Intrinsically disordered PEP-19 confers unique dynamic properties to apo and calcium calmodulin

Abstract

PEP-19 (Purkinje cell protein 4) is an intrinsically disordered protein with an IQ calmodulin (CaM) binding motif. Expression of PEP-19 was recently shown to protect cells from apoptosis and cell death due to Ca²⁺ overload. Our initial studies showed that PEP-19 causes novel and dramatic increases in the rates of Ca²⁺ association and dissociation from the C-domain of CaM. The goal of this work was to study interactions between C-domain of CaM (C-CaM) and PEP-19 by solution NMR to identify mechanisms by which PEP-19 regulates Ca²⁺ binding to CaM. Our results show that PEP-19 causes a greater structural change in apo C-CaM relative to Ca²⁺-C-CaM, and that the 1^st Ca²⁺ binds preferentially to site IV in the presence of PEP-19 with exchange characteristics that are consistent with a decrease in Ca²⁺ binding cooperativity. Relatively weak binding of PEP-19 has distinct effects on chemical/conformational exchange on the μs-ms timescale. In apo C-CaM, PEP-19 binding causes a redistribution of residues that experience conformational exchange, leading to an increase in residues around Ca²⁺ binding site IV that undergo conformational exchange on μs-ms timescale. This appears to be due to an allosteric effect since these residues are not localized to the PEP-19 binding site. In contrast, PEP-19 increases the number of residues that exhibit conformational exchange in Ca²⁺-C-CaM. These residues are primarily localized to the PEP-19 binding site, but also include Asp93 in site III. These results provide working models for the role of protein dynamics in regulation of Ca²⁺ binding to CaM by PEP-19.

PEP-19 (Purkinje cell protein 4) is a small 6.7 kDa polypeptide that was initially identified in the central nervous system (CNS), but is now known to be present in a variety of other tissues (for review see (1–5)). PEP-19 has no known intrinsic activity other than binding to calmodulin (CaM) in the presence or absence of Ca²⁺ via an IQ CaM binding motif. This gives PEP-19 the potential to exert broad cellular effects as a regulator of CaM signaling. This idea is consistent with a general cytoprotective role for PEP-19 put forth based on the fact that Purkinje cells of the cerebellum and granule-cell neurons in the dentate gyrus, which have high levels of PEP-19, are largely spared from the effects of Alzheimer’s Disease, while PEP-19 negative cells are severely affected (6). Conversely, cell types that are most affected by Huntington’s disease exhibit a significant loss of expression of PEP-19 (7). Experimental studies also support a cytoprotective role for PEP-19 since expression of PEP-19 greatly inhibits cell death due to apoptosis (8, 9), or glutamate-induced Ca²⁺ cytotoxicity (9). Interestingly, PEP-19 is greatly increased in human leiomyoma versus matched myometrium (10), and high levels of PEP-19 are found in 58 of 60 NCI-60 tumor cell lines (11), even cells derived from tissues that express low or no PEP-19. This may be related to increased survival potential of transformed cells due to antiapoptotic effects of PEP-19. These observations support the idea that PEP-19 protects cells against apoptosis and Ca²⁺ cytotoxicity due to normal or pathogenic conditions, and would explain the need for high levels of PEP-19 in cells with highly active Ca²⁺ signaling systems.

The potential for PEP-19 to function as a regulator of CaM signaling led us to a series of studies to characterize its structural and biochemical properties, and its effects on CaM. We showed that PEP-19 binds relatively weakly, but selectively to the C-domain of CaM and has novel effects on the Ca²⁺ binding properties of CaM (12). Specifically, PEP-19 causes a 40-fold increase in the rates of association (k_on) and dissociation (k_off) of Ca²⁺ at C-domain of CaM, with little effect on Ca²⁺ binding affinity. Selective effects on the C-domain of CaM are important since the k_on and k_off for Ca²⁺ binding to the C-domain are up to 150-fold slower than for the N-domain, and represent a rate-limiting step for activation of Ca²⁺-dependent target proteins. We also showed that PEP-19 is intrinsically disordered, but with residual structure localized to an acidic/IQ motif that includes the IQ sequence and an adjacent acidic sequence (13). The core IQ motif is required for binding to CaM, but the acidic region is necessary to restrict binding to the C-domain, and for modulating Ca²⁺ binding to CaM (14).

The C-domain of CaM has intriguing dynamic properties that may act in synergy with the intrinsic disorder of PEP-19. Protein dynamics from NMR showed that the C-domain of CaM presents an ensemble of conformational states. The “open” conformation is dominant in the presence of Ca²⁺, and the “closed” conformation dominates in the apo state, but other minor populations of intermediate states exist (15–17). Akke and colleagues showed that the rate of Ca²⁺ exchange in the C-domain CaM is correlated with rates of chemical/conformational exchange (18). Thus, binding targets to CaM provides a mechanism for altering its Ca²⁺ binding properties by changing its dynamic properties and the populations of CaM conformers. For example, binding high-affinity targets to Ca²⁺-CaM, shifts the conformational equilibrium by stabilizing an “open-like” Ca²⁺-bound conformation, which greatly decreases k_off (19). Based on these observations, we propose an inverse mechanism in which low affinity binding of intrinsically disordered PEP-19 modulates the slow dynamics of apo CaM and Ca²⁺-bound CaM to increase Ca²⁺ binding rate constants.

The present study provides experimental evidence to validate the above hypothesis by investigating the effects of PEP-19 on backbone dynamics of the isolated C-domain of CaM (C-CaM). In contrast to high-affinity CaM binding peptides, PEP-19 binding increases slow motions in the C-domain of CaM. In particular, residues located in Ca²⁺ binding sites in both apo- and Ca²⁺-bound forms undergo conformational exchange on us-ms timescale. This provides working models for the structural and dynamic basis for the effect of PEP-19 on the kinetics of Ca²⁺ binding to CaM.

MATERIALS AND METHODS

Protein purification

Recombinant CaM and PEP-19 were cloned, expressed in bacteria and purified as described previously (12, 13, 20, 21). A bacterial expression vector for C-CaM (residues 76–148) was kindly provided by Dr. Madeline Shea (U. Iowa). Proteins were decalcified by addition of 5 mM EGTA and 0.1 mM BAPTA followed by desalting on a Biorad P6DG or P2DG size exclusion column into a buffer that was decalcified by passage over a calcium-sponge column (Molecular Probes). Protein concentrations were estimated using an extinction coefficient of ε_276nm = 0.18 ml⁻¹/mg for CaM, and ε_215nm= 0.59 ml⁻¹/mg for PEP-19, or by using the BCA protein assay (Pierce). More precise concentrations of CaM and C-CaM were determined by measuring the amount of Ca²⁺ released from the C-domain of Ca²⁺-saturated proteins. Briefly, a solution of approximately 2 μM CaM or C-CaM and 30 μM Ca²⁺ was rapidly mixed with a solution of 300 μM of the Ca²⁺ sensitive dye, Quin-2, using a stopped flow fluorimeter to monitor the slow release of Ca²⁺ from the C-domain as described previously (12). The amount of Ca²⁺ released from the sample was determined by calibrating the fluorescence response with EGTA standards. This was then used to calculate the concentration of CaM or C-CaM.

Equilibrium Ca²⁺ titrations monitored by tyrosine fluorescence

Equilibrium Ca²⁺ binding constants for CaM and C-CaM in the presecnce or absence of PEP-19 were determined using tryosine fluoresence as described previously (22). Solutions containing 20 mM MOPS, pH 7.5, 100 mM KCl, 1 mM EGTA, 1 mM HEDTA, 1 mM NTA, 5 μM CaM or C-CaM with or without 30 μM PEP-19 were titrated with a Ca²⁺ stock to achieve a wide range of free Ca²⁺. The concentration of total Ca²⁺ needed to achieve a desired free Ca²⁺ concentration was determined using the online calculator MaxChelator (http://www.stanford.edu/~cpatton/maxc.html). The Ca²⁺ stock was prepared in the same buffer, including proteins, such that only the concentration of Ca²⁺ was changed during the titration. Tyrosine fluorescence intensity was plotted against the free Ca²⁺ concentration and fit to the following form of the Hill equation:

$$F=F_{min}+(F_{max}-F_{min})\left[\frac{{[\text{Ca}]}^n}{{[\text{Ca}]}^n+{[k_{\text{Ca}}]}^n}\right]$$ (1)

where [Ca²⁺] is the free Ca²⁺ concentration, F is the fluorescence intensity at a given free Ca²⁺ concentration, F_min is the initial fluorescence intensity in the absence of added Ca²⁺, F_max is the fluorescence at maximal Ca²⁺, k_Ca is the concentration of Ca²⁺ at which the change in fluorescence is half maximal and n is the Hill coefficient.

NMR methodology

All NMR experiments were performed on a Bruker DRX 600 MHz spectrometer equipped with a 5 mm triple-resonance cryoprobe at 298 K. NMR samples were prepared using decalcified proteins in a buffer containing 10 mM imidazole, 100 mM KCl and 5% D₂O at pH 6.3. Samples used for relaxation measurements and backbone assignments contained 0.5 mM C-CaM uniformly labeled with ¹⁵N or ¹³C/¹⁵N, with or without 0.7 mM PEP-19. Excess EDTA or CaCl₂ was added to maintain apo or Ca²⁺-bound states, respectively.

Backbone assignments for C-CaM/PEP-19 for apo and Ca²⁺-bound states were obtained using the following 3D experiments: HNCO, HNCA, HN(CO)CA, HNCACB, CBCA(CO)NH, ¹⁵N HSQC-TOCSY and ¹⁵N-edited NOESY-HSQC experiments. The ¹⁵N longitudinal (R₁), transverse (R₂) relaxation rate constants and heteronuclear [¹H]-¹⁵N NOE were acquired using published pulse sequences (23). Experiments were recorded at delays of 20, 100, 200, 400, 600, 800, 1000, 1200, and 1400 ms for R₁ measurements, and delays of 16.96, 33.92, 50.88, 67.84, 84.8, 118.72, 135.68, 152.64, and 169.6 ms for R₂ measurements. [¹H]-¹⁵N NOE values were measured from spectra with and without proton saturation recorded in an interleaved manner. Proton saturation was acquired using a 120° ¹H pulse applied every 5.0 s. In the case of the NONOE spectra, a net relaxation delay of 5.0 s was employed, while a relaxation delay of 2.0 s prior to a 3.0 s proton presaturation period was employed for the NOE spectra.

All NMR spectra were processed and analyzed using Topspin 2.0 (Bruker) and FELIX 2004 (MSI, San Diego, CA). ¹H chemical shifts were referenced to DSS (2,2-dimethyl-2-silapentane-5-sulphonate), and ¹⁵N/¹³C chemical shifts were referenced indirectly using their respective gyromagnetic ratios (24). The average amide chemical shift change (Δδ_avg) was calculated using the following equation:

$$\mathrm{\Delta }{\delta }_{\mathit{avg}}=\sqrt{\frac{{(\mathrm{\Delta }\delta \text{H})}^2+{(\mathrm{\Delta }\delta \text{N}/5)}^2}{2}}$$ (2)

Where Δδ H and Δδ N are the changes in ¹H and ¹⁵N chemical shifts, respectively.

Analysis of relaxation data

Values for R₁, R₂ and their uncertainties were derived by plotting peak intensity versus delay time and fitting the data to a single exponential decay equation using non-linear least squares analysis. [¹H]-¹⁵N NOE values were calculated from the ratios of peak intensities with and without presaturation. The standard deviation of the NOE was calculated from the root mean square value of the background noise in the spectra (23).

Reduced spectral density mapping was used to analyze the relaxation data (25, 26). Values for J(0), J(ω_N), and

$$J({\omega }_{\text{H}})\approx J({\omega }_{\text{H}}+{\omega }_{\text{N}})\approx J({\omega }_{\text{H}}-{\omega }_{\text{N}})$$ (3)

$$R_{\text{NOE}}=(\mathit{NOE}-1)R_1({\gamma }_{\text{N}}/{\gamma }_{\text{H}})$$ (4)

$$J(0)=(-0.75R_1+1.5R_2-0.9R_{\text{NOE}})/E$$ (5)

$$J({\omega }_{\text{N}})=(R_1-1.4R_{\text{NOE}})/E$$ (6)

$$J({\omega }_{\text{H}})=R_{\text{NOE}}/5A$$ (7)

$$\begin{array}{lll}E=3A+C^2\hfill & A={({\mu }_{0}h{\gamma }_{\text{H}}{\gamma }_{\text{N}}/16{\pi }^2{r_{\text{NH}}}^3)}^2\hfill & C=({\omega }_{\text{N}}/\sqrt{3})({\sigma }_{\parallel }-{\sigma }_{\perp })\hfill \end{array}$$

Where μ₀ is the permeability of free space, h is Plank’s constant, γ_H and γ_N are the gyromagnetic ratios of ¹H and ¹⁵N, respectively, r_NH is the average N-H bond length (1.02 Å). Uncertainties in the spectral density parameters were estimated using 500 Monte Carlo simulations based on the uncertainties in the measured relaxation parameters.

Residue-specific rotational correlation times (τ_m) were estimated from the ratio of R₂/R₁ (27) (software r2r1_tm, A. G. Palmer III, Columbia University). Residues with R₂/R₁ ratios falling outside one standard deviation (SD) of the mean, as well as residues with NOE < 0.65, were excluded from the analysis (28).

RESULTS

C-CaM is a valid structural model for studying interactions between CaM and PEP-19

We showed previously that application of NMR to study the interactions between PEP-19 and apo CaM was severely limited since resonances for most backbone amides were broadened beyond detection in the apo CaM/PEP-19 complex (13). This was not due to the interaction of PEP-19 with the N-domain of CaM since amide chemical shifts in the N-domain are unaffected by PEP-19 binding (see Supplemental Figure 1). We felt that use of the isolated C-domain of CaM (C-CaM) may overcome this limitation due to different rotational correlation times (time)or other factors relative to intact CaM. This strategy was justified based on previous studies showing that the isolated C-domain of CaM (C-CaM) retains the biochemical properties of the intact protein (29–32). In addition, we showed that PEP-19 binds preferentially to the C-domain of CaM in the absence or presence of Ca²⁺, and that PEP-19 has the same effects on the Ca²⁺ k_off rate of both C-CaM and intact CaM (13, 14). Thus, the initial goal of the current study was to determine if amide resonances in apo C-CaM can be observed upon binding PEP-19. Indeed, addition of PEP-19 to apo C-CaM allowed observation of slow, two-state exchange for most amides. Backbone assignments for apo C-CaM bound to PEP-19 were obtained by HNCA, HN(CO)CA, HNCACB, CBCA(CO)NH, ¹⁵N HSQC-TOCSY and ¹⁵N-edited NOESY-HSQC experiments. All backbone amides were assigned except for Asp131 and Gly134, which are undetectable due to line broadening.

We next determined if C-CaM provided a good structural mimic of the C-domain from intact CaM in the absence of PEP-19. Supplemental Figure 2 compares the ¹H-¹⁵N HSQC spectra of C-CaM versus intact CaM in the presence and absence of Ca²⁺. The spectra of C-CaM are greatly simplified due to the absence of amide cross peaks from residues in the N-domain of CaM. As expected, amide chemical shifts for residues at the N-terminus of C-CaM, including the N-terminal portion of helix E, exhibit greater average amide chemical shift differences (Δδ_avg) relative to CaM due to lack of influence from the tethered N-domain (see Supplemental Figure 3). However, average Δδ_avg for residues 88–148 of C-CaM and intact CaM are 0.014 and 0.012 in the absence and presence of Ca²⁺, respectively, which are within the error of these measurements and calculations. In addition, the secondary structures of C-CaM and the C-domain of intact CaM are in excellent agreement in the presence and absence of Ca²⁺ as determined by ¹³C_α secondary chemical shifts, ¹H-¹⁵N NOESY-HSQC, ³J_HAHA, and the Chemical Shift Index (see Supplemental Table 1).

We next compared ¹H-¹⁵N HSQC spectra for Ca²⁺-bound C-CaM versus intact CaM at increasing concentrations of PEP-19 at 298 K. Residues in the N-domain of CaM are not affected by PEP-19 binding and most resonances in Ca²⁺-bound C-CaM exhibit characteristics of fast to intermediate exchange when titrated with PEP-19. This allowed assignments of backbone amides by following the movement of cross peaks. Assignments were then verified by ¹H, ¹⁵N NOESY-HSQC. All residues were assigned except for Thr110, Asn111, Gly113, Glu114, and Lys115, due to severe resonance line broadening. Supplemental Figure 4 shows that all amide cross peaks for Ca²⁺-C-CaM bound to PEP-19 superimpose on resonances associated with the C-domain of intact CaM bound to PEP-19 except residues 78 to 82 at the N-terminus of C-CaM. Thus, binding PEP-19 induces similar conformations in the Ca²⁺-bound forms of both C-CaM and the C-domain of intact CaM. Together, supplemental Figures 1 to 4 demonstrate that C-CaM is a valid structural model for the study of interactions between CaM and PEP-19 in the presence or absence of Ca²⁺.

Characteristics of slow and fast exchange on the NMR time scale for C-CaM amide resonances during titration with PEP-19 in the absence and presence of Ca²⁺, respectively, are consistent with the kinetics of binding PEP-19. Observed exchange characteristics are dependent on the relationship between chemical exchange (k_ex) and the difference in frequency for amide resonances between the bound and free states (Δω). Fast exchange is observed if k_ex > Δω, while slow exchange is seen if k_ex < Δω Based on this, we can conclude that k_ex > 150 s⁻¹ in the presence of Ca²⁺, and k_ex < 20 s⁻¹ in the absence of Ca²⁺. The k_ex will be largely determined by k_off during titration of C-CaM with PEP-19 since k_ex = [PEP-19]_free*k_on + k_off, and [PEP-19]_free will be very low until its total level exceeds that of C-CaM in the NMR sample. Thus, the above partial limits of k_ex based on Δω are reasonable since we know that k_off values are 400 s⁻¹ and 5 s⁻¹ for binding PEP-19 to CaM in the presence and absence of Ca²⁺, respectively (13, 14).

PEP-19 has a greater effect on the structure of apo versus Ca²⁺-bound C-CaM

Black bars in Figure 1A show that binding PEP-19 causes relatively small Δδ_avg in Ca²⁺-C-CaM at 298 K (see Supplemental Figure 5A for complete spectra). The largest values of Δδ_avg are seen for residues Met144 to Lys148 at the C-terminus, while amide resonances for residues Thr110 to Lys115 in the linker between sites III and IV are broadened beyond detection upon binding PEP-19. Experiments performed previously at 320 K with intact Ca²⁺-CaM (14) showed very similar patterns of Δδ_avg (see grey bars in Figure 1A), except that resonances for residues Thr110 to Lys115 were observed. This temperature dependence indicates that significant conformational exchange occurs on the μs-ms timescale for residues at the end of helix F and the linker between helices F and G of Ca²⁺-CaM when bound to PEP-19.

Figure 1. Effect of PEP-19 on C-CaM amides in the absence and presence of Ca²⁺.

(A) Amide chemical shift perturbations (Δδ_avg) for Ca²⁺-bound C-CaM at 298 K are shown by black bars. Grey bars show Δδ_avg that result from binding PEP-19 to the C-domain of intact Ca²⁺-CaM at 320 K (14). Asterisks denote resonances with significant line-broadening in the presence of PEP-19. (B) Δδ_avg for apo C-CaM at 298 K. Panels (C) and (D) map residues with Δδ_avg >0.2 ppm on the solvent accessible surfaces of Ca²⁺-CaM (pdb:1J7P) and apo-C-CaM (pdb:1F71), respectively. The structures are presented in the same relative orientation. The diagram below panel B shows the relative positions of the helices E to H in the primary sequence of CaM. The 12 amino acids Ca²⁺ binding sites III and IV are also shown in grey. Coil, β-strand and α-helical portions of these structures are indicated by loops, arrows, and boxes, respectively.

Figure 1B shows that binding PEP-19 causes much greater Δδ_avg in apo C-CaM compared to Ca²⁺-bound C-CaM. Similar to Ca²⁺-C-CaM, residues with the largest Δδ_avg in apo C-CaM are localized to helix F and the linker between helices F and G, but changes are also seen for residues 101–103 at the transition between β-strand and helical structures in Ca²⁺ binding site III. Figure 1C and D illustrate the binding interface between C-CaM and PEP-19 in the absence and presence of Ca²⁺ based on residues with Δδ_avg>0.2 ppm.

Experiments were performed to determine if apo and Ca²⁺-bound C-CaM have different secondary structures when bound to PEP-19 that could account for different magnitudes and patterns of chemical shift changes in Figures 1A and 1B. Supplemental Figure 6 and Supplemental Table 1 show that secondary structures in the C-domain of CaM are essentially identical in the presence and absence of Ca²⁺ or PEP-19. Moreover, the antiparallel β-sheet between loops III and IV in apo and Ca²⁺-bound C-CaM remain intact when bound to PEP-19 based on the presence of long-range NOEs (d_NN) between Ile100 and Val136 (data not shown). The largest difference in secondary structure is observed for helix F, which terminates at Val108 in apo C-CaM when bound to PEP-19, but extends to Asn111 under all other conditions. These data, together with Figure 1, suggest that Ca²⁺-bound C-CaM remains predominantly in an “open-like” conformation in the presence of PEP-19, but that binding PEP-19 may alter interhelical angles in apo C-CaM.

Effect of PEP-19 on Ca²⁺ binding cooperativity of C-CaM

We showed previously that PEP-19 increased Ca²⁺ k_off and k_on rates, and decreased the cooperativity of Ca²⁺ binding to the C-domain of intact CaM with out greatly affecting the K_d (14). Specifically, the macroscopic Ca²⁺ dissociation constants for the C-domain of CaM are K_d1 = 17 μM and K_d2 = 0.4 μM. This indicates positive cooperativity since K_d1 > 4K_d2, with a change in free energy due to cooperativity (ΔΔ G_c) of −3.4 kcal/mole. In the presence of PEP-19, K_d1 = 4.7 μM and K_d2 =2.0 μM, respectively, which demonstrates a significant loss of cooperativity, with ΔΔ G_c = −1.3 kcal/mole. Despite these changes in macroscopic binding constants and cooperativity, the overall K_ds for binding Ca²⁺ to the C-domain are very similar in the presence or absence of PEP-19. We also showed that PEP-19 increased, the Ca²⁺ k_off rate of isolated C-CaM (14). Figure 2 uses fluorescence from Tyr99 and Tyr138 to show that PEP-19 also decreases the cooperativity of Ca²⁺ binding to C-CaM as indicated by a decreased Hill coefficient in the presence of PEP-19, but does not greatly affect the K_d for Ca²⁺ binding. Thus, the effects of PEP-19 on the C-domain of intact CaM are also seen for C-CaM.

Figure 2. Effect of PEP-19 on the Ca²⁺ binding constant (K_ca) and Hill coefficient (n).

Intrinsic tyrosine fluorescence was monitored during titration of CaM (○), CaM+PEP-19 (□), C-CaM (●), and C-CaM/PEP-19 (■) with Ca²⁺. The lines show fits of the data to the Hill equation (Eq.1).

To identify potential structural manifestations of decreased cooperativity, amide chemical shifts were monitored by a series of ¹H-¹⁵N HSQC spectra collected during titration of C-CaM with Ca²⁺ in the absence or presence of PEP-19 at 298 K. Figure 3 uses Thr117 to demonstrate slow two-state exchange observed for amides resonances in the absence of PEP-19. At intermediate levels of Ca²⁺ (Figure 3B and Figure 3C), resonances for both 0-Ca²⁺ and 2-Ca²⁺ forms are observed with relative intensities that are proportional to their populations. No evidence for the 1-Ca²⁺ form is observed at intermediate Ca²⁺ levels in the absence of PEP-19, indicating highly cooperative Ca²⁺ binding.

Figure 3. Ca²⁺ -dependent changes in amide chemical shifts of C-CaM show strong cooperativity.

Resonance of Thr117 in the linker region of C-CaM exhibited two-state slow exchange at various molar ratio of Ca²⁺ added into C-CaM, indicating highly cooperative Ca²⁺ binding to sites III and IV. (A) [Ca²⁺]/[C-CaM]=0; (B) [Ca²⁺]/[C-CaM]=0.3; (C) [Ca²⁺]/[C-CaM]=0.6; (D) [Ca²⁺]/[C-CaM]=2.0.

Figure 4 shows that fast to intermediate exchange on the NMR chemical timescale is observed upon titration of C-CaM with Ca²⁺ in the presence of PEP-19. Peaks corresponding to residues in Ca²⁺ binding sites III and IV, and helices E and F have large Ca²⁺-dependent changes in amide chemical shifts that could not be detected at intermediate Ca²⁺ titration points due to line broadening. However, residues such as Thr117 that experience smaller overall Ca²⁺-dependent amide chemical shift changes, could be observed at all Ca²⁺ levels. The NMR spectral characteristics for Thr117 in Figure 4A show resonance line broadening and splitting into multiple peaks at low Ca²⁺ levels, indicating rapid exchange between multiple conformations. Moreover, during titration with Ca²⁺, Thr117 exhibits a biphasic change with an upfield shift in the ¹H dimension between Ca²⁺/C-CaM ratios of 0 and 1.0, and then a downfield shift in the ¹⁵N dimension at Ca²⁺/C-CaM ratios between 1.0 and 2.0. These data are consistent with decreased cooperativity of Ca²⁺ binding in the presence of PEP-19 that results in significant populations of the 1-Ca²⁺ form of C-CaM at substoichiometric levels of Ca²⁺.

Figure 4. Ca²⁺ -dependent changes in amide chemical shifts of C-CaM in the presence of PEP-19 show decreased cooperativity.

Various ratios of [Ca²⁺]/[C-CaM] between 0 and 2.0 are indicated. Exchange characteristics for Thr117 shown in Panel (A) indicate lower degree of cooperativity of Ca²⁺ binding in the presence of PEP-19. Exchange characteristics for Gly98 in Ca²⁺ binding site III and Gly134 in the Ca²⁺ binding site IV shown in Panel (B) indicates preferential binding of the 1^st Ca²⁺ to site IV.

Gly98 and Gly134 at the sixth position of loops III and IV, respectively, were selected to provide information on the sequential nature of Ca²⁺ binding to C-CaM in the presence of PEP-19. These glycines provide markers for Ca²⁺ binding since they facilitate the unusual main chain conformation that allows the Ca²⁺ ligands to take up coordinating positions (33). The large differences in amide chemical shifts of Gly98 and Gly134 between apo and Ca²⁺-bound forms lead to severe resonance broadening at intermediate Ca²⁺ levels. However, Figure 4B shows that the change in chemical shift for Gly134 is maximal at a Ca²⁺/C-CaM ratio of around 1.0, while the change in chemical shift for Gly98 is not maximal until a Ca²⁺/C-CaM ratio of 2.0. These data indicate that PEP-19 promotes preferential binding of the first Ca²⁺ to site IV. However, Figure 4B shows that Gly98 in site III experiences minor conformational change upon binding the first Ca²⁺ to site IV, and binding the second Ca²⁺ to site III sharpens the line width of Gly134 in site IV. The simplest interpretation of these data is that PEP-19 decreases, but does not eliminate cooperativity between sites III and IV.

Effects of PEP-19 on the backbone dynamics of C-CaM

To better understand the effects of PEP-19 on backbone dynamics in C-CaM, ¹⁵N relaxation experiments for apo and Ca²⁺-bound C-CaM were collected in the absence or presence of PEP-19. The [¹H]-¹⁵N NOE is typically most sensitive to motions on the ps –ns timescale, with values near 1.0 indicating a lack of such motions, and lower values indicating increasing local flexibility of the polypeptide. Similar to previous reports (15, 17, 34), Figure 5A and 5F show significant reductions in NOE values for residues at the N- and C- terminus of C-CaM in the presence or absence of Ca²⁺. The two Ca²⁺ binding loops and the linker between helix F and helix G also show lower NOE values. Binding PEP-19 had no significant effect on NOE values for apo- or Ca²⁺-bound C-CaM (see Table 1). This confirms the results in Supplemental Table 1 showing that C-CaM/PEP-19 complexes have well-defined backbone structures.

Figure 5. Backbone ¹⁵N relaxation parameters for apo C-CaM, apo C-CaM/PEP-19, Ca²⁺-bound C-CaM and Ca²⁺-bound C-CaM/PEP-19.

R₁, R₂, [¹H]-¹⁵N NOE, R₂R₁ products and R₂/R₁ ratios are shown for apo- and Ca²⁺-C-CaM in the absence (open circles) and presence (closed circles) of PEP-19. The dashed (with out PEP-19) and solid lines (with PEP-19) in Panels C, D, G and H indicate the mean +1.0 SD. Asterisks denote resonances with significant line-broadening in the presence of PEP-19.

Table 1.

Summary of R₁, R₂, NOE, R₁R₂, R₂/R₁, τ_m, J(0), J(ω_N), and J(ω_H)

Protein	15N relaxation parameters			R1R2	R2/R1	τm	reduced spectral density mapping
	R1	R2	NOE				J(0)	J(ωN)	J(ωH)
	s−1	s−1				ns	ns/rad	ns/rad	ns/rad
Apo C-CaM	2.14±0.13	7.6±2.2	0.67±0.16	16.3±5.2	3.51±0.96	4.83±0.46	1.88±0.64	0.40±0.03	0.011±0.004
Apo C-CaM +PEP-19	1.58±0.085	12.5±3.2	0.70±0.17	19.7±5.2	7.93±2.00	8.33±0.66	3.40±0.92	0.29±0.017	0.008±0.004
Ca2+ C-CaM	2.09±0.17	7.1±1.0	0.68±0.19	14.8±2.7	3.36±0.33	4.81±0.19	1.74±0.26	0.39±0.04	0.010±0.005
Ca2+ C-CaM +PEP-19	1.56±0.091	12.2±2.5	0.70±0.18	18.9±4.1	7.82±1.60	8.38±0.37	3.08±0.79	0.29±0.014	0.009±0.007

Figure 5 and Table 1 show that binding PEP-19 to C-CaM causes an overall decrease in R₁, and an increase in R₂ in the presence or absence of Ca²⁺. A molecular rotational correlation time (τ_m) of 4.8 ns is estimated from R₂/R₁ ratios for free apo and Ca²⁺-bound C-CaM (see Table 1). This is consistent with the τ_m of 4.9 ns determined by Malmendal et al. (15) for apo C-CaM using the mean η_xy/η_z ratio and Model-Free analysis. The estimated τ_m for C-CaM is increased to 8.3 ns when bound to PEP-19 in the presence or absence of Ca²⁺. The τ_m ratio of 1.73 for PEP-19/C-CaM versus free C-CaM is comparable their molecular weight ratio of 1.77, which suggests that the complexes are close to globular in shape. Residues with unusually high R₂/R₁ ratios relative to the average indicate regions that undergo conformational exchange processes on the μs-ms timescale (35). The R₁R₂ product is an effective discriminator of motional anisotropy and chemical/conformational exchange (36). Figure 5D and 5I show the R₂R₁ product, and Figures 5E and 5J show R₂/R₁ ratios for residues in apo and Ca²⁺-bound C-CaM in the presence or absence of PEP-19. Table 2 lists residues with R₂/R₁ and R₁R₂ values that are both greater than 1SD relative to the average.

Table 2.

Summary of Residues that Undergo Conformational Exchange on the μs-ms Timescale

Protein	R2/R1		R2R1		Reduced Spectral Density Mapping		Exchange Broadening
Apo C-CaM	Glu87 Arg90 Phe92 Ile100	Met109 Thr110 Asp131 Glu139	Glu87 Arg90 Phe92 Ile100	Met109 Thr110 Asp131 Glu139	Glu87 Arg90 Phe92 Ile100	Met109 Thr110 Asp131 Glu139
Apo C-CaM +PEP-19	Ala102 Thr110 Glu114 Ile125 Glu127	Ala128 Gly132 Val136 Glu139	Ala102 Thr110 Glu114 Asp122 Ile125 Arg126	Glu127 Ala128 Gly132 Val136 Glu139	Ala102 Thr110 Glu114 Ile125 Glu127	Ala128 Gly132 Val136 Glu139	Asp131 Gly134
Ca2+-C-CaM	Asp93 Glu104 His107 Met109	Leu112 Gly113 Ile130	Ser101 Glu104 His107	Met109 Leu112	Glu104 His107	Leu112 Met 109
Ca2+-C-CaM +PEP-19	Asp93 Met109 Leu116	Thr117 Val121 Thr146	Asp93 Met109 Leu116 Thr117	Val121 Gln143 Thr146	Asp93 Met109 Leu116	Thr117 Val121 Thr146	Thr 110 Asn111 Gly 113	Glu114 Lys115

Residues exhibiting conformational exchange on the μs-ms timescale by all criteria are shown in bold.

The Model-Free approach to predict residues with R_ex was not used since 3D structures of C-CaM/PEP-19 complexes are not available, and isotropic conditions cannot be assumed. Instead, the relaxation data were further analyzed using reduced spectral density mapping. The advantage of this approach lies in the absence of assumptions made a priori regarding internal motions and molecular tumbling (25, 26). Reduced spectral density mapping assumes that the high- frequency spectral density terms are approximately equal in magnitude (i.e. J(ω_H±ω_N) ≈ J(ω_H)) and therefore may be replaced by a single value, J(ω_H). Reduced spectral density terms J(ω_H), J(ω_N) and J(0) were calculated according to Eqs. (3)–(7) (see Materials and Methods) for apo and Ca²⁺-bound C-CaM in the presence and absence of PEP-19, as summarized in Table 1 and Figure 6. J(ω_H) is correlated with fast internal motion of N-H bond vectors on the ps-ns timescale. Since the value of the area under J(ω) curve is constant, highly flexible segments of the backbone are characterized by high J(ω_H) values and low values of J(0) and J(ω_N). This is observed for the N- and C-termini, and the linker between helices F and G of C-CaM. Figure 6A and D show that PEP-19 has relatively little effect on J(ω_H) for residues in C-CaM in the presence or absence of Ca²⁺. This indicates little effect of PEP-19 on backbone dynamics of C-CaM on ps-ns timescale, and is consistent with a lack of effect of PEP-19 on heteronuclear NOEs.

Figure 6. Reduced spectral density mapping for apo C-CaM, apo C-CaM/PEP-19, Ca²⁺-bound C-CaM and Ca²⁺-bound C-CaM/PEP-19.

Reduced spectral density parameters were calculated using Eq. (3) to Eq. (7). J(ω_H), J(ω_N), and J(0) are shown for apo- and Ca²⁺-C-CaM in the absence (open circles) and presence (closed circles) of PEP-19. The mean +1.0 SD of J(0) for apo and Ca²⁺-bound C-CaM in the absence and presence of PEP-19 is indicated by dash and solid line, respectively. Asterisks denote resonances with significant line-broadening in the presence of PEP-19.

An increase in molecular weight due to complex formation will globally increase J(0), however, residues with unusually high J(0) values relative to the average in a given protein or complex indicates significant chemical/conformational exchange on the μs- ms timescale due to contributions from R_ex on R₂ (25, 26, 37). Figure 7 shows residues in C-CaM with J(0) values that are increased relative to the average J(0) based on the expression (J(0)–J(0)_avg)/SD. For example, Figure 7A shows that J(0) values are increased by more than 1 SD relative to J(0)_avg for Arg90, Phe92, Tyr99, Ile100, Thr110, Asp131, and Glu139. These same residues were shown by Malmendal et al. (15) to have the greatest R_ex based on Model-Free analysis. Interestingly, Figure 7 shows that binding PEP-19 causes large variations in J(0) for residues in C-CaM in the presence or absence of Ca²⁺. Residues in Table 2 that are highlighted in bold exhibit conformational exchange based on severe line width broadening., or J(0), R₂/R₁, and R₁R₂ values that are all greater than one SD from the average.

Figure 7. Identification of residues exhibiting conformational exchange on the μs –ms timescale in PEP-19-bound apo C-CaM and Ca²⁺-C-CaM based on reduced spectral density mapping.

Values on the Y-axis are calculated by (J(0)_i – J(0)_avg)/SD. Residues with (J(0)_i – J(0)_avg)/SD >1.0 are considered to undergo significant chemical/conformational exchange. Asterisks indicate residues with extreme line broadening.

DISCUSSION

Calmodulin is an essential and versatile Ca²⁺-sensor capable of interacting with numerous target proteins, however, its intrinsic biochemical properties become limiting in cells that have very rapid Ca²⁺ oscillations, high total Ca²⁺ levels, numerous CaM binding proteins and limiting levels of CaM. This explains the need for proteins that can act as regulators of CaM signaling, or RCS proteins, to allow CaM to respond to diverse Ca²⁺ signals and achieve its wide array of known activities. We showed that the small neuronal IQ-motif proteins PEP-19 (62 a.a.) and RC3 (78 a.a.) modulate the Ca²⁺ binding properties of CaM (12, 38). Another small neuronal protein called ARPP-21 (88 a.a.) was shown to inhibit CaM binding to target proteins (39). These activities could exert broad effects on cell activities. Indeed, ARPP-21 regulates Ca²⁺-channel activity in mammalian brain, and RC3 plays a role in learning and memory (39–45). PEP-19 is thought to have cytoprotective activity based on its pattern of expression in cells that are affected by Alzheimer’s (6) and Huntington’s diseases (7), and expression of PEP-19 inhibits apoptosis and cell death due to Ca²⁺ overload (8, 9). PEP-19 is of particular interest since it is expressed in non-neuronal tissues including endocrine, reproductive, and urinary organs that have highly active Ca²⁺ dynamics (46).

Diverse effects on essential cellular activities highlight the biological significance of PEP-19, ARPP-21 and RC3, and the need to understand their mechanisms of action. We showed previously that PEP-19 greatly increased the k_on and k_off rates for Ca²⁺ binding to the C-domain of CaM, with no change of apparent Ca²⁺ binding affinity (12), and that it decreased the cooperativity of Ca²⁺ binding (14). Changes in the cooperativity and kinetics of Ca²⁺ binding could result from static or dynamic properties of the CaM/PEP-19 complex. Dynamic properties are of particular interest since it is well-known that protein motions on the μs-ms timescale are often coupled with protein function (47–50), and the rate of Ca²⁺ exchange in the C-domain of CaM is correlated with rates of protein conformational exchange (15, 18, 51). The disordered nature of PEP-19 is significant since it could form so-called “Fuzzy” complexes (52), in which bound PEP-19 remains partially disordered to induce conformational exchange in CaM, and since disordered regions in proteins have been shown to enhance allosteric effects (53). These relationships led us to characterize the effect of PEP-19 on backbone dynamics of C-CaM, and to provide the experimental evidence for changes in cooperative Ca²⁺ binding at the atomic level.

It is well established that the C-domain of CaM binds 2 Ca²⁺ with a high degree of cooperativity (12, 31, 54–56). Figure 3 shows that this cooperativity results in apparent slow exchange of amide resonances during titration with Ca²⁺ such that cross peaks for only the 0-Ca²⁺ and 2-Ca²⁺ forms are observed at all Ca²⁺/C-CaM ratios. In contrast, intermediate exchange is observed when C-CaM is titrated with Ca²⁺ in the presence of PEP-19, resulting in severe broadening of most amide cross peaks at intermediate Ca²⁺ levels. Amides for residues in the linker region between Ca²⁺ binding sites III and IV, and near the N- and C-termini are observed at lower contour levels at all Ca²⁺/C-CaM ratios, and many of these residues, including Thr117 shown in Figure 4, exhibit a shift in the direction of the resonance migration at [Ca²⁺]/[C-CaM] > 1.0. The simplest explanation for this is that PEP-19 decreases the cooperativity of Ca²⁺ binding to C-CaM, which allows a significant population of 1-Ca²⁺ form of PEP-19/C-CaM complex to accumulate at low Ca²⁺ levels. The effects of Ca²⁺ on amide cross peaks for Gly98 in Ca²⁺ binding site III and Gly134 in Ca²⁺ binding site IV indicate that Ca²⁺ binds preferentially to site IV in the presence of PEP-19. This Ca²⁺ binding preference is generally consistent with Malmendal et al. (15) who concluded that Ca²⁺ binds preferentially to site IV of unmodified free CaM at very low Ca²⁺ levels. Decreased cooperativity may be intrinsically linked to the effects of PEP-19 on Ca²⁺ k_on and k_off rates based on a mathematical model in which binding the first Ca²⁺ ion to either site III or IV is characterized by fast rate constants, while binding the second Ca²⁺ occurs with much slower rates due to cooperative effects (14). This implies that attenuation of cooperativity by PEP-19 will increase Ca²⁺ binding rate constants by allowing greater expression of rapid rates associated with independent binding to site III or IV.

We used several criteria to identify residues in apo and Ca²⁺-bound C-CaM that experience conformational/chemical exchange (R_ex) in the presence or absence of PEP-19. Residues indicated by bold lettering in Table 2 are those that experience severe exchange broadening, or that show significant conformational exchange based on high J(0) values from reduced spectral density mapping, as well as high values for R_2/R₁ and R₁R₂. Figure 8A shows that residues with significant R_ex in apo C-CaM bound to PEP-19 are localized to Ca²⁺-binding loop IV and the short β-strands between loops III and IV. In contrast, Figure 8B shows that residues 109 to 117 in helix F and the linker between helices F and G of Ca²⁺-C-CaM show conformational exchange in the presence of PEP-19. This pattern of residues with R_ex differs from the effect of a peptide from transcription factor SEF2-1, which induces R_ex primarily in residues 142–147 at the C-terminus of Ca²⁺-CaM (57). It also differs from the effect of binding a CaMKI peptide to Ca²⁺-CaM, which causes relatively small R_ex in residues 102, 105–107 and 109 (58).

Figure 8. Summary of residues that exhibit significant conformational exchange for apo and Ca²⁺-bound C-CaM in the presence of PEP-19.

Panels (A) and (B) show the NMR solution structures for apo (pdb: 1F71) and Ca²⁺-bound C-domain of CaM (pdb: 1J7P), respectively. Colored balls indicate residues that exhibit significant conformational exchange, and are shown by bold lettering in Table 2.

The pattern of residues with R_ex, and the k_on and k_off rates for binding PEP-19 to CaM, suggest that conformational exchange is derived from different mechanisms for apo and Ca²⁺-bound C-CaM. Residues with R_ex in the apo C-CaM/PEP-19 complex are primarily localized to Ca²⁺ binding site IV, however, this does not correspond to the PEP-19 binding site, which is localized primarily to residues 108–117 based on amide chemical shift mapping shown in Figure 1. This suggests that R_ex is not due to intermolecular chemical exchange between apo C-CaM and PEP-19, but to intramolecular conformational exchange within the apo CaM/PEP-19 complex. This is consistent with k_on and k_off rates of 1 μM⁻¹s⁻¹ and 5.6 s⁻¹, respectively, for binding PEP-19 to apo CaM (13). Under conditions used for relaxation measurements, these rates give an estimated k_ex for intermolecular chemical exchange of about 200 s⁻¹, which is at the limit of the time frame that would give rise to R_ex. On the other hand, k_on and k_off rates for binding PEP-19 to Ca²⁺-CaM are least 20 μM⁻¹s⁻¹ and 400 s⁻¹, respectively (13) These rates give a k_ex of least 4,500 s⁻¹, which is well within the range that would give rise to R_ex. In addition, residues with R_ex in the Ca²⁺-C-CaM/PEP-19 complex are localized to the PEP-19 binding site. This suggests that conformational exchange induced by PEP-19 in Ca²⁺-bound C-CaM is due to intermolecular chemical exchange.

The effects of PEP-19 on the cooperativity and k_on and k_off rates of Ca²⁺ binding to CaM may involve residues that experience conformational exchange on the slow timescale. The structural basis for cooperativity Ca2+ binding is primarily associated with two structural features of paired EF hands. The first is the antiparallel β-sheet formed between short β-strands in Ca²⁺ binding sites III and IV. This structure is also formed in other paired EF-hands, and was proposed as a conduit that transmits structural changes between Ca²⁺ binding sites in calbindin D_9k (59). Residues in the β-sheet region of apo C-CaM, but not Ca²⁺-C-CaM, show increased R_ex when bound to PEP-19, including Ala102 in the strand/helix junction of Ca²⁺ binding loop III, and a cluster of residues across EF-hand IV. This supports a model in which PEP-19 increases the Ca²⁺ k_on rate by inducing conformational fluctuations in regions that lead to decreased Ca²⁺ binding cooperativity and preferential binding of Ca²⁺ to site IV. The second structural feature is the polypeptide linker between helix F of Ca²⁺-binding site III and helix G in site IV of CaM. This strong covalent coupling would transmit conformational changes when the first Ca²⁺ binds to either site III or IV (60). PEP-19 induces R_ex for residues in this region of Ca²⁺-bound, but not apo C-CaM, which may reflect coupling between protein dynamics and the Ca²⁺ k_off rate. Another potential mechanism for increasing the Ca²⁺ k_off rate involves Asp93, which has increased R_ex in Ca²⁺-bound C-CaM in the presence of PEP-19. The side chain carboxyl group of Asp93 contributes the +X ligand for coordination of Ca²⁺ at site III, and also forms a hydrogen bond with Phe89 (Asp93H_N→Phe89C′) (61) that stabilizes the otherwise unfavorably close proximity of the negatively charged oxygen atoms in the co-ordination sphere of the bound Ca²⁺. Conformational exchange at Asp93 in the presence of PEP-19 may increase the Ca²⁺ k_off rate.

In summary, the results presented here fill a gap in our understanding of how binding to proteins affects dynamic processes in CaM, by characterizing the effects of an intact, intrinsically disordered protein that binds with relatively low affinity but selectively to the C-domain of CaM in the presence or absence of Ca²⁺. Binding PEP-19 has no effect on CaM backbone dynamics on the fast ns-ps timescale, but increases conformational exchange on the slow μs-ms timescale, and patterns of residues that exhibit conformational exchange differ significantly for apo versus Ca²⁺-CaM. The data provide structural and dynamic evidence to support models in which PEP-19 decreases the degree of Ca²⁺ binding cooperativity in the C-domain of CaM and accelerates Ca²⁺ k_on and k_off rates with no change of apparent Ca²⁺ binding affinity.

Supplementary Material

1_si_001

Supplemental Table 1: Comparison of secondary structures for the C-domain of free and PEP-19 bound CaM.

Supplemental Figure 1: Binding PEP-19 to apo-CaM causes significant resonance line broadening for residues in the C-domain.

Supplemental Figure 2: Comparison of ¹H, ¹⁵N HSQC spectra of C-CaM and intact CaM in the absence and presence of Ca²⁺.

Supplemental Figure 3: Plot of amide chemical shift differences (Δδ_avg) as a function of residue number for C-CaM and the C-domain of intact CaM in the absence and presence of Ca²⁺.

Supplemental Figure 4: Comparison of ¹H, ¹⁵N HSQC spectra of Ca²⁺-bound C-CaM and intact Ca²⁺-bound CaM in the presence of excess PEP-19 at 298 K.

Supplemental Figure 5: Structural perturbations of C-CaM induced by PEP-19 binding at 298K.

Supplemental Figure 6: Identification of secondary structures for Ca²⁺-bound intact CaM/PEP-19.

Abbreviations

CaM

calmodulin

C-CaM

isolated C-domain of CaM

EDTA

2,2′,2″,2‴-(ethane-1,2-diyldinitrilo)tetraacetic acid

EGTA

ethylene glycol-bis(2-aminoethylether)-N,N,N′,N′-tetraacetic acid

HEDTA

N′-(2-hydroxyethyl)ethylenediamine-N,N,N′-triacetic acid

NTA

nitrilotriacetic acid

DSS

2,2-dimethyl-2-silapentane-5-sulphonate

3D

three-dimensional

HSQC

heteronuclear single-quantum coherence spectroscopy

R₁

longitudinal relaxation rate constant

R₂

transverse relaxation rate constant

NOE

Nuclear Overhauser effect

SD

standard deviation

Δδ _avg

average amide chemical shift change

τ_m

rotational correlation time

R_ex

chemical/conformational exchange