.91

0.78 (0.27) 3.35 (0.06)

1.97 (0.07) 0.85 (0.15)

− 0.06

0.14

0.92

3.43 (0.06) 1.22 (0.10)

0.56 (0.17) 1.75 (0.03)

3.56 (0.10) 1.13 (0.11)

0.33 (0.22) 1.77 (0.10)

− 0.04

0.10

0.94

0.21

0.00

0.79avg

0.05

− 0.04

0.99

0.18

0.02

0.80 avg

JCγN

P-60

P60

P180

Max Pop

Pred Popc

1.08 (0.17)

1.80 (0.06)

0.82

0.16

0.02

0.82

0.93

0.63 (0.54) 1.10 (0.15) 0.92 (0.28)

2.02 (0.09) 1.76 (0.05) 1.86 (0.07)

0.96 0.80 0.86

0.01 0.17 0.11

0.03 0.03 0.03

0.96 0.80 0.86

1.00 0.94 0.97

1.23 (0.13) 2.98 (0.07) 3.84 (0.08) 0.87 (0.28)

1.41 (0.06) 0.33 (0.32) 0.39 (0.35) 0.53 (0.23)

0.23

0.08

0.69avg

0.69

0.87

0.96

0.00

0.04

0.96

1.00

0.52

0.34

0.14avg

0.52

0.89

1.01 (0.16) 2.49 (0.07) 1.89 (0.12) 0.81 (0.34) 3.81 (0.08) 0.90 (0.22)

0.98 (0.09) 1.04 (0.09) 0.68 (0.16) 2.06 (0.07) 0.66 (0.20) 1.86 (0.06)

0.52

0.08

avg

0.52

0.87

−0.08

0.05

1.02

1.02

1.00

0.10

0.04

0.86

0.86

ndd

−0.22

0.30

0.91

0.91

1.00

3.34 (0.08) 2.36 (0.07) 2.86 (0.06) 1.35 (0.10)

0.91 (0.14) 1.63 (0.06) 1.46 (0.06) 1.79 (0.10)

0.74

ndd

JCγCO

− 0.05

0.23

0.82

0.14

0.12

0.74avg

− 0.12

0.06

1.05

− 0.01

0.16

0.85

0.11

0.19

0.70avg

0.69 (0.30) 2.99 (0.08) 1.53 (0.28) 2.46 (0.14) 2.90 (0.12) 1.72 (0.34) 3.42 (0.15)

3.05 (0.07) 1.59 (0.06) 2.83 (0.10)

0.79 (0.16) 1.66 (0.04) 0.92 (0.10)

3.76 (0.10) 1.00 (0.13) 3.23 (0.10) 0.85 (0.17) 2.84 (0.06)

0.51 (0.44) 1.81 (0.10) 0.87 (0.08) 1.51 (0.07) 0.61 (0.17)

1.65 (0.50) 2.30 (0.16) 3.01 (0.09)

0.87 (0.85) 1.15 (0.19) 0.86 (0.19)

0.40 0.46 0.26

0.42 0.54 0.74

0.19avg 0.00avg 0.00avg

3.27 (0.12)

0.58 (0.39)

0.13

0.87

0.00

3

0.40

avg

0.74

b

0.10

0.16

0.25

− 0.07

0.82

0.82

0.89

1.56 (0.07) 0.58 (0.20)

0.07

0.20

0.74avg

0.74

0.94

0.55 (0.38) 0.88 (0.20) 0.75 (0.17) 1.93 (0.17) 0.62 (0.41)

0.21 0.36 0.25 0.79 0.12

0.38 0.64 0.75 0.21 0.88

0.41avg 0.00avg 0.00avg 0.00avg 0.00

0.41 0.64

0.98 0.98

0.79 0.88

1.00 1.00

Bold residues indicate a “slow” timescale of motion (less than ms) for the mutant I20F inferred based on the comparison with the maximum predicted rotameric population (see the text). Maximum rotameric population based on coupling constant data analyzed using a three-site jump model.9 The population for valine is the average of both methyl groups. Predicted maximum rotameric population calculated for I20F based on S2axis data and equation (7) of Hu et al.26 The methyl axis order parameter was not determined.

Methyl Dynamics of hPTP1E PDZ2 Mutants

Ile6γ2 Ile20γ2 Ile35γ2 Ile4γ2 Ile52γ2

avg

I20Fa

I20F 3

Methyl Dynamics of hPTP1E PDZ2 Mutants

345

Figure 5. Changes in side-chain chemical shifts do not correlate with methyl order parameter changes. (a) Methyl region from the 1H-13C correlation spectra of the IzCz relaxation experiment for H71Y, I35V, and I20F. The mutant spectra and PDZ WT are colored red and black, respectively. (b) The plot of the weighted chemical shift change for methylcontaining residues for each mutant protein against the absolute value of the methyl order parameter change (ABS (ΔS2axis)) upon mutation for H71Y, I35V, and I20F.

directly from perturbed interactions between the protein and peptide. Neither I35 nor I52 make direct contact with the RA-GEF peptide, but both are within van der Waals (VDW) radius of residues that participate in direct interactions. Namely, Ile52 is within VDW radius of Ile20 and Ile35 is in close proximity to His71. Thus, the changes in binding are likely indirect. To further understand the structural and dynamic consequences of these hot-spot mutations, we selected I20F, I35V, and H71Y for more detailed structural and dynamic analysis. Main-chain dynamics do not change upon mutation The fast (ps–ns) timescale dynamics of the backbone (15N) and side-chain (2H) of I20F, I35V, and H71Y were probed by spin relaxation measurements. The backbone order parameters (S2) and associated timescale (τe) were determined for residues of each mutant and compared to WT. Similar to PDZ2 WT, residues involved in secondary structure are rigid, with S2 at ∼0.8 and deviations from this occurring only at loops

and termini (data not shown). The ps–ns excursions of the backbone are not significantly perturbed upon either the I20F, I35V or H71Y mutation. This may be most easily seen in an overlay plot of 15N T2 (or R2) values for WT and the three mutants; T2 values are uniform throughout the protein, with the exception of residues 1516, 19, 25–31, 34–35, and 68, which show qualitatively similar Rex contributions in all mutants (data not shown). H71Y, I35V, and I20F PDZ2 mutations elicit context-dependent side-chain dynamical changes The fast timescale dynamics of methyl-bearing side-chains were determined for I20F, I35V and H71Y. For each methyl group, an order parameter (S2axis) and associated timescale (τe) was fit and compared to PDZ2 WT. The His71 to Tyr mutation showed a limited dynamic response (Figure 3(a)), dynamic perturbations occurred at Ala12β, Thr28γ2, Val30γ2, and Ala74β. Of these residues only Ala74 interacts directly with the site of mutation (position 71). However, residues in the β2–β3 loop, including

346 Thr28 and Val30, show μs–ms timescale motion on the backbone for all mutants (data not shown) and WT,10 and three-bond scalar coupling constants indicate extensive rotameric averaging for these loop residues. These observations suggest that fast loop motions in the side-chains of β2–β3 may couple to His71 via slower motions in the β2–β3 loop; μs– ms motions of this loop may indeed undergo subtle perturbations, although our data are insufficient to confirm this. Also possible is that the hydroxyl of Thr28 (and hence the β2–β3 loop) is directly connected to the His71 side-chain via a watermediated hydrogen bond that is lost upon mutation to tyrosine. The structural rationale for changes in ΔS2axis at Ala12β is not clear, since it is essentially solvent exposed and nearly 25 Å from residue 71 and changes in the backbone order parameter (S2) were insignificant. The I35V mutation had a larger side-chain dynamic response relative to that for the H71Y mutation (Figure 3(b)). The changes in dynamics for I35V were located at two distinct regions, each able to form a clear contiguous VDW surface (Figure 4(b)). For instance, Val22, Leu78, Val61, and Val85 form a surface that abuts the site of mutation (I35V). On the other hand, Val9 and Leu89 form a second surface that is distally located from residue 35 (∼8 Å). The only intervening residue between this second surface and Ile35 is Val58, which does not have perturbed ps–ns side-chain dynamics (see Figure 3(b)). Similarly, a direct connection between the two surfaces is not apparent as putative linker residues (Leu11 and Leu87) do not show changes in side-chain dynamics. One possibility is that changes in dynamics occur on a slower timescale, which the deuterium relaxation experiments do not access. Although the methyl peaks for Leu 11 are weak in the 2D correlation spectra and the weak peak for the δ2 methyl of Leu11 cannot be explained by strong J-coupling between γ and δ carbon atoms, there is as yet no direct evidence to confirm a conformational exchange process that leads to line broadening. Alternatively, contiguous pathways may not necessarily be required for transmission, especially if there is structural shifting.29 Additional experiments will be required to verify this hypothesis. Irrespective of the spatial pattern of propagated dynamics in I35V, Figure 3(b) shows that all significant changes in side-chain dynamics show an increase in ps–ns flexibility upon mutation. If there were no corresponding changes in the dynamics of the peptide-bound state, this might reflect an entropic basis for the decrease in binding affinity (Table 1). However, with no current data on the I35V complex, this remains speculative. No direct enthalpic affects are expected, as I35 does not make contact with the RA-GEF2 peptide. On the other hand, loss of favorable enthalpic effects might be significant in I35V with respect to protein stability. Further experiments should aid in characterizing the thermodynamic basis of the changes in stability and binding affinity for I35V.

Methyl Dynamics of hPTP1E PDZ2 Mutants

In contrast to H71Y and I35V, the Ile to Phe substitution at position 20 (I20F) had a marked ps– ns dynamic response (Figure 3(c)). Fourteen residues show a significant perturbation in either S2axis or τe. These residues are mapped onto the solution structure of PDZ2 WT19 (Figure 4(c)). The pattern of change that emerges shows dynamical perturbations radiating from position 20 to residues deep into the hydrophobic core of the structure. As seen in Figure 4(c), residues Leu18δ1, Val22γ1, Val37 γ2 , Ile41 γ2 , Ala46 β , Val58 γ1 , Val85 γ1 , and Leu78δ1 form a contiguous VDW surface. Although Val9γ1 is not within VDW contact of the above residues, there is a possibility that a link to them may exist via slow motions of Leu11 on an unidentified timescale (see above). Finally, no direct connection to the methyl groups of Val64γ2 or Ala69β is evident. However, as in our previous study,10 Thr81 and Val61 may provide connectivity from the perturbation site to this region. Here we have less certainty because the ΔS2axis of Thr81γ2 has a large associated error and the methyl groups of Val61 could not be analyzed due to spectral overlap. Thus, it is possible that Val64γ2 and Ala69β are part of the contiguous network. Taken together, the I20F mutation resulted in side-chain dynamic changes that propagate locally through VDW contacts to distances from ∼7 Å (Val9 and V85) to ∼13 Å (Val64). Interestingly, with the exception of Ile41γ2 , all methyl groups had a positive change in ΔS2axis, signifying an overall rigidification relative to WT PDZ2. The I20F mutation recapitulates the PDZ2/RA-GEF2 dynamic response Of particular interest is the comparison between the dynamic response due to the I20F mutation with that previously determined upon peptide binding.10 Residues having significant changes in side-chain dynamics for each case are highlighted in Figures 3, 4 and Table 3. Inspection of these data indicated that peptide binding and the I20F mutation induce a similar pattern of perturbed side-chain dynamics. To assess this more quantitatively, we considered only methyl-bearing residues for which ΔS2axis and Δτe values were obtained from both the bound/free and I20F/WT comparisons, 23 in total. Of the ten residues that showed changes in side-chain dynamics (non-zero ΔS2axis or Δτe) upon peptide binding, eight of these ten were also dynamically perturbed in the I20F mutant. In addition, nine residues were unperturbed by either mutation or peptide binding. Only four residues were perturbed by the I20F mutation that were not perturbed by ligand binding. In summary, 17 (8 + 9) residues were similarly perturbed (i.e. eight perturbed in both and nine unperturbed in both) and six were differentially perturbed. This is analogous to flipping a coin 23 times and seeing heads 17 times; statistically, there is <1% chance of this occurring at random. From the perspective of I20F mimicking ligand binding, since several of these residues make contact with I20 in

Methyl Dynamics of hPTP1E PDZ2 Mutants

WT-PDZ2, these perturbations might be considered trivial. If we exclude the residues within VDW radius of I20F (Leu18, Val22, Val37, Ala46 and Ile52), the likelihood of obtaining a ligand-like response is still low, <2%. A graphical representation of the similarities between the two responses is shown in Figure 6, in which changes in both S2axis and τe have been simplified to show a single perturbation (ζ, see the Figure legend for details) for each residue. Interestingly, chemical shift perturbations due to peptide

Figure 6. Similarity of PDZ2 side-chain dynamic response of I20F mutation to peptide binding. Shown are perturbations in S2axis and τe, recalculated to provide single effective S2axis and τe values for each methylbearing residue. The ζ represents the magnitude of perturbation for a given residue, I20F – WT (gray) or bound – free (black); ζ values can be positive or negative, although absolute values are shown here. For Ala, Thr, and Met residues, ζ(S2) = ΔS2axis and ζ(τe) = Δτe. For Val, Leu, and Ile residues, ζ(S2) = [(ΔS2axis(a) + ΔS2axis(b))/2], where a and b refer to the two methyl groups; ζ(τe) is equal to the largest observed Δτe (positive or negative) of either of the two methyl groups on a given side-chain. The peptide-bound state refers to the complex of PDZ2 with the RA-GEF2 peptide, as previously reported.10 In nearly every case of significant perturbation (ζ > 2× error), ζ values have the same sign and ζ(S2) values are positive. The sole exception is with Val85 for the bound – free PDZ2 comparison, indicated by an arrow; there, the value of ζ(S2) is negative and represents an increase in flexibility upon peptide binding. Asterisks are placed by each of the eight common residues (see Discussion) that are significantly perturbed (in either S2axis or τe) in both the bound – free and I20F – WT comparisons.

347 binding and the I20F mutation are not strongly correlated (data not shown). It is important to note that the changes in dynamics result from a single point mutation (Ile20 → Phe), which corresponds to a change in volume of ∼10 Å3 (or a change in accessible surface area of ∼36 Å2) rather than the significantly larger buried surface of the peptideligand (∼430 Å2) used in our previous study.10 Because the effects of mutating I20 propagate to two distal surfaces composed of a large number of residues, I20 might be considered a hub in the network of residues in this PDZ domain or the PDZ domain family. In summary, the response due to the I20F mutation recapitulates many of the features seen in the PDZ2/RA-GEF2 complex including the perturbation of side-chain dynamics for residues at a distance. The importance of position 20 in PDZ domains has been recently highlighted by a non-equilibrium computational study that used an anisotropic thermal diffusion (ATD) technique.30 This method follows the anisotropic dissipation of a local thermal perturbation throughout the protein structure during the simulation. Using ATD, Ota & Agard30 observed a “signaling pathway” within PDZ3 of PSD-95 between His76 and Ile45 that passed through Ile31 and Phe29 (between His71 and Val40 via Val22 and Ile20 in PDZ2). Interestingly, the pathway obtained was reciprocal in that similar results were obtained whether the local heating occurred at His76 or Phe29. These results are based on simulations performed in the presence of peptide ligand and it is not clear whether the peptide directly contributes to the intramolecular signaling pathway. ATD simulations in the absence of peptide resulted in a weakened signal with large errors. Our results with PDZ2 suggest that the peptide is not required for signaling propagation (I20F), but may be crucial for reciprocity (H71Y). Lack of thermodynamic coupling between I20 and H71 in PDZ2 The hallmark of allostery is the energetic connectivity between distal sites. Our previous studies have revealed a dynamic network in PDZ2 with respect to ligand binding. 10 The present study probed this network through the use of single point mutants that generally have only small effects on the energetics of ligand binding. Neither of these techniques directly addresses the issue of thermodynamic connectivity. A more rigorous test would be to examine if two positions are thermodynamically coupled by performing double-mutant site cycle analyses. 31–33 Here, we have tested for energetic couplings between position 20 and 71, which are separated by ∼11 Å. Table 2 summarizes the results of this analysis with respect to protein unfolding and peptide binding. Mutation of Ile20 to Phe increases protein stability by nearly 1 kcal/mol, while the H71Y mutation does not significantly affect stability. Comparison of the free energy of unfolding for the double mutant to the sum of the

348 free energy of unfolding of single mutants indicates no non-zero free energy coupling (ΔΔΔGi), within the precision of our measurements. On the other hand, the free energy coupling between residue 20 and 71 with respect to peptide binding might be of more consequence, since the side-chains at position 20 and 71 both interact with peptide and a dynamic response at position 20 was previously detected on peptide binding.10 In support of this, studies of the third PDZ domain (PDZ3) of PSD-95 show a free energy coupling between residues at these analogous positions.20 In PDZ2, the I20F mutation decreased binding affinity by nearly twofold. The mutation at His71 (H71Y) markedly reduced the peptide binding affinity by nearly 200fold. However, the double mutant had ∼370-fold reduction in affinity, indicating that within the error of our measurements, no thermodynamic coupling (ΔΔΔGi) occurs between residue 20 and 71 (Table 2). This serves as a reminder that statistical coupling analysis (SCA) reveals trends in coupling patterns and the results need not apply equally to individual members of a protein family. Along the same line, the existence of thermodynamic coupling (e.g. with respect to ligand binding) between two sites in one protein may not translate into a similar coupling in a homologous protein, especially if the mutations used to probe the coupling are different. This appears to be the case here with H71Y/I20F in PDZ2. It will be interesting to see if different mutation pairs induce thermodynamic coupling between these two sites. Mechanism(s) of dynamic propagation The long-term goal of this work is to increase our understanding in mechanisms of allosteric behavior. How does a localized perturbation lead to allosteric transitions in proteins? Or, equivalently, how does a perturbation propagate beyond its neighboring residues and change the structural and motional properties of distal residues? In the classical twostate allosteric mechanism34 a perturbation can switch the equilibrium between two conformations. Recently, attention has been drawn towards this mechanism of allostery, which is essentially a “switch” mechanism: the kinetics of equilibrium transitions between two states are observed/characterized site-specifically using NMR T2 or relaxation dispersion experiments. 35–37 Key to this approach is the idea that proteins exist as a preexisting equilibrium which can be shifted via perturbation. Although exciting because these experiments access switching on timescales commensurate with many biological processes, there is still a reliance on the classical notion of a concerted switching between two (or sometimes three) conformations. What is the mechanism of the switch? Is it truly concerted or can it be decomposed into smaller (and faster) events? The PDZ domain studied here may be useful because propagated effects (due to ligand binding or mutation) are observed in the absence of large, concerted confor-

Methyl Dynamics of hPTP1E PDZ2 Mutants

mational switching on the μs–ms timescale. Rex is observed in two limited regions of the WT and mutant proteins (see above and Fuentes et al.10), but the Rex appears not to be significantly altered by ligand or mutations. CPMG relaxation-dispersion experiments did not reveal additional “slow motions” (data not shown). More importantly, Rex is not observed (in free or perturbed states) in many of the regions of the protein where propagated changes in side-chain dynamics are observed (see Figure 5 of Fuentes et al.10). Thus, this is not a simple case of shifting a pre-existing population of two (or three) states that are in dynamic equilibrium. The I20F mutant reported here is particularly helpful because it elicits a very similar dynamic response to ligand binding, and from RDCs of I20F versus WT it is clear that structural changes are minimal (Figure 2). That is, we have studied propagation effects (i.e. allosteric behavior) within a minimally altered structural ensemble at the backbone level, which may lead to insights into early events that promote large switching events. Alternatively, our observations of I35V, H71Y, and I20F mutants may indicate that a broadened view of allostery may be needed to account for mutants that have long-range functional effects when evidence of conformational change is lacking. Of significant interest is that this propagation is not random, but appears to be directed along a common network despite the perturbation (either ligandbinding or mutation). In the case of the PDZ domain, this dynamic network appears to be consistent with observed “sparse energetic networks”.20 The question of how and why this propagation is directed remains elusive.

Materials and Methods Cloning, expression and purification of hPTP1E PDZ2 mutants Polymerase chain reaction (PCR) oligonucleotide-directed mutagenesis was used to produce all PDZ2 mutants using the wild-type pET21 PDZ2 (1361-1456)10 plasmid as template. All PCR reactions were verified by automated sequencing (UNC sequencing facility). PDZ2 proteins were expressed and purified as described.10 Nitrogen and carbon isotope labeling was achieved by growing BL21(DE3) cells harboring the appropriate plasmid in minimal media containing 15NH Cl (99%) and D-glucose (U-13C -99%). Random, 4 6 fractionally labeled 2H-methyl samples were produced 15 using minimal media containing NH4Cl (99%), Dglucose (U-13C6-99%), and 60% 2H2O. Protein NMR samples were generally ∼1–2 mM protein dissolved in buffer containing 10% 2H2O, 50 mM phosphate, and 150 mM NaCl (pH 6.8). Measurement of peptide binding affinities The peptide binding affinity for PDZ2 WT and mutants were established in most cases by a fluorescence-based peptide binding assay.38 Specifically, the

Methyl Dynamics of hPTP1E PDZ2 Mutants

349

fluorescence emission was used to monitor the binding of a dansylated peptide to the PDZ proteins (∼μM affinities). Dansylsated RA-GEF2 peptide (dansylKENEQVSAV-COOH) (UNC Microprotein Sequencing & Peptide Synthesis Facility) at 0.2 μM was titrated with PDZ2 proteins (1.5−2 mM stock) in a stirred-cell cuvette (1500 μl starting volume). A three-parameter fit (Sigma Plot; SPSS, Inc.) to equation (1) yielded estimates of Kd:38 y ¼ Fo þ

ðFmax  Fo Þ 1þ

x Kd

x Kd

Dynamics ð1Þ

where y is the fluorescence reading, x is the PDZ protein concentration, and Kd is the dissociation constant for the binding reaction. Fo is the initial fluorescence reading and Fmax is the fluorescence value at saturation. The H71Y and the H71Y/I20F mutants had markedly reduced binding affinity for peptide and hence their binding constants were estimated by titrating peptide ligand (Acetyl-ENEQVSAV-COOH) (UNC Microprotein Sequencing & Peptide Synthesis Facility) to 0.5 mM protein (15N-labeled) in NMR buffer. Changes in chemical shift were monitored by use of 15 N-HSQC spectra recorded as a function of peptide concentration. Chemical shifts from significantly perturbed residues were fit to equation (2):39 yobs yP ¼ ðyPL  yP Þ 

ðKd þ LT þ PT Þ 

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðKd þ LT þ PT Þ2 4PT LT 2PT

quired at 10 °C. Amide and side-chain assignments at 25 °C were obtained by tracking resonances as a function of temperature. Stereospecific assignments were previously obtained for the wild-type protein and could be easily tracked for each mutant†. All NMR data were processed using NMRPipe 2.1 and analyzed using NMRView 5.0.4 and 5.2.2 compiled for Linux worksations.

ð2Þ

where PT, LT, and δobs are the total protein, ligand concentrations, and the observed change in chemical shift upon ligand titration, respectively. δPL and Kd, the chemical shift of the protein–peptide complex and the dissociation constant were determined by non-linear regression analysis (Sigma Plot; SPPS, Inc.) to equation (2). Chemical denaturation Chemical denaturation experiments were performed using protocols described.40 Samples of 50 μM protein in 50 mM phosphate and 150 mM NaCl (pH 6.8) were titrated with a solution of equal protein concentration containing 6.8 M guanidine hydrochloride. Protein solutions were equilibrated overnight at 4 °C. Titrations were performed at least in triplicate using an ATF105 dual-channel titrating fluorometer (Protein Solutions, Inc.) at 25 °C by monitoring tyrosine emission of the unfolding reaction at a wavelength of 305 nm (excitation at 280 nm). Gibbs free energy of unfolding (ΔGu) was determined from sixparameter fits to the denaturation data as described.40 NMR spectroscopy NMR experiments were carried out at 25 °C (calibrated with methanol) unless otherwise noted. Varian Unity Inova spectrometers operating at 500 and 600 MHz 1H frequency and equipped with 1H/15N/13C probes and z-axis pulsed-field gradients were used for all experiments. Backbone and side-chain methyl spectral assignments for all mutants were confirmed using standard triple-resonance assignment experiments. Backbone and side-chain assignment experiments for I35V were ac-

Standard backbone (15N) relaxation experiments,41 with minor modifications,42,43 were used to collect 15N T1, T2, and {1H}-15N nuclear Overhauser enhancement (NOE) data at 500 and 600 MHz for WT and PDZ2 mutants. Sidechain 2H-methyl relaxation experiments for CH2D isotopomers were collected for samples containing random fractional 2H-labeling. Three multi-coherence relaxation experiments (IzCz, IzCzDz, and IzCzDy) were collected at 500 and 600 MHz.24 The details of acquisition, processing, and data analysis were identical to those described previously with PDZ WT.10 Determination of rotational correlation times and 15 N-derived model free parameters PDZ domains (∼10 kDa) are expected to have rotational correlation times, τm, ∼6–7 ns.44–46 Consistent with these expectations, our previous backbone characterization of PDZ2 dynamics,10 at a protein concentration of ∼2 mM, indicated a τm of ∼7.0 ns free or bound to a peptide ligand. At ∼2 mM concentration, analysis of I20F and H71Y relaxation data yielded τm values of 6.2 and 6.5 ns, respectively. We investigated the possibility that τm might be dependent on concentration by examining transverse relaxation rates (T2) as a function of protein concentration (0.5 mM–2 mM) for PDZ2 WT. A uniform decrease in T2 values was noted from 1 mM to 2 mM protein concentration (data not shown). T2 values obtained at a protein concentration below 1 mM (∼0.5 mM) had no further change. Global fitting of relaxation data (T1, NOE, and T2) for PDZ2 WT at 1 mM concentration yielded a τm of ∼6.0 ns. A detailed comparison of the derived modelfree parameters for PDZ2 WT from data taken at 1 mM and 2 mM protein concentration indicated that no significant changes occurred in the derived S2 or τe values. Acquisition of T2 relaxation rates for H71Y and I20F as a function of protein concentration indicated no significant changes from 1 mM−2 mM protein. Relaxation data for I35V at 1 mM protein concentration yielded a τm value of 6.1 ns. All further relaxation data were collected at this concentration. Hence, for the H71Y and I20F mutant proteins the 15N-derived parameters (and deuteriumderived methyl data) were obtained at 2 mM concentration while 1 mM samples were used for PDZ2 WT and I35V. Backbone dynamical parameters were fittted to the five standard motional models using ModelFree 4.1 (Dr Art Palmer, Columbia University) and model selection for PDZ WT.47 The data for each mutant were fit to the same model as obtained for PDZ WT.

† Previously reported prochiral methyl S2axis and τe parameters (for valines and leucines in both free and bound states) were swapped in Figures 3 and 4 of Fuentes et al.10 Correct prochiral methyl assignments are given in Table 3.

Methyl Dynamics of hPTP1E PDZ2 Mutants

350 Residual dipolar couplings Residual dipolar couplings were measured for backbone NH groups for PDZ2 WT, I20F, and H71Y mutant proteins. One-bond 1H-15N splittings were measured using the IPAP-HSQC experiment48 for samples in isotropic and partially aligned media. Protein samples were ∼2 mM in NMR buffer (isotropic media). Partial alignment of protein samples (0.5 mM) was achieved through the use of the C12E5/n-hexanol (Fluka) liquid crystal system49 in NMR buffer at 20 °C. An RDC error was estimated by propagating the error determined from the uncertainty in cross-peak positions.

3. 4.

5.

6.

Determination of χ rotamer populations 1

Three bond 3JCγN and 3JCγCO coupling constants were determined for PDZ2 WT, PDZ2/RA-GEF2, and I20F by the use of established methods.50,51 Rotameric populations were estimated using a three-site jump model as described.9 equation (3) was used to estimate relative rotamer populations for each protein sample: 3

Jexp ðCg NÞ ¼ p180 3 Jtrans ðCg NÞ þ ð1  p180 Þ 3 Jgauche ðCg NÞ

3

Jexp ðCg COÞ ¼ p60 3 Jtrans ðCg COÞ þ ð1  p60 Þ

7.

8.

9.

 3 Jgauche ðCg COÞ ð3Þ

10.

where p180, p−60, , and p60 are the populations of the trans and gauche side-chain rotamers. 3Jtrans and 3Jgauche are the expected coupling constants for methyl side chains (Thr, Ile, and Val) exhibiting full rotamer occupancy. The values used here were taken from:9 Thr: 3Jtrans(CγCO) = 3.4 Hz, 3 Jtrans(CγN) = 1.9 Hz, 3Jgauche(CγCO) = 0.4, 3Jgauche(CγN) = 0.2, and Ile/Val: 3Jtrans(CγCO) = 3.6 Hz, 3Jtrans(CγN) = 2.1 Hz, 3 Jgauche(CγCO) = 0.6, 3Jgauche(CγN) = 0.4.

11.

p60 ¼ 1  p180  p60

Acknowledgements The authors thank members of the Lee lab for helpful discussions and comments on the manuscript. We also thank Rama Ranganathan for sharing unpublished data. A.L.L. is supported by NSF grant (MCB-0344354). E.J.F. was an NSF Minority Fellow (DBI-0314399).

12. 13.

14.

15.

16.

Supplementary Data

17.

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/ j.jmb.2006.08.076

18.

References 1. Cooper, A. & Dryden, D. T. F. (1984). Allostery without conformational change - a plausible model. Eur. Biophys. J. 11, 103–109. 2. Wand, A. J. (2001). Dynamic activation of protein

19.

20.

function: a view emerging from NMR spectroscopy. Nature Struct. Biol. 8, 926–931. Gunasekaran, K., Ma, B. & Nussinov, R. (2004). Is allostery an intrinsic property of all dynamic proteins? Proteins: Struct. Funct. Genet. 57, 433. Lee, A. L., Kinnear, S. A. & Wand, A. J. (2000). Redistribution and loss of side chain entropy upon formation of a calmodulin-peptide complex. Nature Struct. Biol. 7, 72–77. Kay, L. E., Muhandiram, D. R., Farrow, N. A., Aubin, Y. & Forman-Kay, J. D. (1996). Correlation between dynamics and high affinity binding in an SH2 domain interaction. Biochemistry, 35, 361–368. Kay, L. E., Muhandiram, D. R., Wolf, G., Shoelson, S. E. & Forman-Kay, J. D. (1998). Correlation between binding and dynamics at SH2 domain interfaces. Nature Struct. Biol. 5, 156–163. Finerty, P. J., Muhandiram, R. & Forman-Kay, J. D. (2002). Side-chain dynamics of the SAP SH2 domain correlate with a binding hot spot and a region with conformational plasticity. J. Mol. Biol. 322, 605–620. Loh, A. P., Pawley, N., Nicholson, L. K. & Oswald, R. E. (2001). An increase in side chain entropy facilitates effector binding: NMR characterization of the side chain methyl group dynamics in Cdc42Hs. Biochemistry, 40, 4590–4600. Schnell, J. R., Dyson, H. J. & Wright, P. E. (2004). Effect of cofactor binding and loop conformation on side chain methyl dynamics in dihydrofolate reductase. Biochemistry, 43, 374–383. Fuentes, E. J., Der, C. J. & Lee, A. L. (2004). Liganddependent dynamics and intramolecular signaling in a PDZ domain. J. Mol. Biol. 335, 1105–1115. Hu, H., Clarkson, M. W., Hermans, J. & Lee, A. L. (2003). Increased rigidity of Eglin c at acidic pH: evidence from NMR spin relaxation and MD simulations. Biochemistry, 42, 13856–13868. Johnson, E. C. & Handel, T. M. (1999). Effect of hydrophobic core packing on sidechain dynamics. J. Biomol. NMR, 15, 135–143. Millet, O., Mittermaier, A., Baker, D. & Kay, L. E. (2003). The effects of mutations on motions of sidechains in protein L studied by 2H NMR dynamics and scalar couplings. J. Mol. Biol. 329, 551–563. Mittermaier, A. & Kay, L. E. (2004). The response of internal dynamics to hydrophobic core mutations in the SH3 domain from the Fyn tyrosine kinase. Protein Sci. 13, 1088–1099. Clarkson, M. W., Lee, A. L., Hu, H. & Hermans, J. (2004). Long-range dynamic effects of point mutations propagate through side chains in the serine protease inhibitor Eglin c. Biochemistry, 43, 12448–12458. Best, R. B., Rutherford, T. J., Freund, S. M. V. & Clarke, J. (2004). Hydrophobic core fluidity of homologous protein domains: relation of side-chain dynamics to core composition and packing. Biochemistry, 43, 1145–1155. Igumenova, T. I., Lee, A. L. & Wand, A. J. (2005). Backbone and side chain dynamics of mutant calmodulin-peptide complexes. Biochemistry, 44, 12627–12639. Zhang, M. & Wang, W. (2003). Organization of signaling complexes by PDZ-domain scaffold proteins. Acc. Chem. Res. 36, 530–538. Kozlov, G., Banville, D., Gehring, K. & Ekiel, I. (2002). Solution structure of the PDZ2 domain from cytosolic human phosphatase hPTP1E complexed with a peptide reveals contribution of the beta 2-beta 3 loop to PDZ domain-ligand interactions. J. Mol. Biol. 320, 813–820. Lockless, S. W. & Ranganathan, R. (1999). Evolutionarily

Methyl Dynamics of hPTP1E PDZ2 Mutants

21. 22.

23.

24.

25.

26.

27. 28.

29.

30. 31.

32. 33. 34. 35. 36. 37.

conserved pathways of energetic connectivity in protein families. Science, 286, 295–299. Fersht, A. R. (1998). Structure and Mechanism in Protein Science: A Guide for Enzyme Catalysis and Protein Folding. W. H. Freeman and Company, New York. Lipari, G. & Szabo, A. (1982). Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules: 1. Theory and range of validity. J. Am. Chem. Soc. 104, 4546–4559. Lipari, G. & Szabo, A. (1982). Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 2. Analysis of experimental results. J. Am. Chem. Soc. 104, 4559–4570. Muhandiram, D. R., Yamazaki, T., Sykes, B. D. & Kay, L. E. (1995). Measurement of 2H T1 and T1ρ relaxation times in uniformly 13C-labeled and fractionally 2Hlabeled proteins in solution. J. Am. Chem. Soc. 117, 11536–11544. Lee, A. L., Flynn, P. F. & Wand, A. J. (1999). Comparison of 2H and 13C NMR relaxation techniques for the study of protein methyl group dynamics in solution. J. Am. Chem. Soc. 121, 2891–2902. Hu, H., Hermans, J. & Lee, A. L. (2005). Relating sidechain mobility in proteins to rotameric transitions: insights from molecular dynamics simulations and NMR. J. Biomol. NMR, 32, 151–162. Kern, D. & Zuiderweg, E. R. (2003). The role of dynamics in allosteric regulation. Curr. Opin. Struct. Biol. 13, 748–757. Formaneck, M. S., Ma, L. & Cui, Q. (2006). Reconciling the “old” and “new” views of protein allostery: a molecular simulation study of chemotaxis Y protein (CheY). Proteins, 63, 846–867. Clarkson, M. W., Gilmore, S. A., Edgell, M. H. & Lee, A.L. L. (2006). Dynamic coupling and allosteric behavior in a non-allosteric protein. Biochemistry, 45, 7693–7699. Ota, N. & Agard, D. A. (2005). Intramolecular signaling pathways revealed by modeling anisotropic thermal diffusion. J. Mol. Biol. 351, 345–354. Carter, P. J., Winter, G., Wilkinson, A. J. & Fersht, A. R. (1984). The use of double mutants to detect structuralchanges in the active-site of the tyrosyl-transfer RNAsynthetase (Bacillus stearothermophilus). Cell, 38, 835–840. Ackers, G. K. & Smith, F. R. (1985). Effects of site-specific amino-acid modification on protein interactions and biological function. Annu. Rev. Biochem. 54, 597–629. Horovitz, A. & Fersht, A. R. (1990). Strategy for analysing the co-operativity of intramolecular interactions in peptides and proteins. J. Mol. Biol. 214, 597–629. Monod, J., Wyman, J. & Changeux, J. P. (1965). On the nature of allosteric transitions: a plausible model. J. Mol. Biol. 12, 88–118. Volkman, B. F., Lipson, D., Wemmer, D. E. & Kern, D. (2001). Two-state allosteric behavior in a singledomain signaling protein. Science, 291, 2429–2433. Beach, H., Cole, R., Gill, M. L. & Loria, J. P. (2005). Conservation of μs-ms enzyme motions in the apo- and substrate-mimicked state. J. Am. Chem. Soc. 127, 9167–9176. Eisenmesser, E. Z., Millet, O., Labeikovsky, W., Korzhnev, D. M., Wolf-Watz, M., Bosco, D. A. et al. (2005). Intrinsic dynamics of an enzyme underlies catalysis. Nature, 438, 117–121.

351 38. Harris, B. Z., Hillier, B. J. & Lim, W. A. (2001). Energetic determinants of internal motif recognition by PDZ domains. Biochemistry, 40, 5921–5930. 39. Lian, L. & Roberts, G. C. K. (1993). Effects of chemical exchange on NMR spectra. In NMR of Macromolecules. A Practical Approach (Roberts, G. C. K., ed). Oxford University Press, Oxford, UK. 40. Edgell, M. H., Sims, D. A., Pielak, G. J. & Yi, F. (2003). High-precision, high-throughput stability determinations facilitated by robotics and a semi-automated titrating fluorometer. Biochemistry, 42, 7587–7593. 41. Farrow, N. A., Muhandiram, R., Singer, A. U., Pascal, S. M., Kay, C. M., Gish, G. et al. (1994). Backbone dynamics of a free and phosphopeptide-complexed Src homology 2 domain studied by 15 N NMR relaxation. Biochemistry, 33, 5984–6003. 42. Lee, A. L. & Wand, A. J. (1999). Assessing potential bias in the determination of rotational correlation times of proteins by NMR relaxation. J. Biomol. NMR, 13, 101–112. 43. Lee, A. L., Sharp, K. A., Kranz, J. K., Song, X. J. & Wand, A. J. (2002). Temperature dependence of the internal dynamics of a calmodulin-peptide complex. Biochemistry, 41, 13814–13825. 44. Tochio, H., Hung, F., Li, M., Bredt, D. S. & Zhang, M. (2000). Solution structure and backbone dynamics of the second PDZ domain of postsynaptic density-95. J. Mol. Biol. 295, 225–237. 45. Walma, T., Spronk, C., Tessari, M., Aelen, J., Schepens, J., Hendriks, W. & Vuister, G. W. (2002). Structure, dynamics and binding characteristics of the second PDZ domain of PTP-BL. J. Mol. Biol. 316, 1101–1110. 46. Walma, T., Aelen, J., Nabuurs, S. B., Oostendorp, M., van den Berk, L., Hendriks, W. & Vuister, G. W. (2004). A closed binding pocket and global destabilization modify the binding properties of an alternatively spliced form of the second PDZ domain of PTP-BL. Structure (Camb), 12, 11–20. 47. Mandel, A. M., Akke, M. & Palmer, A. G. III, (1995). Backbone dynamics of Escherichia coli ribonuclease HI: correlations with structure and function in an active enzyme. J. Mol. Biol. 246, 144–163. 48. Ottiger, M., Delaglio, F. & Bax, A. (1998). Measurement of J and dipolar couplings from simplified two-dimensional NMR spectra. J. Magn. Reson. 131, 373–378. 49. Ruckert, M. & Otting, G. (2000). Alignment of biological bacromolecules in novel nonionic liquid crystalline media for NMR experiments. J. Am. Chem. Soc. 122, 7793–7797. 50. Grzesiek, S., Vuister, G. W. & Bax, A. (1993). A simple and sensitive experiment for measurement of Jcc couplings between backbone carbonyl and methyl carbons in isotopically enriched proteins. J. Biomol. NMR, 3, 487–493. 51. Vuister, G. W., Wang, A. C. & Bax, A. (1993). Measurement of three-bond nitrogen-carbon J couplings in proteins uniformly enriched in 15N and 13C. J. Am. Chem. Soc. 115, 5334–5335. 52. Koradi, R., Billeter, M. & Wuthrich, K. (1996). MOLMOL: a program for the visualization and analysis of macromolecular structures. J. Mol. Graph. 14, 51–55.

Edited by A. G. Palmer III (Received 25 May 2006; received in revised form 17 August 2006; accepted 26 August 2006) Available online 1 September 2006