Effects of the Arg9Cys and Arg25Cys Mutations on Phospholamban’s Conformational Equilibrium in Membrane Bilayers

Abstract

Approximately, 70% of the Ca²⁺ ion transport into the sarcoplasmic reticulum is catalyzed by the sarcoplasmic reticulum Ca²⁺-ATPase (SERCA), whose activity is endogenously regulated by phospholamban (PLN). PLN comprises a TM inhibitory region and a cytoplasmic regulatory region that harbors a consensus sequence for cAMP-dependent protein kinase (PKA). The inhibitory region binds the ATPase, reducing its apparent Ca²⁺ binding affinity. β-adrenergic stimulation activates PKA, which phosphorylates PLN at Ser 16, reversing its inhibitory function. Mutations and post-translational modifications of PLN may lead to dilated cardiomyopathy (DCM) and heart failure. PLN’s cytoplasmic region interconverts between a membrane-associated T state and a membrane-detached R state. The importance of these structural transitions on SERCA regulation is emerging, but the effects of natural occurring mutations and their relevance to the progression of heart disease are unclear. Here we use solid-state NMR spectroscopy to investigate the structural dynamics of two lethal PLN mutations, R9C and R25C, which lead to DCM. We found that the R25C mutant enhances the dynamics of PLN and shifts the conformational equilibrium toward the R state confirmation, whereas the R9C mutant drives the amphipathic cytoplasmic domain toward the membrane-associate state, enriching the T state population. The changes in membrane interactions caused by these mutations may explain the aberrant regulation of SERCA.

Graphical Abstract

INTRODUCTION

Phospholamban (PLN) is a central regulator of cardiac contractility[1]. This 52-amino acid single-pass membrane protein forms a complex with the sarcoplasmic reticulum Ca²⁺-ATPase (SERCA), handling 70–90% of the cytoplasmic Ca²⁺ ions[2]. In fact, the SERCA/PLN complex is responsible for Ca²⁺ reuptake into the sarcoplasmic reticulum (SR), regulating the diastolic phase of heart muscle[2–4]. When unphosphorylated, PLN binds SERCA via intramembrane interactions and reduces its apparent Ca²⁺ affinity[5]. Upon β-adrenergic stimulation, cAMP-dependent protein kinase A phosphorylates PLN at Ser16, reversing its inhibitory effects and augmenting cardiac contractility[2]. Protein phosphatase 1 dephosphorylates PLN, reestablishing its basal inhibitory activity[6].

Functional and sequence analyses show that PLN structure is divided into two regions: a cytoplasmic (regulatory) region consisting of domain Ia (residues 1–16), a loop region (residues 17–22), and domain Ib (residues 23–30), and a transmembrane (TM) inhibitory region that includes domain II (residues 31–52)[6]. In cell and reconstituted membrane systems, PLN exists in equilibrium between a monomeric and an oligomeric state[7, 8]. The homopentamer is arranged in a pinwheel topology[7, 9]. Although initial studies suggested that the pentamer functions as a Ca²⁺ or Cl⁻ channel, more recent studies revealed that oligomerization is a mechanism for storage of active PLN monomers, playing a key role in SERCA regulation [10, 11]. Structural studies in lipid membranes show that upon de-oligomerization, PLN’s helical transmembrane (TM) domain changes its tilt angle with respect to the lipid bilayer by approximately 5 degrees with a slight rotation of the azimuthal angle[7, 12]. The amphipathic cytoplasmic region, on the other hand, samples a range of conformations, including two major states: a T state, more populated, helical, and membrane associated, and an R state, more dynamic, membrane detached, and unfolded [13–15]. By correlating biophysical data with functional assays and site-directed muta-genesis, we discovered that the T-to-R equilibrium regulates the extent of SERCA inhibition by PLN[16]. In fact, by shifting the conformational equilibrium toward the R state, we were able to generate PLN loss-of-function (LOF) mutants that mimic the effects of Ser16 phosphorylation by PKA [17, 18]. When bound to SERCA, the T-to-R equilibrium is further shifted towards a SERCA-bound (B) state [7, 12, 19]. While the T and R states are inhibitory, the B state is non-inhibitory. Functional assays carried out on a truncated version of PLN lacking the cytoplasmic region, support this regulatory model[14].

In the past years, several pathological mutations or post-translational modifications have been identified in patients diagnosed with familial dilated cardiomyopathy (DCM). The first human mutation identified was R9C (PLN^R9C), located in domain Ia of the regulatory region[20]. This autosomal dominant, missense mutation at nucleotide 25 causes the typical DCM phenotype, with a severe dilation of the left ventricle and decreased contractile function of the heart muscle[21]. The later sequencing of R9H and R9L mutations in DCM patients further emphasized the importance of PLN’s position 9 for SERCA regulation [22]. More recently, Kranias and co-workers identified another mutation of PLN at Arginine 25 (PLN^R25C). As for the R9C, the arginine 25 was mutated into a cysteine with cardiotoxic effects associated with SERCA super-inhibition and arrhythmia[23]. To date, the molecular mechanisms for dysregulation of Ca²⁺ transport by these mutants are still unclear.

In an effort to elucidate the structural basis of the PLN^R9C and PLN^R25C cardiotoxicity, we investigated the effects of these single-site mutations on the conformational equilibrium of PLN in lipid bilayers using solid-state NMR techniques. We found that these mutants affect substantially the conformational equilibrium of PLN. Specifically, the R9C mutation shifts PLN’s conformational equilibrium toward the membrane associated T state, supporting earlier conclusions that increased hydrophobicity at position 9 of domain Ia can be used as a predictor for the development and progression of DCM[22]. In contrast, the R25C mutant enhances the R state population, shifting the equilibrium of the cytoplasmic region of PLN toward more dynamic ensemble. The structural and topological changes of these mutants in the membrane are interpreted in the light of our proposed regulatory model and represent a first step to explain the different manifestations in DCM patients carrying these two mutations.

MATERIAL AND METHODS

Protein expression and purification.

The R9C and R25C mutation was cloned onto the pMal c2E PLN (monomeric C36A, C41F, and C46A mutant) and wt-PLN (wild-type) background using the QuikChange II mutagenesis kit (Agilent). Uniformly ¹³C/¹⁵N labeled PLN^AFA, PLN^R9C, and PLN^R25C were expressed in BL21(DE3) E. coli [24]. The only modification with PLN^R9C and PLN^R25C was that expression was limited to four hours. Purification followed the previously described protocol for maltose binding protein (MBP) tagged PLN, except that the cleaved PLN^R9C in SDS was incubated at 55°C during HPLC purification with a C4 column (Agilent) while PLN^R25C was incubated at 45°C. PLN^R9C and PLN^R25C were also purified in the presence of TCEP-HCl (pH 7.3) to prevent oxidation.

Solution NMR in isotropic bicelles.

Samples for solution NMR experiments were prepared in isotropic bicelles with a q ratio of 0.33 using DHPC (1,2-dihexanoyl-sn-glycero-3-phosphocholine) and DMPC (1,2-dimyristoyl-sn-glycero-3-phosphocholine) purchased from Avanti Polar Lipids. Chloroform stocks for DHPC and DMPC were aliquoted into glass tubes and dried down under nitrogen. The dried film was reconstituted in water, flash frozen in liquid nitrogen and lyophilized overnight. The lyophilized lipids were reconstituted in buffer containing 20 mM HEPES (pH 7.0), 2.5% glycerol, 100 mM KCl, 1 mM MgCl₂, and 0.02% NaN₃. The samples of PLN^R9C and PLN^R25C also contained 10 mM TCEP. Lyophilized PLN was solubilized in the reconstituted DHPC and then added to the DMPC to form the bicelle. All the spectra in iso-tropic bicelles were acquired at the Minnesota NMR center using an Agilent 600 MHz spectrometer operating a 5mm triple resonance cryoprobe at 25 °C.

Magic Angle Spinning (MAS) ssNMR.

MAS samples of PLN, PLN^R25C and PLN^R9C were prepared in ²H DMPC (1,2-dimyristoyl-d54-sn-glycero-3-phosphocholine, Avanti Polar Lipids®) at a protein to lipid ratio of approximately 1:100 following previously described protocols [13]. Briefly, ²H DMPC suspended in chloroform was dried down under nitrogen flux. The dried lipids were reconstituted in a buffer containing 20 mM HEPES (4-(2-Hydroxyethyl)piperazine-1-ethanesulfonic acid, N-(2-Hydroxyethyl)piperazine-N′-(2-ethanesulfonic acid)), 50 mM KCl, 2.5 mM MgCl₂, and 1 mM NaN₃ at pH 7. Samples containing PLN^R9C and PLN ^R25C were prepared in the presence of 10 mM TCEP (Tris(2-carboxyethyl)phosphine hydrochloride). Lyophilized uniformly labeled PLN variants (~2 mg) were reconstituted in 1% C₁₂E₈, prior to mixing with the ²H DMPC. The reconstituted sample was incubated with Biobeads® to remove the C₁₂E₈ (60 mg Biobeads®/mg C₁₂E₈). The hydrated membrane protein pellet, obtained by a series of centrifugation steps, was packed into a 3.2 mm MAS rotor. The hydration level was approximately 120% as calculated according to Saitô and Naito [25] and Zhang et al. [26].

All of the solid-state NMR experiments were acquired at the Minnesota NMR center using an Agilent ssNMR spectrometer operating at a ¹H Larmor frequency of 700 MHz and equipped with 3.2 mm MAS probe with reduced RF heating technology. The temperature was held constant for all of the experiments at 25 °C. MAS rate of 12 kHz was used in all the experiments with a recycle delay of 3 seconds and 50 kHz spectral width in the t₂ dimension. The 90° pulse lengths for ¹H, ¹³C, and ¹⁵N were set to 2.5, 7, and 7 μs, respectively. The direct dimension (t₂) acquisition time was set to 100 ms for INEPT-HETCOR and riHSDQC experiments, and 20 ms for CP and DARR experiments. The ¹H detected riHSQC 2D spectra were acquired with 80 t₁ increments with 5 kHz t₁ spectral width. Whereas ¹³C detected 2D DARR spectra were acquired with 256 t₁ increments with 33 kHz t₁ spectral width using 100 ms DARR mixing period. The Hartmann–Hahn (HH) contact time for ¹H-¹³C CP was set to 1 ms, during t₁ and t₂ periods TPPM decoupling was applied on ¹H with 100 kHz RF amplitude. The 1D CP and INEPT-HETCOR ¹⁵N spectra of Figure 1A were acquired using 2k to 10k scans. All the samples were acquired with identical experimental parameters. The relative R-state populations (Figure 1B) were obtained by measuring the integrated intensities of CP and INEPT-HETCOR spectra between 100 and 140 ppm at the same noise level, where the CP intensity was normalized to 100 [27]. The ¹H detected ¹⁵N HSQC spectra of Figure 2A were obtained from riHSQC pulse sequence [27]. Water suppression was obtained from phase switched spin-lock pulses with phases x and y with RF amplitude of 30 kHz and duration (τ₁) set to 200 to 250 ms. The τ value was set to 5.4 ms (1/2 J_NH), and during t₂ acquisition ¹⁵N heteronuclear decoupling was obtained using the WALTZ-16 sequence with the RF amplitude set to 10 kHz. All the spectra were processed using NMRPipe and analyzed using Sparky [28, 29].

Figure 1:

A. Primary sequence of PLN^AFA (monomeric PLN mutant). B. ¹⁵N spectral mapping of T- and R-states of PLN^AFA, PLN^R9C, and PLN^R25C using CP and INEPT ssNMR experiments respectively. B. Relative sensitivity of R state confirmation obtained from integrated intensities for PLN^AFA, PLN^R9C, and PLN^R25C are respectively given by 3.10 (±0.02), 0.34 (±0.01), and 20.40 % (±0.04), where the errors are calculated from the standard deviations.

Figure 2:

¹⁵N-HSQC spectra of PLN^AFA, PLN^R25C, and PLN^R9C reconstituted in lipid vesicles (A), and isotropic lipid bicelles (B). The spectra in (A) and (B) were obtained from solid-state, and solution-NMR techniques, respectively.

RESULTS

Unless specified, the NMR experiments were carried out on the monomeric state of PLN, where the three transmembrane (TM) cysteines are mutated into Ala, Phe, and Ala residues to disrupt the pentameric assemblies [30–32]. This mutant (PLN^AFA) has identical activity to wt-PLN and eliminates the monomer-pentamer equilibrium, simplifying our spectroscopic analysis and data interpretation. To analyze the changes in the structural dynamics of PLN’s cytoplasmic domain, we recorded both cross-polarization (CP) and insensitive nuclei enhanceed by polarization transfer (INEPT) based solid-state NMR experiments on PLN and its R9C and R25C mutants. Figure 1A shows the ¹⁵N signatures of the backbone amides for PLN^AFA, PLN^R9C, and PLN^R25C reconstituted in zwitterionic DMPC lipid bilayers. While the CP-based experiment detects the most rigid residues of membrane proteins, the INEPT-based HETCOR experiment images the more dynamic residues. In fact, for the INEPT experiments, we generally use relatively long J-coupling evolution periods (~10 ms), which select for mobile residues with long transverse relaxation time (T₂)[33]. As a result, the CP spectra map both TM and membrane-associated residues; whereas the INEPT-HETCOR spectra image residues corresponding to the R state undergoing fast motions. Although it is difficult to estimate the absolute R-state population, we utilized the relative integrated intensities from the INEPT-HETCOR peaks with respect to the corresponding CP experiments to obtain a semi-quantitative estimate of the R state population for the three different PLN variants. By comparing the histogram of the relative populations (Figure 1B), it is possible to appreciate a significant difference in the relative R state for PLN^R9C and PLN^R25C mutants with respect to PLN^AFA. In the R9C mutant, the population of the R state is significantly reduced relative to that of PLN^AFA (approximately one tenth). In contrast, the R state population in the R25C mutant is approximately 7 times higher than that of PLN^AFA. To further validate the shift in populations for these mutants, we carried out 2D riHSQC MAS ssNMR experiments in vesicles. Figure 2A shows the HSQC spectra of PLN^AFA, PLN^R9C, and PLN^R25C. While the fingerprints for both PLN^AFA and PLN^R9C show significantly broader peaks indicative of a higher degree of conformational heterogeneity, the R25C resonances are notably narrower with higher relative peak intensities. These data support the conformational shift toward the R state caused by the R25C mutation in domain Ib, and at the same time, reveal its highly dynamic nature for the cytoplasmic region. Similar HSQC patterns were observed for PLN^AFA and PLN^R25C reconstituted in isotropic lipid bicelles using solution NMR (Figure 2B). Due to the longer correlation times of isotropic bicelles, the more rigid TM resonances are either weak or undetectably broad; whereas the peaks corresponding to the dynamic cytoplasmic region display higher intensities. The favorable relaxation conditions in isotropic bicelles made it possible to assign the majority of these peaks, which correspond to the 1–20 region of PLN^AFA, encompassing domain Ia and the dynamic loop. The heteronuclear correlated spectra are more resolved in the isotropic bicelles than in the vesicles, suggesting that the dynamics of PLN^AFA in different membrane-mimetic systems occur on different time scales.

To analyze the secondary structure adopted by the cytoplasmic domains in vesicles, we carried out ¹³C-¹³C TOBSY and DARR experiments on all three PLN variants. The TOBSY experiment correlates aliphatic intra-residue carbon peaks of dynamic regions, transferring the polarization through ¹³C-¹³C resonances via J-couplings; whereas the DARR experiment utilizes dipolar coupling based polarization transfer that is sensitive to immobile transmembrane and membrane attached cytoplasmic residues. As expected from the analysis of the ¹⁵N-¹H HSQC spectra, the TOBSY spectra for both PLN^AFA and AFA^R9C display only a few correlations, due to the restricted motions of the cytoplasmic domain, i.e., lower R state population. In contrast, the TOBSY spectrum of the R25C mutants shows a significantly higher number of correlations that enabled us to assign 16 out of the 22 residues encompassing the cytoplasmic region. Note that the ¹³C resonances in this spectrum have chemical shifts identical to those in the solution NMR experiments carried out in isotropic bicelles[34, 35]. The chemical shift index (CSI) obtained from the CA-CB correlations (Figure 3B) shows a residual helical secondary structure the R25C resonances. As reported previously[16], the chemical shifts of the resonances in the cytoplasmic domain represent a weighted average between the population of the completely unfolded state of PLN and its helical membrane-adsorbed T state; therefore, the R25C mutation still possess a significant fraction of the membrane adsorbed helical conformation. Figure 4 shows residue-specific changes as probed by DARR and TOBSY experiments for the different PLN mutants, emphasizing the T-to-R state equilibrium. Due to spectral overlap, only Ala, Ser, and Thr residues are resolved (the full 2D spectra are reported in the Supporting Information, Figure 1S). It is apparent that the cytoplasmic Ala residues (A11, and A15) are in slow conformational exchange between T and R states (Figure 4)[14]. These two populations are apparent for PLN^AFA and PLN^R25C, but, the R-state peaks are missing or undetectable for the R9C mutant, due to the low population of the R state. For R25C mutant, Ser and Thr residues located in the short loop connecting domains Ia and Ib are essentially missing in the DARR spectrum, but they are present in the TOBSY spectrum, again, indicating their higher mobility. The same residues are present although with much lower intensities in the corresponding experiments for PLN^AFA and PLN^R9C. Taken all together, these observations demonstrate that the loop region is more mobile for the PLN^R25C than PLN^AFA and PLN^R9C mutants.

Figure 3:

A. ¹³C-¹³C TOBSY ssNMR spectra for dynamic R state of PLN samples reconstituted in DMPC vesicles. (B) Chemical shift index for the cytoplasmic region of PLN^R25C (residues 2 to 17).

Figure 4:

DARR and TOBSY experiments map the immobile and dynamic states of PLN. Spectral regions of DARR and TOBSY spectra for PLN^AFA (A), PLN^R25C (B), and PLN^R9C (C), respectively, showing alanine, serine and threonine residues.

DISCUSSION

Our previous NMR and EPR studies unveiled the importance of PLN’s conformational equilibrium in the regulatory cycle of SERCA[14, 16, 36, 37]. We found that PLN adopts three different conformational states (T, R, and B) and their relative populations determine the extent of inhibition of SERCA. We also discovered that post-translational phosphorylation of PLN at Ser16 by cAMP- dependent protein kinase shifts the conformational equilibrium of the SERCA/PLN complex, augmenting the apparent affinity for Ca²⁺ and muscle contractility[14, 36]. Phosphorylation does not detach PLN from the ATPase; rather its cytoplasmic domain rearranges its SERCA-bound conformation upon phosphorylation[14, 36, 38], with PLN acting as a subunit[39]. These structural changes are mostly localized in the cytoplasmic region of PLN, while the TM domain of the protein remains essentially unperturbed[36]. Mutagenesis studies carried out by our group showed that it is possible to modulate SERCA function by engineering single or double mutations in the short intervening loop between domain Ia and the juxtamembrane domain Ib[17, 18] that affect the folding of the cytoplasmic region. Remarkably, a single mutation (P21G) nearly mimics the dynamic state of Ser16 phosphorylated PLN[17, 18]. Parallel EPR and NMR studies also showed that the nature as well as the composition of lipid bilayers affects PLN’s conformational equilibrium[16, 38], whose existence has been detected even in native lipids extract from the sarcoplasmic reticulum membrane[13, 40]. The folding-unfolding mechanism of the cytoplasmic region probably plays an important role for PLN interactions with other binding partners[41]. For instance, we found that the unfolding of domain Ia, which contains the consensus sequence for protein kinase, is kinetically important for kinase recognition of PLN and high efficiency of phosphoryl transfer[42–44].

Based on these data, we hypothesized that the dysfunctional folding and unfolding mechanism of the PLN’s cytoplasmic domain is also involved in the progression of cardiac disease, and that structural disruptions of the amphipathic helix of domain Ia may have effects on SERCA regulation. In fact, Young and co-workers have proposed that the higher hydrophobicity caused by the R9C mutation has a direct effect on regulation of SERCA[22]. In our first attempt to characterize the structural effects of this mutant, we performed solution NMR spectroscopy using DPC micelles as a membrane mimetic environment (Figures S2–S4). Unfortunately, this membrane-mimetic system revealed its considerable limitations. Specifically, solution NMR did not detect any substantial difference in the structure and dynamics of PLN cytoplasmic domain caused by the R9C mutation. In contrast, our structural and dynamic NMR analysis in lipids highlights substantial changes caused by both R9C and R25C mutations, suggesting that the DPC micelle is an inadequate system to investigate the complexity of PLN’s regulatory mechanism. In particular, solid-state NMR experiments in lipid membranes show that the R9C mutation shifts the conformational equilibrium toward the T state, the inhibitory and membrane associated conformation. Moreover, functional assays carried out in the presence of SERCA show that the R9C mutant is partially inhibitory, while kinetic phosphorylation assays reveal slower kinetics of phosphorylation by protein kinase A[45]. The latter can be explained by a stronger association of the regulatory region of PLN with the membrane bilayer, rendering the phosphorylation site virtually inaccessible to the kinase[45]. Also, Ha et al.[45] showed that R9C pentamer is unable to regulate SERCA’s activity, due to the formation of disulfide bridges between the protomers, a situation that is exacerbated under oxidizing conditions[45, 46].

On the other hand, the R25C mutant has been associated with super-inhibition of SERCA and Ca²⁺ transport, and unlike the R9C mutant, gives rise to sarcoplasmic Ca²⁺ leaks and causes arrhythmogenesis under stress conditions[23]. The R25C mutation is located in the juxtamembrane domain Ib, which is more dynamic than the TM domain II[13, 47]. In the wild-type PLN, Arg25 interacts electrostatically with the lipid membrane head groups[7, 12], anchoring the protein to the membrane. The cysteine mutation abolishes these interactions, causing a structural rearrangement of the lipid-protein interactions, which leads to topological changes of the TM domain, with a concomitant increase of the R state population. The super-inhibitory character of the R25C mutant can be probably explained by an increase of affinity between SERCA and PLN and a possible topological rearrangement within the inhibitory complex with SERCA. On the other hand, the increased population of the R state might affect the interactions with other binding partners, resulting in dysfunctional effects such as arrhythmogenesis or increase in the frequency of Ca²⁺ sparks and waves that were not observed in the R9C mutations. Additional structure-function correlations need to be carried out to test these hypotheses. While these studies provide a first structural look at the effects of R9C and R25C mutations on PLN’s interactions with lipid bilayers, only the characterization of the binding interactions with SERCA will provide more definitive insights into how these pathogenic mutations influence Ca²⁺ cycling.

CONCLUSIONS

In conclusion, we used a combination of solid-state NMR techniques to analyze the effects of two deadly mutations of PLN that lead to heart disease. Both CP and INEPT-based experiments show that these mutations affect PLN’s conformational equilibrium in opposite directions. The R9C mutation shifts the equilibrium toward the membrane-associated T state, which prevents phosphorylation and causes its LOF character; whereas the R25C mutants shifts the equilibrium toward the unfolded R state, which preferentially binds SERCA[35] and might interfere with the recognition mechanism of PLN by its multiple binding partners. Functionally, both mutants result in similar phenotypes; however, this new structural evidence suggests that the interactions with lipid bilayers also contribute to dysregulation of SERCA and make this mutant less responsive to β-adrenergic regulation in vivo.

HIGHLIGHTS.

• Solid-state NMR reveal the conformational equilibrium of phospholamban in lipid membranes.

• Arg9Cys mutation drives the equilibrium toward the T state of phospholamban

• Arg25Cys mutation shifts the conformational equilibrium toward the R state

• Changes in membrane interactions might be linked to the dysfunctional regulation of SERCA by phospholamban.