Synchronous Opening and Closing Motions are Essential for cAMP-Dependent Protein Kinase A Signaling

SUMMARY

Conformational fluctuations play a central role in enzymatic catalysis. However, it is not clear how the rates and the coordination of the motions affect the different catalytic steps. Here, we used NMR spectroscopy to analyze the conformational fluctuations of the catalytic subunit of the cAMP-dependent protein kinase (PKA-C), a ubiquitous enzyme involved in a myriad of cell signaling events. We found that the wild-type enzyme undergoes synchronous motions involving several structural elements located in the small lobe of the kinase, which is responsible for nucleotide binding and release. In contrast, a mutation (Y204A) located far from the active site desynchronizes the opening and closing of the active cleft without changing the enzyme’s structure, rendering it catalytically inefficient. Since the opening and closing motions govern the rate-determining product release, we conclude that optimal and coherent conformational fluctuations are necessary for efficient turnover of protein kinases.

INTRODUCTION

Protein kinase A (PKA) is a ubiquitous phosphoryltransferase involved in a myriad of cellular events. PKA was one of the first kinases to be discovered and characterized (Krebs et al., 1959), and represents a benchmark for understanding the structure-function relationship and catalytic mechanism for the entire kinase superfamily (Taylor et al., 2012; Taylor and Kornev, 2011). Unlike metabolic enzymes optimized for fast turnover rates, PKA has evolved to orchestrate cell signaling events, mediating the transfer of external stimuli inside the cell (Taylor and Kornev, 2011). Although it targets several hundred substrates (Manning et al., 2002), this kinase is highly specific, due to a tight spatio-temporal control that ensures physiological levels of phosphorylation of its specific cellular targets (Scott and Pawson, 2009). β-adrenergic stimulation activates the PKA holoenzyme (two catalytic and two regulatory subunits, C- and R- subunits) (Zhang et al., 2012), which unleashes the active C-subunits (PKA-C), a catalytic core that PKA shares with other serine/threonine- and tyrosine-specific protein kinases. PKA-C has a bilobate structure, comprised of a small β-strand-rich lobe with a short α-helix (C-helix), which harbors the nucleotide binding pocket, and a large, mostly helical lobe containing the substrate binding site (Knighton et al., 1991). PKA-C’s catalytic cycle is a multistep process that involves nucleotide binding, substrate recognition and positioning, phosphoryl transfer and product release (ADP and phosphorylated substrate) (Lew et al., 1997). Crystal structures have captured this enzyme in the major conformational states along the reaction coordinates: apo, intermediate (nucleotide-bound), and closed (ternary complex with nucleotide and pseudo-substrate) (Johnson et al., 2001; Meharena et al., 2013). More recently the peptide substrate and product complexes have been trapped in crystals (Bastidas et al., 2013). However, these structures offers static images that do not fully describe the conformational changes and the role of motions occurring during allosteric regulation and catalysis (Huse and Kuriyan, 2002). The prominent role of conformational fluctuations for PKA-C’s catalytic cycle was inferred from the analysis of mutants aimed at disrupting the proposed allosteric network (Aimes et al., 2000; Iyer et al., 2005a; Iyer et al., 2005b; Moore et al., 2003; Yang et al., 2005; Yang et al., 2004). These studies, coupled with kinetic assays, suggest that the rate-determining step of the turnover is the ADP release following a fast phosphoryl transfer (~500 s⁻¹), with the conformational transition from closed to open state being rate-limiting (Adams, 2001). Among these mutants, Y204A (PKA-C^Y204A), located in the large lobe of the enzyme and distant from the active site (Figure 1), has been shown to substantially affect both k_cat/K_m and k_cat (Moore et al., 2003; Yang et al., 2005). Interestingly, the X-ray structure of PKA-C^Y204A is virtually superimposable to that of the wild-type in the presence of PKI and ATP-Mg₂, with an overall rmsd of ~0.5 Å (Yang et al., 2004). From the comparison of the two X-ray structures, it emerges that Y204 couples the P+1 loop in the activation segment to both E230 and the hydrophobic core. Interestingly, Y204 does not have any direct contact with the peptide substrate, and yet this alanine substitution reduces dramatically the enzyme’s catalytic efficiency. Because the mutation induces no significant structural changes in the enzyme, let alone the active site, PKA-C^Y204A represents an ideal benchmark to test the relevance of the rates of motions in the PKA-C catalytic cycle.

Figure 1.

Electrostatic node Y204 mapped on the crystal structure (PDB: 1ATP). The mutation is located approximately at 15 Ǻ from the active site and does not involve direct interactions with the substrate.

To analyze the effects of the Y204A mutation on the conformational dynamics of the enzyme, we carried out NMR spin relaxation measurements both in fast (ps-ns) and slow (μs-ms) timescales (Boehr et al., 2006a; Mittermaier and Kay, 2006; Palmer, 2004), focusing on the nucleotide-bound, dynamically-activated and substrate-primed forms of PKA-C^WT and PKA-C^Y204A. We found that by disrupting the electrostatic network at the fringe of the small and large lobes, the enzyme responds with global dynamical changes; structural fluctuations in the ps-ns time-scale show regions with both enhanced and reduced mobility. Importantly, relaxation dispersion measurements for quantifying chemical exchange in the μs-ms timescale range show synchronous motions in the wild-type protein that are obliterated by the mutation. These effects are particularly evident for the Gly-rich loop and hinge region, those that mediate nucleotide binding, substrate recognition, and the rate-determining-step of the catalysis, i.e., ADP release. We conclude that synchronous (i.e., concerted) conformational fluctuations are essential for catalytic efficiency.

RESULTS

PKA-C^Y204A displays lower binding cooperativity

Kinetic data reported in the literature show the Y204A mutation affects both k_cat/K_m and k_cat (Yang et al., 2004). To dissect the thermodynamics of nucleotide and substrate binding from the kinetics data, we monitored the binding of the nucleotide ATPγN and the pseudo-substrate PKI_5-24 using Isothermal Titration Calorimetry (ITC) for both PKA-C^WT and PKA-C^Y204A (Figure S1, Table S1). We found that Y204A exhibits a slightly higher affinity for ATPγN (K_d ~ 23 μM) compared to that of the wild-type (K_d ~ 58 μM), with a higher enthalpy of binding (ΔH ≈ −4.2 versus −2.7 kcal/mol). However, the affinity for PKI_5-24 binding to the ATPγN-saturated enzymes is substantially different. While for PKA-C^WT we observed K_d ~ 120 nM consistent with previous literature data (Herberg et al., 1999), for PKA-C^Y204A K_d is ~14 μM, corresponding to a loss of affinity of nearly 100 fold and a reduction of binding enthalpy of 10 kcal/mol. Since the k_cat/K_M and substrate affinities for the Y204A mutant is different from the wild-type enzyme, we chose to focus on the analysis of the dynamics for the nucleotide bound state.

Nucleotide-bound PKA-C^WT is dynamically activated

To monitor the structure and dynamics of PKA-C^WT and PKA-C^Y204A, we collected [¹H,¹⁵N]-TROSY-HSQC spectra of the apo, nucleotide-bound (ATPγN), and ternary (ATPγN and PKI) forms (Figure S2). Saturating the apo forms of both wild-type and mutant PKA-C with ATPγN resulted in significant broadening of numerous amide resonances, several of which broadened beyond detection, indicating the presence of conformational heterogeneity (Figure 2). Importantly, we detected substantial resonance broadening in the Gly-rich, DFG, and activation loops, which play major roles in the catalytic cycle (Johnson et al., 2001). Residues in the C-terminal loop flanking the Gly-rich loop and the C-helix also display notable amide resonance broadenings. Some regions with significant line broadening coincide with several loops displaying high B-factors in the crystal structure 1BKX (Figure S3).

Figure 2.

Residues broadened in PKA-C^WT upon binding of ATPγN. Several broadened residues are located close to the active site (Gly-rich loop, DFG loop, C-helix, acidic residues in the C-terminal tail), others are distant (Peptide-positioning loop, activation loop, D and H helices).

Chemical shifts report only minimal structural effects for Y204A

Overall, the chemical shift differences incurred by mutation at Y204A in all three states are minimal (Figure 3A), with the most prominent changes close to the mutation site, indicating that the average solution structures of PKA-C^WT and PKA-C^Y204A are nearly identical (Figure 3B). However, tracing these subtle chemical shift changes offers useful information regarding local and global allosteric phenomena (Selvaratnam et al., 2011). Building on the chemical shift covariance analysis (Selvaratnam et al., 2011), we developed CONCISE (COordiNated ChemIcal Shifts bEhavior) that defines the probability density of each state and traces changes in the conformational equilibria upon ligand titration or mutation (Cembran et al., 2014). For both PKA-C^WT and PKA-C^Y204A, CONCISE identified 94 residues interspersed throughout the whole structure that follow linear paths of the amide resonances upon titration with both the nucleotide and PKI_5-24 (Figure S4). This indicates that the entire enzyme is involved in the open to closed transition. The probability density curves of the apo, intermediate, and closed states for the mutant are similar to those of the wild-type, with only slight differences for the apo and nucleotide-bound forms, which appear slightly more closed in the mutant. For the ternary complexes, however, the average conformations are essentially the same, with a complete overlay of the probability density curves (Figure 3C). This indicate that PKI_5-24 (although with reduced binding affinity) traps for Y204A in a inhibited, dynamically quenched state (Masterson et al., 2011).

Figure 3.

Chemical shift differences between PKA-C^WT and PKA-C^Y204A for Apo, ATPγN-bound and ternary (ATPγN and PKI_5-24-bound). (A) ¹H,¹⁵N combined chemical shift differences mapped on the corresponding X-ray structures (B) Superposition of the active sites in the crystal structures of PKA-C^WT (orange/red) and PKA-C^Y204A (cyan/blue). C. CONCISE analysis of amide resonances.

Fast timescale dynamics are redistributed throughout Y204A

In PKA-C^WT, nucleotide binding drives the enzyme conformation toward a state committed to catalysis (i.e., dynamically committed state) (Masterson et al., 2010; Masterson et al., 2011). The adenine ring binds in a deep cleft located underneath the Gly-rich loop, completing the C-spine, engaging both lobes, and promoting backbone fluctuations in the ps-ns time scale (Masterson et al., 2010; Masterson et al., 2011). The total entropy changes dramatically upon binding the substrate peptide or PKI_5-24. At the same time, the conformational fluctuations of the ternary complexes with the substrate and inhibitor are radically different, with the substrate binding re-distributing the motions throughout the enzyme and the pseudo-substrate quenching the conformational dynamics both in the slow and fast time scale. Based on these data, we hypothesized that the motions in the intermediate (nucleotide-bound) state of the kinase may be responsible for substrate binding cooperativity (Masterson et al., 2011). To test this hypothesis, we probed the conformational dynamics for the PKA-C^Y204A mutant in complex with ATPγN over a range of timescales (Summarized in Table 1). Specifically, we measured the ¹⁵N relaxation parameters R₁ (longitudinal relaxation rate constant), R₂ (longitudinal relaxation rate constant), and ¹H-¹⁵N heteronuclear steady-state NOE (HX-NOE) (Palmer, 2004). From the R₁ and R₂ relaxation rates, we estimated an overall rotational correlation time of approximately 24 and 23 ns for the wild-type and Y204A mutant, respectively, values that are indicative of a monomeric enzyme. The average NOE values for wild-type PKA-C and the mutant Y204A were found to be 0.79 ± 0.13 and 0.81 ± 0.12, respectively, indicating that on average the overall fluctuations in the fast time scale are very similar in both enzymes. However, we identified local differences in the HX-NOEs between the two proteins (Figure S5). Overall, no clear patterns emerged from the data, suggesting that the conformational freedom gained by disrupting the electrostatic node impinged on Y204 is redistributed across the entire enzyme (Figure S5). This reorganization of fast timescale dynamics may contribute to the drastically reduced binding affinity for PKI_5-24 as measured by ITC.

TABLE 1.

Summary of ¹⁵N relaxation and ITC data for PKA-C^WT and PKA-C^Y204A.

	PKA-CWT:ATPγN	PKA-CY204A: ATPγN
Kd for PKI5-24 (μM)a	0.12 ± 0005	14.6 ± 2.1
Average R1 (s−1)	0.51 ± 0.08	0.51 ± 0.06
Average R2 (s−1)	51 ± 5	46 ± 4
Average HX-NOEb	0.79 ± 0.13	0.81 ± 0.12
kex from CPMG (s−1)c	1020 ± 150	NDd
pclosed (%)c	93.5 ± 8.5	NDd

^aDetermined by ITC. Additional results found in Table S1.

^bResidue specific HX-NOE values found in Figure S5.

^cComplete group fit parameters found in Table S2.

^dDynamics are no longer concerted; therefore overall parameters are not available.

Slow conformational dynamics are no longer synchronized in Y204A

According to our proposed model (Masterson et al., 2010), the nucleotide-bound enzyme still undergoes opening and closing transitions, although with a smaller amplitude. We hypothesize that these motions are necessary for organizing the active site for substrate binding and phosphoryl transfer to occur, but also for product release. To test this hypothesis, we measured the rates of conformational exchange for those residues manifesting fast/intermediate exchange in both the wild-type and the mutant. To quantify the rates of these motions, we measured the exchange contribution to the total transverse relaxation rate constant (R_ex) with two different NMR techniques: ¹⁵N Carr-Purcell-Meiboom-Gill (CPMG) relaxation dispersion measurements (Wang et al., 2001) and TROSY-Hahn echo (Wang et al., 2003). The rates obtained from the CPMG relaxation dispersion experiments depend on the equilibrium populations (p_closed and p_open), the chemical shift difference between the two interconverting species (Δω), and the kinetics of exchange (k_ex=k_f+k_r, where the f and r subscripts indicate the forward and reverse rates) (Palmer et al., 2001). From the individual fitting of the dispersion curves, we clustered fifteen residues manifesting similar exchange rates (Figure 4, Tables S2 and S3). These residues are distributed mainly in the small lobe, comprising G55 in the Gly-rich loop, M58 and L59 in β-sheet 2, L77 and K78 in the B-helix, E91 and Q96 in the C-helix, A124 and F318 in the hinge region between the two lobes, and E332 and E446 located in the C-terminal that wraps around the enzyme. Also included from the large lobe are I163, in the catalytic loop, and Y306. The clustering of the relaxation rates for these residues distributed among several regions indicates that in the wild-type in the nucleotide-bound state the small lobe undergoes concerted motions between the open and closed conformations present. Fitting the dispersion profiles using the full Richard-Carver equation (see Supplementary Information) resulted in populations of 94 and 6% for the closed and open states, respectively, with an exchange constant of ~1020±150 s⁻¹ (k_f ~ 66 and k_r ~ 953 s⁻¹). While the broadened resonances for both the wild-type and mutated enzymes are identical, the dispersion curves for the fast/intermediate exchange residues are quite different (Figure 5). Remarkably, the individual fitting carried out on the resonances did not result in similar k_ex values, and the cluster of residues identified for the wild-type as responsible for the synchronous motions is no longer present (Table S4). As a result, a global fit is not possible for multiple residues, and only sparse pairs of residues in the small lobe can be coupled. Therefore, mutation of Y204A induces a dramatic change in the dynamics of PKA-C on the μs-ms timescale in which a large region of the protein involved in active site motions is no longer synchronized. Moreover, several residues of the mutant enzyme show faster exchange dynamics as detected from the individual fits and uncorrelated motions in the small lobe (Tables S2–S4), suggesting that there is a shift in the motional regime. To support the latter, we used the ¹⁵N TROSY Hahn-echo (TrHE) experiment (Wang et al., 2003) for both PKA-C^WT and PKA-C^Y204A (Figure S6). The window of observation of the chemical exchange for CPMG and TrHE is different, with the TrHE more sensitive to faster conformational dynamics (Wang and Palmer, 2003; Wang et al., 2003). For the wild-type, we found a good agreement between the R_ex values measured by CPMG experiments and the corresponding values measured using the TrHE experiments (Figure 6). In sharp contrast, R_ex values measured with these two techniques are markedly different for the mutant, as the R_ex values measured with the TrHE are systematically higher than the corresponding values measured by CPMG experiments. Overall, these data suggest that the mutation increases the rates of motions throughout the enzyme with the most prominent effects localized in the small lobe (β₂- and β₃-sheet as well as B- and C-helix) and between the activation and positioning loop. Therefore, the disruption of the electrostatic node shifts the range of the conformational dynamics to a faster timescale that affects both substrate binding and product release.

Figure 4.

Conformational motions in the slow timescale probed by ¹⁵N CPMG relaxation dispersion. Left: residues showing concerted motions in the wild-type enzyme are mapped on the structure. Right: ¹⁵N CPMG dispersion curves for residues of PKA-C^WT and PKA-C^Y204A reveal a change in the nature of the dynamics upon Y204A mutation.

Figure 5.

Comparison of the R_ex measured by CPMG relaxation dispersion and TrHE. Only the residues with both CPMG and TrHE values are compared. All of the TrHE data are reported in Figure S6.

Figure 6.

Molecular dynamics simulations (5–6 μs) community analysis of PKA-C^WT and PKA-C^Y204A. Left: The community map is colored on the structure. Middle: A graph representation of the community map where the thickness of the lines connecting the communities indicates the presence of correlation of the motions. Right: The C-spine and R-spine anchored to the F-helix. Y204A mutation disrupts semi-rigid communities and causes the dihedral populations to shift in a number of areas related to activation, ATP-binding, and substrate binding. Definition of the communities: ComA: ATP binding and substrate coordination; ComB: αC-helix adjusting; ComC: regulatory and assembly the R spine; ComD: ATP binding and catalytic site; ComE: C-spine supporting; ComF: Activation and substrate binding. ComF1 and ComG: substrate binding.

Molecular dynamics simulations highlight loss of correlated motions in Y204A

To interpret the motions highlighted by the nuclear spin relaxation measurements and visualize how the Y204A mutation alters inter-residue contacts and correlated motions at a global level, we performed multiple-microsecond timescale molecular dynamics simulations for both the wild-type (5.7 μs) and Y204A (5.1 μs) kinases. The MD trajectories were analyzed using the community analysis (Kappel et al., 2012; Newman, 2004; Sethi et al., 2009), which identifies structurally-contiguous regions exhibiting correlated motions. The community analysis views motions in the protein as a dynamical network and highlights correlated motions in proteins as semi-rigid, semi-flexible units of three-dimensional sub-structures that are grouped together and connected to other units by physical contacts and correlated motions. Community analysis of PKA^WT clusters the kinase residues in eight major communities (Figure 6). A prominent feature of the wild-type enzyme is the regulatory ComC, which acts as a hub and is strongly connected to the rest of the community (thick lines in Figure 6). In the mutated enzyme, these correlations are completely rearranged, with ComC splitting into two different sub-communities (ComC and ComC1), losing the role of central communication hub. As a result, the αC-β4 linker, DFG-motif, and several R-spine residues pull away from ComC that contains the αC-helix, a critical domain for catalysis. The loss of correlations within and between the communities parallels the effects of the mutation on the slow time scale motion as probed by NMR and is consistent with the loss of concerted motion in Y204A. Overall the changes in the distribution of motions and the reduced couplings within and between communities throughout the entire enzyme upon mutation support the conclusions reached by the NMR analysis and identify subtle but important changes that reduce the binding affinity and catalytic efficiency for the enzyme

DISCUSSION

There is a passionate debate about the role of conformational fluctuations in enzymatic catalysis (Garcia-Viloca et al., 2004; Hammes-Schiffer, 2002; Hammes-Schiffer and Benkovic, 2006; Pisliakov et al., 2009; Schwartz and Schramm, 2009). Recent studies have inferred that motions in the NMR-detectable timescale might affect the chemical step of catalysis (Bhabha et al., 2011; Boehr et al., 2006b; Whittier et al., 2013). On the other hand, theoretical calculations seem to dismiss this argument (Pisliakov et al., 2009). Nonetheless, enzyme motions are implicated in several steps of catalysis, including allosteric activation (Cui and Karplus, 2008; Lipchock and Loria, 2010; Popovych et al., 2006; Tzeng and Kalodimos, 2009; Tzeng and Kalodimos, 2013), active site optimization (Boehr et al., 2006b; Henzler-Wildman et al., 2007; Mauldin et al., 2009), co-factors or substrate binding, as well as product release (Masterson et al., 2010). An intriguing hypothesis is that enzyme structure and conformational dynamics may have co-evolved (Bhabha et al., 2013; Chao et al., 2012; Jackson et al., 2009; Wolf-Watz et al., 2004). Indeed, the importance of conformational flexibility in protein kinases is emerging in several different instances. In particular, motions in the intermediate-exchange timescale (μs-to-ms) have been detected by NMR in several different kinases, e.g., p38a mitogen-activated protein kinase (Honndorf et al., 2008), Abelson kinase (Skora et al., 2013), the CheA histidine kinase (Wang et al., 2014), and extracellular signal-regulated kinase 2 (ERK2) (Xiao et al., 2014). In past years, we have used nuclear spin relaxation measurements to trace the conformational changes as well as structural fluctuations of kinase A, mimicking major states along the reaction coordinates (Masterson et al., 2010; Masterson et al., 2011). We found that the apo state is malleable and water accessible, but does not display conformational fluctuations that are synchronous to the enzymatic turnover. Nucleotide binding shifts the population of the enzyme toward an intermediate state, providing the phosphoryl group for the reaction and connecting the small lobe to the large lobe via the hydrophobic C-spine. Upon substrate binding, the conformational fluctuations of the enzyme are redistributed, allowing the enzyme to adopt a dynamically competent state (Masterson et al., 2010). In sharp contrast, binding of a pseudo substrate peptide inhibitor such as PKI obliterates the conformational fluctuations, trapping the enzyme in a dynamically quenched state (Masterson et al., 2011).

Although these studies have advanced our understanding of the role of motions on enzymatic turnover, they did not clearly demonstrate the importance of the rates and the correlations of the conformational fluctuations with respect to the structural changes. The X-ray structures of the different states (Taylor et al., 2005) as well as molecular dynamics simulations (Masterson et al., 2011) suggest that motion is a driving force in the catalytic cycle (Lu et al., 2005). Fluctuations in kinase A involve a mechanical hinge-bending motion of the two lobes, featuring an asymmetric twist that brings the glycine-rich loop close to the triphosphate moiety, thus causing the complete closure of the active cleft over the substrate (Li et al., 2002). The kinetics of exchange for these large amplitude motions occurring during the turnover were estimated at a k_ex of ~200 s⁻¹ (Masterson et al., 2010), a value remarkably similar to that of ERK2 (Xiao et al., 2014). However, the importance of the faster rates of motions that we measured in the nucleotide bound form (i.e., k_ex ~ 1020 s⁻¹) has not been fully appreciated. The Y204A mutant offers a unique opportunity to test the role of conformational dynamics as the structures of the PKA-C^WT and PKA-C^Y204A are virtually superimposable (Yang et al., 2004). Yet, the mutation causes a 400-fold decrease of the catalytic efficiency (Moore et al., 2003) and approximately a 100-fold reduction in the affinity for the pseudo-substrate, two phenomena essentially inexplicable from the structural data alone. Thermocalorimetric studies and deuterium exchange mass spectrometry data showed an overall reduction of the PKA-C^Y204A stability, inferring an increase of conformational dynamics (Yang et al., 2005). Indeed, our data show that the key to understanding the anomalous behavior of the mutant is the nucleotide binding event. In contrast to ERK2 (Xiao et al., 2014), where the nucleotide binding seems to have a marginal role in modulating the conformational fluctuations, our data emphasize the central role of the nucleotide in orchestrating phosphoryl transfer in PKA-C. Binding of the nucleotide first links the two hydrophobic spines so that all of the catalytic machinery is integrated. In addition, the nucleotide provides the phosphoryl group necessary for chemistry to take place; however it also pre-organizes the substrate binding site, enhancing its binding affinity, shifting the range of the dynamics in a regime that is competent for catalysis. Importantly, the changes in the rates of the opening and closing motions detected for the nucleotide-bound form of PKA-C^Y204A by the relaxation dispersion measurements are probably responsible for both the decreased affinity for the substrate and the decrease in k_cat, affecting product release. In the wild-type enzyme, the opening and closing motions of the nucleotide-bound form are synchronous for several residues in the small lobe and in the hinge regions. For the Y204A mutant, these conformational fluctuations are no longer synchronous, reducing the enzyme’s catalytic efficiency. These dysfunctional conformational fluctuations of PKA-C^Y204A are too fast to be captured by the CPMG experiments, but are detectable by TrHE experiments. Since the chemical step of the catalysis is too fast to be determined by NMR experiments, a logical explanation for the reduced catalytic turnover in PKA-C^Y204A is that these dysfunctional dynamics reduce the probability of the enzyme reaching a conformation competent for phosphoryl transfer (Daily et al., 2010; Mauldin et al., 2009). While our previous studies have shown that absence of motion renders the kinase inactive, relaxation dispersion data show that the presence of asynchronous motions affects the opening and closing of the enzyme, reducing the enzyme catalytic turnover.

Using a mutation of PKA-C that removes a crucial electrostatic node far from the active site, we demonstrated the relevance of both exchange rates and coordination of the conformational fluctuations for the kinase A catalytic turnover. When the fluctuations of the small lobes are synchronous, the enzyme functions with proper turnover rates. Upon removal of the electrostatic node, the rates of opening and closing are no longer synchronous and the enzyme becomes sluggish. Since different kinases display a wide range of catalytic efficiencies while maintaining the same fold, modulation of the conformational fluctuations across the entire kinome are probably responsible for tuning the signaling efficiency of these enzymes.

EXPERIMENTAL PROCEDURES

ITC

Both PKA-C^WT and PKA-C^Y204A were dialyzed into 20 mM MOPS, 90 mM KCl, 10 mM DTT, 10 mM MgCl₂, 1 mM NaN₃ at pH 6.5. PKA-C concentrations for ITC measurements were between 11.8 – 32 μM. ATPγN was added to reach a final concentration of 2 mM. All ITC measurements were performed with a Microcal VP-ITC instrument at 27°C. Approximately 1.7 mL of PKA-C was used for each experiment and 280 μL of 2–4 mM of ATPγN or 140–210μM of PKI_5-24 in the titrant syringe. All experiments were performed in triplicate. The heat of dilution of the ligand to the buffer was taken into account by measuring the heat of dilution of the ligand to the buffer and was subtracted from the experiment accordingly. The binding curves were analyzed with the NanoAnalyze software (TA Instruments) assuming 1:1 stoichiometry and using the Wiseman Isotherm:

$$\frac{d[MX]}{d[X_{\mathit{tot}}]}=\mathrm{\Delta }H°V_0\left[\frac{1}{2}+\frac{1-(1+r)/2-R_m/2}{{(R_m^2-2R_m(1-r)+{(1+r)}^2)}^{1/2}}\right]$$

where the change of the total complex, d[MX] with respect to the change of the ligand concentration, d[X_tot] is dependent on r, the ratio of the K_d with respect to the total protein concentration, and R_m, the ratio between the total ligand and total protein concentration. The free energy of binding was determined using the following:

$$\mathrm{\Delta }G=\mathit{RT}ln{K}_d$$

where R is the universal gas constant and T is the temperature at measurement (300K). The entropic contribution to binding was calculated using the following:

$$T\mathrm{\Delta }S=\mathrm{\Delta }H-\mathrm{\Delta }G$$

NMR Spectroscopy

Uniformly ¹⁵N-labeded perdeuterated wild-type PKA-C and its Y204A mutant were overexpressed and purified as described elsewhere (Masterson et al., 2008). The NMR samples contained 90 mM KCl, 20 mM KH₂PO₄, 10 mM MEGA-8 (Octanoyl-N-methylglucamide), 5% glycerol, 12 mM ATPγN, 10 mM MgCl₂, 20 mM DTT, 1 mM NaN₃ in a mixed solvent of 5% D₂O in water, pH 6.5. The protein concentrations for PKA-C^WT and PKA-C^Y204A were ~400 μM, as determined from A₂₈₀ measurements. Experiments were performed at 27 °C on Varian Inova spectrometers operating at ¹H frequency of 800 MHz and 900 MHz. The [¹H,¹⁵N] TROSY-HSQC spectra of PKA-C^WT and PKA-C^Y204A in the apo, nucleotide bound (ATPγN), and inhibitor bound (ATPγN and PKI) forms were re-referenced to remove any shifting artifacts. To improve our statistics we carried out the CONCISE analysis using a total of six states (Cembran et al., 2014). The ¹⁵N shifts were then scaled by 0.154 to balance the weight of perturbations along the nitrogen and proton dimensions, and principal component analysis (PCA) was applied independently to each residue to identify the direction of largest variance across the six states defined above. The ratio between the standard deviations along PC1 and PC2 constitutes a measure of how well the different states follow a linear pattern for a given residue. Residues characterized by poor linearity (σ-PC1/σ-PC2) < 3.0) were discarded from our analysis, together with residues with small perturbations (range of chemical shift along PC1 < 0.05 ppm) and thus more heavily affected by experimental error. These thresholds reduced the number of residues available for all states from 155 to 94 (Figure S4). Nuclear spin (¹⁵N) relaxation experiments performed with TROSY detected T₁, T₁ρ and ¹⁵N[¹H]-NOE, TROSY-Hahn-echo, and TROSY-CPMG experiments for the PKA-C^WT and PKA-C^Y204A samples were carried out using 50×1600 points and spectral widths of 2403×10504 Hz in the indirect and direct dimensions. The ¹H and ¹⁵N carrier frequencies were set on water resonance and 120.5 ppm, respectively.

Molecular Dynamics Simulations

The details of the systems simulated are described in the Supporting Information. The systems were prepared using AMBER12 (Gotz et al., 2012; Le Grand et al., 2013). Production simulations (wild-type, 5.7 μs, and Y204A, 5.1 μs) were run on a 512-node Anton supercomputer, and the trajectories were analyzed using mutual information values to capture correlated motions involving semi-rigid regions (Morcos et al., 2010).

Community Analysis

For our community analysis, we computed Cartesian mutual information values instead of dihedral mutual information values, as the former better capture correlated motions involving semi-rigid regions (Morcos et al., 2010), while the latter are more suitable for analyzing couplings between surface sites (McClendon et al., 2009; Wan et al., 2013). We calculated the Cartesian mutual information for backbone Cα atoms, as well as for an additional representative terminal atom for each residue. The matrix of mutual information among these atoms was calculated with the MutInf software package (McClendon et al., 2009; McClendon et al., 2012), splitting each trajectory into six blocks of equal size and using uniformly binned histograms with 24 bins per degree of freedom. We used 20 sets of scrambled data to estimate the distribution of mutual information under the null hypothesis of independence, and calculated the excess mutual information with the previously-described under sampling correction (McClendon et al., 2009).

Supplementary Material

supplement

Figure S1: Isothermal Titration Calorimetry (ITC) with ATPγN to the apo form and the pseudo-substrate PKI_5-24 to the ATPγN-bound form of PKA-C^WT and PKA-C^Y204A. Top: ATPγN binding to Apo PKA-C^WT and the PKA-C^Y204A mutant. Bottom: Pseudo-substrate, PKI_5-24, binding to PKA-C^WT and PKA-C^Y204A in the presence of 2 mM ATPγN. All measurements were performed in triplicate and the error reflects the standard deviation.

Figure S2: Overlays of the [¹H, ¹⁵N]-TROSY-HSQC spectra for PKA-C^WT and PKA-C^Y204A for the three forms of the kinase: apo (top), ATPγN-bound (middle), and ATPγN and PKI_5-24-bound (bottom).

Figure S3: Qualitative comparison of the B-factor from the crystal structure 1BKX (left) with the resonances undergoing broadening beyond detection upon binding the nucleotide (also shown in Figure 2). B-factors are shown in a rainbow scale from blue (1) to red (100).

Figure S4. Left: List of residues that defined linear trajectories for the 6 states analyzed with CONCISE using the thresholds reported in the Methods section. Right: The linear residues are mapped onto the kinase structure (1ATP).

Figure S5: ¹H-¹⁵N HX-NOE as probe of fast conformational dynamics. Top: The HX-NOE is plotted as a function of the PKA sequence where the red colored values represent NOEs above or below one standard deviation from the average. Bottom: HX-NOE values reported on the crystal structure 1ATP of the kinase. Bulges and colors show regions with higher mobility. S10, Q12, E13, E17, T37, I46, K61, K105 are more rigid in the mutant (higher values of HX-NOEs), while K7, N67, K78 are more dynamic (lower HX-NOE values). Residues near or belonging to the linker regions and preceding the D-helix (Y117, M120 and Y122) are relatively rigid; whereas G125, located near a C-spine residue, and A188, right after the DFG-loop, are more mobile. In the peptide positioning loop, A206 becomes rigid, while E208 becomes mobile in the Y204A enzyme.

Figure S6: ¹⁵N TROSY Hahn-echo as a probe for slow conformational dynamics. Top: R_ex values obtained from the TrHE experiments are plotted as a function of the PKA sequence where the red colored values represent R_ex > 7 s⁻¹ and |ΔR_ex| > 7 s⁻¹. Due to the magnitude and orientation of the ¹⁵N chemical shift tensor, the R_ex values for the non-exchanging residues are scattered around zero with a standard deviation of approximately 2 s⁻¹. Bottom: R_ex values are shown on the crystal structure 1ATP of the kinase with bulges and colors highlighting the regions with higher mobility.

TABLE S1, related to Table 1: Thermodynamics of ligand binding to PKA-C^WT and PKA-C^Y204A determined by ITC. Binding of the ATP mimic, ATPγN, to Apo form of the enzyme (top) was measured as well as PKI_5-24 binding of the nucleotide bound form of the enzyme (bottom).

TABLE S2, related to Table 1. ¹⁵N CPMG relaxation dispersion group fit of PKA-C^WT.

TABLE S3, related to Figure 4. ¹⁵N CPMG relaxation dispersion individual fit of PKA-C^WT.

TABLE S4, related to Figure 5. ¹⁵N CPMG relaxation dispersion individual fit of PKA-C^Y204A.

HIGHLIGHTS.

• Synchronous motions are essential for kinase A catalytic efficiency

• Allosteric mutation globally de-synchronizes motions

• Optimal conformational fluctuations are central to kinase turnover.