doi:10.1016/j.jmb.2010.11.014 J. Mol. Biol. (2011) 405, 707–723

Contents lists available at www.sciencedirect.com

Journal of Molecular Biologyj ourna l homepage: ht tp : / /ees .e lsev ie r.com. jmb

Phosphorylation and Mutation of Phospholamban AlterPhysical Interactions With the SarcoplasmicReticulum Calcium Pump

John Paul Glaves1,2, Catharine A. Trieber1,2, Delaine K. Ceholski1,David L. Stokes3,4 and Howard S. Young1,2⁎1Department of Biochemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2H72National Institute for Nanotechnology, University of Alberta, Edmonton, Alberta, Canada T6G 2H73Skirball Institute of Biomolecular Medicine, School of Medicine, New York University, New York, NY 10016, USA4New York Structural Biology Center, New York, NY 10027, USA

Received 19 July 2010;received in revised form2 November 2010;accepted 8 November 2010Available online23 November 2010

Edited by W. Baumeister

Keywords:SERCA;phospholamban;phosphorylation;electron crystallography;2D crystals

*Corresponding author. E-mail addrhyoung@ualberta.ca.Abbreviations used: SERCA, sarco

calcium ATPase; SR, sarcoplasmic rephospholamban; PKA, cAMP-depen

0022-2836/$ - see front matter © 2010 E

Phospholamban physically interacts with the sarcoplasmic reticulumcalcium pump (SERCA) and regulates contractility of the heart in responseto adrenergic stimuli. We studied this interaction using electron microscopyof 2D crystals of SERCA in complex with phospholamban. In earlier studies,phospholamban oligomers were found interspersed between SERCA dimerribbons and a 3D model was constructed to show interactions with SERCA.In this study, we examined the oligomeric state of phospholamban and theeffects of phosphorylation and mutation of phospholamban on theinteraction with SERCA in the 2D crystals. On the basis of projection mapsfrom negatively stained and frozen-hydrated crystals, phosphorylation ofSer16 selectively disordered the cytoplasmic domain of wild type phospho-lamban. This was not the case for a pentameric gain-of-function mutant(Lys27Ala), which retained inhibitory activity and remained ordered in thephosphorylated state. A partial loss-of-function mutation that altered thecharge state of phospholamban (Arg14Ala) retained an ordered state,while acomplete loss-of-function mutation (Asn34Ala) was also disordered. Thefunctional state of phospholamban was correlated with an order-to-disordertransition of the phospholamban cytoplasmic domain in the 2D co-crystals.Furthermore, co-crystals of the gain-of-functionmutant (Lys27Ala) facilitateddata collection from frozen-hydrated crystals. An improved projection mapwas calculated to a resolution of 8 Å, which supports the pentamer as theoligomeric state of phospholamban in the crystals. The 2D co-crystals withSERCA require a functional pentameric form of phospholamban, whichphysically interactswith SERCAat an accessory site distinct from that used bythe phospholamban monomer for the inhibitory association.

© 2010 Elsevier Ltd. All rights reserved.

ess:

plasmic reticulumticulum; PLB,dent protein kinase.

lsevier Ltd. All rights reserve

Introduction

Cation transport by the P-type ion pumps is anessential process in all eukaryotic cells, wherechanges in intracellular cation concentrations arelinked to precise physiological responses. The bestunderstoodmembers of this transport family include

d.

708 Co-crystals of SERCA and Phospholamban

the sarcoplasmic reticulum calcium ATPase(SERCA) found in muscle cells and the plasmamembrane sodium-potassium ATPase (Na+-K+

pump) found in all cell types. These two P-type ionpumps are particularly important in cardiac con-tractility and are major drug targets for the clinicalimprovement of heart disease. An extensive series ofX-ray and electron crystallographic studies haveresulted in structures of a variety of SERCA reactionintermediates, thus revealing how ATP hydrolysis iscoupled to calcium transport across the sarcoplasmicrecticulum (SR) membrane in order to achievemuscle relaxation.1–14 These studies show thatSERCA is composed of a transmembrane domainthat contains the calcium-binding sites and threecytosolic domains that are responsible for nucleotidebinding, phosphorylation, and communication withthe transmembrane domain. It has been shown thatthe intermediate states (calcium binding, phosphor-ylation, calcium transport, dephosphorylation andproton counter-transport) involve coupled domainmovements that link the cation-binding sites withthe phosphorylation state of the enzyme.Despite this wealth of structural information, the

regulation of the calcium pump in cardiac muscleremains an elusive target of study. In cardiac andsmooth muscle, SERCA is regulated by phospho-lamban (PLB), a 52 residue integral membraneprotein. PLB engages in an inhibitory interactionwith SERCA that reduces its apparent calciumaffinity. This is a dynamic process that depends onthe cytosolic calcium concentration, as well as thephosphorylation and oligomeric states of PLB, whichis in dynamic equilibrium between monomeric andhomo-oligomeric states,with pentameric forms beingdominant in SDS-PAGE.15,16 Mutation of key leucineand isoleucine residues in the transmembrane do-main of PLB destabilizes the pentameric structureand has been shown to shift this equilibrium in favorof the monomer. These pentamer-disrupting muta-tions are associated with increased inhibition ofSERCA, leading to the speculation that the PLBmonomer is the active inhibitory species,17–19 andthat the pentamer is an inactive storage form.20,21

Unfortunately, it has not been possible to test thismodel directly with well defined PLB oligomericstates within a lipid membrane. In any case, SERCAinhibition by the PLBmonomer canbe reversed eitherby elevated cytoplasmic calcium concentrations or byphosphorylation of PLB. The primary physiologicalmechanism for relieving SERCA inhibition is throughthe phosphorylation of PLB at Ser16 by cAMP-dependent protein kinase (PKA), and PLB can bephosphorylated at Thr17 by either calcium/calmod-ulin-dependent protein kinase II22 or by Akt.23

The functional effect of PLB phosphorylation onSERCA regulation is clear but the mechanism forthis effect is less certain. The original model forSERCA regulation suggested that monomeric PLB

binds to and inhibits SERCA, and phosphorylationdisrupts this inhibitory complex.24,25 However,there is contradictory evidence about whether PLBis physically dissociated from SERCA followingphosphorylation. Fluorescence energy transferexperiments suggest that PLB inhibits and aggre-gates SERCA, and that phosphorylation reversesthis process and causes dissociation of PLB andSERCA.26 Similarly, cross-linking experiments in-dicate that phosphorylation weakens the physicalassociation of PLB with SERCA and makes thecomplex more susceptible to dissociation bysub-saturating concentrations of calcium.27,28 Incontrast, studies using co-immunoprecipitation,29

fluorescence30,31 and EPR spectroscopy32 all sug-gest that PLB remains associated with SERCAfollowing phosphorylation. Rather than dissocia-tion of PLB from SERCA after phosphorylation,EPR and NMR studies point to a transition fromorder to disorder in the cytoplasmic domain ofPLB.32,33 Such a transition is consistent with theresults of a variety of biophysical studies showingthat phosphorylation causes a partial unwindingand disordering of the PLB N-terminal α-helixaround the Ser16 phosphorylation site.34–38 Regu-lation by phosphorylation is thought to occur in thecontext of a complex between monomeric PLB andSERCA yet it has been suggested that the PLBpentamer is necessary for regulation of cardiaccontractility in a physiological context;39 and adirect interaction has been proposed between thePLB pentamer and SERCA.40–42 EPR measurementsof boundary lipids suggest that phosphorylation ofPLB shifts the population towards the oligomericstate.16 These results raise questions about the roleof PLB oligomeric states in the regulation of SERCA.In earlier work, we observed a direct interaction

between an oligomeric form of PLB and SERCA in 2Dcrystals.42 Specifically, we characterized co-crystals ofSERCA and a super-inhibitory mutant of PLB(Ile40Ala; I40A).42 While SDS-PAGE indicated thatthis mutant of PLB was monomeric,17 our projectionmap revealed that I40A formed an oligomer, whichwas later supported by fluorescence resonance energytransfer experiments.43,44 These results reinforce theconclusion reached by Jones and co-workers, thatSDS-PAGEmight indicate the relative stability of PLBoligomers but it is not a definitive means of assessingthe oligomeric species adopted by PLB within thelipid bilayer.18 A 3D model based on our projectionmap suggested that PLB pentamers interact withSERCA at two potential sites; one near transmem-brane segment M3 and another near the C-terminus.These contact sites are distinct from the inhibitory siteoccupied by the PLB monomer, which is adjacent toM2, M4 and M6 of SERCA according to the results ofmutagenesis and cross-linking studies.45–49 In thisstudy, we investigated the effects of PLB phosphory-lation and mutation on the interaction between a PLB

Table 1. Apparent calcium affinity (KCa) determined forSERCA in the absence and in the presence of wild type,mutant and phosphorylated forms of PLB

KCa (μM calcium) ΔKCa

SERCA (n=28) 0.41±0.01 –PLBwt (n=9) 0.69±0.01 0.28Phospho-PLBwt (n=3) 0.43±0.02 0.02K27A (n=7) 1.00±0.03 0.59Phospho-K27A (n=3) 0.65±0.03 0.24R14A (n=3) 0.59±0.03 0.18N34A (n=6) 0.46±0.03 0.05

ΔKCa is the change in calcium concentration at half-maximalATPase activity of SERCA in the presence of the wild type,mutant and phosphorylated forms of PLB; calculated as thedifference in KCa values for SERCA in the absence and in thepresence of PLB.

709Co-crystals of SERCA and Phospholamban

oligomer and SERCA in the context of 2D crystals.Our results show a correlation between PLB functionand crystal formation, suggesting that the physicalinteractions that stabilize the crystal are sensitive tophysiologically relevant perturbations. Our data areconsistent with an order-to-disorder transition in thePLB cytoplasmic domain, where the inhibitory formsof PLB (e.g.wild type and the gain-of-functionmutantLys27Ala) retain an ordered state in the crystals andthe non-inhibitory formsof PLB (phosphorylatedwildtype and a loss-of-functionmutant Asn34Ala) adopt adisordered state.

Results

Co-reconstitution of SERCA and PLB

Methods for co-reconstituting SERCA and PLBhave been established for both functional50,51 andstructural studies.42,52,53 Co-reconstituted proteoli-posomes have been shown to have a lipid/proteinmolar ratio of approximately 120:1 and a PLB/SERCA molar ratio of 3.5:1.42,51 These conditionsmimic cardiac SR and, under the appropriate bufferconditions, they promote formation of 2D crystals.Measurements of ATPase activity have been used todemonstrate the regulatory interactions betweenwild type and mutants forms of PLB and SERCA.For the current studies, wild-type PLB, a gain-of-function mutant (Lys27Ala), a partial loss-of-function mutant (Arg14Ala), and a complete loss-of-function mutant (Asn34Ala) were chosen to repre-sent different functional forms of PLB.17,19,51,53–55

Importantly, the oligomeric stability of the mutantsused for 2D crystallization was similar to that ofwild type PLB, despite differences in their ability toregulate SERCA.17,55 Measurements of ATPaseactivity were used to determine the apparentcalcium affinity of SERCA in the absence and inthe presence of PLB (Table 1). Reconstituted SERCAin the absence of PLB had a KCa value of 0.41±0.01 μM calcium, and co-reconstituted SERCA in thepresence of wild-type PLB had a KCa value of 0.69±0.01 μM calcium. The KCa values (μM calcium) in thepresence of mutant forms of PLB were 1.00±0.03 forK27A, 0.59±0.03 for R14A and 0.46±0.03 for N34A.We tested the effect of PKAphosphorylation on the

inhibitory capacity of PLB. After co-reconstitution,phosphorylation of wild type PLB restored theapparent calcium affinity of SERCA to control levels(Fig. 1a; Table 1). However, phosphorylation of theK27A mutant by PKA did not completely restorethe apparent calciumaffinity of SERCA (Fig. 2a; Table1). The KCa values in the presence of phosphorylatedforms of PLB were 0.43±0.02 μM calcium for wildtype (nearly complete (93%) reversal of inhibition)and 0.65±0.03 μM calcium for K27A (partial (59%)reversal of inhibition). Both SDS-PAGE (Figs 1b and

2b) andmatrix-associated laser desorption ionizationtime-of-flight mass spectrometry (MALDI-TOF; datanot shown) were used to demonstrate stoichiometricphosphorylation of wild type and K27A PLB. TheR14A and the N34A mutants were not treated withPKA because the former is not recognized by PKA56

and the latter (loss-of-function) is not affected byphosphorylation. Finally, our observation that phos-phorylated K27A remains partially inhibitory (Fig.2a) is consistent with earlier work demonstrating thatphosphorylated N27C also retains its inhibitorycapacity and can be cross-linked to SERCA.28 Notethat residue 27 is lysine in the human protein (usedhere) and asparagine in the canine protein.28

Co-crystallization of SERCA and PLB

We next tested the ability of wild type, mutant andphosphorylated forms of PLB to interact withSERCA in large 2D co-crystals.42 The proteolipo-somes described above were capable of formingcrystals after treatment with decavanadate, EGTA,and a freeze-thaw procedure designed to enhanceproteoliposome fusion and crystal growth. Howev-er, the PLB mutants varied markedly in their abilityto form co-crystals. In order to characterize the effectof mutation and phosphorylation on crystal forma-tion, care was taken to ensure that the crystallizationconditions were equivalent between the varioussamples. The number of crystals in a grid square(2500 μm2) was counted in negatively stainedsamples (Table 2). Wild type PLB produced amoderate frequency of about three crystals in agrid square and other samples revealed a correlationbetween the functional state of PLB and its propen-sity to form co-crystals with SERCA. For example,the K27A gain-of-function mutant formed about fivecrystals in a grid square, and the N34A loss-of-function mutant formed about one crystal in a gridsquare. This analysis was not done for the R14Amutant. Despite changes in crystal frequency andorder, the crystal morphology and lattice parameters

Fig. 1. Co-reconstitution and co-crystallization of SERCA with non-phosphorylated and phosphorylated wild typePLB. (a) ATPase activity of SERCA reconstituted in the absence (●) and in the presence of wild type (▼) andphosphorylated wild type (◼) PLB. The calcium affinity values (KCa) are reported in the text, and all curves have beennormalized to the maximal activity (V/Vmax). (b) An example of staining with Coomassie brilliant blue after SDS-PAGE ofthe co-reconstituted proteoliposomes used in the crystallization studies. SERCA and the wild type PLB pentamer (PLB5)are labeled. Co-reconstituted proteoliposomes were run on 10% (top) and 16% (bottom) polyacrylamide gels. A fivefoldlarger amount of sample was loaded onto the 16% gel for display purposes. A characteristic shift was observed in themobility of the PLB pentamer after phosphorylation (ph-PLB5) with protein kinase A (PKA). (c) Projection map ofnegatively stained co-crystals of SERCA in the presence of wild type PLB. A single unit cell (a≈345 Å, b≈70 Å) andsymmetry operators are indicated for the p22121 plane group. The green densities indicate a single SERCA molecule,where the negative stain reveals only the cytoplasmic domain. The relative locations of the actuator (A) and nucleotide-binding (N) domains are indicated. The densities associatedwith PLB are interspersed between the SERCA dimer ribbons.(d) The projection map from negatively stained co-crystals of SERCA in the presence of phosphorylated wild-type PLB.The projection maps (c and d) are contoured showing all negative (b0; broken lines) and positive (≥0; continuous lines)densities; each contour level corresponds to 0.25 σ.

710 Co-crystals of SERCA and Phospholamban

were similar to one another and to those reportedearlier.42 In particular, all crystals exhibited p22121plane group symmetry with approximate latticedimensions of a=345 Å and b=70 Å (γ=90°).

Projection maps from negatively stained2D co-crystals

For each form of PLB, projection maps ofnegatively stained samples were calculated after

averaging Fourier data from five different crystalimages at a resolution of ~20 Å. To rule outdifferences in negative staining, two or threeprojection maps were calculated for each form ofPLB, where each map represented an independentco-reconstitution, crystallization and negative stainEM grid. Typical projection maps are shown fornon-phosphorylated and phosphorylated wild typePLB (Fig. 1). Consistent with what was observedearlier,42 the projection maps were dominated by

Fig. 2. Co-reconstitution and co-crystallization of SERCA with non-phosphorylated and phosphorylated K27A PLB.(a) ATPase activity of SERCA reconstituted in the absence (●) and in the presence of K27A (▼) and phosphorylated K27A(◼) PLB. The calcium affinity values (KCa) are given in the text, and all curves have been normalized to the maximalactivity (V/Vmax). (b) An example of a Coomassie brilliant blue-stained SDS-PAGE of the co-reconstitutedproteoliposomes used in the crystallization studies. SERCA and the K27A PLB pentamer (PLB5) are labeled. Co-reconstituted proteoliposomes were run on 10% (top) and 16% (bottom) polyacrylamide gels. A fivefold greater amountof sample was loaded onto the 16% gel for display purposes. A characteristic shift was observed in the mobility of the PLBpentamer after phosphorylation (ph-PLB5) with PKA. (c) Projection map from negatively stained co-crystals of SERCA inthe presence of K27A PLB. (d) Projection map from negatively stained co-crystals of SERCA in the presence ofphosphorylated K27A PLB. The projection maps (c and d) are contoured showing only positive (continuous lines)densities; each contour level corresponds to 0.25 σ.

Table 2. Lattice parameters and crystal propensity for negatively stained crystals of SERCA with and withoutphospholamban

p22121 lattice parameters

Characteristics a b γ Crystal frequency Crystal quality

PLB Pentamer, inhibitory 341.3±2.7 70.3±0.3 90.2±0.5 3.1±0.5 (n=7) IntermediatePhospho-PLB Pentamer, non-inhibitory 341.6±6.2 70.2±0.9 90.1±1.0 1.4±0.4 (n=7) PoorK27A Pentamer, gain-of-function 344.9±4.2 70.9±0.9 85.6±1.5 5±1 (n=6) HighPhospho-K27A Pentamer, inhibitory 339.5±1.5 71.3±0.5 90.1±0.4 1.7±0.5 (n=7) Poor to intermediateN34A Pentamer, loss-of-function 339.1±2.0 70.6±0.7 89.5±0.5 1±0.4 (n=7) Poor

Crystal frequency is the average number of crystals observed per grid square (400-mesh grids) for a minimum of six independent co-reconstitutions and crystallization trials. This was not done for the R14A mutant of PLB. For each independent co-reconstitution andcrystallization trial (n), at least 30 grid squares were examined for crystal frequency.

711Co-crystals of SERCA and Phospholamban

Fig. 3. Co-reconstitution and co-crystallization of SERCA with R14A PLB. (a) ATPase activity of SERCA reconstitutedin the absence (●) and presence of R14A (▼) PLB. The calcium affinity values (KCa) are given in the text, and all curveshave been normalized to the maximal activity (V/Vmax). (b) The projection map from negatively stained co-crystals ofSERCA in the presence of R14A PLB. The projection map is contoured showing only positive (continuous lines) densities;each contour level corresponds to 0.25 σ.

712 Co-crystals of SERCA and Phospholamban

rows of 2-fold related densities that corresponded toanti-parallel dimer arrays of SERCA. In the presenceof wild type PLB, additional density was inter-spersed between the SERCA arrays, consistent withthe presence of PLB oligomers (Fig. 1c). Thesepatches of extra density were relatively small inthe maps of negatively stained crystals, owing to thelow contrast generated by negative stain within thelipid bilayer and the small size of the cytoplasmicdomain of PLB relative to SERCA. Nonetheless, thePLB density in the projection maps was similar tothose reported earlier.42 Upon phosphorylation ofSer16 with PKA, the PLB density was no longerpresent, consistent with disordering of the cytoplas-mic domain (Fig. 1d). As mentioned above, phos-phorylation of wild type PLB also reduced thefrequency of co-crystal formation (Table 2).We studied the effect of phosphorylation on the

co-crystals with K27A PLB, which has the sameoligomeric state as wild type PLB.17,55 Like wildtype PLB, the projection maps of non-phosphory-lated K27A co-crystals show PLB density lyingbetween the anti-parallel dimer ribbons of SERCA(Fig. 2). The frequency of the K27A co-crystals ishigher than that of the wild type, reflecting the factthat the K27A mutation produces a super-inhibitoryPLB molecule (gain of function). Interestingly,

phosphorylation of K27A had no effect on the PLBdensity (compare Figs. 1d and 2d). This behavior isconsistent with the fact that stoichiometric phos-phorylation of K27A is unable to fully reverse itsinhibition of SERCA (Fig. 2a and b). We reasonedthat if phosphorylation disrupted a crystal contactor altered contrast produced by negative stain (e.g.,by adding a negative charge to the cytoplasmicdomain), then phosphorylation of K27A co-crystalsshould produce a result similar to the wild type PLB(Fig. 1c and d). However, if co-crystallization relieson a functional interaction between PLB andSERCA, then the inhibitory properties of thephosphorylated K27A mutant should correlatewith its behavior during co-crystallization. Basedon our observations, we believe that the cytoplasmicdomain of wild type PLB becomes disordered uponphosphorylation, whereas the cytoplasmic domainof K27A does not. This comparison rules out thepossibility that the disappearance of the PLB densitywas a result of simply adding the negative charge(PO4

2–) to its cytoplasmic domain.

Additional mutants of PLB

Since the phosphorylation of wild type PLBreverses SERCA inhibition and alters density in the

Fig. 4. Co-reconstitution and co-crystallization of SERCA with N34A PLB. (a) ATPase activity of SERCA reconstitutedin the absence (●) and in the presence of N34A (▼) PLB. The calcium affinity values (KCa) are given in the text, and allcurves have been normalized to the maximal activity (V/Vmax). (b) Projection map from negatively stained co-crystals ofSERCA in the presence of N34A PLB. The projection map is contoured showing all negative (broken lines) and positive(continuous lines) densities; each contour level corresponds to 0.25 σ.

713Co-crystals of SERCA and Phospholamban

2D co-crystals,we tested the effects of (i)mutation of acharged residue (Arg14) proximal to the phosphory-lation site (Ser16) and (ii) a well-characterized loss-of-functionmutant (N34A). As reported earlier,17,56 R14Aretains substantial inhibitory capacity (Fig. 3a), where-asN34A is a complete loss-of-functionmutant (Fig. 4a).The R14A mutant changes the net charge of the PLBcytoplasmic domain adjacent to the phosphorylatedresidue Ser16, while the N34Amutant is located at theinterface between the membrane and the cytosol. Inprojection maps of negatively stained co-crystals withR14A PLB, additional density was observed betweenthe rows of SERCA molecules (Fig. 3b). However, inco-crystals with N34A PLB, the additional density wasabsent (Fig. 4b). These results further support the ideathat the functional state, rather than the net charge ofthe cytoplasmic domain, determines the stability ofco-crystals and whether PLB-associated densities arevisible in the projection maps. Interestingly, the N34Amutation appears to cause a disordering of thecytoplasmic domain that is similar to the effect ofphosphorylation of the wild type PLB.

Projection maps from frozen-hydrated2D co-crystals

Our observations of negatively stained crystalsindicated that the co-crystallization of PLB and

SERCA was negatively impacted by PLB phospho-rylation and that the cytoplasmic domain of PLBtended to become disordered. To test if thisdisordering affected the transmembrane domainof PLB, we imaged the co-crystals in the frozen-hydrated, unstained state. Based on molecularmodels for PLB,41,57,58 the density associated withPLB in projection maps from frozen-hydrated co-crystals is dominated by the pentameric transmem-brane coiled coil. This is due to the alignment of thetransmembrane helices along the imaging direction,which produces very strong density in thecorresponding projection maps. Therefore, compar-ison of the density observed in negatively stainedco-crystals with that observed in frozen-hydratedcrystals allows us to evaluate the relative orderingof the cytoplasmic and transmembrane domains,respectively. In particular, if phosphorylation dis-orders primarily the cytoplasmic domain of PLB aspredicted,32 density attributable to PLB should beweak in the negatively stained samples andunaffected in frozen-hydrated samples. For thiscomparison, projection maps were calculated fromimages of frozen-hydrated co-crystals with wildtype PLB before and after PKA phosphorylation.Following merging and averaging of data from atleast five crystal images, diffraction amplitudeswith high signal-to-noise ratios and low phase

Table 3. Summary of crystallographic data for frozen-hydrated co-crystals of SERCA and phospholamban

Phospholamban

I40Aa K27A

Number of images 5 15Cell parameters

a (Å) 359.2 346.5b (Å) 71.9 70.7γ (°) 90.3 89.7

Overall weighted phase residual (°)b 16.8 16.1a Data from Stokes et al. (2006).42b Including data to IQ 7.

714 Co-crystals of SERCA and Phospholamban

residuals were observed for all resolution shells to aresolution of 10 Å (Table 3). The resulting mapswere nearly identical, both containing anti-paralleldimer ribbons of SERCA molecules interspersedwith density consistent with pentameric PLB (Fig. 5).This similarity suggests that intramembrane inter-actions between SERCA and the transmembranedomain of PLB mediate contacts in the crystalsand that these contacts persist after phosphoryla-tion of the PLB cytoplasmic domain. Loss ofdensity for these cytoplasmic domains suggeststhat they are disordered and thus not stronglybound to SERCA after phosphorylation. Given thenegative effect of phosphorylation on crystal order,the PLB cytoplasmic domain probably provides

Fig. 5. Projection maps from frozen-hydrated co-crystalsphosphorylated wild type PLB. Statistics for merging five cryresolution of approximately 10 Å. The region of the map shownorientations of SERCAmolecules are indicated by arrows and tThe projection maps are contoured showing only positive (co

additional interactions in the non-phosphorylatedstate.

What is the oligomeric state of PLB in theco-crystals?

To further characterize the physical interactionbetween SERCA and PLB and to evaluate itsoligomeric state in the co-crystals, we used frozen-hydrated preparations to improve the resolution ofthe existing projection map.42 We chose to image co-crystals of K27A PLB because the relative abun-dance and high quality of these crystals facilitateddata collection. Images from frozen-hydrated co-crystals displayed computed diffraction to a resolu-tion of approximately 15 Å. Following merging andaveraging of data from 15 crystal images, diffractionamplitudes with high signal-to-noise ratios and lowphase residuals were observed for all resolutionshells to a resolution of 8 Å (Fig. 6; Table 3).The resulting projection map for SERCA in the

presence of K27A PLB is similar to that deter-mined for SERCA in the presence of I40A PLB.42

The size and shape of the additional densitiesseen in our projection maps were consistent withthe pentamer, which is the principle oligomericform of PLB observed by SDS-PAGE. However,neither map from frozen-hydrated co-crystals(I40A reported earlier42 or K27A reported here)provided direct evidence of pentameric assembly

of SERCA in the presence of (a) wild type and (b)stal images indicated that phase residuals were b35° to ais approximately 600 Å×159 Å. The relative locations andhe locations of the PLB densities are indicated by brackets.ntinuous lines) densities.

Fig. 6. Projection map from frozen-hydrated co-crystals of SERCA in the presence of K27A PLB. (a) Statistics formerging 15 crystal images indicate that phase residuals are b26° to a resolution of 8 Å. (b) Projection map generated forco-crystals of SERCA and K27A PLB. The region of the map shown is approximately 600 Å×159 Å. The PLB densities aresimilar to those characterized for I40A PLB,42 reflecting an identical oligomeric state andmode of interaction with SERCA.The projection map is contoured showing only positive (continuous lines) densities.

715Co-crystals of SERCA and Phospholamban

of PLB, presumably due to limited resolution. Toaddress this issue, the high-resolution terms ofthe projection map were enhanced by applyinga negative B-factor (temperature factor) of500 Å– 2.59 The truncation of Fourier data at 8 Åresolution ensured that the contribution of noisein our map was minimal. This procedure im-proved contrast and detail for the densitiesassociated with both SERCA and PLB (Fig. 7).Significantly, the density assigned to the PLBoligomer resolved into a five-lobed density con-sistent with the PLB pentamer. However, ratherthan the symmetric structure predicted by NMRstructural models,41,58 our density resembles adistorted pentamer (Fig. 7b–d), where two of thefive observed lobes are stronger and betterdelineated than the others. Although the shapeof the pentamer might be affected by theincreased noise in the “sharpened” projectionmap, a physical distortion of the pentamerwould be consistent with its asymmetric interac-tion with SERCA. Interestingly, the strongerdensities in the pentamer are proximal toSERCA, and one appears next to transmembranesegment M3 (asterisk in Fig. 7a). An interactionbetween M3 of SERCA and PLB was alsosuggested by our earlier studies of the complexusing electron cryo-microscopy of helical crystals(Fig. 4 in 53).

Discussion

Physical interactions between SERCA and PLB

There is general consensus that the monomericform of PLB interacts with M2, M4 and M6 ofSERCA,17,19,20,25,48,49 producing the inhibition thatcharacterizes the resting state of cardiac muscle.However, oligomeric forms of PLB have beenobserved repeatedly both in detergent and mem-branous environments. While the pentamer appearsto be the most stable oligomer, tetramers, trimersand dimers have all been observed and the balancebetween them can be influenced by single-sitemutations.17,19 Furthermore, there is dynamic ex-change between oligomeric andmonomeric forms ofPLB and evidence that phosphorylation of PLB orincreased cytosolic calcium concentrations increasethe proportion of the pentameric pool at the expenseof the monomeric pool.16 This evidence has led tothe hypothesis that the monomer represents the“active” inhibitory form and that oligomeric formsrepresent an inactive reserve.20,21 While severalcongruent molecular models explaining the inhibi-tory properties of PLB are described in theliterature,47,48,60,61 there are a number of inconsis-tencies in published observations that are notfully explained. First, Kranias and colleagues

Fig. 7. The projection map recalculated with an applied negative B-factor. (a) In the sharpened projection map, thecontrast and level of detail is enhanced for both SERCA and PLB. The asterisk (⁎) indicates the region of closest contactbetween the PLB pentamer and transmembrane segment M3 of SERCA. (b) A close up view of the density associated withPLB. The size and shape of the PLB densities are now compatible with a pentamer. (c) A simulated projection for atransmembrane pentamer is shown for comparison. (d) Superimposition of the maps shown in (b) and (c), indicating aslightly distorted pentameric arrangement.

716 Co-crystals of SERCA and Phospholamban

demonstrated that a mutant form of PLB (Cys41Phe)was insufficient for proper regulation of cardiaccontractility in a mouse model.39 Because thismutant was reported to be monomeric with thesame inhibitory potency as the wild type protein,62

the authors inferred a physiological role for the PLBpentamer. Despite the fact that the oligomeric stateof the C41F mutant has not been demonstrateddirectly in a membranous environment, this notionis consistent with independent observations thatSERCA might interact with oligomeric forms ofPLB.40–42,63 Thus, the existence of oligomers appearsto offer a functional advantage for the SERCA–PLBinteraction,40 and this advantage is not explained bycurrent molecular models (i.e., PLB oligomers asinactive storage forms). Second, there are disparate

observations on how the physical associationbetween SERCA and PLB responds to phosphory-lation. One group of studies postulate that PLBremains associated with SERCA and phosphoryla-tion alters the structural interaction between the twoproteins,29,30–32,64 whereas contradictory crosslink-ing studies have led to the notion that intermolec-ular interactions at the M2, M4 and M6 interface arelost following phosphorylation27,28 and PLB dis-sociates from SERCA.More specifically, chemical crosslinking experi-

ments indicated that PKA-mediated phosphoryla-tion decreases the efficiency of crosslinking toSERCA at multiple sites distributed throughoutboth the cytoplasmic and the transmembranedomains of PLB.27,28 The inference then is that

717Co-crystals of SERCA and Phospholamban

phosphorylation decreases binding affinity of PLB forSERCA and causes it to dissociate at sub-saturatingconcentrations of calcium. Contradictory evidencein favor of persistent association comes from thefollowing literature. Antibodies recognizingPLB phosphorylated at Ser16 were shown to co-immunoprecipitate SERCA1a and SERCA2a afterco-expression inHEK-293 cells.29 Fluorescence, spin-label EPR spectroscopy and other biophysicalstudies demonstrated that SERCA restricts themotional freedom of PLB both before and afterphosphorylation.30,31 More recent EPR studies alsosupport persistent association and suggest thatphosphorylation induces a dynamically disorderedstate in the PLB cytoplasmic domain and supportedthe idea that phosphorylated PLB remains associatedwith SERCA in a non-inhibitory state.32 Despite thisgrowing body of evidence, it is difficult to envisionhow PLBmight remain bound to theM2,M4 andM6interface of SERCA, given the large conformationalchanges caused by calcium binding that occlude thisinteraction interface.These apparent contradictions can be reconciled

by hypothesizing a physical interaction betweenSERCA and oligomeric PLB at a secondary, non-inhibitory site. Specifically, we have observed aninteraction between M3 of SERCA and the PLBpentamer in the 2D co-crystals. Such an interactioncould explain the mutual effects that SERCA andPLB have on spectroscopic analysis of each other'saggregation state42,63 and could provide a direct rolefor the PLB oligomer under physiologicalconditions.39 In addition, distinct functional con-sequences and binding sites of the monomer andoligomer could explain the experimental discrepan-cy regarding the persistence of the interaction. ThePLB monomer might dissociate from the M2, M4and M6 inhibitory site under the appropriatephysiological stimuli and there could be a highprobability for interaction with the PLB oligomer atthe secondary site on the other side of the SERCAmolecule (e.g., M3). Compared to the dramaticmovements of M2, M4 and M6 during calciumbinding, the transmembrane helix M3 of SERCA isless mobile and might represent a stable interactionpoint for a PLB oligomer that is insensitive tophosphorylation or the level of cytosolic calcium.In the present work, our data support a function-

ally relevant interaction between SERCA and a PLBoligomer in 2D co-crystals. In particular, the effectsof PLB mutants on crystallization are correlatedwith their effects on the inhibition of SERCA. TheK27A and N34A mutations were chosen for thesimilar stability of their pentameric form, yet widelydifferent inhibitory behavior. Thus, it is significantthat the gain-of-function K27A mutation enhancedcrystallization, whereas the loss-of-function N34Amutation interfered with crystallization. We con-clude that the structural properties of PLB that

govern the association of the monomer with theinhibitory site of SERCA are directly related to thoseused in the interaction of a PLB oligomer at thesecondary, accessory site.

Effect of phosphorylation on physicalinteraction between SERCA and PLB

In studies of the PLB monomer, there is someconsistent evidence that phosphorylation of Ser16causes localized changes in the structure of itscytoplasmic domain.32,33,35,65,66 These results pro-vide a physical basis for disruption of the productivestructural interaction between SERCA and PLB,which might lead to dissociation of the complex.Early NMR and CD spectroscopy studies showed acontinuous α-helix in the N-terminal portion of PLB(residues 1–16), which partially unwound uponphosphorylation (residues 12–16 were no longerhelical).35 However, these studies utilized anN-terminal fragment of PLB (residues 1–25), ratherthan the full-length protein, and did not includeSERCA. Since then, there have been many struc-tural models for full-length PLB based on NMRmeasurements in the non-phosphorylated,41,61,67–71

phosphorylated65 and pseudo-phosphorylated72

state under a variety of experimental conditions.These models differ in the amount of secondarystructure in the N-terminal cytoplasmic domain ofPLB, suggesting that this domain can adopt a varietyof conformational states. Furthermore, most studiesagree that phosphorylation alters the dynamicsof this domain, the NMR structure of a pseudo-phosphorylated form of PLB notwithstanding.72

A variety of biophysical studies indicate thatPLB undergoes a conformational change uponphosphorylation.12,34,35,73 Many of these ideas coa-lesced in recent EPR32 and NMR33,66 studies, whichproposed an order-to-disorder transition in thecytoplasmic domain of PLB that disrupts intermo-lecular contacts and reverses SERCA inhibition, butdoes not dissociate the SERCA–PLB complex.Consistent with this idea, we found that phos-

phorylation of wild type PLB at Ser16 selectivelydisordered the cytoplasmic domains of the penta-mer (Fig. 1) and reduced its ability to mediate 2Dcrystallization with SERCA (Table 2). In maps fromnegatively stained crystals, density attributable tothe cytoplasmic domain disappeared when PLB wasphosphorylated, whereas maps from frozen-hydrat-ed crystals showed that the transmembrane helicesof PLBwere still clearly visible. We conclude that thetransmembrane domain of phosphorylated PLBremains associated with SERCA in the co-crystals,even though the cytoplasmic domain becamedisordered. This disordering did not occur for theK27A gain-of-function mutant (Fig. 2) but it didoccur for the N34A loss-of-function mutant, even in

Fig. 8. A partial sequence alignment and a diagram forthe interaction of PLB with SERCA. Upper: The potentialsequence similarity between the transmembrane domainof PLB (residues 32–52) and transmembrane segment M3of SERCA (residues 254–274). Leu44, Ile47 andLeu51makeup part of the leucine-isoleucine zipper that stabilizes thepentameric state of PLB. Lower: The pentamer andmonomer are in dynamic equilibrium,where themonomeris postulated to interact with and inhibit SERCA (right-hand side). We hypothesize that the pentamer alsointeracts with SERCA,which leads to the active associationor dissociation of a monomer (left-hand side). The activedissociation of a monomer leads to a physical interactionwith and inhibition of SERCA and this process is reversedby phosphorylation of PLB. These two pathways are notmutually exclusive and might operate simultaneously orunder disparate physiological conditions.

718 Co-crystals of SERCA and Phospholamban

the non-phosphorylated state (Fig. 4). Furthermore,the propensity to induce co-crystallization wasretained by the K27A mutant after phosphorylation,whereas the N34A mutant was marginal in co-crystallization, even in the non-phosphorylatedstate (Table 2). It is interesting to consider wherethese three residues lie in the structure of PLB. Ser16flanks the N-terminal end of domain Ib, Lys27 isfound in the middle of this domain and Asn34 flanksthe C-terminal end. These residues also have twoopposed functional effects; PLB function is lostwhen phosphorylated at Ser16 or mutated atAsn34 and PLB function is enhanced when mutatedat Lys27. Thus, domain Ib might be the keystructural element responsible for phosphorylation-or mutation-dependent conformational changes inthe PLB cytoplasmic domain that impact SERCAinhibition. Indeed, many of the structural differ-ences between existing PLB models involvedomain Ib,41,61,65,67–72 suggesting that this domainmight be a less structured, more dynamic regionof the protein. Therefore, it is reasonable tosuppose that phosphorylation alters the dynamicsof domain Ib, thereby controlling an order-to-disorder transition in the PLB cytoplasmic domainand reversing SERCA inhibition.29,31,32

Model for the interaction of the PLB pentamer

The major findings in this study are that the PLBpentamer is capable of a physical interaction withSERCA and that this interaction is sensitive tofunctional modification of PLB through phospho-rylation or mutation. In particular, PKA-mediatedphosphorylation of Ser16 caused the cytoplasmicdomain of PLB to become disordered, yet thepentamer remained associated with SERCA throughintramembrane interactions. This disordering oc-curred also for a well characterized loss-of-functionmutation, Asn34Ala, suggesting that the loss ofinhibition might occur through a molecular mech-anism similar to phosphorylation. These data areconsistent with a functional interaction betweenSERCA and PLB oligomers,40 and inconsistent withthe notion of PLB oligomers as inactive storageforms.21 The inhibitory site involving the PLBmonomer and transmembrane helices M2, M4 andM6 of SERCA45,48,61 is distinct from the physicalinteractions that stabilize our 2D crystals, whichcenter on M3 of SERCA and involve the transmem-brane domain of PLB. The inhibitory site of SERCAalternately opens and closes during the calciumtransport cycle, owing to large movements oftransmembrane helices M2, M4 and M6. This isnot the case for the accessory site of SERCA, becauseM3 is less mobile during the transport cycle. Thus,M3 could act as an interaction point between SERCAand the PLB pentamer (and perhaps other oligo-meric forms). Interestingly, a primary structure

comparison between SERCA and PLB reveals aregion of potential sequence similarity that spans theC-terminal portion of the PLB transmembrane helixand M3 of SERCA (Fig. 8). Residues Leu266, Val269and Leu273 of SERCA face the lipid environment

719Co-crystals of SERCA and Phospholamban

and adopt a side chain orientation similar to that ofLeu44, Ile47 and Leu51 of PLB. These residues formpart of the leucine-isoleucine zipper that stabilizesthe pentameric form of PLB.74

Through its interaction with M3 of SERCA, thePLB pentamer might play a more active role incapturing monomeric or phosphorylated PLB spe-cies. Specifically, this SERCA–pentamer interactionmight facilitate the exchange of PLB monomerswith the SERCA inhibitory site in response tophysiological cues (elevated cytosolic calcium and/or phosphorylation). This interaction would alsoexplain how SERCA is able to influence theoligomeric state of PLB.63 Based on their NMRstructure of a PLB pentamer, Oxenoid and Chousuggested that individual PLB monomers couldinitiate binding to SERCA without dissociationfrom the pentamer.41 We suggest a modificationof this hypothesis, where a PLB pentamer interactswith the membrane domain of SERCA and servesas a reservoir for directed diffusion of monomers toand from the inhibitory site (Fig. 8). This pathwaymight be important in efficiently delivering mono-meric PLB to its binding site on SERCA, ensuringthat the inhibited state is maintained following acycle of calcium release in cardiac muscle. Specif-ically, a site for the PLB pentamer on SERCA mightposition the monomer in a conformation that iscompatible with formation of the inhibitory com-plex, as suggested earlier.41 Furthermore, theresultant depolymerization of the PLB pentamerwould leave behind a tetramer that might remainassociated with the pump. There is ample evidenceof intermediate oligomeric states for PLB50 andsuch intermediates could be poised to reacquire aPLB monomer once it dissociates from the inhibi-tory complex with SERCA. Finally, phosphoryla-tion has been shown to increase the oligomericpropensity of PLB16 and an interaction of thesespecies with the pump could ensure a rapid returnto the inhibited state upon dephosphorylation.Otherwise, if allowed to diffuse away during aperiod of β-adrenergic stimulation, there could be asubstantial delay before SERCA randomly encoun-ters a non-phosphorylated PLB molecule within themembrane plane.

Materials and Methods

Octaethylene glycol monododecyl ether (C12E8) wasobtained from Barnet Products (Englewood Cliff, NJ).SM-2 Biobeads were obtained from Bio-Rad (Hercules,CA). Egg yolk phosphatidylcholine (EYPC), egg yolkphosphatidylethanolamine (EYPE) and egg yolk phos-phatidic acid (EYPA) were obtained from Avanti PolarLipids (Alabaster, AL). All reagents used in the coupledenzyme assay for measuring ATPase activity were of thehighest purity available (Sigma-Aldrich, Oakville, ONCanada).75

Reconstitution of SERCA with PLB

SERCA was prepared from rabbit hind leg muscle76 byaffinity chromatography.9 Recombinant human PLB wasprepared as described.77 The followingproteinswere used inthis study: wild type PLB, a gain-of-function mutantLys27Ala (K27A), a partial loss-of-functionmutantArg14Ala(R14A), and a loss-of-functionmutantAsn34Ala (N34A). Co-reconstitution of SERCA with PLB followed establishedmethods for the formation of large 2D co-crystals.42 Briefly,100 μg of PLB and 600 μg of lipids (EYPC/EYPE/ EYPA in aweight ratio of 8:1:1) were solubilized in a chloroform-trifluoroethanol mixture, dried to a thin film under nitrogengas and lyophilized. Buffer (20 mM imidazole, pH 7.0,100mMKCl, 0.02% (w/v)NaN3) anddetergent (C12E8)wereadded to solubilize the mixture, followed by the addition of500 μg of detergent-solubilized, purified SERCA. It has beenshown that this method of co-reconstitution yields completerecovery of SERCA transport activity andPLB inhibition.50,78

The final concentrations were adjusted to a 1:1:2 weight ratioof protein/lipid/detergents (final molar ratio of SERCA/PLB/lipids of approximately1:3.5:180). The detergent wasremoved by the slow addition of SM-2 Biobeads (25 mg ofwet beads) over 4 h. For the best results, crystallization wasperformed immediately on the co-reconstituted proteolipo-somes containing SERCA and PLB. Reconstituted proteoli-posomes containing SERCA in the absence of PLB wereprepared simultaneously under identical conditions. ATPaseactivity of the proteoliposomes was measured by a coupled-enzyme assay over a range of calcium concentrations of0.1 μM–10 μM.51,75 The apparent calcium affinity, Kca, wasdetermined by fitting the data to the Hill equation usingSigma Plot software (SPSS Inc., Chicago, IL). The functionalcharacterization of these mutants has been described byothers17,55 as well as by our laboratory.51

For studies of the effect of PLBphosphorylation,wild typeandK27A PLBwere solubilized in detergent and phosphor-ylated with the catalytic subunit of PKA; Sigma-Aldrich, St.Louis, MO) before co-reconstitution with SERCA intoproteoliposomes. PKA was added to a concentration of100 units/mg of detergent-solubilized PLB and the reactionwas incubated at 30 °C for 3 h. It was concluded that thistreatment resulted in complete phosphorylation of PLB,because theunphosphorylated protein couldnot be detectedeither by MALDI-TOF mass spectrometry or by SDS-PAGEand western blotting (data not shown).

2D crystallization

Co-reconstituted proteoliposomes were collected by cen-trifugation in crystallization buffer (20 mM imidazole, pH7.4, 100 mM KCl, 35 mM MgCl2, 0.5 mM EGTA, 0.25 mMNa3VO4, 30 μM thapsigargin).79 The pellet was subjected totwo freeze-thaw cycles, suspended with a micropipette,followed by two more freeze-thaw cycles. Reconstitutedsampleswere incubated at 4 °C for up tooneweek.Althoughcrystallization occurred quickly, three to five days wasoptimal for the highest frequency and quality of 2D crystals.

Electron microscopy

Crystals were imaged in a Tecnai F20 electron micro-scope (FEI Company, Einhoven, Netherlands) in the

720 Co-crystals of SERCA and Phospholamban

Microscopy and Imaging Facility (University of Calgary)or a JEOL 2200FS electron microscope (JEOL Ltd., Tokyo,Japan) in the Electron Microscopy Facility (NationalInstitute for Nanotechnology, University of Alberta andNational Research Council of Canada). Both microscopeswere operated at 200 kV. A standard room temperatureholder was used for negatively stained samples and aGatan 626 cryoholder (Gatan Inc., Pleasanton, CA) wasused for frozen-hydrated samples. Low-dose images wererecorded either on film at a magnification of 50,000×(Tecnai F20) or image plates at a magnification of 35,800×(JEOL 2200FS). For film, the best images were digitized at6.35 μm/pixel with a Nikon Super Coolscan 9000 followedby pixel averaging to achieve a final resolution of 2.54 Å/pixel. The image plates were scanned at 15 μmpixel for afinal resolution of 4.44 Å/pixel. All data were recordedwith defocus levels of 0.5–2 μm, with an emphasis on low-defocus images (0.5 μm and 1 μm) for the frozen-hydratedsamples.Projection maps were determined using the MRC image

processing suite.80 Two rounds of lattice unbending weredone before extracting amplitudes and phases from eachimage. Data from frozen-hydrated crystals were correctedfor the contrast transfer function after estimating defocuslevels using the program PLTCTFX.81 Data from nega-tively stained crystals were not corrected for the contrasttransfer function. Common phase origins for mergingwere determined in the p22121 plane group using theprogram ORIGTILT (considering reflections with signal-to-noise ratio (IQ) b4). For averaging, data were weightedaccording to the IQ including data with IQ b7, and thecorresponding phase residuals represent the inverse cosineof the figure of merit from this averaging. Projection mapswere determined by Fourier synthesis from the averageddata using the CCP4 software suite,82 followed by normal-ization of density levels to enable comparison of projectionmaps originating from different SERCA-PLB samples.

Acknowledgements

This work was supported by a grant from theCanadian Institutes of Health Research to HSY(MOP53306) and a grant from NIH (GM56960) toDLS. JPG is the recipient of a Canada GraduateScholarship Doctoral Award from the CanadianInstitutes of Health Research and the AlbertaIngenuity Fund. HSY is a Senior Scholar of theAlberta Heritage Foundation for Medical Research.

References

1. Toyoshima, C., Nakasako, M., Nomura, H. & Ogawa,H. (2000). Crystal structure of the calcium pump ofsarcoplasmic reticulum at 2.6 Å resolution. Nature,405, 647–655.

2. Toyoshima, C. & Nomura, H. (2002). Structuralchanges in the calcium pump accompanying thedissociation of calcium. Nature, 418, 605–611.

3. Toyoshima, C. & Mizutani, T. (2004). Crystal structureof the calcium pump with a bound ATP analogue.Nature, 430, 529–535.

4. Toyoshima, C., Nomura, H. & Tsuda, T. (2004).Lumenal gating mechanism revealed in calciumpump crystal structures with phosphate analogues.Nature, 432, 361–368.

5. Sorensen, T., Moller, J. & Nissen, P. (2004). Phosphoryltransfer and calcium ion occlusion in the calciumpump. Science, 304, 1672–1675.

6. Olesen, C., Picard, M., Winther, A. M., Gyrup, C.,Morth, J. P., Oxvig, C. et al. (2007). The structural basisof calcium transport by the calcium pump. Nature,450, 1036–1042.

7. Toyoshima, C., Norimatsu, Y., Iwasawa, S., Tsuda, T.& Ogawa, H. (2007). How processing of aspartylpho-sphate is coupled to lumenal gating of the ionpathway in the calcium pump. Proc. Natl Acad. Sci.USA, 104, 19831–19836.

8. Olesen, C., Sorensen, T., Nielsen, R., Moller, J. &Nissen, P. (2004). Dephosphorylation of the calciumpump coupled to counterion occlusion. Science, 306,2251–2255.

9. Stokes, D. L. & Green, N. M. (1990). Three-dimensional crystals of Ca-ATPase from sarcoplasmicreticulum: symmetry and molecular packing. Biophys.J. 57, 1–14.

10. Stokes, D. L. & Green, N. M. (1990). Structure ofCaATPase: electron microscopy of frozen-hydratedcrystals at 6 Å resolution in projection. J. Mol. Biol. 213,529–538.

11. Shi, D., Hsiung, H. H., Pace, R. C. & Stokes, D. L.(1995). Preparation and analysis of large, flat crystalsof Ca2+-ATPase for electron crystallography. Biophys.J. 68, 1152–1162.

12. Li, M., Cornea, R. L., Autry, J. M., Jones, L. R. &Thomas, D. D. (1998). Phosphorylation-induced struc-tural change in phospholamban and its mutants,detected by intrinsic fluorescence. Biochemistry, 37,7869–7877.

13. Ogawa, H., Stokes, D. L., Sasabe, H. & Toyoshima, C.(1998). Structure of Ca2+ pump of sarcoplasmicreticulum: a view along the lipid bilayer at 9-Åresolution. Biophys. J. 75, 41–52.

14. Xu, C., Rice, W., He, W. & Stokes, D. (2002). Astructural model for the catalytic cycle of Ca2+-ATPase. J. Mol. Biol. 316, 201–211.

15. Jones, L. R., Simmerman, H. K. B., Wilson,W.W., Gurd,F. R. N. & Wegener, A. D. (1985). Purification andcharacterization of phospholamban from canine cardiacsarcoplasmic reticulum. J. Biol. Chem. 260, 7721–7730.

16. Cornea, R. L., Jones, L. R., Autry, J. M. & Thomas,D. D. (1997). Mutation and phosphorylation changethe oligomeric structure of phospholamban in lipidbilayers. Biochemistry, 36, 2960–2967.

17. Kimura, Y., Kurzydlowski, K., Tada,M.&MacLennan,D. H. (1997). Phospholamban inhibitory function isenhanced by depolymerization. J. Biol. Chem. 272,15061–15064.

18. Cornea, R., Autry, J., Chen, Z. & Jones, L. (2000).Re-examination of the role of the leucine/isoleucinezipper residues of phospholamban in inhibition of theCa2+-pump of cardiac sarcoplasmic reticulum. J. Biol.Chem. 275, 41487–41494.

721Co-crystals of SERCA and Phospholamban

19. Autry, J. & Jones, L. (1997). Functional co-expression ofthe canine cardiac Ca2+ pump and phospholamban inSpodoptera frugiperda (Sf21) cells reveals new insights onATPase regulation. J. Biol. Chem. 272, 15872–15880.

20. MacLennan, D. & Kranias, E. (2003). Phospholamban:a crucial regulator of cardiac contractility. Nat. Rev.Mol. Cell Biol. 4, 566–577.

21. Becucci, L., Cembran, A., Karim, C. B., Thomas, D. D.,Guidelli, R., Gao, J. & Veglia, G. (2009). On thefunction of pentameric phospholamban: ion channelor storage form. Biophys. J. 96, L60–L62.

22. Wegener, A. D., Simmerman, H. K. B., Lindemann, J. P.& Jones, L. R. (1989). Phospholamban phosphorylationin intact vesicles. Phosphorylation of serine 16 andthreonine 17 in response to beta-adrenergic stimula-tion. J. Biol. Chem. 264, 11468–11474.

23. Catalucci, D., Latronico, M. V., Ceci, M., Rusconi, F.,Young, H. S., Gallo, P. et al. (2009). Akt increasessarcoplasmic reticulum Ca2+ cycling by direct phos-phorylation of phospholamban at Thr17. J. Biol. Chem.284, 28180–28187.

24. Stokes, D. (1997). Keeping calcium in its place: Ca2+-ATPase and phospholamban. Curr. Opin. Struct. Biol.7, 550–556.

25. MacLennan, D. H., Kimura, Y. & Toyofuku, T. (1998).Sites of regulatory interaction between calciumATPases and phospholamban. Ann. N. Y. Acad. Sci.853, 31–42.

26. Thomas, D., Reddy, L., Karim, C., Li, M., Cornea, R.,Autry, J. et al. (1998). Direct spectroscopic detection ofmolecular dynamics and interactions of the calciumpump and phospholamban. Ann. N. Y. Acad. Sci. 853,186–194.

27. James, P., Inui, M., Tada, M., Chiesi, M. & Carafoli, E.(1989). Nature and site of phospholamban regulationof the Ca2+ pump of sarcoplasmic reticulum. Nature,342, 90–92.

28. Chen, Z., Akin, B. L. & Jones, L. R. (2007). Mechanismof reversal of phospholamban inhibition of the cardiacCa2+-ATPase by protein kinase A and by anti-phospholamban monoclonal antibody 2D12. J. Biol.Chem. 282, 20968–20976.

29. Asahi, M., McKenna, E., Kurzydlowski, K., Tada, M.& MacLennan, D. (2000). Physical interactions be-tween phospholamban and sarco(endo)plasmic retic-ulum Ca2+-ATPases are dissociated by elevated Ca2+,but not by phospholamban phosphorylation, vana-date, or thapsigargin, and are enhanced by ATP.J. Biol. Chem. 275, 15034–15038.

30. Fowler, C., Huggins, J. P., Hall, C., Restall, C. J. &Chapman, D. (1989). The effects of calcium, temper-ature, and phospholamban phosphorylation on thedynamics of the calcium-stimulated ATPase of caninecardiac sarcoplasmic reticulum. Biochim. Biophys. Acta,980, 348–356.

31. Negash, S., Yao, Q., Sun, H., Li, J., Bigelow, D. J. &Squier, T. C. (2000). Phospholamban remains associ-ated with the Ca2+- and Mg2+-dependent ATPasefollowing phosphorylation by cAMP-dependent pro-tein kinase. Biochem. J. 351, 195–205.

32. Karim,C. B., Zhang,Z.,Howard, E. C., Torgersen, K.D.& Thomas, D. D. (2006). Phosphorylation-dependentconformational switch in spin-labeled phospholambanbound to SERCA. J. Mol. Biol. 358, 1032–1040.

33. Metcalfe, E. E., Traaseth, N. J. & Veglia, G. (2005).Serine 16 phosphorylation induces an order-to-disorder transition in monomeric phospholamban.Biochemistry, 44, 4386–4396.

34. Terzi, E., Poteur, L. & Trifilieff, E. (1992). Evidence fora phosphorylation-induced conformational change inphospholamban cytoplasmic domain by CD analysis.FEBS Lett. 309, 413–416.

35. Mortishire-Smith, R. J., Pitzenberger, S.M., Burke, C. J.,Middaugh, C. R., Garsky, V. M. & Johnson, R. G.(1995). Solution structure of the cytoplasmic domain ofphospholamban: phosphorylation leads to a localperturbation of secondary structure. Biochemistry, 34,7603–7613.

36. Tatulian, S. A., Jones, L. R., Reddy, L. G., Stokes, D. L.& Tamm, L. K. (1995). Secondary structure andorientation of phospholamban reconstituted in sup-ported bilayers from polarized attenuated totalreflection FTIR spectroscopy. Biochemistry, 34,4448–4456.

37. Pollesello, P. & Annila, A. (2002). Structure of the 1-36N-terminal fragment of human phospholambanphosphorylated at Ser-16 and Thr-17. Biophys. J. 83,484–490.

38. Paterlini, M. G. & Thomas, D. D. (2005). The alpha-helical propensity of the cytoplasmic domain ofphospholamban: a molecular dynamics simulationof the effect of phosphorylation and mutation.Biophys. J. 88, 3243–3251.

39. Chu, G., Li, L., Sato, Y., Harrer, J. M., Kadambi, V. J.,Hoit, B. D. et al. (1998). Pentameric assembly ofphospholamban facilitates inhibition of cardiac func-tion in vivo. J. Biol. Chem. 273, 33674–33680.

40. Colyer, J. (1993). Control of the calcium pump ofcardiac sarcoplasmic reticulum. A specific role for thepentameric structure of phospholamban? Card. Res.27, 1766–1771.

41. Oxenoid, K. & Chou, J. (2005). The structure ofphospholamban pentamer reveals a channel-likearchitecture in membranes. Proc. Natl Acad. Sci. USA,102, 10870–10875.

42. Stokes, D., Pomfret, A., Rice, W., Glaves, J. & Young,H. (2006). Interactions between Ca2+-ATPase and thepentameric form of phospholamban in two-dimen-sional co-crystals. Biophys. J. 90, 4213–4223.

43. Robia, S. L., Campbell, K. S., Kelly, E. M., Hou, Z.,Winters, D. L. & Thomas, D. D. (2007). Forster transferrecovery reveals that phospholamban exchangesslowly from pentamers but rapidly from the SERCAregulatory complex. Circ. Res. 101, 1123–1129.

44. Kelly, E. M., Hou, Z., Bossuyt, J., Bers, D. M. & Robia,S. L. (2008). Phospholamban oligomerization, quater-nary structure, and sarco(endo)plasmic reticulumcalcium ATPase binding measured by fluorescenceresonance energy transfer in living cells. J. Biol. Chem.283, 12202–12211.

45. Chen, Z., Akin, B., Stokes, D. & Jones, L. (2006).Cross-linking of C-terminal residues of phospholam-ban to the Ca2+ pump of cardiac sarcoplasmicreticulum to probe spatial and functional interactionswithin the transmembrane domain. J. Biol. Chem. 281,14163–14172.

46. Chen, Z., Stokes, D. & Jones, L. (2005). Role of leucine31 of phospholamban in structural and functional

722 Co-crystals of SERCA and Phospholamban

interactions with the Ca2+ pump of cardiac sarcoplas-mic reticulum. J. Biol.Chem. 280, 10530–10539.

47. Chen, Z., Stokes, D., Rice, W. & Jones, L. (2003).Spatial and dynamic interactions between phospho-lamban and the canine cardiac Ca2+ pump revealedwith use of heterobifunctional cross-linking agents.J. Biol. Chem. 278, 48348–48356.

48. Toyoshima, C., Asahi, M., Sugita, Y., Khanna, R.,Tsuda, T. & MacLennan, D. (2003). Modeling of theinhibitory interaction of phospholambanwith theCa2+

ATPase. Proc. Natl Acad. Sci. USA, 100, 467–472.49. Morita, T., Hussain, D., Asahi, M., Tsuda, T.,

Kurzydlowski, K., Toyoshima, C. & Maclennan, D. H.(2008). Interaction sites among phospholamban, sarco-lipin, and the sarco(endo)plasmic reticulum Ca2+-ATPase. Biochem. Biophys. Res. Commun. 369, 188–194.

50. Reddy, L. G., Jones, L. R., Cala, S. E., O'Brian, J. J.,Tatulian, S. A. & Stokes, D. L. (1995). Functionalreconstitution of recombinant phospholamban withrabbit skeletal Ca-ATPase. J. Biol. Chem. 270,9390–9397.

51. Trieber, C., Douglas, J., Afara, M. & Young, H. (2005).The effects of mutation on the regulatory properties ofphospholamban in co-reconstituted membranes. Bio-chemistry, 44, 3289–3297.

52. Young, H., Reddy, L., Jones, L. & Stokes, D. (1998).Co-reconstitution and co-crystallization of phospho-lamban and Ca2+-ATPase. Ann. N.Y. Acad. Sci. 853,103–115.

53. Young, H. S., Jones, L. R. & Stokes, D. L. (2001).Locating phospholamban in co-crystals with Ca2+-ATPase by cryoelectron microscopy. Biophys. J. 81,884–894.

54. Reddy, L., Cornea, R., Winters, D., McKenna, E. &Thomas, D. (2003). Defining the molecular compo-nents of calcium transport regulation in a reconsti-tuted membrane system. Biochemistry, 42, 4585–4592.

55. Kimura, Y., Asahi, M., Kurzydlowski, K., Tada, M. &MacLennan, D. H. (1998). Phospholamban domain Ibmutations influence functional interactions with theCa2+-ATPase isoform of cardiac sarcoplasmic reticu-lum. J. Biol. Chem. 273, 14238–14241.

56. Toyofuku, T., Kurzydlowski, K., Tada, M. &MacLennan,D. H. (1994). Amino acids Glu2 to Ile18 in thecytoplasmic domain of phospholamban are essentialfor functional association with the Ca2+-ATPase ofsarcoplasmic reticulum. J. Biol. Chem. 269, 3088–3094.

57. Robia, S. L., Flohr, N. C. & Thomas, D. D. (2005).Phospholamban pentamer quaternary conformationdetermined by in-gel fluorescence anisotropy. Bio-chemistry, 44, 4302–4311.

58. Traaseth, N. J., Verardi, R., Torgersen, K. D., Karim, C.B., Thomas, D. D. & Veglia, G. (2007). Spectroscopicvalidation of the pentameric structure of phospho-lamban. Proc. Natl Acad. Sci. USA, 104, 14676–14681.

59. Auer, M., Scarborough, G. & Kuhlbrandt, W. (1998).Three-dimensional map of the plasma membrane H+-ATPase in the open conformation. Nature, 392,840–843.

60. Hutter, M., Krebs, J., Meiler, J., Griesinger, C.,Carafoli, E. & Helms, V. (2002). A structural modelof the complex formed by phospholamban and thecalcium pump of sarcoplasmic reticulum obtained bymolecular mechanics. ChemBiochem, 3, 1200–1208.

61. Seidel, K., Andronesi, O. C., Krebs, J., Griesinger, C.,Young, H. S., Becker, S. & Baldus, M. (2008). Structuralcharacterization of Ca2+-ATPase-bound phospholam-ban in lipid bilayers by solid-state nuclear magneticresonance (NMR) spectroscopy. Biochemistry, 47,4369–4376.

62. Fujii, J., Maruyama, K., Tada, M. & MacLennan, D. H.(1989). Expression and site-specific mutagenesis ofphospholamban: studies of residues involved inphosphorylation and pentamer formation. J. Biol.Chem. 264, 12950–12955.

63. Reddy, L., Jones, L. & Thomas, D. (1999). Depolymer-ization of phospholamban in the presence of calciumpump: a fluorescence energy transfer study. Biochem-istry, 38, 3954–3962.

64. Hou, Z., Kelly, E. M. & Robia, S. L. (2008).Phosphomimetic mutations increase phospholambanoligomerization and alter the structure of its regula-tory complex. J. Biol. Chem. 283, 28996–29003.

65. Chu, S., Abu-Baker, S., Lu, J. & Lorigan, G. A. (2010).(15)N Solid-state NMR spectroscopic studies onphospholamban at its phosphorylated form at Ser-16in aligned phospholipid bilayers. Biochim. Biophys.Acta, 1798, 312–317.

66. Traaseth, N. J., Thomas, D. D. & Veglia, G. (2006).Effects of Ser16 phosphorylation on the allosterictransitions of phospholamban/Ca2+-ATPase com-plex. J. Mol. Biol. 358, 1041–1050.

67. Lamberth, S., Schmid, H., Muenchbach, M., Vorherr,T., Krebs, J., Carafoli, E. & Griesinger, C. (2000). NMRsolution structure of phospholamban. Helv. Chim.Acta, 83, 2141–2152.

68. Zamoon, J., Mascioni, A., Thomas, D. & Veglia, G.(2003). NMR solution structure and topologicalorientation of monomeric phospholamban in dode-cylphosphocholine micelles. Biophys. J. 85, 2589–2598.

69. Andronesi, O., Becker, S., Seidel, K., Heise, H., Young,H. & Baldus, M. (2005). Determination of membraneprotein structure and dynamics by magic-angle-spinning solid-state NMR spectroscopy. J. Am. Chem.Soc. 127, 12965–12974.

70. Potluri, S., Yan, A. K., Chou, J. J., Donald, B. R. &Bailey-Kellogg, C. (2006). Structure determination ofsymmetric homo-oligomers by a complete search ofsymmetry configuration space, using NMR restraintsand van der Waals packing. Proteins: Struct. Funct.Genet. 65, 203–219.

71. Traaseth, N. J., Shi, L., Verardi, R., Mullen, D. G.,Barany, G. & Veglia, G. (2009). Structure andtopology of monomeric phospholamban in lipidmembranes determined by a hybrid solution andsolid-state NMR approach. Proc. Natl Acad. Sci. USA,106, 10165–10170.

72. Oxenoid, K., Rice, A. J. & Chou, J. J. (2007).Comparing the structure and dynamics of phospho-lamban pentamer in its unphosphorylated and pseudo-phosphorylated states. Protein Sci. 16, 1977–1983.

73. Pollesello, P., Annila, A. & Ovaska, M. (1999).Structure of the 1–36 amino-terminal fragment ofhuman phospholamban by nuclear magnetic reso-nance andmodeling of the phospholamban pentamer.Biophys. J. 76, 1784–1795.

74. Simmerman, H., Kobayashi, Y., Autry, J. & Jones, L.(1996). A leucine zipper stabilizes the pentameric

723Co-crystals of SERCA and Phospholamban

membrane domain of phospholamban and forms acoiled-coil pore structure. J. Biol. Chem. 271, 5941–5946.

75. Warren, G. B., Toon, P. A., Birdsall, N. J. M., Lee, A. G.& Metcalfe, J. C. (1974). Reconstitution of a calciumpump using defined membrane components. Proc.Natl Acad. Sci. USA, 71, 622–626.

76. Eletr, S. & Inesi, G. (1972). Phospholipid orientation insarcoplasmic reticulum membranes: spin-label ESR andprotonNMRstudies.Biochim. Biophys.Acta, 282, 174–179.

77. Douglas, J., Trieber, C., Afara, M. & Young, H. (2005).Rapid, high-yield expression and purification of Ca2+-ATPase regulatory proteins for high-resolution struc-tural studies. Protein Expr. Purif. 40, 118–125.

78. Reddy, L. G., Jones, L. R., Pace, R. C. & Stokes, D. L.(1996). Purified, reconstituted cardiac Ca2+-ATPase isregulated by phospholamban but not by direct

phosphorylation with Ca2+/calmodulin-dependentprotein kinase. J. Biol. Chem. 271, 14964–14970.

79. Dux, L. & Martonosi, A. (1983). Two-dimensionalarrays of proteins in sarcoplasmic reticulum andpurified Ca2+-ATPase vesicles treated with vanadate.J. Biol. Chem. 258, 2599–2603.

80. Crowther, R. A., Henderson, R. & Smith, J. M. (1996).MRC image processing programs. J. Struct. Biol. 116,9–16.

81. Toyoshima, C., Yonekura, K. & Sasabe, H. (1993).Contrast transfer for frozen-hydrated specimens: II.Amplitude contrast at very low frequencies. Ultra-microscopy, 48, 165–176.

82. Potterton, E., McNicholas, S., Krissinel, E., Cowtan, K.& Noble, M. (2002). The CCP4 molecular-graphicsproject. Acta Crystallogr. D, 58, 1955–1957.