|Home | About | Journals | Submit | Contact Us | Français|
This chapter reviews the molecular biology, biochemical, and NMR methods that we used to study the structural dynamics, membrane topology and interaction of phospholamban (PLN), a small regulatory membrane protein involved in the regulation of the sarcoplasmic reticulum Ca-ATPase (SERCA). In particular, we show the progression of our research from the initial hypotheses toward understanding the molecular mechanisms of SERCA’s regulation, including the effects of PLN oligomerization and posttranslational phosphorylation. Finally, we show how the knowledge of the molecular mechanism of the structural dynamics and topology of free and bound proteins can lead to the rational design of PLN analogs for possible use in gene therapy.
Phospholamban (PLN) is a small integral membrane protein (52 residues) localized in the sarcoplasmic reticulum (SR) membrane that regulates the flow of calcium ions into the SR lumen of cardiac muscle(1). Specifically, PLN binds to SR calcium ATPase (SERCA), the enzyme responsible for calcium re-uptake into the SR lumen. In its unphosphorylated form, PLN inhibits SERCA by reducing its affinity for calcium. Upon β-adrenergic stimulation, PLN is phosphorylated by protein kinase A at S16 with concomitant relief of SERCA inhibition and restoration of SERCA activity. These cyclic events account for the relaxation (diastolic phase) of the heart muscle and if disrupted, may evolve into heart failure(2). Due to their central role in cardiac muscle contractility, PLN and SERCA have become targets for new, alternative therapies for heart failure(3). What can be learned from the study of the structural and dynamic transitions of PLN free and when bound to SERCA? More importantly, can these studies contribute in the development of new clinical approaches to heart failure?
Much progress has been made toward understanding the molecular mechanism of SERCA within the enzymatic cycle(4, 5). Chapters 9 and 15 of this volume give exquisite portraits of the latest findings, outlining the successes of x-ray crystallography in reconstructing the structural transitions of SERCA during enzyme turnover. Undeniably, the first crystal structure of SERCA (E1-Ca2, calcium bound form) solved by Toyoshima and co-workers paved the way for the atomic level understanding of calcium translocation in the SR(6). To date, there are several crystal structures of SERCA under different experimental conditions that are allowing enzymologists to model the major conformational states of the enzyme during turnover(7). However, very few studies have been dedicated to the determination of the complexes between SERCA and its endogenous inhibitor PLN. This is probably due to the inherent dynamics of the SERCA/PLN complex that might complicate the formation of large, diffracting crystals. This hypothesis is reinforced by several fluorescence and EPR studies published by the Thomas and Squier Laboratories underscoring the dynamic nature of the SERCA/PLN complex(8-14). So far, the labs of Stokes and Young (15, 16) have produced the best images of this complex, although the resolution of the crystals does not define the interaction surface between the two proteins at the atomic level. NMR studies, on the other hand, have been focused on the structural dynamics of PLN in the unbound form. This is mainly due to the difficulties in producing isotopically labeled SERCA and analyzing its complex with PLN by solution-state NMR (Mw ~ 116 kDa).
Our laboratory, in collaboration with the Thomas laboratory, has used a combination of solution and solid-state NMR methods to map the structural dynamics, topology, and interactions between PLN and SERCA. Based on our chemical shift perturbation data from solution NMR, we have constructed an allosteric model for SERCA regulation by PLN(11, 13, 17-19). This regulatory model has been now adopted by several groups in the fields of structural biology and muscle physiology to explain the effects of naturally occurring and engineered mutations in PLN. More importantly, this model has allowed us to rationally manipulate the structural dynamics of PLN and design new mutants that are possible candidates for gene therapy in heart failure treatment.
In the following sections, we briefly review the molecular biology, biochemistry, and NMR approaches we have used to study SERCA’s regulation by looking at PLN free (monomeric and pentameric) and bound to SERCA. We show the progression of our research from our initial hypotheses on the mechanism of SERCA regulation by PLN to our initial attempts to control PLN structural dynamics to tune SERCA’s function.
For structural studies of membrane proteins by NMR, it is necessary to find a robust expression system that allows for the purification of milligram quantities of pure protein. For this purpose, we optimized a fusion protein construct for both monomeric (cysteine null PLN with the following mutations C36A, C41F, C46A, which we refer to as AFA-PLN) and pentameric species(21).
To maximize the expression, all of the codons were optimized for E. coli. The nucleic acid sequence was divided into four overlapping primers and three polymerase chain reactions (PCR) were performed, with incorporation of a TEV protease cleavage site between MBP and PLN, an EcoRI restriction site at the 5′end, and a HindIII site at the 3′ end (a map of the plasmid is available from ref. (21)). Each PCR was performed with a total volume of 50 μL containing 0.06 μM of each oligonucleotide, 80 μg of parental template, 200 μM dNTP mix, and 5.0 U Pfu Turbo DNA polymerase. Each PCR reaction was carried out in an Eppendorf Master Cycler system with standard temperature cycling suggested for Pfu Turbo DNA polymerase. AFA-PLN PCR product was amplified and then ligated into EcoRI and HindIII sites on the pMal c2E vector using T4 DNA ligase (Rationale for fusion partner is described in Note 1). The ligated product was used to transform XL1-Blue competent cells with transformants selected by growth on LB/ampicillin plates. For single-site mutations of PLN, the forward and reverse primers were designed with optimized codons for E. coli three residues before and after the target residue to be mutated. PCR was performed with Pfu DNA polymerase using the protocol as described above. PCR products were then digested with Dpn-I in order to remove non-mutated methylated dsDNA, and then were transformed into XL-Blue competent cells. Constructs were amplified and extracted using the Qiaquick Spin Miniprep kit.
E. coli BL21(DE3) cells were transformed with MBP-PLN fusion constructs. 100 ml of LB growth media with 50 μg/ml ampicillin was inoculated with a single colony and grown ~16 h at 25 °C shaking at 250 RPM to reach an OD600 of ~1.0. The growth was then diluted (1:50) into M9 media containing mineral and vitamin cocktails and grown at 37 °C to an OD600 of ~1.0. Protein expression was induced with 1 mM IPTG for 4.5 h at 37 °C. All cells were harvested by centrifugation at 6,370 × g for 20 min at 4 °C. Pellets were stored at −20 °C.
PLN pellets (~10-15 g) were resuspended in ~400 ml of lysis buffer and agitated by blending using a Sorval Omni-Mixer 17105. Cell lysates were then sonicated on ice for 15 min using a Branson Sonifier 450 sonicator at a output setting of 4 and a duty cycle of 45%. After sonication, lysates were spun down at 45,700 × g for 20 min at 4 °C. The supernatant containing MBP/PLN fusion protein was applied to an amylose resin affinity chromatography column at 4 °C. The column was washed with 12 column volumes of wash buffer until the OD280 reached ≤ 0.05. The protein was eluted using a maltose solution, which MBP preferentially binds, separating the protein from the amylose resin. PLN was cleaved from MBP with TEV protease at 100 U/mg fusion protein at 30°C for ~5-8 h. At this point PLN can be further purified in two ways: 1) an S-100 sephacryl gel filtration column using an AKTAprime liquid chromatography system (Amersham Pharmacia Biotech), or 2) using reversed-phase HPLC. While we originally used the gel filtration method, we have abandoned this method in favor of the faster and higher resolving HPLC method as reported by Young and co-workers(22). For the HPLC method, the cleavage reaction was dialyzed against 4 L of ddH2O in a 1000 Da cutoff membrane until PLN precipitated (~12 hr). The suspension was centrifuged at 6,370 × g for 20 min at 4 °C. The supernatant containing soluble proteins (MBP, TEV protease, uncleaved fusion protein) was discarded. Pellets were solubilized in 1-2 ml of 10% (w/v) SDS and stored at 4 °C. This crude solution was purified by reversed phase HPLC using an Agilent 1100 system (Agilent Technologies, Inc. Santa Clara, CA, USA) equipped with a Vydac 208TP C8 monomeric reversed phase column (250 mm × 1.0 mm i.d. × 10 μm particle size, Grace Vydac, Hesperia, CA, USA). The mobile phases used were: A – H2O with 0.1% TFA and B – isopropanol. The column (incubated at 50 °C) was equilibrated with 10% solvent B and proteins were eluted at 2 ml/min flow rate using the following gradient: 10% to 30% solvent B in 10 min, 30% to 50% in 20 min, 50% to 70% in 30 min, and 70% to 100% in final 10 min. The column was re-equilibrated with 10% B for 10 min after each run. With these conditions the retention time for PLN was ~59 min. PLN fractions were pooled and isopropanol was evaporated under a stream of N2. The resulting suspension was lyophilized and stored at −20 °C.
Protein concentration at each purification step was determined using protein absorbance at 280 nm, or assayed with densitometry measurements of Coomassie or silver stained gels (Figure 1A). Densitometry data was collected on a Bio-Rad Molecular Imager FX using Bio-Rad Quantity One quantitation software. Recombinant protein concentrations were determined by comparison to a standard curve of 1-3 μg synthetic PLN peptides previously quantified by amino acid analysis. PLN was also identified with MALDI-MS and amino acid sequence analysis.
HPLC-purified PLN was reconstituted to a concentration of 0.25 mg/mL for phosphorylation. 1 mg PLN was phosphorylated with 1000 U of protein kinase A catalytic subunit. Phosphorylation was confirmed using a gel-shift from a western blot with anti-pS16 PLN antibody 285 and by MALDI-TOF MS.
SERCA was extracted from rabbit skeletal muscle, purified, and tested for activity as reported previously (15, 17, 18). A solution containing 0.45 mM of SERCA, 5 mM DPC, 1 mM CaCl2, 1 mM MgCl2, 20 mM MOPS, 20% glycerol, 0.25 mM DTT, 0.4 mM ADP, pH of 7.0 was used for titration experiments (see below).
Lyophilized PLN protein was solubilized in 240 μL of chloroform containing 2.4 mg of lipids in a 4:1 molar ratio of 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE). Samples were then dried by removing the organic solvent under a stream of N2 gas. The dried film of lipid and PLN was then hydrated in 120 μL of 25 mM imidazole (pH 7.0) by vortexing followed by brief (30-60 sec) water-bath sonication. The lipid/PLN vesicles were then adjusted to contain a final concentration of 20 mM imidazole (pH 7.0), 0.1 M KCl, 5 mM MgCl2, and 10% glycerol. 4.8 mg of β-octylglucoside was added, followed by 60 μg of purified SERCA with the final volume adjusted to 300 μL with buffer. Removal of the detergent was performed by incubation with 120 mg of wet Biobeads for 3 h at 25 °C. The SERCA/PLN lipid vesicles were separated from Biobeads using a gel-loading tip and micropipette and immediately assayed for function.
SERCA activity was assayed by an enzyme-linked assay (23) performed in a 96-well plate. Each assay was performed at 37 °C in a Thermomax microplate reader (Molecular Devices) in triplicate at different free calcium concentrations in a total volume of 175 μL. Between 1 and 3 μg of SERCA was added to the samples to start the assay, and the absorbance of NADH was monitored at 340 nm to determine the rate of ATP hydrolysis. Data was plotted in Origin 7.5 and fit using the Hill Equation (see below) to determine the shift in pKCa (Figure 1B).
To measure the hydrolysis of ATP by SERCA under solution NMR conditions (DPC detergent micelles), assays were performed by monitoring the inorganic phosphate 31P NMR signal. The assay mixture consisted of 6 μM SERCA, 137 μM PLN, assay mix, and concentrations of CaCl2 required to reach free Ca2+ concentration calculated based on the method of Fabiato and Fabiato(24). Upon each addition of nucleotide, eight free induction decays (FIDs) were collected, each consisting of 16 single pulse transients on 31P. Data were collected on a Varian Inova spectrometer operating at a 1H Larmor frequency of 500 MHz (Figure 1). After monitoring the rate, the inorganic phosphate signal was normalized to the known amount of DPC in the sample. This normalized amount was then plotted (V vs pCa) and fit by the Hill equation (Figure 1C).
V is the initial ATPase rate and n is the Hill coefficient. The data were normalized to the maximal rate, Vmax, which was obtained from the fit, and then replotted to determine the shift in pKCa (Figure 1B).
NMR samples were prepared by dissolving isotopically labeled PLN in NMR sample buffer containing 300 mM DPC and 10% D2O to a final concentration of 0.23 mM. For the preparation of wt-PLN NMR samples (or other “difficult” samples), a slightly different procedure was used: wt-PLN was solubilized in 300 mM DPC, 20 mM PBS (pH 6.0) and 6 M guanindinum-HCl. The sample was dialyzed overnight in a mini-dialysis chamber against 1 L dialysis buffer. For the NMR titration, SERCA at a concentration of 0.51 mM was added incrementally to the PLN sample as previously reported (17). To correct for this dilution, and potential pH and salt effects, the same titration experiment was performed with SERCA buffer in the absence of enzyme. After each SERCA addition, a 1H-15N heteronuclear single quantum coherence (HSQC) experiment was collected on a Varian Inova spectrometer operating with a 1H Larmor frequency of 600 MHz at 37 °C, using an inverse detection triple-resonance and triple-axis gradient probe. The HSQC pulse program was equipped with pulsed field gradients for both coherence selection and sensitivity enhancement (25).
For measuring the binding constants, we assumed 1:1 binding between SERCA and PLN, and that the intensity reduction (Iretention) of the transmembrane domain residues in Figure 2C was directly related to the fraction of PLN bound to SERCA (fb). For a complete derivation of Eqn. 3 see ref. (18)(Eqns. 4-10).
The dissociation constant (Kd) was calculated using a non-linear fit to Eqn. 3.
PLN was reconstituted into lipid vesicles using two different preparations: 1) detergent mediated reconstitution and 2) organic solvent mediated reconstitution.
80 mg (4/1, w/w) of DOPC/DOPE were dissolved in chloroform and thoroughly vortexed. Chloroform was evaporated under a stream of N2, after which the lipids were resuspended in 40 mL of ddH2O. The lipid suspension was sonicated on ice until small unilamellar vesicles (SUVs) formed (the suspension became transparent). After sonication, the lipids were centrifuged (6,370 × g for 10 min at 4°C) to remove larger vesicles and metal particles. PLN (~4 mg) was solubilized in 1 mL of SDS (10%, w/v) and dissolved in the lipid mixture and subjected to one freeze-thaw cycle. Samples were extensively dialyzed against ddH2O to remove the detergent and subsequently concentrated to 2 ml using an ultrafiltration device (10 kDa cutoff membrane). Approximately 100 μL of sample were transferred onto each of 20 glass plates. Samples were slowly dried at 40 °C, rehydrated to the liquid-crystalline phase and finally sealed in a rectangular glass cell.
DOPC/DOPE 80 mg (4/1, w/w) were dissolved in chloroform and thoroughly mixed. PLN (~4 mg) was solubilized in 50 μL of trifluoroethanol and added to the lipid mixture. PLN (~4 mg) was solubilized in 50 μL of trifluoroethanol and added to the lipid mixture. Solvents were first evaporated under a stream of N2 and then lyophilized overnight to ensure complete removal of organic solvents. The lipid/protein mixture was resuspended in 40 mL of ddH2O and carefully vortexed. Small unilamellar vesicles were prepared by repeatedly extruding the lipid mixture through polycarbonate filters of decreasing pore size (200, 100, 50 μm) using a bench-top extruder. This SUV suspension was concentrated to 2 ml with samples dried and rehydrated as reported above. The final molar ratio of lipid/protein for all samples was ~200/1.
The 2D polarization inversion spin exchange at the magic angle (PISEMA) was performed (26, 27) with TPPM decoupling during acquisition (28). A phase modulated Lee-Goldberg (PMLG) was used in the indirect dimension to decouple 1H-1H interactions and allow for the evolution of 1H−15N dipolar coupling. The initial 90° 1H pulse, cross-polarization, PMLG (1H effective field), and TPPM decoupling during acquisition were applied at ~60 kHz RF field strength. Spectra of uniformly 15N samples were acquired with 1 k scans and 30 t1 increments while selectively labeled samples required 4-12 k scans and ~8-16 increments. All experiments were performed at the National High Magnetic Field Laboratory at a 14.1 T magnetic field strength (1H frequency of 600.1 MHz) equipped with a Bruker DMX spectrometer using a low-E probe built by the RF program(29).
Solution and solid-state NMR methods have been used as a complementary approach to the structure determination of small and medium size membrane proteins(30, 31). For solution NMR, the proteins are reconstituted in detergent micelles with a meticulous choice of experimental conditions, which is based on the NMR spectral features as well as preservation of SERCA function. For solid-state, we reconstitute membrane proteins in synthetic lipid bilayers, which are more amenable to enzyme function. The major advantage of solution NMR is to offer high-resolution spectra (membrane proteins up to hundreds of kDa have now been studied and assigned) and to obtain secondary and tertiary structures through NOE measurements. On the other hand, when membrane proteins are solubilized in detergent micelles, the topological information of the protein within the lipid bilayer is lost. This orientation information regarding the molecular topology can be recovered by solid-state NMR measurements using aligned samples.
Recent reports verify that the protein adopts a similar structure in both lipid bilayers and micelles; therefore, we decided to pool the information derived from these two techniques and combine the structural restraints into a unique molecular modeling protocol. The total potential energy function (ETotal) is defined as a combination of an empirical energy function (Echem) and two penalty functions that include solution (EsolNMR) and solid-state NMR (EssNMR) restraints:
The energy terms of Echem are included into standard force field of XPLOR-NIH(32). EssNMR contains dipolar coupling and chemical shift anisotropy values derived from PISEMA or HETCOR experiments. As a result, the simulated annealing protocol minimizes the energy function for both distance and angular restraints as well as orientational restraints derived from solid-state NMR PISEMA experiments. Figures 3D shows the average structures from the lowest energy conformational ensembles for both the monomer and the pentamer(33).
The outcomes of our investigation are summarized within the schematic in Figure 4. In agreement with both in vitro and in vivo studies, we found that PLN adopts a pentameric, L-shaped conformation both in micelles and in lipid bilayers. In particular, both solution and solid-state NMR data show that the cytoplasmic domain (which is amphipathic) interacts with the surface of the membrane mimicking environment (both for micelles and lipid bilayers), while the transmembrane domain form a tight hydrophobic bundle stapled together by a leucine-isoleucine zipper. PLN depolymerization into active monomers occurs in the presence of SERCA. In the monomeric state (prelude to the interaction with SERCA), PLN is predominantly in a resting T state with a small population in the R state. Using EPR spectroscopy, Thomas and co-workers have estimated that ~10% of PLN exists in the R state in lipid bilayers at 4 °C. In the presence of SERCA, the equilibrium is shifted toward the R state that is probably selected to bind the enzyme. While this initial detection of PLN ground (T) and excited ® states shed light inot the dynamic regulatory process of SERCA, more experiments are needed to characterize the excited states both structurally and dynamically. Our ongoing investigation is revealing a more complex equilibrium than previously anticipated. Although dynamic in nature, PLN forms a stable inhibitory complex with SERCA (Kd in DPC detergent micelles ~60 μM). This inhibition is reversed by phosphorylation at S16, which changes the conformational dynamics and results in an unwinding of the cytoplasmic helix. In the presence of SERCA, phosphorylation at S16 PLN causes a slight rearrangement within domain Ib (residues 23-30) and changes the dynamics at the binding interface. These structural dynamics affect the conformational transitions of the enzyme, and ultimately calcium translocation.
A fundamental aim of structural biology is to move from understanding structure and dynamics to controlling molecular function. With this in mind, we attempted to manipulate the structural dynamics of PLN and to promote the formation of the R state through mutagenesis. Recently, we found that by mimicking the S16 phosphorylated state of PLN (i.e. enhancing the local dynamics in the hinge region) it is possible to tune the extent of SERCA inhibition(36). This work was originally inspired by in vivo studies carried out by Chien and co-workers, who demonstrated that a pseudo-phosphorylated variant of PLN can relieve SERCA inhibition and increase heart contractility, ultimately reversing the damages of myocardial infarction. We found that by changing Pro21 to Gly it is possible to generate a loss-of-function mutant with characteristics similar to S16E (pseudo-phosphorylated species tested by Chien), preserving the posttranslational control by β-adrenergic stimuli(36). This new generation of mutants possesses a residual inhibitory power that could be important in adjusting the pathophysiology of diseased hearts leading to reversal or hindrance of cardiac remodeling. We are currently working on different sites to improve the characteristic of loss-of-function mutants.
This work was supported by grants to G.V. from the National Institutes of Health (GM64742, HL80081, GM072701) and NJT (AHA 0515491Z). PISEMA spectra were acquired at the NHMFL, Tallahassee, FL (DMR-0084173). NMR instrumentation at the University of Minnesota High Field NMR Center was funded by the National Science Foundation (BIR-961477) and the University of Minnesota Medical School.