Time-resolved FRET Reveals the Structural Mechanism of SERCA-PLB Regulation

Abstract

We have used time-resolved fluorescence resonance energy transfer (TR-FRET) to characterize the interaction between phospholamban (PLB) and the sarcoplasmic reticulum (SR) Ca-ATPase (SERCA) under conditions that relieve SERCA inhibition. Unphosphorylated PLB inhibits SERCA in cardiac SR, but inhibition is relieved by either micromolar Ca²⁺ or PLB phosphorylation. In both cases, it has been proposed that inhibition is relieved by dissociation of the complex. To test this hypothesis, we attached fluorophores to the cytoplasmic domains of SERCA and PLB, and reconstituted them functionally in lipid bilayers. TR-FRET, which permitted simultaneous measurement of SERCA-PLB binding and structure, was measured as a function of PLB phosphorylation and [Ca²⁺]. In all cases, two structural states of the SERCA-PLB complex were resolved, probably corresponding to the previously described T and R structural states of the PLB cytoplasmic domain. Phosphorylation of PLB at S16 completely relieved inhibition, partially dissociated the SERCA-PLB complex, and shifted the T/R equilibrium within the bound complex toward the R state. Since the PLB concentration in cardiac SR is at least 10 times that in our FRET measurements, we calculate that most of SERCA contains bound phosphorylated PLB in cardiac SR, even after complete phosphorylation. 4 μM Ca²⁺ completely relieved inhibition but did not induce a detectable change in SERCA-PLB binding or cytoplasmic domain structure, suggesting a mechanism involving structural changes in SERCA’s transmembrane domain. We conclude that Ca²⁺ and PLB phosphorylation relieve SERCA-PLB inhibition by distinct mechanisms, but both are achieved primarily by structural changes within the SERCA-PLB complex, not by dissociation of that complex.

1. Introduction

A hallmark of heart failure is dysregulation of intracellular Ca²⁺ handling [1], primarily caused by inadequate removal of Ca²⁺ from the cytosol by the sarco(endo)plasmic reticulum Ca²⁺-ATPase (SERCA) [2], which actively sequesters Ca²⁺ back into the sarcoplasmic reticulum (SR) at a ratio of 2 Ca²⁺ per ATP hydrolyzed [3]. In the heart, SERCA is regulated by phospholamban (PLB), a single-pass integral membrane protein. Unphosphorylated PLB inhibits SERCA by decreasing its apparent Ca²⁺ affinity [4]. Under physiological conditions, this inhibition is relieved by either micromolar Ca²⁺ or by phosphorylation of PLB, primarily by protein kinase A (PKA) at S16 [5]. Recent gene therapies targeting increased SERCA activity show promise for alleviation of heart failure [6]. Ablation of PLB or introduction of phosphomimetic mutant S16E-PLB suppress progression of heart failure in animal models [7,8,9]. Overexpression of SERCA2a, the cardiac isoform, has successfully completed Phase IIa clinical trials [10]. These studies validate the SERCA-PLB complex as a potent therapeutic target for heart failure. However, rational design of improved therapies is hampered by uncertainty regarding the mechanism by which PLB regulates SERCA.

Two structural mechanisms for relief of SERCA inhibition have been proposed. The dissociation model hypothesizes that PLB dissociates from SERCA to relieve inhibition. This model is supported by crosslinking and co-immunoprecipitation studies that found decreased physical interaction between SERCA and PLB at either micromolar Ca²⁺ or after phosphorylation at S16 by PKA [11,12,13,14,15,16]. However, recent spectroscopic studies support the subunit model, in which inhibition of SERCA is relieved by structural rearrangements within the SERCA-PLB complex, not by dissociation. Electron paramagnetic resonance (EPR) and fluorescence resonance energy transfer (FRET) studies showed that PLB remains bound to SERCA when the pump is activated [17,18,19,20]. EPR and NMR studies showed that the cytoplasmic domain of PLB exists in equilibrium between a T state that is ordered and an R state that is dynamically disordered [21,22,23]. Phosphorylation shifts the equilibrium toward the R state and relieves inhibition [24]. FRET studies showed that variation of lipid headgroup charge shows a strong correlation between the population of the R state and SERCA-PLB activation, without dissociation, further validating the subunit model [25]. That study showed the power of time-resolved (TR) FRET to distinguish between changes in structure and association. In the present study, we have used TR-FRET, using fluorophore-labeled SERCA and PLB reconstituted in lipid bilayers, to resolve the effects of both micromolar Ca²⁺ and PLB phosphorylation on the structure and stability of the SERCA-PLB complex. These results provide definitive insights into the molecular mechanisms underlying relief of inhibition in cardiac SR.

2. Materials and methods

2.1 SERCA purification and labeling

Crude SR vesicles were prepared from the fast-twitch skeletal muscle of New Zealand white rabbits [26]. SERCA was further purified from crude SR vesicles using reactive-red chromatography [27]. For FRET studies, purified SERCA was labeled with 5-iodoacetamidofluorescein (IAF) (Invitrogen, CA) specifically and completely at C674 [28].

2.2 Expression, purification, phosphorylation and labeling of PLB

Native PLB equilibrates between monomers and homopentamers [29]. To simplify the analysis and focus on the SERCA-PLB interaction, a monomeric mutant of PLB was used, with the three cysteine residues (C36, C41 and C46) in the transmembrane domain mutated to alanine, phenylalanine, and alanine, respectively [30]. Site-directed mutagenesis was performed to mutate Y6 to C for thiol-reactive fluorophore attachment. This site was chosen because Y6 is not involved in the interaction with SERCA [31]. Recombinant PLB was expressed in E. coli and purified as previously published [32]. For site-directed fluorophore labeling, lyophilized PLB powder was dissolved at a concentration of 0.2 mM in 20 mM MOPS, 1% octyl β-D-glucopyranoside (OG), pH 7.0. Alexa Fluor® 350 C₅ maleimide (Invitrogen, CA) freshly dissolved in DMSO was then added at 10-fold molar excess. The reaction was allowed to proceed at room temperature for 1 hour, and the labeled PLB was purified by reversed-phase HPLC. For phosphorylation studies, labeled PLB was phosphorylated as described previously [29] and purified by reversed-phase HPLC. Complete labeling and phosphorylation of PLB was confirmed by ESI-MS. The concentration of PLB was measured by the BCA assay.

2.3 Co-reconstitution of SERCA and PLB

SERCA and PLB were co-reconstituted into lipid vesicles using 4:1 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC)/1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) (mol/mol) to yield unilamellar vesicles [33]. Ca²⁺-ATPase activity and FRET measurements were performed immediately after co-reconstitution. The final buffer composition of FRET samples was 50 mM MOPS, 50 mM KCl, 5 mM MgCl₂, 1 mM EGTA, pH 7.0, with varying concentrations of CaCl₂ to yield the desired free Ca²⁺ concentration (pCa 8.0, 6.4, or 5.4).

2.4 Ca²⁺-ATPase functional measurements

To ensure that labeled SERCA and PLB remain functional, we measured the ATPase activity of co-reconstituted samples at 25 °C, using an NADH-linked enzyme-coupled assay [33,34]. The Ca²⁺-dependent rate of ATP consumption was fitted to the Hill equation,

$$\mathrm{V}={\mathrm{V}}_{\max }∕[1+{10}^{-\mathrm{n}(\mathrm{pKCa}-\mathrm{pCa})}],$$ Eq 1

where pCa = −log₁₀ [Ca²⁺], V_max is the limiting activity at saturating calcium, pK_Ca is the pCa value where V = 0.5V_max, and n is the Hill coefficient. The inhibition of SERCA by PLB is shown as ΔpK_Ca, the shift of pK_Ca upon addition of PLB.

2.5 Time-resolved fluorescence resonance energy transfer (TR-FRET) measurements

SERCA and PLB were labeled with fluorophores at the sites shown in Fig. 1A. PLB was labeled with Alexa Fluor 350 maleimide (donor) at Y6C, and SERCA was labeled with IAF (acceptor) at C674. The quantum yield of bound Alexa Fluor 350 maleimide was measured in 20 mM MOPS, 1% OG, pH 7.0, using quinine sulfate dehydrate (AnaSpec, CA) as the standard, yielding a quantum yield of 0.48 for PLB, 0.80 for phosphorylated PLB (pPLB). The corresponding R₀ values [35] are calculated to be 4.6 nm and 5.0 nm, respectively. The time-resolved fluorescence decay of co-reconstituted samples was measured by time-correlated single-photon counting (Becker-Hickl, Berlin, Germany), following excitation at 385 nm using a subnanosecond pulsed diode laser (PicoQuant, Berlin, Germany), filtering the emitted light using a 440/40 filter (Semrock, NY), and detection with a PMH-100 photomultiplier (Becker-Hickl). The instrument response function (IRF, Fig. 1B) was recorded from water. TR-FRET data was analyzed as previously published [25]. The observed donor-only waveform (Fig. 1B) was fitted by a simulation F_Dsim(t), consisting of a multi-exponential decay F_D(t) convoluted with the IRF

$$\begin{array}{cc}\hfill F_D\left(t\right)=& \sum _{i=1}^n{A}_i\exp (t∕{\tau }_{Di})\hfill \\ \hfill F_{Dsim}\left(t\right)=& {\int }_{-\infty }^{+\infty }IRF(t-t^{\prime })F_D\left(t^{\prime }\right)d{t}^{\prime }\hfill \end{array}$$ Eq 2

where τ_Di is the fluorescence lifetime of the ith component. In this study, two exponentials were sufficient to fit the donor-only waveform (Fig. S1). The waveform of the samples containing donor and acceptor (Fig. 1B), was fitted by F_D+A,sim(t):

$$\begin{array}{cc}\hfill F_{D+A}\left(t\right)=& (1-x_{DA})F_D\left(t\right)+x_{DA}\sum _{i=1}^m{F}_{DA,j}\left(t\right)\hfill \\ \hfill F_{D+A,sim}\left(t\right)=& {\int }_{-\infty }^{\infty }IRF(t-t^{\prime })F_{D+A}\left(t^{\prime }\right)d{t}^{\prime },\hfill \end{array}$$ Eq 3

where x_DA is the mole fraction of donor-labeled PLB that is bound to acceptor-labeled SERCA, and F_DA,j(t) is the fluorescence waveform of the jth donor-acceptor complex. We found that m=2 was sufficient (Fig. S2).

$$F_{DA}\left(t\right)={\int }_{-\infty }^{+\infty }\rho \left(R\right)\sum _{i=1}^n{A}_i\exp \left\{\right(-t∕{\tau }_{Di}\left)\right[1+{(R_{oi}∕R)}^6\left]\right\}dR$$ Eq 4

where R_oi is the Förster distance of Alexa Fluor 350 and IAF, and ρ(R) is the distance distribution function, assumed to be sum of Gaussians, each with a mole fraction x_j, a center r_j and a full width at half maximum (FWHM_j) (Fig. 1C):

$$\begin{array}{cc}\hfill \rho \left(R\right)=& \sum _{j=1}^2{x}_j\left(\frac{1}{{\sigma }_j\sqrt{2\pi }}\right)\exp \left[-\frac{{(R-r_j)}^2}{2{\sigma }_j^2}\right]\hfill \\ \hfill \text{where}& {\sigma }_j=\frac{FWH{M}_j}{2\sqrt{2\ln 2}}\hfill \end{array}$$ Eq 5

Fig. 1.

TR-FRET experimental design. A Models of SERCA-PLB complex. SERCA (silver)-bound PLB (blue) is proposed to exist in an equilibrium between T and R states, as shown previously by EPR [24]. There is no high-resolution structure of the SERCA-PLB complex, but a model for the R state (right) is from [36], and the T state model (left) is based on the NMR structure of PLB. The labeling sites on SERCA (C674) and PLB (Y6C) are shown as green and magenta spheres, respectively. The interprobe distances in T and R states are denoted as r_T and r_R, respectively. Image created in VMD [45]. B Representative fluorescence waveforms. D. donor only. D+A, donor plus acceptor. Grey indicates instrument response function. C Interprobe distance distributions resolved from fluorescence waveforms in B. Two Gaussian distance distributions were resolved. The shorter and longer distances correspond to R and T states, respectively [25].

We assumed that PLB and pPLB populate the same two states, because the distance parameters resolved from independent fittings are similar for PLB and pPLB (Fig. S2) and this assumption does not change the goodness of the fit (Fig. S3). Our previous FRET studies showed that the shorter distance observed between probes on the SERCA and PLB cytoplasmic domains is the R state (correlating with decreased inhibition), while the longer distance is due to the T state, in which the PLB cytoplasmic domain interacts with membrane surface and hence increases interprobe distance [25]. Therefore, we refer the parameters of the shorter distance with a subscript R, and those of the longer distance with a subscript T (Fig. 1).

2.6 Statistical Analysis

Data is presented as the mean ± SE. For comparison between two groups, a student’s t-test was performed. A p-value less than 0.05 was taken as significant.

3. Results

3.1. Function of labeled SERCA and PLB

To confirm that the labeled proteins preserve their functions, we co-reconstituted SERCA and PLB into lipid vesicles and measured the Ca²⁺-dependent ATPase activity. The function of unlabeled PLB is characterized by inhibition of SERCA (decreasing pK_Ca) and reversal after phosphorylation at S16 by PKA (Fig. 2). For unlabeled SERCA, pK_Ca = 6.43 ± 0.02, unlabeled PLB shifted pK_Ca to 6.32 ± 0.003, and unlabeled pPLB restored pK_Ca to 6.39 ± 0.02. Labeled proteins demonstrated similar results (Fig. 2). The pK_Ca of acceptor-labeled SERCA is 6.45 ± 0.02, donor-labeled PLB shifted it to 6.32 ± 0.003, and phosphorylated donor-labeled PLB shifted it back to 6.45 ± 0.01. Therefore, the fluorophore-labeled SERCA and PLB retained their functions, and these samples should provide reliable information of the conformational rearrangement induced by phosphorylation of S16 by PKA.

Fig. 2.

Functional effect of PLB phosphorylation (as measured by ΔpK_Ca) is not affected by FRET labeling of SERCA and PLB. The molar ration of Lipid:PLB:SERCA is 700:5:1. n ≥ 3.

3.2 TR-FRET within the SERCA-PLB complex

Time-resolved fluorescence of donor-labeled PLB was measured in the presence of either unlabeled or acceptor-labeled SERCA with a PLB to SERCA molar ratio of 0.2. The fluorescence waveforms were then fitted by the equations described in Materials and Methods. TR-FRET data was fitted to two Gaussian distance distributions [25], assuming that PLB and pPLB populate the same two structural states (Fig. S2 and S3). The shorter distance, r_R, is 3.52 ± 0.42 nm and the longer distance, r_T, is 8.27 ± 0.27 nm (Fig. 3A&B). The distance between the labeling sites predicted by the SERCA-PLB model is 2.69 nm (Fig. 1A right) [36]. Crosslinking showed previously that K3 of PLB was within 1.5 nm of SERCA K400 [11]. The distance between C674 and K400 predicted by a crystal structure in the absence of Ca is 3.48 nm (3B9R) [37], consistent with our measurement for r_R. The fraction of pPLB that binds to SERCA (x_DA) decreased significantly from that of PLB, from 0.99 ± 0.01 to 0.50 ± 0.06 (Fig. 3C), indicating that the SERCA-PLB complex partially dissociates after PLB phosphorylation. Within the bound population, the fraction of the R state, x_R, is substantially increased after PLB phosphorylation, from 0.33 ± 0.04 to 0.68 ± 0.04 (Fig. 3A&C), indicating a substantial shift of the T/R equilibrium toward the R state.

Fig. 3.

TR-FRET shows that phosphorylation of PLB partially dissociates the SERCA-PLB complex and shifts the population to the shorter distance distribution. A: Two Gaussian distance distributions (two structural states of SERCA-PLB, R and T) were needed to fit the TR-FRET data. Red: PLB. Blue: pPLB. B: Centers (r) and widths (FWHM) of distance distributions. C: Compared to PLB, phosphorylated PLB decreased the mol fraction of SERCA-bound PLB (x_DA), but did not abolish it. D: The mol fraction of the short distance component (x_R) increased after phosphorylation. SERCA: PLB = 5. Lipid : SERCA = 200. n ≥ 3. * p<0.05, ** p<0.01, ***p<0.001.

Interprobe distances were measured at three Ca²⁺ concentrations, and no significant effects of Ca²⁺ were observed in binding or structure (Fig. 3, Table S1). These results show clearly that PLB remains bound to SERCA after the E2 (Ca-free)) to E1 (Ca-bound) transition. This result is consistent with previous FRET findings that the SERCA-PLB complex does not dissociate at physiological Ca²⁺ concentrations [17,19].

4. Discussion

We have performed TR-FRET on fluorophore-labeled SERCA and PLB co-reconstituted into lipid vesicles, to probe the structural mechanism for relief of inhibition by PLB phosphorylation and by micromolar Ca²⁺. The functions of the proteins were preserved after covalent modification with fluorophores (Fig. 2). TR-FRET analysis showed that phosphorylation of PLB partially dissociates the SERCA-PLB complex and shifts the SERCA-bound PLB toward the less inhibitory R state. Ca²⁺ does not dissociate the complex or perturb the T-R equilibrium (Fig. 3).

The mechanism of relief of inhibition due to PLB phosphorylation has been controversial. Phosphorylation of PLB has been found to perturb the physical interaction between SERCA and PLB, as detected by decreases in crosslinking [14] and co-immunoprecipitation [15]. One study in living cells, using fluorescent fusion proteins, reported that phosphomimetic PLB mutations decreased FRET between SERCA and PLB [38], while another study reported increased FRET upon PLB phosphorylation [20]. Due to the steady-state nature of these FRET measurements, it is not clear whether changes in FRET were due to changes in SERCA-PLB association or to changes in the donor-acceptor distance within the complex. In the present study, this ambiguity was removed by the detection of time-resolved fluorescence, which simultaneously and independently resolves changes in binding and structure [25]. TR-FRET showed that 40-50% of PLB remains bound to SERCA after phosphorylation (Fig. 3C), even though phosphorylation completely relieves SERCA inhibition (Fig. 2). Therefore, the activity of the SERCA-pPLB complex is comparable to that of SERCA alone, and the relief of inhibition in the bound complex must be due to a structural change within the SERCA-pPLB complex.

To assess the physiological significance of the two observed effects – dissociation and structural change, it is important to compare the composition of our reconstituted samples with those in cardiac SR. In the present study with a fluorescent donor on PLB, acceptor-labeled SERCA was in 5-fold excess over PLB, to maximize the precision of FRET measurements. However, in cardiac SR, as in our functional measurements (Fig. 2), PLB is in large excess (3-5 fold) over SERCA [39]. Based on the affinity of SERCA for pPLB indicated from Fig. 3, and on the known concentrations of SERCA and PLB in cardiac SR [39,40], 76 ± 2% of SERCA in cardiac SR contains bound pPLB, even after complete phosphorylation of PLB (Table S2). This is consistent with EPR and NMR findings that pPLB remains bound to SERCA [18,23], with allosteric relief of inhibition achieved by shifting the T/R equilibrium toward the R state [23,24]. The present study clearly supports this “subunit” model for SERCA regulation (Fig. 4, left) over the dissociation model.

Fig. 4.

Two mechanisms to relieve SERCA inhibition by PLB in cardiac SR. Phosphorylation shifts the PLB cytoplasmic domain toward the dynamically disordered R state. Micromolar Ca²⁺ induces a structural change in the SERCA transmembrane domain.

Although both dissociation and subunit models have been proposed to explain the mechanism of relief of inhibition by micromolar Ca²⁺ [12,13,16,17,19], our TR-FRET data showed no Ca-dependent changes in either binding or structure (Fig. 3B, C, D), clearly ruling out the subunit model, but also showing clearly that Ca²⁺ relieves SERCA-PLB inhibition by a different mechanism from that of phosphorylation. This result is consistent with previous steady-state FRET studies showing little or no dissociation of PLB from SERCA due to Ca²⁺ [17,19]. Ca²⁺ is known to induce structural changes in its activation of SERCA [41,42], but these changes apparently do not perturb the distance between our labeling sites in the cytoplasmic domains of SERCA and PLB. This conclusion is consistent with structural models, since the distance between SERCA C674 (our acceptor labeling site) and K400 (proposed to lie in the binding groove of the PLB cytoplasmic domain, where our donor is located), changes only from 3.48 nm in the Ca-free crystal structure to 3.41 nm in the Ca-bound structure [37,43]. Since SERCA inhibition by PLB is caused by interaction of the transmembrane domains of the two proteins [44], we propose that Ca-induced structural changes in the SERCA transmembrane domain disrupt this inhibitory interaction, without changing interactions of the cytoplasmic domains, and without producing a substantial decrease in SERCA-PLB affinity (Fig. 4, right). This rearrangement of transmembrane helices would explain the Ca-induced decrease in crosslinking between the transmembrane domains of SERCA and PLB [12,13,16].

In summary, we conclude that phosphorylation of PLB and Ca²⁺ relieve SERCA inhibition through different mechanisms (Fig. 4), neither of which relies primarily on dissociation of the SERCA-PLB complex. PLB phosphorylation acts by promoting the dynamic disorder of the PLB cytoplasmic domain, while Ca²⁺ acts by changing the structure of the transmembrane domain of SERCA. Both mechanisms must disrupt the inhibitory interactions between the two transmembrane domains, by distinct mechanisms that remain to be elucidated (Fig. 4).

HIGHLIGHTS.

• What is the mechanism by which PLB inhibition of SERCA is relieved in the heart?

• Time-resolved FRET was measured between PLB and SERCA cytoplasmic domains.

• Binding and structure were simultaneously resolved.

• Both PLB phosphorylation and Ca²⁺ relieve SERCA inhibition without dissociation.

• PLB and Ca²⁺ induce distinct structural changes in the SERCA-PLB complex.

Abbreviations

IAF

5-iodoacetamidofluorescein

DOPC

1,2-dioleoyl-sn-glycero-3-phosphocholine

DOPE

1,2-dioleoyl-sn-glycero-3-phosphoethanolamine

EGTA

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

ESI-MS

electrospray ionization mass spectroscopy

FRET

fluorescence resonance energy transfer

TR-FRET

time-resolved FRET

MOPS

3-(N-morpholino)propanesulfonic acid

pCa

−log₁₀ [Ca²⁺ ]

pK_Ca, −log(K_Ca)

calcium concentration at half-maximal ATPase activity

PLB

phospholamban

SERCA

Sarco(endo)plasmic reticulum Ca²⁺-ATPase

SR

sarcoplasmic reticulum

OG

octyl β-D-glucopyranoside

PKA

protein kinase A