Phosphorylation releases constraints to domain motionin ERK2Yao Xiaoa, Thomas Leea,b, Michael Parker Lathama,1, Lisa Rose Warnera,2, Akiko Tanimotoa,3, Arthur Pardia,4,and Natalie G. Ahna,b,c,4

aDepartment of Chemistry and Biochemistry, bHoward Hughes Medical Institute, and cBioFrontiers Institute, University of Colorado Boulder, Boulder,CO 80309

Edited* by Melanie H. Cobb, University of Texas Southwestern Medical Center, Dallas, TX, and approved January 8, 2014 (received for review October 6, 2013)

Protein motions control enzyme catalysis through mechanismsthat are incompletely understood. Here NMR 13C relaxation disper-sion experiments were used to monitor changes in side-chainmotions that occur in response to activation by phosphorylationof the MAP kinase ERK2. NMR data for the methyl side chains onIle, Leu, and Val residues showed changes in conformational ex-change dynamics in the microsecond-to-millisecond time regimebetween the different activity states of ERK2. In inactive, unphos-phorylated ERK2, localized conformational exchange was ob-served among methyl side chains, with little evidence forcoupling between residues. Upon dual phosphorylation by MAPkinase kinase 1, the dynamics of assigned methyls in ERK2 werealtered throughout the conserved kinase core, including many res-idues in the catalytic pocket. The majority of residues in activeERK2 fit to a single conformational exchange process, with kex ≈300 s−1 (kAB ≈ 240 s−1/kBA ≈ 60 s−1) and pA/pB ≈ 20%/80%, sug-gesting global domain motions involving interconversion betweentwo states. A mutant of ERK2, engineered to enhance conforma-tional mobility at the hinge region linking the N- and C-terminaldomains, also induced two-state conformational exchange through-out the kinase core, with exchange properties of kex ≈ 500 s−1 (kAB ≈15 s−1/kBA ≈ 485 s−1) and pA/pB ≈ 97%/3%. Thus, phosphorylationand activation of ERK2 lead to a dramatic shift in conforma-tional exchange dynamics, likely through release of constraintsat the hinge.

The MAP kinase, extracellular signal-regulated kinase 2 (ERK2),is a key regulator of cell signaling and a model for protein

kinase activation mechanisms (1). ERK2 can be activated by MAPkinase kinases 1 and 2 (MKK1 and 2) through dual phosphor-ylation of Thr and Tyr residues located at the activation loop(Thr183 and Tyr185, numbered in rat ERK2) (1, 2). Phosphor-ylation at both sites is required for kinase activation, resultingin increased phosphoryl transfer rate and enhanced affinity forATP and substrate (3).Conformational changes accompanying the activation of ERK2

have been documented by X-ray structures of the inactive, un-phosphorylated (0P-ERK2) and the active, dual-phosphorylated(2P-ERK2) forms (4, 5). Phosphorylation rearranges the activa-tion loop, leading to new ion-pair interactions between phospho-Thr and phospho-Tyr residues and basic residues in the N- andC-terminal domains of the kinase core structure. This leads toa repositioning of active site residues surrounding the catalyticbase, enabling recognition of the Ser/Thr-Pro sequence motif atphosphorylation sites and exposing a recognition site for inter-actions with docking sequences in substrates and scaffolds (6).Less is known about how changes in internal motions con-

tribute to kinase activation. Previous studies using hydrogen-exchange mass spectrometry (HX-MS) and electron paramagneticresonance spectroscopy (7–9) led to a model where conforma-tional mobility at the hinge linking the N- and C-terminal domainsis increased by phosphorylation, therefore releasing constraintsneeded for activation. Such a model differs from other types ofautoinhibitory mechanisms in protein kinases, which involve

interactions with domains outside the kinase core (10, 11). How-ever, how hinge flexibility regulates ERK2 is unknown.NMR relaxation dispersion methods enable protein dynamics

to be monitored by measuring exchange between conformationalstates (12). In particular, Carr–Purcell–Meiboom–Gill (CPMG)relaxation dispersion experiments report on motions on slow(100–2,000 s−1) timescales (13), which are often important for en-zymatic function (13–16). In the CPMG experiment, exchangebetween different conformational states is probed with varyingtimes between “refocusing” pulses. Conformational exchangeleads to imperfect refocusing, thus decreasing the intensity of theNMR signal. Increasing the pulse frequency allows less chancefor conformational exchange, and therefore increased NMRsignal intensity. For a given pulse frequency, analysis of the signalintensity yields the effective relaxation rate for the resonance,R2,eff. This is typically plotted as a relaxation dispersion curve, whichcan be fit to a two-state conformational exchange process (e.g.,A ⇌ B interconversion). Fitting extracts the populations and theexchange rates between states, thus reflecting the thermodynamicsand kinetics of the system (17, 18).Here we performed CPMG relaxation dispersion experiments

at multiple field strengths to compare the dynamic properties of[13C]methyl-labeled ERK2 in its phosphorylated and unphos-phorylated states. The results demonstrate that phosphorylationcauses a significant change in exchange dynamics throughout thekinase core, consistent with a global domain motion. Increasinghinge mobility by introducing mutations at the hinge also pro-motes domain motion within the core but with differing kineticsand populations. Taken together, the results show that large

Significance

This paper uses NMR methods to compare the dynamics of aprotein kinase in its active and inactive states. The results showthat domain movements in the MAP kinase ERK2 are inherentlyconstrained until the enzyme is activated by phosphorylation,with the constraint located at the hinge region. This representsan important mode for dynamical regulation in ERK2, not an-ticipated from previous X-ray structural analyses.

Author contributions: Y.X., T.L., A.P., and N.G.A. designed research; Y.X., T.L., M.P.L.,L.R.W., A.T., and A.P. performed research; M.P.L. contributed new reagents/analytic tools;Y.X., T.L., M.P.L., L.R.W., A.T., A.P., and N.G.A. analyzed data; and Y.X., M.P.L., A.P., andN.G.A. wrote the paper.

The authors declare no conflict of interest.

*This Direct Submission article had a prearranged editor.1Present address: Department of Molecular Genetics, University of Toronto, Toronto, ON,Canada M5S 1A8.

2Present address: Institute of Structural Biology, Helmholtz Zentrum München, 85764Neuherberg, Germany.

3Present address: Department of Chemistry and Biochemistry, The Ohio State University,Columbus, OH 43210-1173.

4To whom correspondence may be addressed. E-mail: natalie.ahn@colorado.edu orarthur.pardi@colorado.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1318899111/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1318899111 PNAS Early Edition | 1 of 6

BIOCH

EMISTR

Y

changes in dynamics accompany ERK2 phosphorylation, whichare influenced by conformational mobility at the hinge. We pro-pose that the activation of ERK2 involves removing inhibitoryconstraints to domain motion, which are conferred by the internalarchitecture of the kinase.

ResultsNMR Peak Assignments and Chemical Shift Behavior. ERK2 wasselectively labeled with [methyl-1H,13C]Ile, -Leu, and -Val (ILV) asdescribed inMaterials and Methods (19, 20) and activated by in vitrophosphorylation using active MKK1 (21). Fig. 1A overlays 2D(13C,1H) methyl transverse relaxation-optimized spectroscopy(TROSY) heteronuclear multiple-quantum coherence (HMQC)spectra for 0P- and 2P-ERK2 at 25 °C. In 0P-ERK2, 140 of 144predicted ILV methyl resonance peaks were observed. Of these,70 methyl peaks were assigned to 60 residues by combining datafrom site-directed mutagenesis, through-space 1H-1H nuclearOverhauser effects (NOEs) between 1H,13C-labeled methylsanalyzed using the X-ray structure of 0P-ERK2, and through-bond intraresidue (HMCM[CG]CBCA) experiments (Fig. S1)(4, 5, 22). Most ILV residues in the hydrophobic core of the kinasewere assigned, including those in the structurally conservedα-helices (αC–αH) and β-strands (β1–β5, β7, β8) as well as non-conserved helices in the MAP kinase insert (α1L14, α2L14) and Cterminus (αL16). Assignments were limited in the activation loopand surface-accessible loops, due to the lack of methyl-methylNOEs in these regions. In 2P-ERK2, 137 of 144 predicted methylresonances were observed, of which 67 were assigned by trans-ferring assignments from 0P-ERK2 and confirmed by methyl-methyl NOE measurements of 2P-ERK2.The differences in [13C]methyl chemical shifts between 0P-

and 2P-ERK2 (in ppm), jΔδ13Cj, provide a sensitive probe ofchanges in the local environment. Eight residues showed signif-icant chemical shift differences between the two forms of ERK2,ranging between 0.16 and 0.47 ppm, whereas 16 residues showedmoderate changes, ranging between 0.1 and 0.15 ppm (Fig. 1Band Fig. S2). Two of the three residues with the largest jΔδ13Cjwere I82 and L154 (0.47 and 0.29 ppm, respectively), which in-teract with the hinge that links the N- and C-terminal domains.L154 is located in the β7-strand, and I82 is located in the αC-β4loop. These form hydrophobic contacts with each other as well aswith M106 at the hinge (Fig. 1C). Large chemical shift changes at

these residues were surprising, given their distance from the siteof phosphorylation. The X-ray structures showed no significantconformational differences between 0P- and 2P-ERK2 aroundthese residues (Fig. 1C). Significant jΔδ13Cj were also observedat I72 near the ATP binding site in helix αC, I241 in helix αG,and I253 in the MAP kinase insert (0.41, 0.26, and 0.25 ppm,respectively). Other residues with measurable jΔδ13Cj are shownin Fig. S2. Based on the X-ray structures, chemical shift changeswere expected in the activation loop and C-terminal L16 loop,but could not be determined due to incomplete assignments.Overall, the NMR chemical shift data reported significant changesin the chemical environment of ILV methyl side chains at thehinge, ATP binding site, and MAP kinase insert upon phosphory-lation of ERK2.

Residues in 0P-ERK2 Show Local Conformational Dynamics. Confor-mational dynamics of ILV methyl groups were examined in0P-ERK2 using (13C,1H) multiple-quantum CPMG relaxationdispersion experiments (18). Methyl groups undergoing confor-mational exchange on the microsecond-to-millisecond timescaleshowed changes in their effective relaxation rate R2,eff, measuredas a function of the frequency of the refocusing pulses (νCPMG).Relaxation dispersion curves, collected at 25 °C and at three fieldstrengths (600, 800, and 900 MHz), are shown in Fig. 2 and Fig.S3. Dynamics were indicated when Rex, the contribution to R2,efffrom exchange, was significant (>4 s−1), yielding curvature in theplot of R2,eff vs. νCPMG (Fig. 2 and Table S1). No evidence fordynamics was indicated when Rex ≈ 0 s−1. For individual ILVmethyl groups, the dispersion curves at 600, 800, and 900 MHzwere fit to a two-state exchange model (A ⇌ B), yielding ex-change rate constants (kex = kAB + kBA) (17, 18). Under optimalconditions, the populations (pA and pB) and the 13C chemical shiftdifferences between the two states, jΔω13Cj (in Hz, or jΔδCPMG

13Cjin ppm), were obtained for individual methyl groups (17, 18).Methyl peaks from 13 residues in 0P-ERK2 showed significant

conformational exchange dynamics (Fig. 3A and Table S1). Theseresidues were clustered in three regions located in the cleft be-tween the N- and C-terminal domains, including the αC-β4 loop,β7-β8 loop, and helix αE. Other regions with significant exchangewere located near the activation loop and the P+1 loop, helix αG,and helices in the MAP kinase insert, α1L14/α2L14. Individualfits of the relaxation dispersion data yielded kex ranging between

BN-terminal domain

C-terminal domain

Hinge

Activation lip

ATP binding site

MAP kinase

αG

DEJL binding site

αC

αE

0.1

0.4

0.2

0

0.5

0.3

L16

αF

α2L14

αH

αD

αL16

β1β2β3β4

β7β8

α1L14

β9β6

β5

C

M106

L154

I82

1H ppm

13C

ppm

0.00.51.01.5

15

20

25

A

I72

I131I241I345

I82

I101

L155L242

V143

V271

L288

L67

L73

L74

L220

I29

4.13.94.4

0.60.7

7.08.0

I241

I82L154

I72

I253

P+1 site

|Δδ13C| (ppm)

Fig. 1. ERK2 phosphorylation induces chemical shift changes at the hinge region. (A) Two-dimensional (13C,1H) HMQC spectra of [methyl-13C]ILV-labeled 0P-ERK2 (blue) and 2P-ERK2 (red) recorded at 800 MHz and 25 °C. (B) X-ray structure of 0P-ERK2 (Protein Data Bank ID code 1ERK) showing spheres representingpositions of ILV methyls, with 13C chemical shift differences (jΔδ13Cj) between 0P-ERK2 and 2P-ERK2 indicated by the color scale (also see Fig. S2). UnassignedILV methyls are represented by white spheres. DEJL is the docking site motif for ERK, JNK, and LXL. (C ) Structural alignment of 0P-ERK2 (blue) and2P-ERK2 (red). The expansion shows that the side chains of L154 and I82, which have two of the largest jΔδ13Cj (B), form hydrophobic interactionswith each other and with M106 at the hinge. Structure images were prepared using PyMOL (www.pymol.org).

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1318899111 Xiao et al.

900 and 2,200 s−1 (Table S1). These exchange processes were faston the NMR chemical shift timescale (i.e., kex > jΔω13Cj), andtherefore it was not possible to confidently extract populationsand chemical shift differences for individual methyls. Attemptsto globally fit all or subsets of residues resulted in poorly definedpopulations. Thus, relaxation dispersion measurements on 0P-ERK2 were inconsistent with ILV residues undergoing a singleexchange process. Instead, they reported fast conformationalexchange processes in subdomains of the kinase, with little or noevidence for coupling between residues, including those withinteracting methyl groups.

Phosphorylation of ERK2 Induces Large-Scale Conformational ExchangeDynamics. Significant changes in conformational exchange dynamicswere observed between 0P-ERK2 and 2P-ERK2, as reflected by

differences in their dispersion curves (Fig. 2). Methyls from 22residues in 2P-ERK2 showed significant Rex. For many ILVmethyls, the values of kex, populations, and

13C chemical shift dif-ferences between the two states in ppm jΔδCPMG

13Cj could beconfidently fit for individual residues (Fig. 3B and Table S2).Nineteen residues throughout the kinase core and in the catalyticpocket could be globally fit (Fig. 3C), consistent with a single ex-change process with kex = 300 ± 10 s−1 and pA and pB populationsof 20% and 80% (± 0.6%). They included residues at the hinge,αC-β4 loop, and β5, which formed the hydrophobic cluster at thecleft between the N- and C-terminal domains; N-terminal domainresidues located at the interface between αC and β3-β4 strands; C-terminal domain residues at the interface between αE and αF; andC-terminal domain residues between αF and αG (Fig. 3C). Incontrast, residues in the MAP kinase insert could not be fit globally,but individual fits yielded kex values of 640–750 s−1, significantlylower than the corresponding kex of 1,200–2,200 s

−1 in 0P-ERK2.The results revealed that phosphorylation of ERK2 induces adramatic change in dynamics, where internal motions becomedominated by a slow-exchange process characterized by in-terconversion between two major conformational states withpopulations of 20% and 80%.

Phosphorylation Shifts the Equilibrium Between Two ConformationalStates. We next examined the temperature dependence of theHMQC spectra of 2P-ERK2. Several methyls appeared as twodiscrete peaks at 5 °C, indicating two conformational states inslow exchange on the NMR chemical shift timescale (Fig. 4A).These methyls showed similar populations of the two states asa function of temperature, varying from 50%:50% at 5 °C to80%:20% at 25 °C. Importantly, the populations at 25 °Cmatched those obtained from the global fits of the relaxationdispersion experiments (Table S2). This demonstrates that theconformational exchange process reflects two discrete conformers,which shift their populations with temperature.Thermodynamic parameters determined from van’t Hoff plots

were most accurately measured for residue I72, yielding ΔH° =9.7 ± 1.1 kcal/mol and ΔS° = 36 ± 2 cal·mol−1·K−1 (Fig. S4A).Thermodynamic parameters estimated for other residues (e.g.,I72, I131, and L288) were consistent with these values, indicatingthat the same conformational exchange processes underlie dy-namics throughout most of the molecule. The exception was I241,located within helix αG, which gave ΔH° = 5.8 ± 1.1 kcal/moland ΔS° = 22 ± 3 cal·mol−1·K−1 (Fig. S4B), indicating a dis-tinct process.In 0P-ERK2, each methyl group appeared as one peak in the

HMQC spectra. In each case, peaks from 0P-ERK2 overlappedwell with those corresponding to the minor conformational statein 2P-ERK2 (i.e., with 20% population at 25 °C) (Fig. 4B). Thus,

30

50

70

90

R2ef

f (s−

1 )ef

f−

1

30

50

70

90

R2

(s

)

30

50

70

90

R2ef

f (s−

1 )

0 200 400 600 800 1000 0 200 400 600 800 1000νCPMG (Hz)

Ile72

Ile345

Leu155

600 800 900

Rex

0P-ERK2 2P-ERK2

Ile72

Leu155, Rex≈ 0 s-1

Ile345, Rex≈ 0 s-1

Fig. 2. CPMG dispersion experiments show changes in methyl dynamics fol-lowing ERK2 phosphorylation. The dispersion curvesmeasured at 25 °C and 900(green), 800 (blue), and 600 (red) MHz are shown for representative residues in0P-ERK2 (Left) and 2P-ERK2 (Right). The Rex is shown for I72 in 2P-ERK2 at 800MHz. Lines show individual fits to the Carver–Richards equation, and the errorbars were estimated from peak intensities of duplicate measurements.

L155

I345I70I101

I87

I82L154**

L288

L135*L148I138*

I81*

L235**

I241

L67

A B

I72V143*

I131

I241

L198

I81I138**

L73*

I163**

L242L235

L262I253

C

400800

2000kex (s-1)

12001600

Fig. 3. ERK2 phosphorylation induces global confor-mational exchange in the kinase core. (A) Assignedmethyls in 0P-ERK2 are shown as spheres, with kexvalues indicated by the color scale for individual fits ofmethyls that have Rex > 4 s−1. The asterisks and doubleasterisks indicate methyls having errors in kex > 50%and kex > 100%, respectively, with the latter methylsshown as black spheres. Methyls with no observabledynamics (Rex ≈ 0 s−1) are shown as white spheres.Thr183 and Tyr185 are shown as black sticks. (B) Resi-dues in 2P-ERK2 with observable dynamics each fittedindividually, colored as in A. (C) In 2P-ERK2, 19methyls could be fit globally to a single rate con-stant (kex = 300 ± 10 s−1) and population (pA/pB = 80/20 ± 0.6%), demonstrating global domain motionfor residues throughout the kinase core. Residues inthe MAP kinase insert subdomain could not be fit bya global process.

Xiao et al. PNAS Early Edition | 3 of 6

BIOCH

EMISTR

Y

based on chemical shifts, the major state in 0P-ERK2 corre-sponds to the minor state in 2P-ERK2. In 0P-ERK2, the singleobserved conformer (population >99.5%) exchanges with anunobservable conformer (population <0.5%), from which ΔG°can be estimated as >+3.1 kcal/mol. In 2P-ERK2, interconvert-ing conformers represent two states with populations of 80% and20%, yielding ΔG° = −0.8 kcal/mol. Two-dimensional (13C,1H)HMQC experiments performed on 0P-ERK2 and 2P-ERK2 inthe presence and absence of saturating Mg2+-adenosine-5′-(β,γ-imido)triphosphate (AMP-PNP) showed little or no effecton the populations in either kinase, indicating that the majorstate in 2P-ERK2 is not affected by nucleotide binding (Fig. S5).

The Conformational Switch Is Controlled by Hinge Residues.We nextasked what type of dynamics might be reflected by a large-scaletwo-state exchange process involving residues throughout theconsensus kinase core. Prior HX-MS studies suggested thatphosphorylation of ERK2 leads to enhanced backbone flexibilityat the hinge, reducing the constraint for interdomain motionbetween the N- and C-terminal domains (9). We therefore askedwhether increased hinge motion could affect the conformationalexchange process, as measured by NMR.To address this, mutations were engineered to increase con-

formational mobility at the hinge, by replacing residues M106-E107 with G-G to create an “ME/GG-ERK2” mutant. HMQCspectra were acquired on the [methyl-13C]ILV-labeled ME/GG-ERK2 and assigned as described in Materials and Methods. Thechemical shifts of the methyls in ME/GG-ERK2 overlaid wellwith wild-type 0P-ERK2, except for residues near the mutationsite, where structural changes would be expected (Fig. 5 A and B).Relaxation dispersion experiments on unphosphorylated

ME/GG-ERK2 showed that methyls on 21 ILV residues had sig-nificant dynamics, located throughout the enzyme (Fig. 5C andTable S3). These corresponded to 12 of the 13 residues withsignificant Rex in 0P-ERK2, and 14 of 22 residues with significantRex in 2P-ERK2. Importantly, 16 residues throughout the kinasecore could be fit to a single exchange model, indicating a globalmotion with kex = 500 ± 60 s−1 and populations of 97 and 3 ±0.2% (Fig. 5C). The similar chemical shifts for ME/GG-ERK2and 0P-ERK2 mean that the dominant state is similar in both

proteins. Within the MAP kinase insert of ME/GG-ERK2, kexvalues of 1,200–1,600 s−1 were observed, comparable to the ex-change rate constants in 0P-ERK2. Therefore, ME/GG par-tially mimics the shift in conformational exchange dynamicsinduced by phosphorylation and activation of ERK2.

DiscussionOur study demonstrates that the catalytic activation of a eukary-otic protein kinase elicits significant changes in protein dynam-ics. In 0P-ERK2, side chains show fast, localized motions on amicrosecond-to-millisecond timescale, with little evidence of cou-pling. In the phosphorylated enzyme, side-chain motions are dom-inated by a global exchange process involving residues throughoutthe kinase core. The NMR data demonstrate that large-scaleinterdomain motion, with an exchange rate constant of 300 s−1,accompanies phosphorylation and activation. Motions withinthe MAP kinase insert indicated a separate process, illustratingthat independent subdomain motion occurs outside the con-sensus kinase core. The results support a model, illustrated inFig. 6A, in which unphosphorylated ERK2 is stabilized in theinactive conformer by an inhibitory constraint. This is releasedupon phosphorylation, allowing a global shift in equilibriumbetween conformational states, illustrated here as a hypotheti-cal domain movement.Lateral or rotational movements between the N- and C-terminal

domains provide one conceptual model to describe the globalmotions observed in active ERK2. Interconversions between“open” and “closed” domain conformations have been noted inother enzymes, including protein kinases (23, 24). In the cAMP-dependent protein kinase catalytic subunit (PKA-C), the X-raystructure of the apo form shows an open conformation, whereasa binary complex with Mg2+-AMP-PNP shows a closed confor-mation formed by rotation of the N-terminal domain and closureof the catalytic cleft around the nucleotide (25, 26). NMRrelaxation measurements of backbone amides in the nucleotide–PKA-C binary complex show global exchange behavior in residueslining the catalytic core (27, 28), which are absent in apo PKA-C,and have been interpreted as an equilibrium shift to a closedconformation. Thus, PKA-C and active ERK2 share characteristics

0.550.65

7.5

8.0

0.550.65 0.550.65 0.550.65 0.550.65 0.550.65

0.300.40

13.013.514.0

0.300.40 0.300.40 0.300.40 0.300.40 0.300.40

0.850.95

12.0

12.5

13.00.850.95 0.850.95 0.850.95 0.850.95 0.850.95

-0.25-0.15 -0.25-0.15 -0.25-0.15 -0.25-0.15 -0.25-0.15

A 5 C 10 C 15 C 20 C 25 C 25 CB

1H ppm

13C

ppm

Ile72

Ile131

Ile241

-0.25-0.15

26.0

26.5

27.0

Leu288

Fig. 4. Methyl peaks reveal a slow conformational exchange process in2P-ERK2. (A) Two-dimensional (13C,1H) methyl HMQC spectra of 2P-ERK2from 5 to 25 °C show that many methyls have two distinct peaks that areundergoing slow conformational exchange on the NMR chemical shifttimescale. The black lines on the 5-°C spectra highlight the two exchangingpeaks, indicating two well-populated states. (B) Overlay of spectra from0P-ERK2 (black) and 2P-ERK2 (red) at 25 °C shows that the higher-populatedstate in 0P-ERK2 corresponds to the lower-populated state in 2P-ERK2.

0.300.350.40

13.5

14.0

-0.20-0.16

26.426.626.8

0.580.62

7.4

7.8

8.2

Ile72

Ile131

Leu288

Val99

400800

2000kex (s-1)

12001600

0.850.90

21.5

22.0

1H ppm

13C

ppm

A CB

0.1

0.4

0.2

|Δδ13C| (ppm)

0

0.5

0.3

Fig. 5. ME/GG hinge mutations induce conformational exchange in thekinase core but not the MAP kinase insert. (A) Overlays of several methylregions of the 2D HMQC spectra showing similar chemical shifts for thesemethyls in 0P-ERK2 (blue) and ME/GG-ERK2 (red). (B) The jΔδ13Cj between0P-ERK2 and ME/GG-ERK2 are indicated by the color scale, and unassignedILV methyls are represented by white spheres. The jΔδ13Cj between 0P-ERK2and ME/GG-ERK2 are small throughout the protein, except for residues closeto the mutated M106 and E107 (shown in pink). (C ) Sixteen methyls inME/GG-ERK2 could be globally fit with kex (500 ± 60 s−1; shown as cyan spheres)and population (97 and 3 ± 0.2%), indicating a single exchange processthroughout the kinase core, similar to 2P-ERK2. The assignedmethyls in theMAPkinase insert are shown as spheres, with kex values indicated by the color scale.

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1318899111 Xiao et al.

consistent with interconversion between open and closed con-formers, with comparable rates of conformational exchange (kex ≈200 s−1 in PKA-C and 300 s−1 in 2P-ERK2). However, in PKA-C,the closed conformation is stabilized by nucleotide binding (28),whereas Mg2+-AMP-PNP has little or no effect on the equilibriumbetween conformers in ERK2 (Fig. S5). This suggests that thesimple domain closure observed in PKA-C may not adequatelydescribe the dynamics in ERK2, where global exchange is allo-sterically stabilized by phosphorylation but not by ligand binding.Other models are possible, for example, one involving rotation

of secondary structures within the N-terminal domain. In X-raystructures of ERK2, phosphorylation induces rearrangement ofthe activation loop, which in turn directs movement of helices αCand αL16 and refolding of the C-terminal L16 extension (4, 5). Anetwork of hydrogen bond and hydrophobic residue interactionsthroughout the N-terminal domain enables communication be-tween these structures as well as connectivity with residues inthe β3–β5 strands. It is possible that this network extends tothe hinge through connections with the αC-β4 loop, and that thelarge chemical shift changes induced by phosphorylation withinthis region (Fig. 1B) reflect environmental changes due to ro-tational movements in the N-terminal domain.An important finding from our study was that the unphos-

phorylated ME/GG hinge mutant induced a global exchangeprocess similar to that of 2P-ERK2. This involved residues aswidely spread across the kinase core as in 2P-ERK2, stronglyimplying that the constraints to global exchange in 0P-ERK2 canbe released by enhancing hinge flexibility. Mechanically, ERK2might be conceptualized as having a strong spring constant at thehinge, which is weakened upon either phosphorylation or muta-tion, thus leading to a change in population of the conformationalstates. Fig. 6A illustrates this model, where the ME/GG mutationbypasses the constraint to domain movement in 0P-ERK2 bymimicking the effect of phosphorylation on conformational mo-bility at the hinge. This increased hinge flexibility lowers the bar-rier to domain movement relative to 0P-ERK2. However, unlike2P-ERK2, ME/GG-ERK2 showed a smaller population (3%,instead of 80%) of the “active-like” conformer (Fig. 6).The specific activity of ME/GG-ERK2 (0.06 nmol·min−1·mg−1)

was comparable to that of 0P-ERK2 (0.08 nmol·min−1·mg−1), bothsignificantly lower than 2P-ERK2 (264 nmol·min−1·mg−1). TheNMR relaxation dispersion data summarized in Fig. 6A show nocorrelation of the kinase activity with the kinetics or populationsfor any of the studied forms of ERK2. Instead, the data supporta model where multiple events take place following the phos-phorylation of ERK2. One event involves removal of a constraint

to global exchange dynamics, illustrated by domain movement,which can be induced by either ME/GG or phosphorylation. Asecond event involves stabilization of the “active” conformationin 2P-ERK2, illustrated by ionic interactions that promoteinteractions between the N- and C-terminal domains (Fig. 6C).Further work is needed to understand the details of the con-formational transitions, as well as the kinetic and thermodynamiccontributions, that occur upon activation of ERK2.Compared with the large amount of structural information on

protein kinases, the understanding of how kinases are regulatedat the level of internal protein motions is much less well de-veloped. A major conclusion of this study is that, before phos-phorylation, ERK2 is maintained in its inactive form by amechanism that imposes constraints on protein dynamics. Unlikeautoinhibitory mechanisms that involve intra- or intermolecularocclusion at the catalytic site in other kinases, the constraintsin ERK2 involve the hinge region, distal from the site of phos-phorylation. The importance of hinge mobility has been suggestedin other protein kinases. For example, allosteric communicationbetween the activation loop and hinge was reported in fibroblastgrowth factor receptor 2 (FGFR2), which forms an autoinhibitory“molecular brake” involving a triad of interacting residues atthe hinge, αC-β4 loop, and β8 strand (29). Structural studies alsoshowed an extensive hydrogen-bond network accompanied by re-duced hinge flexibility in zeta-chain–associated protein kinase 70(ZAP-70), which was proposed to maintain the kinase-inactiveform (30). Finally, molecular dynamics calculations suggestedlocal unfolding of the hinge in the catalytic domain of epidermalgrowth factor receptor (EGFR), which was proposed as an in-termediate step toward the transition from the inactive to the activestate (31, 32). However, in contrast to FGFR2, ZAP-70, andEGFR, the X-ray structures of ERK2 show no significant con-formational changes around the hinge upon phosphorylation (4,5). Likewise, both inactive and active X-ray structures of ERK2show a properly formed catalytic site and an intact “hydrophobicspine” network, whose misalignment underlies the mechanism ofinactivation in many protein kinases (33, 34). Thus, the mechanismof communication from the activation loop to the hinge mostlikely involves subtle reorganization of residues, suggesting thatthe autoinhibitory regulation in ERK2 involves architecturalfeatures that may be unique to this kinase. Our findings now setthe stage for future studies to define the details of this archi-tecture and understand its prevalence across the protein kinasesuperfamily.

R

pTpTpY pY

R

TY

T Y

M E

G G

ME

G G

T Y

ME

0P-ERK2

ME/GG

2P-ERK2

60 s-1

240 s-1

485 s-1

15 s-1

M E

>99.5%

97%

20% 80%

3%

<0.5%

T Y

A B

C

pThr183

Arg68 Arg65

Arg170

Arg146 pTyr185

Arg189

Arg192Tyr185

Thr183

> 3.1 2.1

0.8

0P-ERK2 2P-ERK2 ME/GG

G°

(kca

l/mol

)

Reaction coordinates

“inactive”

“active”

“active-like”

Fig. 6. Model for activation of ERK2, involving therelease of constraints to domain movement andchange in energy landscape induced by phosphor-ylation. (A and B) Before phosphorylation, 0P-ERK2is in an inactive conformation that is constrainedfrom domain motion involving rigidity at thehinge (illustrated by the straight thick line betweenhinge residues M and E). The CPMG dispersion datashow no evidence for additional conformations,indicating they must have low populations (<0.5%),reflecting ΔG° > +3.1 kcal/mol. Phosphorylationenhances hinge mobility (illustrated by the wavythin line) and shifts the equilibrium to favor theactive conformer with ΔG° = −0.8 kcal/mol and rateconstants of kAB = 240 s−1 and kBA = 60 s−1. TheME/GGmutation partially relieves the constraint to domainmotion by increasing mobility at the hinge, sam-pling of the active-like conformer (3%), and re-ducing ΔG° to +2.1 kcal/mol. The activation energy is not known for 0P-ERK2, indicated by the dashed line in B. (C) 0P-ERK2 and 2P-ERK2 structures show howphosphorylation at the activation loop promotes interactions between the N- and C-terminal domains through ion pairing between pT183 and Arg65/68 inhelix αC, which may stabilize the active form. The ERK2 backbone is colored white, phosphates are colored orange, Thr183 and Tyr185 are colored red, Argside chains are represented as sticks, and surfaces are shown for the guanidinium groups.

Xiao et al. PNAS Early Edition | 5 of 6

BIOCH

EMISTR

Y

Materials and MethodsProtein Preparation and Enzyme Assays. Wild-type or mutant rat His6-ERK2was expressed from the pET23a plasmid in Escherichia coli BL21(DE3) cells(35). [methyl-13C]ILV-labeled samples of 0P-ERK2 were prepared and purifiedas described in SI Materials and Methods. 2P-ERK2 was prepared from0P-ERK2 in vitro using constitutively active mutant MKK1 as described (21). Thebuffer used for ERK2 in NMR experiments contained 50 mM Tris (pH 7.4),150 mM NaCl, 5 mM MgSO4, 0.1 mM EDTA, 5 mM DTT, 100% D2O, and 2.5%(vol/vol) glycerol. Kinase activities of 0P-ERK2, ME/GG-ERK2, and 2P-ERK2were measured as described in a previous study (36). Nucleotide bindingexperiments were performed by titrating AMP-PNP (Roche Applied Science)into the NMR samples of 0P-ERK2 and 2P-ERK2, to final AMP-PNP concen-trations of 1.35 mM and 1.15 mM, respectively.

NMR Spectroscopy. De novo nonstereospecific Ile, Leu, and Val methylassignments ([1Hm,

13Cm]) were made for methyl peaks of 0P-ERK2 using 3D(13C,13C,1H) HMQC-NOESY, 2D (13C,1H) methyl TROSY HMQC, and 3D(13C,13C,1H) HMCM[CG]CBCA experiments (SI Materials and Methods), car-ried out at 25 °C on a Varian 800-MHz NMR spectrometer with a cryoprobe.Leu and Val methyls were distinguished using HMCM[CG]CBCA experimentsto facilitate accurate identification of ILV clusters (Fig. S1A). Spatially adja-cent ILV methyls were identified in 3D (13C,13C,1H) HMQC-NOESY spectra,and assigned by mapping onto the X-ray structure of 0P-ERK2 (Fig. S1 B andC). Two-dimensional (13C,1H) methyl TROSY HMQC spectra of 10 ERK2 I, L,or V to A mutants were used to help assign individual methyls (Fig. S1D).Methyl assignments for ME/GG-ERK2 were made by transferring assignmentsfrom 0P-ERK2.

Methyl side-chain dynamics were measured using 2D (13C,1H) multiple-quantum (MQ) CPMG relaxation dispersion experiments (17, 18) on 0P- and2P-ERK2 (SI Materials and Methods), performed at 25 °C on Varian 600-, 800-,

and 900-MHz NMR spectrometers with cryoprobes. The CPMG dispersionprofiles were fit on a per-methyl basis to a two-state exchange model usingthe Carver–Richards equation (18, 37), obtaining the exchange rate constant(kex), populations (pA, pB), and

13C chemical shift changes (in ppm) betweenstates A and B (jΔδCPMG

13Cj, assuming jΔδCPMG1Hj = 0). Fittings were per-

formed using the program CATIA (http://pound.med.utoronto.ca/software).The estimated errors of the individual fittings (Tables S1–S3) assumed a two-state exchange process, and could be underestimated due to the existenceof more than two states.

Slow exchange in 2P-ERK2 was probed from analysis of methyl peaks inHMQC spectra, collected on a Varian 800-MHz spectrometer with a cryoprobeat 5, 10, 15, 20, and 25 °C. The peak volumes of slow-exchanging peaks wereestimated using CcpNmr Analysis software (38). Equilibrium constants oftwo-state exchange (Keq) were calculated from the volume ratios of the twoslow-exchanging peaks.

Two-dimensional (13C,1H) single-quantum CPMG measurements (39) on0P- and 2P-ERK2 were carried out at 900 MHz, in parallel with 2D (13C,1H) MQCPMG experiments. The results showed dispersion curves (Fig. S6) comparableto MQ CPMG experiments, validating the assumption that jΔδCPMG

1Hj is ≈ 0.

ACKNOWLEDGMENTS.We are indebted to Lewis Kay (University of Toronto)for NMR pulse sequences, analysis software, and insightful discussions. Wealso thank Geoffrey Armstrong and Richard Shoemaker for help with NMRdata acquisition, and Rebecca Page and Wolfgang Peti (Brown University)for their gift of the ERK2 expression plasmid. These studies were supportedby National Institutes of Health Grants R01GM074134 (to N.G.A.), T32GM008759(to L.R.W.), and T32GM065103 (to M.P.L.). NMR instrumentation was purchasedwith partial support from NIH grants GM068928, RR11969, and RR16649;National Science Foundation Grants 9602941 and 0230966; and the W. M.Keck Foundation.

1. Roskoski R, Jr. (2012) ERK1/2 MAP kinases: Structure, function, and regulation.Pharmacol Res 66(2):105–143.

2. Pearson G, et al. (2001) Mitogen-activated protein (MAP) kinase pathways: Regulationand physiological functions. Endocr Rev 22(2):153–183.

3. Prowse CN, Lew J (2001) Mechanism of activation of ERK2 by dual phosphorylation.J Biol Chem 276(1):99–103.

4. Canagarajah BJ, Khokhlatchev A, Cobb MH, Goldsmith EJ (1997) Activation mecha-nism of the MAP kinase ERK2 by dual phosphorylation. Cell 90(5):859–869.

5. Zhang F, Strand A, Robbins D, Cobb MH, Goldsmith EJ (1994) Atomic structure of theMAP kinase ERK2 at 2.3 Å resolution. Nature 367(6465):704–711.

6. Lee T, et al. (2004) Docking motif interactions in MAP kinases revealed by hydrogenexchange mass spectrometry. Mol Cell 14(1):43–55.

7. Lee T, Hoofnagle AN, Resing KA, Ahn NG (2005) Hydrogen exchange solvent pro-tection by an ATP analogue reveals conformational changes in ERK2 upon activation.J Mol Biol 353(3):600–612.

8. Hoofnagle AN, Stoner JW, Lee T, Eaton SS, Ahn NG (2004) Phosphorylation-dependentchanges in structure and dynamics in ERK2 detected by SDSL and EPR. Biophys J86(1 Pt 1):395–403.

9. Hoofnagle AN, Resing KA, Goldsmith EJ, Ahn NG (2001) Changes in protein confor-mational mobility upon activation of extracellular regulated protein kinase-2 asdetected by hydrogen exchange. Proc Natl Acad Sci USA 98(3):956–961.

10. Smock RG, Gierasch LM (2009) Sending signals dynamically. Science 324(5924):198–203.11. Nagar B, et al. (2003) Structural basis for the autoinhibition of c-Abl tyrosine kinase.

Cell 112(6):859–871.12. Baldwin AJ, Kay LE (2009) NMR spectroscopy brings invisible protein states into focus.

Nat Chem Biol 5(11):808–814.13. Mittermaier AK, Kay LE (2009) Observing biological dynamics at atomic resolution

using NMR. Trends Biochem Sci 34(12):601–611.14. Bhabha G, et al. (2011) A dynamic knockout reveals that conformational fluctuations

influence the chemical step of enzyme catalysis. Science 332(6026):234–238.15. Eisenmesser EZ, et al. (2005) Intrinsic dynamics of an enzyme underlies catalysis.

Nature 438(7064):117–121.16. Fraser JS, et al. (2009) Hidden alternative structures of proline isomerase essential for

catalysis. Nature 462(7273):669–673.17. Korzhnev DM, Kloiber K, Kanelis V, Tugarinov V, Kay LE (2004) Probing slow dynamics

in high molecular weight proteins by methyl-TROSY NMR spectroscopy: Applicationto a 723-residue enzyme. J Am Chem Soc 126(12):3964–3973.

18. Korzhnev DM, Kloiber K, Kay LE (2004) Multiple-quantum relaxation dispersion NMRspectroscopy probing millisecond time-scale dynamics in proteins: Theory and appli-cation. J Am Chem Soc 126(23):7320–7329.

19. Gardner KH, Kay LE (1997) Production and incorporation of N-15, C-13, H-2 (H-1-delta1 methyl) isoleucine into proteins for multidimensional NMR studies. J Am Chem Soc119(32):7599–7600.

20. Goto NK, Gardner KH, Mueller GA, Willis RC, Kay LE (1999) A robust and cost-effectivemethod for the production of Val, Leu, Ile (delta 1) methyl-protonated 15N-, 13C-, 2H-labeled proteins. J Biomol NMR 13(4):369–374.

21. Shapiro PS, et al. (1998) Activation of the MKK/ERK pathway during somatic cell

mitosis: Direct interactions of active ERK with kinetochores and regulation of the

mitotic 3F3/2 phosphoantigen. J Cell Biol 142(6):1533–1545.22. Tugarinov V, Kay LE (2003) Ile, Leu, and Val methyl assignments of the 723-residue

malate synthase G using a new labeling strategy and novel NMR methods. J Am Chem

Soc 125(45):13868–13878.23. Sinha N, Kumar S, Nussinov R (2001) Interdomain interactions in hinge-bending

transitions. Structure 9(12):1165–1181.24. Johnson DA, Akamine P, Radzio-Andzelm E, Madhusudan M, Taylor SS (2001) Dy-

namics of cAMP-dependent protein kinase. Chem Rev 101(8):2243–2270.25. Robbins DJ, et al. (1993) Regulation and properties of extracellular signal-regulated

protein kinases 1 and 2 in vitro. J Biol Chem 268(7):5097–5106.26. Akamine P, et al. (2003) Dynamic features of cAMP-dependent protein kinase re-

vealed by apoenzyme crystal structure. J Mol Biol 327(1):159–171.27. Masterson LR, et al. (2011) Dynamically committed, uncommitted, and quenched

states encoded in protein kinase A revealed by NMR spectroscopy. Proc Natl Acad Sci

USA 108(17):6969–6974.28. Masterson LR, et al. (2010) Dynamics connect substrate recognition to catalysis in

protein kinase A. Nat Chem Biol 6(11):821–828.29. Chen H, et al. (2007) A molecular brake in the kinase hinge region regulates the

activity of receptor tyrosine kinases. Mol Cell 27(5):717–730.30. Deindl S, et al. (2007) Structural basis for the inhibition of tyrosine kinase activity of

ZAP-70. Cell 129(4):735–746.31. Shan Y, et al. (2012) Oncogenic mutations counteract intrinsic disorder in the EGFR

kinase and promote receptor dimerization. Cell 149(4):860–870.32. Shan Y, Arkhipov A, Kim ET, Pan AC, Shaw DE (2013) Transitions to catalytically in-

active conformations in EGFR kinase. Proc Natl Acad Sci USA 110(18):7270–7275.33. Kornev AP, Haste NM, Taylor SS, Eyck LF (2006) Surface comparison of active and

inactive protein kinases identifies a conserved activation mechanism. Proc Natl Acad

Sci USA 103(47):17783–17788.34. Taylor SS, Kornev AP (2011) Protein kinases: Evolution of dynamic regulatory proteins.

Trends Biochem Sci 36(2):65–77.35. Francis DM, et al. (2011) Resting and active states of the ERK2:HePTP complex. J Am

Chem Soc 133(43):17138–17141.36. Emrick MA, et al. (2006) The gatekeeper residue controls autoactivation of ERK2 via

a pathway of intramolecular connectivity. Proc Natl Acad Sci USA 103(48):18101–18106.37. Palmer AG III (2004) NMR characterization of the dynamics of biomacromolecules.

Chem Rev 104(8):3623–3640.38. Vranken WF, et al. (2005) The CCPN data model for NMR spectroscopy: Development

of a software pipeline. Proteins 59(4):687–696.39. Skrynnikov NR, Mulder FA, Hon B, Dahlquist FW, Kay LE (2001) Probing slow time

scale dynamics at methyl-containing side chains in proteins by relaxation dispersion

NMR measurements: Application to methionine residues in a cavity mutant of T4

lysozyme. J Am Chem Soc 123(19):4556–4566.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1318899111 Xiao et al.

Supporting InformationXiao et al. 10.1073/pnas.1318899111SI Materials and MethodsProtein Expression and Purification. A pET-23a plasmid containingthe rat His6-ERK2 sequence was transformed into Escherichiacoli BL21(DE3) cells (1). A single colony was used to inoculate1 mL of LB growth media containing 100 μg/mL carbenicillinand 34 μg/mL chloramphenicol, which was shaken overnight at37 °C. Cells were resuspended in 25 mL of unlabeled M9 mini-mal medium in H2O. At an OD600 of 0.8, the cells were spundown (1,600 × g, 10 min) and resuspended in 200 mL of M9minimal medium with D-glucose-d7 in D2O, shaking at 37 °Cuntil OD600 ≈ 0.6. Cells were then spun down (1,600 × g, 10 min)and resuspended in 1 L of M9 minimal medium with 3 g/L D-glucose-d7 and 1 g/L 15NH4Cl for U-[2H,15N] labeling, shaking at37 °C. Precursors (α-keto-3-methyl-d3-butyric acid-4-13C and 2-ketobutyric acid-4-13C) for Ile (13CδH3), Leu (13CδH3,

12CδD3),and Val (13C γH3,

12C γD3) labeling were added to cells once theOD600 reached 0.8. After 1 h, the cells were induced with iso-propyl-β-D-1-thiogalactopyranoside (2, 3) and shaken at 18 °Cfor 16 h, after which cells were centrifuged at 6,000 × g for10 min. The cell pellets were resuspended in lysis buffer (50 mMNaH2PO4/Na2HPO4, pH 8.0, 300 mM NaCl, 0.1% β-mercap-toethanol, 5 mM imidazole, 1× Halt protease inhibitor, 1 mg/mLlysozyme, and 1 mM EDTA). The cell suspension was incubatedat room temperature for 20 min and then sonicated and centri-fuged at 24,000 × g for 45 min at 4 °C. The supernatant was col-lected as cell lysate. Proteins were purified from the cell lysate byNi2+-NTA affinity chromatography (Bio-Rad), PD-10 desalting(GE Healthcare), MonoQ FPLC (GE Healthcare), and SephadexS200 size-exclusion chromatography (GE Healthcare) (4). Dual-phosphorylated (2P)-ERK2 was prepared from unphosphorylated(0P)-ERK2 after the MonoQ step, phosphorylated in vitro usingconstitutively active mutant MAP kinase kinase 1 (MKK-G7B:ΔN4/S218D/M219D/N221D/S222D) as described (5), and fur-ther purified by Sephadex S200 size-exclusion chromatography.The NMR samples of 0P-ERK2 and 2P-ERK2 were concentratedto 0.3–0.4 mM and exchanged into NMR buffer [50 mM Tris, pH7.4, 150 mM NaCl, 5 mM MgSO4, 0.1 mM EDTA, 5 mM DTT,100% D2O, and 2.5% (vol/vol) glycerol] for NMR experiments.Mutants of 0P-ERK2 I101A, L105A, M106E107GG, I124A,I131A, L154A, L155A, L161A, L198A, L235A, and I238A werecreated by site-directed mutagenesis of ERK2 using the Quik-Change Mutagenesis Kit (Stratagene), confirmed by DNA se-quencing, and purified in the same way as WT 0P-ERK2. Themutants were concentrated down to a final concentration of0.1–0.4 mM in NMR buffer containing 5% (vol/vol) D2O.

NMR Experiments for Assigning Residues.All of the NMR data usedfor making assignments were performed at 25 °C on a VarianVNMRS 800-MHz NMR spectrometer with a z-axis gradientcryoprobe and were processed using the software packageNMRPipe (6). A 3D (13C,13C,1H) HMCM[CG]CBCA (“out-and-back”) experiment was performed on a 0.4 mM 0P-ERK2sample that was 13C-labeled and methyl-partially protonated onIle, Leu, and Val side chains [Ile (13CδH3), Leu (13CδH3,

12CδD3),Val (13C γH3,

12C γD3), and U-[15N,13C,2H]]. This sample wasexpressed and purified as described above, except that a dif-ferent set of precursors was used: 2-keto-3-(methyl-d3)-1,2,3,4-

13C4-3-d1-butyrate and 2-keto-3,3-d2-1,2,3,4-

13C4-butyrate (7). The3D (13C,13C,1H) HMCM[CG]CBCA experiments were acquiredwith 64 and 42 complex points (4 and 10.5 ms, respectively) inthe t1 (Caliph) and t2 (Cm) dimensions and 1,024 complex points inacquisition period. 13C decoupling was applied with WALTZ-16

during the 71-ms acquisition period (8). A 1.2-s delay period wasused with 20 scans, leading to a total acquisition time of 83 h. Time-domain data in the 1H dimension were apodized by a cosine win-dow function and zero-filled before Fourier transformation. Theindirect dimensions were also apodized by a cosine window functionand zero-filled before Fourier transformation. Three-dimensional(13C,13C,1H) heteronuclear multiple-quantum coherence (HMQC)-NOESY experiments were performed on Ile, Leu, and Val (ILV)-labeled samples of both 0P-ERK2 and 2P-ERK2 using a 350-msmixing time. The experiments on 0P-ERK2 were acquired with 54and 52 complex points (15.9 and 15.3 ms, respectively) in the t1(13C) and t2 (

13C) dimensions and 1,024 complex points in the ac-quisition period. GARP1 13C decoupling was applied during the73-ms acquisition period (9). A 1.8-s delay period was used witheight scans, leading to a total acquisition time of 56 h. The NOESYexperiments for 2P-ERK2 were essentially the same, except that 64and 64 complex points were used (18.8 and 18.8 ms, respectively) inthe t1 (

13C) and t2 (13C) dimensions, resulting in a total acquisition

time of 82 h. Time-domain data in the 1H dimension wereapodized by a squared cosine window function and zero-filledbefore Fourier transformation. The number of points in bothindirect dimensions was doubled using forward–backward linearprediction (10). The indirect dimensions were then apodized bya cosine window function and zero-filled before Fourier trans-formation and baseline corrections were applied to both indirectdimensions.A 2D (13C,1H) methyl transverse relaxation-optimized spec-

troscopy (TROSY) HMQC spectrum was collected on each 0P-ERK2 mutant (0.1–0.4 mM) with a total experimental time of 1–4 h. Parameters varied with protein concentration; for example,the spectrum of I124A at 0.2 mM was acquired with 160 complexpoints (29.5 ms) in the t1 (13C) dimension and 1,024 complexpoints in the acquisition period. WALTZ-16 13C decoupling wasapplied during the 85-ms acquisition period. A 1.5-s delay periodwas used with eight scans, leading to a total acquisition time of1 h. The NMR data were processed as above.

Procedure for Assigning Residues. The procedure for making non-stereospecific resonance assignments of the ILV methyls([1Hm,

13Cm]) in 0P- or 2P-ERK2 primarily involved analysis ofdata from through-space and through-bond 3D NMR experi-ments in combination with the previously determined X-raystructures [Protein Data Bank (PDB) ID code 1ERK or 2ERK].A series of mutants was then used to confirm the assignments.The first step in the assignment procedure of 0P-ERK2 was

to distinguish the Leu and Val methyls from each other using3D (13C,13C,1H) HMCM[CG]CBCA (out-and-back) experi-ments (7). Three resonances (Cγ, Cβ, Cα) were observed foreach Leu methyl, whereas only two resonances (Cβ, Cα) wereobserved for each Val methyl (Fig. S1A). Methyl peaks from thesame residue (L/V) shared the same Cα and Cβ chemical shifts,although this property was not sufficient to assign two methylpeaks to the same residue.The next step was to identify spatially clustered ILV methyl

groups from analysis of the 3D (13C,13C,1H) HMQC-NOESYspectra (7), where the nuclear Overhauser effect (NOE) cross-peak intensities were qualitatively compared with distances be-tween methyl protons in 0P-ERK2. Protons were added to PDBID code 1ERK using the PDB Utility Servers (http://spin.niddk.nih.gov/bax/nmrserver) to calculate the shortest distances be-tween the protons of two methyls. For example, a cluster of sixIle residues was identified from six Ile peaks that showed a set of

Xiao et al. www.pnas.org/cgi/content/short/1318899111 1 of 9

interconnecting NOEs in the 3D NOESY (Fig. S1B). Two Valpeaks each showed NOEs to at least five Ile methyls in the Ilecluster (two of the Ile methyls had the same 13C chemical shift;Fig. S1B). The fact that these two Val methyls did not showNOEs to each other, and had the same Cα and Cβ chemicalshifts in the HMCM[CG]CBCA spectrum (Fig. S1A), stronglysupports their assignments to the same residue. Next, the X-raystructure of 0P-ERK2 was screened for a large cluster of Ilemethyls. The largest Ile cluster is composed of I54, I70, I84, I87,I101, and I345 (Fig. S1C), consistent with the six-Ile clusterfound in the NOESY spectra. The surrounding Leu and Valmethyls were used to make further assignments, where two Leuand one Val methyls were found close to this Ile cluster. Bothmethyls of V99 were close to most of the six Ile methyls, leadingto direct or spin-diffusion NOEs to these Ile methyls. Thus, thesetwo Val methyl peaks were assigned to V99. The next step in theassignment of the Ile methyls in the six-Ile cluster was mea-surement of the NOE cross-peak intensity for each of the Ilemethyls to both methyls of V99. Two of the Ile methyls showedweak NOEs to V99, and both methyls showed NOEs to anotherLeu. Comparison of the NOE pattern and intensities with the X-ray structure led to tentative assignments for the other Ilemethyls in this cluster as well as methyls on L67 and L74. Asimilar strategy was also used to assign methyls in other regionsof the protein.Once tentative assignments were made from the NOE data, 2D

(13C,1H) methyl TROSY HMQC spectra of 10 ERK2 mutants,each substituting Ile, Leu, or Val with Ala, were used to adjustand confirm the tentative assignments. The absence of a peak inthe spectra is used to help assign the resonance to a mutatedresidue (Fig. S1D). In some cases the mutation led to chemicalshift changes in other methyl resonances, and therefore theHMQC spectra of the mutants by themselves did not generallyprovide unambiguous resonance assignments. The set of as-signments of 0P-ERK2 was finalized by several iterations ofthese procedures, yielding 70 assignments of methyls to 62 res-idues. The assignments of 0P-ERK2 were then transferred to2P-ERK2 and confirmed with 3D (13C,13C,1H) HMQC-NOESYexperiments on 2P-ERK2 (Fig. S1E), yielding assignments of 67methyls to 60 residues.

Constant Time Carr–Purcell–Meiboom–Gill Relaxation DispersionExperiments. Carr–Purcell–Meiboom–Gill (CPMG) relaxationdispersion experiments were performed on 0.3 mM 0P-ERK2and 2P-ERK2 (the same samples that were used in NOESYexperiments) at 25 °C using three field strengths (600, 800, and900 MHz). The datasets of 800 and 900 MHz were collectedon Varian VNMRS 800- and 900-MHz NMR spectrometers

equipped with z-axis gradient cryoprobes. The 600-MHz datasetswere collected on a Varian INOVA 600-MHz NMR spectrom-eter equipped with a room temperature probe. The experimentwas arrayed with different delays (2τ) between 13C refocusingpulses, with a total time of 20 ms (Trelax) (11). The experimentfor 2P-ERK2 at 800 MHz was acquired with 12 different τ timescorresponding to CPMG frequencies (νCPMG) ranging from 50 to1,000 Hz. Several τ times were collected twice to estimate errorsin the measurements. A set of 160 complex points in the t1 (

13C)dimension and 1,024 complex points in the acquisition periodwas collected. A 2-s delay period was used with eight scans,leading to a total acquisition time of 21 h. Wideband UniformRate Excitation Smooth Truncation (WURST)-40 13C decou-pling was applied during the 85-ms acquisition period (12). Thisexperiment, with slightly different values of τ and numbers ofcomplex points in the t1 period, was repeated for 2P-ERK2 at600 and 900 MHz, as well as for 0P-ERK2 at all three fieldstrengths. Spectra were processed using NMRPipe (6), where thedata for each τ were processed similar to a 2D HMQC spectrumdescribed above. Peak intensities were measured using the pro-gram FuDA (http://pound.med.utoronto.ca/software). The ef-fective decay rate (R2,eff) was calculated by the equation (13)R2,eff = −1/T ln[I(νCPMG)/I(0)], where νCPMG = 1/(4τ), 2τ isthe interval between successive 180° 13C refocusing pulses, andI(νCPMG) and I(0) are the intensities of peaks recorded with andwithout the CPMG period, respectively. The CPMG dispersionprofiles of 0P-, 2P-, and ME/GG-ERK2 were fit on a per-methylbasis to a two-state exchange process using the generalizedCarver–Richards equation (11, 14), obtaining exchange rateconstant (kex = kAB + kBA), populations of the two exchangingstates (pA, pB), and

13C chemical shift changes (in ppm) betweenstates A and B (jΔδCPMG

13Cj, assuming jΔδCPMG1Hj = 0). Fittings

were performed using the program CATIA (http://pound.med.utoronto.ca/software). In addition, nonlinear least-squared fit-ting using the simplified, fast-exchange limit equation (14)

R2;eff = R2;0 +pApBΔω2

kex

�1−

2 tanh�kexτCP

2

�

kexτCP

�

was performed to estimate the term pApBΔω2 for methyls in0P-ERK2 and ME/GG-ERK2. R2,0 is the relaxation decay ratein the absence of exchange, τCP is the interval between successive180° 13C refocusing pulses (2τ), and Δω is the chemical shiftdifference in Hz between the two exchanging states A and B,which have populations of pA and pB, respectively.

1. Francis DM, et al. (2011) Resting and active states of the ERK2:HePTP complex. J AmChem Soc 133(43):17138–17141.

2. Gardner KH, Kay LE (1997) Production and incorporation of N-15, C-13, H-2 (H-1-delta1 methyl) isoleucine into proteins for multidimensional NMR studies. J Am Chem Soc119(32):7599–7600.

3. Goto NK, Gardner KH, Mueller GA, Willis RC, Kay LE (1999) A robust and cost-effectivemethod for the production of Val, Leu, Ile (delta 1) methyl-protonated 15N-, 13C-,2H-labeled proteins. J Biomol NMR 13(4):369–374.

4. Lee T, Hoofnagle AN, Resing KA, Ahn NG (2005) Hydrogen exchange solventprotection by an ATP analogue reveals conformational changes in ERK2 uponactivation. J Mol Biol 353(3):600–612.

5. Shapiro PS, et al. (1998) Activation of the MKK/ERK pathway during somatic cellmitosis: Direct interactions of active ERK with kinetochores and regulation of themitotic 3F3/2 phosphoantigen. J Cell Biol 142(6):1533–1545.

6. Delaglio F, et al. (1995) NMRPipe: A multidimensional spectral processing systembased on UNIX pipes. J Biomol NMR 6(3):277–293.

7. Tugarinov V, Kay LE (2003) Ile, Leu, and Val methyl assignments of the 723-residuemalate synthase G using a new labeling strategy and novel NMR methods. J Am ChemSoc 125(45):13868–13878.

8. Shaka AJ, Keeler J, Frenkiel T, Freeman R (1983) An improved sequence forbroadband decoupling: WALTZ-16. J Magn Reson 52(2):335–338.

9. Shaka AJ, Lee CJ, Pines A (1988) Iterative schemes for bilinear operators; applicationto spin decoupling. J Magn Reson 77(2):274–293.

10. Zhu GA, Bax A (1992) 2-Dimensional linear prediction for signals truncated in bothdimensions. J Magn Reson 98(1):192–199.

11. Korzhnev DM, Kloiber K, Kay LE (2004) Multiple-quantum relaxation dispersion NMRspectroscopy probing millisecond time-scale dynamics in proteins: Theory andapplication. J Am Chem Soc 126(23):7320–7329.

12. Kupce E, Freeman R (1995) Adiabatic Pulses for Wide-Band Inversion and Broad-BandDecoupling. Journal of Magnetic Resonance Series A 115(2):273–276.

13. Skrynnikov NR, Mulder FA, Hon B, Dahlquist FW, Kay LE (2001) Probing slow timescale dynamics at methyl-containing side chains in proteins by relaxation dispersionNMR measurements: Application to methionine residues in a cavity mutant of T4lysozyme. J Am Chem Soc 123(19):4556–4566.

14. Palmer AG III (2004) NMR characterization of the dynamics of biomacromolecules.Chem Rev 104(8):3623–3640.

Xiao et al. www.pnas.org/cgi/content/short/1318899111 2 of 9

21.66 22.11 12.75 14.66 13.70 14.37 15.55 15.40

0.9 0.8 0.7 0.7 1.0 0.7 0.70.8 0.8

25

Red: I101A 0P-ERK2Blue: WT 0P-ERK2

13C

ppm

1H ppm

1H ppm

13C

ppm

A B

* *

2P-ERK2

15

Ile101

Ile345Ile70Ile54Ile84/87

Ile101 Ile345 Ile87 Ile84Val99a Val99b Ile70Ile54

*

**

** *

0P-ERK2

Val99aVal99b

Leu74b

Leu74a

Leu67a

C

I345I54

L74bL74a

L67aL67bV99b

V99a

I70I101

I87

I84

≈90°

4.8

3.13.3

4.87.22.4

5.04.2

6.4 3.44.9

3.16.9

5.1V99b

V99a

I345I70

I87

I84I101

I54L67a

L67bL74a

0.20.40.60.81.0

14

16

13C

ppm

1H ppm

Ile101D

23.70 21.60 22.091.1 1.0 1.0

30

40

50

60

V99a V99bL74b

Cα Cα

Cβ Cβ

Cγ

Cβ

Cα

E

15

21.66 22.11 12.88

0.9 0.8 0.7

25

1H ppm

Ile101Val99a Val99bIle101

Val99aVal99b

13C

ppm

*

* *

Fig. S1. Methods used for assigning ILV methyls in 0P- and 2P-ERK2. (A) Through-bond intraresidue (HMCM[CG]CBCA) experiments allowed distinction of Valfrom Leu methyls. Strip plots [1H,13Caliph] of the 3D (13C,13C,1H) HMCM[CG]CBCA spectra on an Ile (13CδH3), Leu (13CδH3,

12CδD3), Val (13CγH3,

12CγD3), andU-[15N,13C,2H] sample of 0P-ERK2 acquired at 800 MHz at 25 °C as described in SI Materials and Methods, showing examples of Leu (L74) and Val (V99) residues.The chemical shifts (ppm) for the 13Cmethyl dimension are labeled at the bottom of each strip. The signs of the peaks alternate as magnetization is transferredalong the carbon chain (i.e., Cα is 180° out of phase with Cβ, which is 180° out of phase with Cγ) as indicated by red (negative) and green (positive) contours. Valresidues can be distinguished from Leu residues because the former only show two carbon resonances (Cα and Cβ), whereas the latter show three resonances(Cα, Cβ, and Cγ). Both methyl groups in V99 share the same Cα and Cβ chemical shifts, which helps assign them to the same residue. (B) Strip plots [1H,13C] fromthe 350-ms 3D (13C,13C,1H) HMQC-NOESY spectrum of 0P-ERK2 illustrate that two Val methyl peaks (Val99a and Val99b) show NOEs to at least five Ile methyls(two of these Ile methyls have the same 13C chemical shift). In addition, most of these Ile methyls show NOEs to each other. Three Leu methyl peaks, comingfrom at least two Leu residues, also show NOEs to a subset of the Ile methyls in this cluster. The chemical shifts (ppm) of the 13Cmethyl dimension are labeled atthe bottom of each strip, and the asterisks highlight methyl self-peaks. Cross-peaks that were used in the assignment were aligned with the self-peak asindicated by the horizontal lines. Cross-peaks for V99b and L74b are highlighted in red and green boxes, respectively. (C) X-ray structure of 0P-ERK2, wherespheres represent methyls from Ile (red), Leu (yellow), and Val (green). The largest Ile cluster (I54, I70, I84, I87, I101, and I345) in the molecule is highlighted bythe dashed box. The expansions illustrate this Ile cluster and adjacent Leu and Val methyls. Interresidue methyl proton–methyl proton distances are shown forV99b if a corresponding NOE cross-peak was observed in the NOESY spectrum in B. An ≈90° rotated view is shown to highlight L67 and L74, which are also closeto the Ile cluster, and which provide additional information for assigning the Ile methyls in this cluster. L74b is relatively close to I84, I87, I70, and I101, and allthese direct or spin diffusion-induced NOEs for L74b are observed in the NOESY spectrum in B. L67a is close to I345, and the corresponding NOE cross-peak isseen in B. (D) An overlay of the Ile methyl region of the 2D (13C,1H) methyl TROSY HMQC spectra of the wild-type (blue) and mutant I101A (red) sample of0P-ERK2. Although multiple peaks showed clear chemical shift perturbations in this mutant, the absence of the I101 peak in the I101A 0P-ERK2 mutant wasconsistent with the assignment made from NOE data. (E) A similar pattern of NOE cross-peaks is seen in 2P-ERK2; strip plots are shown only for NOEs betweenV99 and I101.

Xiao et al. www.pnas.org/cgi/content/short/1318899111 3 of 9

21 61 91 6 2 92 74 94 15 45 76 07 27 37 47 18 28 78 88 991 015010113119114 211315318313418413514515513618919022125126120225222325328 3293214224235226217267 27 8288 229200320363314344 3543

0.0

0.1

0.2

0.3

0.4

0.5

)mpp( egnahc tfihs laci

mehC

Residue Number

|Δδ13C||Δδ1H|

Fig. S2. Changes in chemical shifts and chemical environment between 0P-ERK2 and 2P-ERK2. Plot of the changes in carbon jΔδ13Cj (solid bars) and protonjΔδ1Hj (open bars) chemical shifts for the assigned methyl groups in 0P- and 2P-ERK2. The dashed line indicates a threshold of jΔδj = 0.15 ppm, which highlightsmethyls with the largest chemical shift changes induced by phosphorylation.

R2ef

f (s−

1 )

30

50

70

90

0 200 400 600 800 1000

2P-I70

R2ef

f (s−

1 )

30

50

70

90

0 200 400 600 800 1000

0P-I72

R2ef

f (s−

1 )

30

50

70

90

0 200 400 600 800 1000

0P-L73

R2ef

f (s−

1 )

30

50

70

90

0 200 400 600 800 1000

2P-L74

R2ef

f (s−

1 )

νCPMG (Hz)

10

30

50

70

90

0 200 400 600 800 1000

2P-I82

30

50

70

90

0 200 400 600 800 1000

2P-I101

R2ef

f (s−

1 )

30

50

70

90

0 200 400 600 800 1000

0P-I81

30

50

70

90

0 200 400 600 800 1000

2P-I72

30

50

70

90

0 200 400 600 800 1000

2P-I73

30

50

70

90

0 200 400 600 800 1000

2P-I81

30

50

70

90

0 200 400 600 800 1000

0P-I131

30

50

70

90

0 200 400 600 800 1000

2P-L135

30

50

70

90

0 200 400 600 800 1000

0P-I138

30

50

70

90

0 200 400 600 800 1000

0P-V143

νCPMG (Hz)

30

50

70

90

0 200 400 600 800 1000

2P-L148

30

50

70

90

0 200 400 600 800 1000

2P-L154

30

50

70

90

0 200 400 600 800 1000

2P-L155

30

50

70

90

0 200 400 600 800 1000

2P-I131

30

50

70

90

0 200 400 600 800 1000

2P-I138

30

50

70

90

0 200 400 600 800 1000

2P-V143

0P-ERK22P-ERK2 2P-ERK2

30

50

70

90

0 200 400 600 800 1000

0P-I163

30

50

70

90

0 200 400 600 800 1000

0P-I241

30

50

70

90

0 200 400 600 800 1000

2P-I241

30

50

70

90

0 200 400 600 800 1000

νCPMG (Hz)

0P-L242

30

50

70

90

0 200 400 600 800 1000

2P-L288

30

50

70

90

0 200 400 600 800 1000νCPMG (Hz)

2P-I345

30

50

70

90

0 200 400 600 800 1000

0P-I198

10

30

50

70

90

0 200 400 600 800 1000

0P-I253

30

50

70

90

0 200 400 600 800 1000

0P-L262

30

50

70

90

0 200 400 600 800 1000

0P-L235

30

50

70

90

0 200 400 600 800 1000

2P-L198

10

30

50

70

90

0 200 400 600 1000 800

2P-I253

30

50

70

90

0 200 400 600 800 1000

2P-L262

30

50

70

90

0 200 400 600 800 1000

2P-L235

0P-ERK2 0P-ERK22P-ERK2 2P-ERK2

νCPMG (Hz)

10

30

50

70

90

0 200 400 600 800 1000

Rex≈ 0 s-10P-I82

30

50

70

90

0 200 400 600 800 1000

2P-L67 600 800 900

30

50

70

90

0 200 400 600 800 1000

Rex≈ 0 s-10P-L67

R2ef

f (s−

1 )

30

50

70

90

0 200 400 600 800 1000

Rex≈ 0 s-10P-I70

0P-ERK2

30

50

70

90

0 200 400 600 800 1000

Rex≈ 0 s-1

0P-L74

30

50

70

90

0 200 400 600 800 1000

2P-I87

30

50

70

90

0 200 400 600 800 1000

Rex≈ 0 s-10P-I87

30

50

70

90

0 200 400 600 800 1000

Rex≈ 0 s-10P-I101

30

50

70

90

0 200 400 600 800 1000

Rex≈ 0 s-10P-L135

νCPMG (Hz)

30

50

70

90

0 200 400 600 800 1000

Rex≈ 0 s-10P-L148

30

50

70

90

0 200 400 600 800 1000

Rex≈ 0 s-10P-L154

30

50

70

90

0 200 400 600 800 1000

Rex≈ 0 s-10P-L155

30

50

70

90

0 200 400 600 800 1000

νCPMG (Hz)

2P-L242

30

50

70

90

0 200 400 600 800 1000

30

50

70

90

0 200 400 600 800 1000νCPMG (Hz)

30

50

70

90

0 200 400 600 800 1000

Overlapping peak2P-I163

Rex≈ 0 s-1

0P-L288

Rex≈ 0 s-10P-I345

Fig. S3. CPMG relaxation dispersion curves for 0P- and 2P-ERK2. Relaxation dispersion curves at 25 °C and 900 (green), 800 (blue), and 600 (red) MHz areplotted for all residues that showed significant Rex in either 0P-ERK2 (Left) or 2P-ERK2 (Right). Curves labeled “Rex ≈ 0 s−1” are residues where the dispersionwas too small to confidently fit, meaning they had Rex < 4 s−1. Lines show individual fits to the Carver–Richards equation, and the error bars were estimatedfrom peak intensities of duplicate experiments.

Xiao et al. www.pnas.org/cgi/content/short/1318899111 4 of 9

Ile72 Ile241A BΔH° = 9.7 ± 1.1 kcal/molΔS° = 35.5 ± 3.7 cal/(mol*K)

ΔH° = 5.8 ± 1.1 kcal/molΔS° = 21.7 ± 2.9 cal/(mol*K)

0.0034 0.0036 0.00380.0

0.5

1.0

1.5

2.0

ln(K

eq)

0.0034 0.0036 0.00380.0

0.5

1.0

1.5

2.0

ln(K

eq)

1/T (K-1) 1/T (K-1)

Fig. S4. Enthalpies and entropies estimated from 2D (13C,1H) HMQC spectra. van’t Hoff plots for (A) I72 and (B) I241, where the Keq values were estimatedfrom the volumes of the two peaks that showed slow exchange for individual methyls in the HMQC spectra at different temperatures (Fig. 4).

0.550.600.65

7.5

8.0

0.550.60

7.5

8.0

-0.25-0.20-0.15

26.5

27.0

Leu288

-0.25-0.20-0.15

26.5

27.0

Ile72

-0.25-0.20-0.15

26.0

26.5

0.550.600.65

7.0

7.5

8.0

1H (ppm)

13C

(ppm

)

0P-ERK2, 25 ºC 2P-ERK2, 25 ºC 2P-ERK2, 5 ºC

Fig. S5. Nucleotide binding does not affect populations of the two states in 0P- and 2P-ERK2. (Left) Two-dimensional (13C,1H) HMQC peaks of 0P-ERK2 at 25 °Cwith and without 1.35 mM adenosine-5′-(β,γ-imido)triphosphate (AMP-PNP) colored in red and black, respectively. (Center) 2P-ERK2 at 25 °C with and without1.15 mM AMP-PNP, colored in magenta and blue, respectively. (Right) 2P-ERK2 at 5 °C with and without 1.15 mM AMP-PNP, colored in orange and sky blue,respectively.

Xiao et al. www.pnas.org/cgi/content/short/1318899111 5 of 9

R2ef

f (s−

1 )R

2eff (

s−1 )

A

B

Fig. S6. Single-quantum CPMG experiments show dispersion profiles similar to multiple-quantum CPMG experiments. CPMG dispersion curves were com-parable between 2D (13C,1H) single-quantum CPMG experiments (SQ; closed circles) and 2D (13C,1H) multiple-quantum CPMG experiments (MQ; open circles),for (A) 0P-ERK2 and (B) 2P-ERK2. Each set of experiments was carried out at 900 MHz and 25 °C.

Xiao et al. www.pnas.org/cgi/content/short/1318899111 6 of 9

Table S1. Exchange parameters for methyl groups in 0P-ERK2, fitted individually*

Parameters

Residue kex, s−1† Rex(800), s

−1‡ pApB(Δω13C)2, s−2§ χ2/DF{

Kinase coreIle72 1,400 ± 400 23 32,000 ± 5,000 2.2Leu73 900 ± 700 12 16,000 ± 8,000 3.2Ile81 2,000 ± 600 9 26,000 ± 9,000 0.8Ile131 1,700 ± 700 9 26,000 ± 10,000 0.9IIe138jj N.D. 7 N.D. N.D.Val143 1,100 ± 600 9 11,000 ± 4,000 1.0Ile163jj N.D. 10 N.D. N.D.Ile241 1,100 ± 200 26 40,000 ± 2,000 1.0Leu242 1,500 ± 300 13 19,000 ± 3,000 0.8

MAP kinase insertLeu198 1,300 ± 300 15 10,000 ± 2,000 0.4Leu235 1,200 ± 200 29 45,000 ± 3,000 1.0Ile253 2,200 ± 300 20 62,000 ± 15,000 0.9Leu262 1,400 ± 300 22 34,000 ± 3,000 0.5

*These parameters were all obtained from MQ CPMG dispersion data collected at 25 °C.†The kex values were obtained by fitting CPMG dispersion data collected at 600, 800, and 900 MHz to a two-stateexchange process using the Carver–Richards equation on a per-methyl basis. The errors were estimated from fitsusing the CATIA program, and may be underestimated if the data are not well-described by a two-state exchangeprocess (Materials and Methods).‡Rex(800) was estimated using the equation Rex = R2,eff(50 Hz) – R2,eff(1,000 Hz) from 800-MHz CPMG dispersiondata.§This term was estimated assuming a two-state exchange model under fast-exchange limit using 800-MHz CPMGdispersion data (Materials and Methods).{χ2/DF (reduced χ2) describes the quality of the fit for the experimental data. DF is the degrees of freedom and wascalculated by DF = (number of experimental data points – number of parameters – 1) for each methyl in thedataset.jjIndividual fits were not determined (N.D.) due to large errors, but the Rex could still be estimated.

Xiao et al. www.pnas.org/cgi/content/short/1318899111 7 of 9

Table S2. Exchange parameters for methyl groups in 2P-ERK2, fitted individually*

Parameters

Residue kex, s−1† pB, %

† jΔδCPMG13Cj, ppm† Rex(800), s

−1‡ χ2/DF§ jΔδCPMG13Cglobalj, ppm{

Kinase coreLeu67 550 ± 400 2 ± 1 0.6 ± 0.3 9 2.3 0.1 ± 0.01Ile70 210 ± 100 8 ± 2 0.4 ± 0.09 15 2.3 0.2 ± 0.01Ile72 150 ± 60 28 ± 8 0.7 ± 0.07 40 2.7 0.6 ± 0.02Leu73 230 ± 80 13 ± 2 0.4 ± 0.06 19 1.3 0.3 ± 0.01Leu74jj N.D. N.D. N.D. 10 N.D. 0.2 ± 0.01Ile81 110 ± 60 17 ± 6 0.5 ± 0.09 19 0.7 0.3 ± 0.01Ile82 140 ± 60 35 ± 12 0.7 ± 0.08 50 8.3 0.6 ± 0.02Ile87 200 ± 90 19 ± 4 0.4 ± 0.1 22 0.9 0.3 ± 0.01Ile101 220 ± 50 13 ± 2 0.4 ± 0.05 18 0.6 0.3 ± 0.01Ile131 190 ± 50 11 ± 2 0.4 ± 0.05 13 0.5 0.2 ± 0.01Leu135 180 ± 90 18 ± 3 0.4 ± 0.1 24 0.6 0.3 ± 0.01IIe138 110 ± 90 8 ± 6 0.9 ± 0.1 10 2.1 0.3 ± 0.01Val143 190 ± 50 16 ± 6 0.3 ± 0.06 14 0.8 0.2 ± 0.01Leu148 290 ± 60 16 ± 6 0.3 ± 0.04 19 1.1 0.2 ± 0.01Ile154jj N.D. N.D. N.D. 20 N.D. 0.3 ± 0.01Leu155 290 ± 50 13 ± 3 0.4 ± 0.05 17 1.6 0.3 ± 0.01Leu241 160 ± 50 21 ± 3 0.5 ± 0.06 33 0.8 0.4 ± 0.01Leu288 260 ± 220 15 ± 18 0.3 ± 0.3 16 1.7 0.2 ± 0.01Ile345 380 ± 30 20 ± 20 0.1 ± 0.01 19 0.6 0.3 ± 0.01Global fitting 300 ± 10 80 ± 0.6 — — 1.0 See above

MAP kinase insertLeu198 750 ± 340 20 ± 100 0.3 ± 0.7 25 2.0 —

Leu235jj N.D. N.D. N.D. 5 N.D. —

Ile253 750 ± 130 5 ± 3 0.5 ± 0.2 11 0.8 —

Leu262 640 ± 80 60 ± 70 0.1 ± 0.04 11 0.8 —

*These parameters were all obtained from MQ CPMG dispersion data collected at 25 °C.†These parameters were all obtained by fitting CPMG dispersion data collected at 600, 800, and 900 MHz to a two-state exchange process using the Carver–Richards equation on a per-methyl basis. The errors were estimated fromfits using the CATIA program, and may be underestimated if the data are not well-described by a two-stateexchange process (Materials and Methods).‡Rex(800) was estimated using the equation Rex = R2,eff(50 Hz) – R2,eff(1,000 Hz) from 800-MHz CPMG dispersiondata.§χ2/DF (reduced χ2) describes the quality of the fit for the experimental data. DF is the degrees of freedom and wascalculated by DF = (number of experimental data points – number of parameters – 1) for each methyl in thedataset.{jΔδCPMG

13Cglobalj is obtained by a global fitting of the methyls in the kinase core that showed Rex(800) > 4 s−1

using CATIA.jjIndividual fits were not determined due to large errors, but the Rex could still be estimated.

Xiao et al. www.pnas.org/cgi/content/short/1318899111 8 of 9

Table S3. Exchange parameters for methyl groups in ME/GG-ERK2, fitted individually*

Parameters

Residue kex, s−1† Rex(800), s

−1‡ pApB(Δω13C)2, s−2§ χ2/DF{ jΔδCPMG13Cglobalj, ppmjj

Kinase coreIle54 900 ± 500 9 7,000 ± 1,000 1.2 0.4 ± 0.05Ile72 1,100 ± 800 13 28,000 ± 12,000 0.9 1 ± 0.1Ile82 500 ± 100 6 4,400 ± 800 0.3 0.3 ± 0.04Ile84** N.D. 7 N.D. N.D. 0.3 ± 0.04Val99** N.D. 6 N.D. N.D. 0.3 ± 0.04Ile101 1,000 ± 300 8 10,000 ± 2,000 0.9 0.5 ± 0.05Leu105** N.D. 4 N.D. N.D.Leu110 800 ± 300 13 31,000 ± 10,000 0.4 1.1 ± 0.1Ile113** N.D. 4 N.D. N.D.Ile124 200 ± 170 4 3,400 ± 800 0.2Ile131 1,700 ± 700 12 50,000 ± 45,000 4 0.9 ± 0.1Leu135** N.D. 6 N.D. N.D. 0.5 ± 0.04Ile138** N.D. 7 N.D. N.D. 0.4 ± 0.04Val143 500 ± 200 8 7,500 ± 3,200 1.0 0.3 ± 0.05Leu148** N.D. 4 N.D. N.D.Leu154 1,400 ± 800 8 16,000 ± 3,000 0.9 0.7 ± 0.05Leu155 1,900 ± 600 8 31,000 ± 15,000 0.8 0.4 ± 0.04Leu209 200 ± 100 6 3,600 ± 400 2.7 0.3 ± 0.04Leu225 400 ± 100 6 4,700 ± 700 0.2 0.3 ± 0.04Ile241 700 ± 200 17 26,000 ± 2,000 0.1 0.9 ± 0.1Leu242** N.D. 8 N.D. N.D. 0.4 ± 0.03Global fitting 500 ± 60 — pB = 3 ± 0.2% 1.0 See above

MAP kinase insertLeu198 1,200 ± 300 11 21,000 ± 3,000 0.9 —

Leu235 1,400 ± 800 20 34,000 ± 4,000 1.4 —

Ile253 1,600 ± 1,200 11 36,000 ± 18,000 0.6 —

Leu262 1,300 ± 600 28 95,000 ± 25,000 4.6 —

*These parameters were all obtained from MQ CPMG dispersion data collected at 25 °C.†These parameters were all obtained by fitting CPMG dispersion data collected at 600, 800, and 900 MHz toa two-state exchange process using the Carver–Richards equation on a per-methyl basis. The errors were esti-mated from fits using the CATIA program, and may be underestimated if the data are not well-described bya two-state exchange process (Materials and Methods).‡Rex(800) was estimated using the equation Rex = R2,eff(50 Hz) – R2,eff(1,000 Hz) from 800-MHz CPMG dispersion data.§This term was estimated assuming a two-state exchange model under fast-exchange limit using 800-MHz CPMGdispersion data (Materials and Methods).{χ2/DF (reduced χ2) describes the quality of the fit for the experimental data. DF is the degrees of freedom andwas calculated by DF = (number of experimental data points – number of parameters – 1) for each methyl in thedataset.jjjΔδCPMG

13Cglobalj is obtained by a global fitting of the methyls in the kinase core that showed Rex(800) > 4 s−1

using CATIA.**Individual fits were not determined due to large errors, but the Rex could still be estimated.

Xiao et al. www.pnas.org/cgi/content/short/1318899111 9 of 9