Controlling the Inhibition of the Sarcoplasmic Ca2+-ATPase by Tuning Phospholamban Structural Dynamics*

Cardiac contraction and relaxation are regulated by conformational transitions of protein complexes that are responsible for calcium trafficking through cell membranes. Central to the muscle relaxation phase is a dynamic membrane protein complex formed by Ca2+-ATPase (SERCA) and phospholamban (PLN), which in humans is responsible for ∼70% of the calcium re-uptake in the sarcoplasmic reticulum. Dysfunction in this regulatory mechanism causes severe pathophysiologies. In this report, we used a combination of nuclear magnetic resonance, electron paramagnetic resonance, and coupled enzyme assays to investigate how single mutations at position 21 of PLN affects its structural dynamics and, in turn, its interaction with SERCA. We found that it is possible to control the activity of SERCA by tuning PLN structural dynamics. Both increased rigidity and mobility of the PLN backbone cause a reduction of SERCA inhibition, affecting calcium transport. Although the more rigid, loss-of-function (LOF) mutants have lower binding affinities for SERCA, the more dynamic LOF mutants have binding affinities similar to that of PLN. Here, we demonstrate that it is possible to harness this knowledge to design new LOF mutants with activity similar to S16E (a mutant already used in gene therapy) for possible application in recombinant gene therapy. As proof of concept, we show a new mutant of PLN, P21G, with improved LOF characteristics in vitro.

Calcium translocation is a major signaling mechanism for mammalian cells (1). In cardiac myocytes, calcium regulates contractility and relaxation through a "calcium-induced calcium release" cycle (2)(3)(4). At low concentrations, calcium ions enter the myocytes through voltage-dependent L-type calcium channels that are embedded in the sarcolemmal sarcoplasmic reticulum (SR) 3 junctions, acting as a trigger for the release of calcium ions by the ryanodine receptors. Subsequently, free intracellular calcium ions bind the myofilaments, initiating muscle contraction. Relaxation of the muscle fibers occurs when calcium ions dissociate from the myofilaments and are transported out of the cytosol into the SR vesicles, a process that is primarily controlled by SR Ca 2ϩ -ATPase (SERCA) and phospholamban (PLN) (5)(6)(7).
Growing evidence suggests that PLN acts as a subunit of SERCA (15), with its transmembrane helix bound in the hydrophobic groove between transmembrane domains M2, M4, M6, and M9 of the enzyme and its cytoplasmic helix interacting with the cytoplasmic domains of SERCA (16) see (Fig. 1). ␤-Adrenergic stimuli activate protein kinase A, resulting in PLN phosphorylation at Ser-16 (pS16-PLN) with complete relief of inhibition and restoration of the calcium flux into SR vesicles (7). Ser-16 phosphorylation is necessary and sufficient to relieve the inhibitory effect of PLN (17,18), although Thr-17 and Ser-16/ Thr-17 double phosphorylation are also present in intact myocardial tissue (19,20).
Defective cellular calcium handling by the SERCA⅐PLN complex is responsible for the progression of heart failure (5). In fact, aberrant mutations in the PLN primary sequence have been directly linked with specific heart conditions. R9C mutation, Arg-14 deletion, and Leu-39 truncation of PLN have been detected in patients affected by dilated cardiomyopathy (21)(22)(23). One promising therapeutic strategy is to augment cardiac contractility by inducing a positive inotropic effect, targeting PLN function directly (i.e. reducing its inhibitory effects) (24). This can be accomplished by (a) decreasing the expression level of PLN, (b) increasing phosphorylation of PLN, (c) using PLN inhibitors that interfere with the SERCA⅐PLN complex formation (e.g. quercitin, tannin, and ellagic acid), or (d) using mutants with dominant negative effects, loss-of-function (LOF) mutants, that compete with endogenous wild-type PLN (24). The last approach has received significant attention after the success of Chien and co-workers in the design and delivery of a pseudo-phosphorylated mutant of PLN (S16E-PLN). Using a new recombinant adeno-associated virus vector, these researchers were able to reverse the effects of progressive dilated cardiomyopathy (25,26). In particular, they showed that expression of S16E-PLN in animal models with chronic heart failure and dilated cardiomyopathy as well as with post-cardiac infarction resulted in a substantial increase of both cardiac diastolic and systolic functions (27). More recently, using percutaneous cardiac recirculation-mediated gene transfer with S16E-PLN, the Power and Chien groups demonstrated the reversal of heart failure progression even in large animals (28).
Prompted by the success of the S16E-PLN, we began our structural investigation with a view toward rational design of PLN LOF mutants. Previous mutagenesis studies (29,30) gave invaluable insight into PLN function. However, they did not distinguish whether a point mutation causes changes in the interaction surface between SERCA and PLN, in PLN structural dynamics, or both. As a consequence, rational design of single or double mutations with increasing or decreasing inhibitory effects have been quite difficult to achieve (31)(32)(33). From these studies, Young and coworkers concluded that the inhibitory properties of each amino acid are context-dependent (34).
For our rational design of mutants, we postulated that by manipulating PLN structural dynamics it could be possible to regulate SERCA function. This hypothesis is substantiated by a compilation of studies that we have carried out in the last few years on PLN (phosphorylated and unphosphorylated) free and bound to SERCA and culminated in the formulation of the allosteric mechanism of regulation of SERCA by PLN (13,(35)(36)(37). This model assumes that PLN exists in a dynamic equilibrium between T and R states, with SERCA shifting the equilibrium toward the R state prior to forming the inhibitory complex. Phosphorylation at Ser-16 affects PLN structural dynamics and, in turn, the allosteric equilibrium, causing the formation of the non-inhibitory SERCA⅐PLN complex (13,(35)(36)(37).
In this report, we demonstrate that it is possible to tune PLN structural dynamics and mimic PLN phosphorylation to gain control of SERCA function. As proof of concept, we present a new PLN mutant, P21G-PLN, with improved in vitro characteristics with respect to S16E-PLN for possible use in recombinant adeno-associated virus vector gene therapy.

EXPERIMENTAL PROCEDURES
NMR Sample Preparation-15 N-Labeled monomeric PLN mutants were expressed as a fusion protein with maltose-binding protein and purified as described previously (38). Primer synthesis and DNA sequencing were performed at the Bio-Medical Genomics Center of the University of Minnesota. Sequences of the forward and reverse primers consisted of the optimized codons for Escherichia coli for three residues before and after the target residue to be mutated. PCR was performed with a total volume of 50 l containing 0.06 M of each oligonucleotide, 80 g of parental template, 200 M dNTP mix (Promega), 10ϫ buffer (10 l per reaction), and 5.0 unit of Pfu Turbo DNA polymerase (Stratagene). PCR was carried out in a PerkinElmer Life Sciences GeneAmp PCR system 2400 at 95°C for 3 min followed by 16 temperature cycles of 95°C for 15 s to denature the DNA, 55°C for 1 min to anneal, and 68°C for 12 min to elongate. PCR products were then digested with Dpn-I (Stratagene) to remove non-mutated methylated doublestranded DNA. After digestion, the products were transformed into XL-1 Blue-competent cells (Stratagene) and grown on Luria-Bertani medium (LB)/ampicillin plates. DNA purification was performed with Quick-Spin Minipreps (Qiagen). The purified DNA was quantitated by measuring the UV absorption. DNA sequencing confirmed the single site mutations on the PLN background. Purified plasmid constructs were then transformed into the BL21(DE3) strain of E. coli, and protein purification was performed as previously described (38). Phosphorylation at Ser-16 by protein kinase A of the P21G mutant was performed as previously described (36). NMR Samples were prepared by dissolving the lyophilized proteins (PLN, P21A, S16E, P21G, P21L, P21I, S16EP21G, pS16-PLN, and pS16-P21G) in 300 l of phosphate-buffered saline solution, pH 6.0, containing 300 mM dodecylphosphocholine (DPC) (Avertec) and 10% D 2 O. The protein final concentration in the NMR samples was ϳ0.25 mM as estimated by gel densitometry.
SERCA was extracted from rabbit skeletal muscle, purified, and tested for activity as reported previously (13,35). NMR Experiments-NMR experiments were carried out on a 600-MHz Varian INOVA spectrometer operating at 1 H Larmor frequency of 600.13 MHz and equipped with an inverse detection triple resonance and triple axis gradient probe. All of the experiments were performed at 37°C. [ 1 H, 15 N] heteronuclear single quantum coherence (HSQC) spectra (39) were acquired with 6000-Hz spectral width in the direct dimension ( 1 H) and 1500 Hz in the indirect dimension ( 15 N). 1024 and 64 complex points were collected for the direct and indirect dimensions, respectively. Two-dimensional data were processed to a final matrix size of 2048 ϫ 256 after zero filling and Fourier transformation. Resonance assignments for all the mutants were carried out using a combination of three-dimensional [ 1 H, 15 (40,41). The threedimensional experiments were acquired with 6000-Hz spectral width in both 1 H dimensions and 1500 Hz in the 15 N dimension. The three-dimensional matrices were acquired with 1024 ϫ 80 ϫ 40 complex points with a final matrix size of 2048 ϫ 320 ϫ 160 after final processing. The recycling delay used was 1.2 s for all of the experiments. Data processing was carried out using NMRPipe software (42). The secondary struc-ture was assessed using H␣ chemical shift index (CSI), which is the difference between the PLN H␣ chemical shifts and those from random coil values (43). Because of the high rotational anisotropy of PLN and its mutants, the analysis of the order parameters (the customary index for protein internal dynamics, picosecond to nanosecond timescale) does not give good convergence; therefore, we limited our analysis to heteronuclear NOE data (44). [ 1 H, 15 N] steady-state NOE enhancement experiments were acquired using standard pulse sequences based on Farrow et al. (45) using a spectral width of 6000 Hz in 1 H, 1500 Hz in 1 H, with 1024 points collected in the direct ( 1 H) dimension and 64 points collected in the indirect ( 15 N) dimension. For the saturated spectrum, a 3-s 1 H presaturation period was used. Peak intensities were analyzed and quantified using NMRView software (46).
Analysis of the Correlations between Mutations and Structural Dynamics-The local and long range perturbations caused by each mutation (or phosphorylation) on the structural (CSI) and dynamical (NOE) properties (P) of PLN were analyzed for each residue by computing the correlation coefficients ( PM ) between the property P and each mutation M (47), where and are the average values and standard deviations for P and M, respectively, and n M is the total number of species. To assign M, we ordered the different PLN mutants based on increasing average NOE values for residues 17-30, corresponding to the loop and domain Ib. The heteronuclear NOEs for the specified range change from 0.50 to 0.69 in the following order: Then the incremental values of the heteronuclear NOE across the species were associated to the variable M for a total of 7 samples (n M ϭ 7). In the histogram of Fig. 3 we report all the correlations calculated. The gradation of gray in the figures indicates the relevance of each correlation. Meaningful correlations respond to the following criteria: (a) have an absolute value much greater than the corresponding error (greater than one standard deviation, and (b) show a significant change in CSI (Ͼ0.05 ppm) or heteronuclear NOE (Ͼ0.1). Given the small sample size (n M ϭ 7), the errors are computed using the nonparametric bootstrap method calculated with the R software package with a number of bootstrap samples of 200.
NMR Titrations of PLN Mutants with SERCA: Determination of K d and T to R Transitions-NMR titrations were carried out monitoring the chemical shifts of the amide fingerprint regions of PLN variants using [ 1 H, 15 N] HSQC experiments. A total of 13 different experiments were acquired after each SERCA addition up to a final molar ratio of 1:1 (SERCA:PLN). The K d values were determined by following the disappearance of the resonances corresponding to domain II of PLN upon addition of incremental amounts of SERCA as described previously (13,35). Assuming 1:1 ratio between SERCA and PLN, the intensity retention (I retention ) of domain II residues is directly related to the fraction of PLN bound to SERCA (f b ), Then I retention is correlated to a dissociation constant (K d ) through the following derivation (13).
Taking the negative quadratic solution, the fraction bound is as follows.
Finally, the K d values were calculated using a best-fit non-linear regression of Equation 8. The transitions from T to R state for each mutant were quantified by following the disappearance of the T state and the appearance of R state in the [ 1 H, 15 N] HSQC spectra (13,35). The percentage of R state was then plotted as a function of the SERCA:PLN molar ratio. The resulting curves were fit using the Hill equation. Hill coefficients were used to estimate the degree of cooperativity of each construct. Buffer titrations (PLN titrations with SERCA buffer only) were used as a control to account for the effects of dilution to peak intensity and chemical shift perturbation (13,35). Preparation of EPR Samples-PLN and mutants were prepared with the spin-labeled amino acid 2,2,6,6-tetramethylpiperidine-1-oxyl-4-amino-4-carboxylic acid (TOAC) substituted for Ala-11, using solid-phase peptide synthesis, as reported previously (48,49). Unlike conventional spin labels attached flexibly to cysteine side chains, this spin label couples the nitroxide moiety rigidly to the ␣ carbon, providing direct correlation to peptide backbone dynamics. This spin label has no effect on PLN function (48). Phosphorylation of spin-labeled PLN at Ser-16 was accomplished as described previously (37). For EPR experiments, TOAC⅐PLN was reconstituted into lipid vesicles containing DOPC:DOPE (4:1, 200 lipids per PLN). EPR experiments were conducted in low calcium (pCa 6.5) buffer (50 mM KCl, 5 mM MgCl 2 , 0.5 mM EGTA, 210 M CaCl 2 , 50 mM MOPS, pH 7.0), where the inhibitory effect of PLN is maximal.
EPR Experiments on PLN in Lipid Bilayers-EPR spectra were acquired using a Bruker EleXsys 500 spectrometer with the SHQ cavity. Samples (20 l in a 0.6-mm inner diameter quartz capillary) were maintained at 4°C using the Bruker temperature controller with a quartz Dewar insert. The field modulation frequency was 100 kHz, with a peak-to-peak amplitude of 1 G. The microwave power was 12.6 milliwatts, producing moderate saturation (so that signal intensity was at least 50% of maximum) without significant effect on the spectral line shape. Spectra were analyzed in terms of peptide backbone rotational dynamics and mole fractions of two resolved conformations using computational simulation as described previously (37).
Activity Assays-The inhibitory potency of PLN and its mutants was measured in lipid bilayers (EPR conditions). To measure activity in membranes, each PLN mutant was co-reconstituted in lipid bilayer membranes (DOPC:DOPE, 4:1) with purified SERCA at molar ratios of 10:1 PLN:SERCA and 700:1 lipids:SERCA. The calcium dependence of the ATPase activity was measured at 25°C using coupled enzyme assays (50). The consumption of NADH was monitored by the decrease in its absorbance at 340 nm using a Spectromax microplate reader (Molecular Devices). The initial ATPase rate (V) was measured as a function of pCa, and the data were fit using the Hill equation, to determine V max , pK Ca (the pCa value when V ϭ V max /2), and n (Hill coefficient). V max was obtained from the fit, and the data were plotted as V/V max versus pCa. Because the maximal velocity is not reproducible for the SERCA⅐PLN complex, we used the pK Ca values to quantify the inhibition caused by PLN mutants.

NMR Structural Dynamics of PLN Mutants-Our central
hypothesis is that PLN regulates SERCA through an allosteric mechanism (35). According to this mechanism, free PLN exists in equilibrium between T and R states, where the T state is the predominant L-shape topology and the R state is more dynamic. SERCA selects and binds the R state (37). Phosphorylation at Ser-16 by protein kinase A reverses PLN inhibition. This single phosphoryl transfer introduces a negative charge in the cytoplasmic domain, partially unwinding the helix, increasing its picosecond to millisecond dynamics, and promoting the transition from the T to the R state. This orderto-disorder transition causes the disruption of the SERCA-PLN interaction surface (13,36). Can mutations mimic the effects of phosphorylation?
To answer this question, we designed a series of PLN mutants with a gradation of helicity and structural rigidity. These mutagenesis studies were carried out on monomeric PLN in which the three non-essential cysteines 36, 41, and 46 were mutated into Ala, Phe, and Ala (51). The monomeric species is biologically relevant for two reasons: first, it has been shown that monomeric PLN is the inhibitory species (31,52), and second, it retains the same activity as wild-type PLN (51).
As a starting point for our mutagenesis studies, we chose Pro-21, a pivotal residue that breaks the transmembrane helix of PLN, determining its L-shaped conformation (12) and dynamics (Fig. 1A) (44). Nuclear spin relaxation measurements revealed that Pro-21 bridges the two most dynamic regions: loop and domain Ib (Fig. 1A). The new mutants were designed using JUFO (53) and PrDOS software packages that predict both secondary structure and dynamics based on protein pri-mary sequences. From the JUFO and PrDOS scores, we selected the following mutants with increasing propensity to form a continuous, more rigid helix throughout the entire PLN sequence: P21G Ͻ P21A Ͻ P21I Ͻ P21L.
We probed the structural dynamics of the different mutants reconstituted in DPC micelles, a membrane-mimicking system that preserves SERCA activity (35). Using a combination of [ 1 H, 15 Fig. 2. The red and black traces correspond to unphosphorylated (inhibitory form) and phosphorylated (non-inhibitory form) PLN that we have previously published (36,44). These are shown in each panel to guide the reader. Visual inspection reveals a gradation of structural dynamics of PLN mutants going from the most flexible and least helical P21G to the most rigid and helical P21L. However, to rationalize the effects of the point mutations on the struc- tural dynamics, we analyzed our data using a correlation function as described under "Experimental Procedures." Fig. 3 shows the correlation plots for CSI and heteronuclear NOE for all of the different mutations. In black are those residues with an absolute value of the correlation coefficient greater than three standard deviations in dark-gray are those between two and three standard deviations, and in white are those less than two standard deviations. In addition, the extent of the variations of CSI and heteronuclear NOE are indicated on top of the most significant residues. A negative correlation (or anti-correlation) for the CSI graph in Fig. 3A indicates that moving from P21G to P21L the resonance for a given residue shifts upfield (the residue becomes more helical), a positive correlation indicates a correlated shift downfield toward random coil values. As expected, the residues adjacent or in the immediate proximity of the mutated site show strong anti-correlations (see residues 18 -20). Interestingly, residue 27 (and to a lesser extent residue 30) exhibits a noticeable positive correlation or decrease in helical content. Note that these residues are neighbors to Gln-26, which we identified as the gatekeeper between the cytoplasmic and the transmembrane domains and is responsible for the transmission of the phosphorylation signal at Ser-16 down to the membrane-embedded region (13). Structural Dynamics of S16E-PLN-The analysis of the CSI and heteronuclear NOE plots indicates that the strength and the dynamics of the cytoplasmic helix of this mutant are intermediate between those of P21G and P21A mutants (Fig. 2). The   DECEMBER 21, 2007 • VOLUME 282 • NUMBER 51 CSI of the S16E mutant has a profile similar to that of PLN for both domains Ib and II, while showing significant differences from Glu-2 to Met-20. The heteronuclear NOE values also follow this trend, showing values that are intermediate between the unphosphorylated and phosphorylated forms of PLN. Therefore, the structural dynamics of S16E pseudo-phosphorylated mutant of PLN is similar but not identical to that of pS16-PLN.

Tuning PLN Structural Dynamics Controls SERCA Activity
EPR of Selected Mutants in Lipid Bilayers-To test whether the effects of PLN mutation on dynamics in DPC micelles are similar to those observed in lipid environments, we carried out EPR experiments on TOAC-spin-labeled PLN reconstituted into lipid bilayers (Fig. 4) of the same lipid composition used to measure functional regulation of SERCA by PLN. In all four PLN variants tested, the EPR probe, reporting the backbone dynamics at position 11 in the cytoplasmic domain, shows clear evidence for two resolved states, a broad component corresponding to the ordered T state, and a narrow component corresponding to the dynamically disordered R state (48). The vertical dashed lines in Fig. 4 show that the spectral line positions, thus the rotational dynamics, of the T and R states are essentially unaffected by phosphorylation or mutation. Therefore, the principal effect observed is a change in the distribution (i.e. mole fractions) between the two states.
Phosphorylation at Ser-16 induces a dramatic increase in the mole fraction in the R state, at the expense of the T state, whereas the S16E mutation produces an intermediate effect and the P21A mutation produces an opposite effect, decreasing the mole fraction in the R state (Fig. 4). Thus, the effects of mutation on these conformational transitions are consistent with the NMR observations in DPC micelles (Fig. 2).
Structural Dynamics-Function Correlations: Effects of Mutations on SERCA Activity-To characterize the functional effects of these mutations, we reconstituted each mutant with SERCA in lipid membranes and measured ⌬pK Ca , which corresponds to the difference between the pK Ca measured from the normalized calcium affinity curve in the absence and presence of PLN. Typical PLN inhibition of SERCA is characterized by ⌬pK Ca (change in calcium affinity) ϳ0.30 for unphosphorylated PLN (inhibitory complex) and a value of 0 for phosphorylated PLN (pS16-PLN, non-inhibitory complex), indicating no inhibition of the enzyme (29). Optimal enzyme inhibition is obtained with PLN (a proline at residue 21), which gives a ⌬pK Ca shift of 0.30. We found that either increased rigidity or increased flexibility of the loop and cytoplasmic regions of PLN decreases the apparent affinity of SERCA for calcium (Fig. 5). Increasing PLN rigidity results in smaller ⌬pK Ca values, i.e. a LOF mutant. Both P21A and P21I were found to have a ⌬pK Ca shift of 0.21, whereas P21L, the most rigid mutant, displayed the greatest loss in function with a ⌬pK Ca shift of 0.17. Increasing PLN mobility also causes decreased inhibitory potency, lowering ⌬pK Ca with respect to PLN. For instance, the P21G mutation showed a decrease in ⌬pK Ca similar to that of P21A and P21I. S16E, which is dynamically very similar to P21G, also showed a decrease in ⌬pK Ca to 0.15, whereas pS16-PLN (the most dynamic PLN form) shows complete relief of inhibition (⌬pK Ca ϳ 0). Therefore, similar to the structure and dynamics (see CSI and NOE values) the inhibitory potencies of S16E-PLN and P21G-PLN are intermediate between those of PLN and pS16-PLN.
Effects of SERCA on the Structural Transitions of PLN Mutants-To map the effects of each mutation on the T to R state transitions of PLN upon interaction with SERCA, we carried out a series of NMR titrations by adding purified SERCA incrementally to each mutant. Addition of SERCA causes the appearance of two different states in slow chemical exchange in the [ 1 H, 15 N] HSQC spectrum (Fig. 6A). In analogy with our previous work (13,35), the resonances that exchanged between two states were assigned to the free T and R states of PLN. Although qualitatively the behavior of the fingerprint resonances in the [ 1 H, 15 N] HSQC of the different mutants upon addition of SERCA were similar, there were marked quantitative differences in the allosteric transition for each mutant. Fig.  6B shows the percentage of the R state for the cytoplasmic residues present with increasing amounts of SERCA (13). For PLN, we found that the transition from the T to R state is gradual, whereas for pS16-PLN this transition occurs in a cooperative manner, with a combined Hill coefficient of ϳ3-4 (13). For the more dynamic mutants (S16E and P21G), we observed an increase in cooperativity with respect to PLN, but a substantial reduction in cooperativity with respect to pS16-PLN (see Hill coefficient in Table 1). Interestingly, the mutants (P21L, P21I, and P21A) that are more motion-restricted show cooperativity similar to that of pS16-PLN, meaning that SERCA:PLN molar ratios smaller than those required for PLN are sufficient for triggering the T to R transition.
In addition to characterizing the transition to the R state, the reduction in transmembrane domain intensities allowed for the determination of a dissociation constant (K d ) for each mutant (Fig. 5B). This was accomplished by using the decrease of the signal intensity of domain II upon titration with SERCA (13). Although the more flexible variants of PLN display K d values ϳ60 M (in agreement with our previous NMR and EPR data in DPC micelles (13,35)), the values of K d increase with increasing helicity and rigidity up to ϳ170 M for P21L-PLN. This supports our earlier hypothesis that PLN molds into the hydropho-bic core of the enzyme with its plasticity, an essential requirement for binding (44).
The Gatekeeper Role of the Gln-26 Side Chain-When analyzing the effects of phosphorylation at Ser-16, we proposed that the side chain of Gln-26 located in domain Ib is crucial for PLN inhibition (13). Using site-directed mutagenesis Mac-Lennan and co-workers (33) showed that the Q26A mutation generates a LOF mutant, supporting our original hypothesis that alanine is unable to form intermolecular hydrogen bonds. Recent molecular modeling calculations carried out by the same group support the involvement of Gln-26 side chain in the SERCA⅐PLN binding interface (33). Fig. 7 shows the intensity retention of the Gln-26 NH 2 groups at the end point of SERCA titrations with all PLN species. Upon addition of SERCA, the majority of the mutants analyzed retain ϳ50% of the intensity of the Gln-26 NH 2 groups, whereas the intensity retention was much greater (up to 100%) for both phosphorylated species (pS16-P21G-PLN and pS16-PLN). This suggests that changes in the dynamics induced by mutations are not sufficient to affect the gatekeeper role of the Gln-26 side chain, and only phosphorylation causes this side chain to swing away from the protein-protein interface.
Phosphorylation of P21G Mutant-We also tested whether P21G still has post-translational control, a crucial factor in the view of developing a new generation of LOF mutants. Indeed, we found that P21G can be fully phosphorylated by protein kinase A and, like pS16-PLN, is a complete LOF species (Fig. 8). Fig. 8 shows that the calcium affinity curves for pP21G-PLN and pS16-PLN are superimposable, demonstrating the complete reversal of the residual inhibitory effect of P21G. Interestingly, pS16P21G-PLN was not as dynamic as pS16-PLN. However, the dynamic-function correlation graph shows that it must have passed the mobility threshold to become a complete LOF species.
P21GS16E Double Mutant-Are the effects of mutations additive? What is the role of electrostatics in PLN phosphorylation? To answer these questions, we cloned a monomeric mutant of PLN with both P21G and S16E mutations. If the effects of the double mutation were additive, we would expect an augmentation of the LOF effects with respect to the single P21G mutation, with a relief of inhibition similar to that of pS16-PLN. On the contrary, we found that this double mutant still has a residual inhibitory potency, which is intermediate between those of P21G and S16E single mutations. Although this behavior is counterintuitive at first, this double mutant is slightly more rigid and helical than both P21G and S16E and has a lower affinity for SERCA (K d ϳ 80 M), making its inhibitory activity similar to that of P21A. We conclude that the effects of the two mutations are not additive and PLN dynamics has a dominant effect over electrostatics at Ser-16.

DISCUSSION
In our previous work, we hypothesized that there exists an allosteric control mechanism for SERCA by PLN (35). We showed the existence of a dynamic equilibrium between two states of PLN: the T state, which is more static and L-shaped, and the R state, which is more dynamic. Based on the effect of phosphorylation at Ser-16, we also suggested that the structural dynamics of PLN may have a direct influence on the calcium FIGURE 5. Structure-dynamics-function correlations diagrams for the different mutants of PLN. A, two-dimensional projection of the change in activity correlated with change in dynamics. B, three-dimensional diagram showing the increasing K d with backbone rigidity of the different PLN species. The diagrams are divided into two regions: high affinity (with K d comparable to that of the monomeric PLN) and low affinity (K d values higher than that of the monomeric PLN). Note that heteronuclear NOE data and CSI for monomeric PLN both phosphorylated and unphosphorylated are taken with permission from Zamoon et al. (12) and Metcalfe et al. (36,44).
affinity and the allosteric transitions of the enzyme (13,37). In the current work, we demonstrate that (a) the structural dynamics of PLN plays a crucial role in the allosteric regulation of SERCA and (b) it is possible to gain control of SERCA function through the modulation of PLN structural dynamics. NMR data provide detailed insight into the conformational dynamics of PLN in DPC micelles, whereas EPR data (Fig. 4) confirm that similar conclusions are valid in lipid bilayers and show clearly that the principal effect of phosphorylation or mutation in the loop region of PLN is to regulate the equilibrium between the R and T states.
In agreement with previous studies (55), we found that proline 21 is pivotal for optimal inhibition of SERCA. Point mutations in this site induce long-range changes in PLN structural dynamics, affecting PLN interactions with SERCA (Fig. 3). Interestingly, the changes in the structural dynamics observed have short and long range correlations. In other words, single point mutations not only affect residues in the immediate proximity of the mutated site (loop) but also distant residues. In particular, CSI correlation plots show a strong positive correlation for Asn-27 and a weaker positive correlation for Asn-30, in antithesis to the trend observed for the other residues around the Pro-21 site. Both residues are proximal to Gln-26, a key residue that was predicted to swing away from the SERCA⅐PLN interface upon phosphorylation at Ser-16 (13,16). It is possible that tuning the structural dynamics of domain Ib with concomitant breakage of intermolecular hydrogen bonds is necessary for complete relief of inhibition.
These changes in structure and dynamics of PLN induced by single mutations have direct effects on its inhibitory potency. To interpret the complex energy landscape of SERCA-PLN interactions, we built a simplified energetic diagram that summarizes the behavior of the mutants analyzed (Fig. 9). In the absence of SERCA, PLN is in equilibrium between a more populated T state where the cytoplasmic domain is in direct contact with the lipid membrane and a less populated R (activated) state. In the presence of SERCA, this equilibrium is shifted toward the R state. Unphosphorylated PLN binds SERCA to form an inhibitory complex (low energy), whereas Ser-16 phosphorylation leads to a non-inhibitory complex (high energy) without completely dissociating from the enzyme. Our results indicate that this transition from an inhibitory to a non-inhibitory complex is strongly influenced by structural dynamics.
Mutations change the activation energy for the T to R transition, a primary difference between pS16-PLN and PLN (Fig.  9). In the presence of SERCA, the most rigid mutants (P21A, P21I, and P21L) have a lower activation energy to transition from the T to R state. Interestingly, they show the same cooperativity (similar Hill coefficient) as the most dynamic species (pS16-PLN), suggesting that the R state could be an ensemble of different conforma-  Can we use this knowledge to design a new generation of LOF mutants? Based on our data, we found that more helical and motionally restricted mutants of PLN are LOF with cooperative T to R state transitions. However, the more rigid mutants have lower binding affinities (K d ϳ 2 times larger for the most motionally restricted P21L) and would not be good candidates to compete with wild-type PLN for possible recombinant adeno-associated virus vector gene therapies (56). On the contrary, more mobile mutants such as S16E and P21G have binding affinities similar to those of the non-mutated PLN and would constitute good LOF candidates. Remarkably, P21G has both an inhibitory effect and affinity for SERCA similar to that of S16E. The residual inhibitory effects are ⌬pK Ca ϳ 0.22 and 0.15 for P21G and S16E, respectively. Because P21G is more dynamic than S16E, this small but significant difference in the inhibitory potency might be attributable to the difference in charge between the two mutants and the different interaction surfaces with SERCA.
A significant limitation of the S16E mutant is its lack of posttranslational control by protein kinase A, i.e. position 16 is no longer phosphorylatable. Unlike S16E, P21G has position 16 still accessible to protein kinase A, enabling post-translational control. This would allow further augmentation of the LOF character of the mutant in vivo. Of course, the ideal LOF species is pS16-PLN, which still binds SERCA with high affinity without inhibiting its function. Rationally designed LOF mutants should mimic the structural dynamics of pS16-PLN. However, complete LOF mutants would correspond to ablation of PLN (i.e. continuous activation of SERCA), which can result in dilated cardiomyopathy (23). Therefore, it is necessary that the new mutants possess a residual inhibitory effect like P21G.
Whereas in vivo testing using recombinant adeno-associated virus vector is necessary to completely validate the therapeutic possibility of P21G, our in vitro assays show that this mutant is a foundation for the next generation of rationally designed PLN species with potential therapeutic application. Unlike S16E, these new mutants would simultaneously preserve post-translational control in addition to increased internal dynamics.
Indeed, the current study is limited to the correlation between backbone internal dynamics and the inhibition of SERCA by PLN. As shown for Gln-26 (a residue involved in the SERCA⅐PLN interface), the involvement of this side chain in the protein-protein interface is not affected by mutations at position 21, but rather it is sensitive only to phosphorylation (see Fig. 7). To fully understand how the phosphorylation signal is relayed through the mutation sites in the loop to the transmembrane protein interface, future studies on the structural dynamics of PLN side chains with and without phosphorylation in the presence of SERCA will be necessary.  . Functional assays in lipid bilayers for P21G and phosphorylated P21G. Note that phosphorylated P21G has the same effect as phosphorylated PLN. Data sets were fit by Equation 9 and plotted as V/V max . Each point represents the mean (n Ն 6); in most cases, the error measured was smaller than the plotted symbol. FIGURE 9. Proposed energy diagram to interpret the effect of mutation and phosphorylation on SERCA activity. All the possible states are reported (T, R, inhibitory, and non-inhibitory complexes). The difference in energy levels of the free PLN species reflects the relative propensity of PLN species to undergo cooperative transition from the T to R state upon addition of SERCA (see Fig. 6 and the Hill coefficients in Table 1). The energy gap between the free and bound states reflects both the K d values while the relative energies of the bound states reflects the inhibitory potency.
In summary, we have demonstrated that, by fine-tuning the structural dynamics of PLN, it is possible to modulate its allosteric regulation of SERCA. The possibility of controlling SERCA activity will help design new PLN mutants that can be utilized in the correction and reversal of calcium cycling abnormalities in heart failures. From a general structural biology point of view, the possibility of fine-tuning a membrane enzyme's function by modifying its subunits' structural dynamics opens up the application of similar strategies in the regulation of other membrane-bound proteins that are governed by similar mechanisms. For instance, similar approaches can be used in the modulation of the Na/K-ATPase, which is regulated by phospholemman, a single pass membrane protein with a function similar to that of PLN (3).