Role of Cysteine Residues in Structural Stability and Function of a Transmembrane Helix Bundle*

To study the structural and functional roles of the cysteine residues at positions 36, 41, and 46 in the transmembrane domain of phospholamban (PLB), we have used Fmoc (N-(9-fluorenyl)methoxycarbonyl) solid-phase peptide synthesis to prepare α-amino-n-butyric acid (Abu)-PLB, the analogue in which all three cysteine residues are replaced by Abu. Whereas previous studies have shown that replacement of the three Cys residues by Ala (producing Ala-PLB) greatly destabilizes the pentameric structure, we hypothesized that replacement of Cys with Abu, which is isosteric to Cys, might preserve the pentameric stability. Therefore, we compared the oligomeric structure (from SDS-polyacrylamide gel electrophoresis) and function (inhibition of the Ca-ATPase in reconstituted membranes) of Abu-PLB with those of synthetic wild-type PLB and Ala-PLB. Molecular modeling provides structural and energetic insight into the different oligomeric stabilities of these molecules. We conclude that 1) the Cys residues of PLB are not necessary for pentamer formation or inhibitory function; 2) the steric properties of cysteine residues in the PLB transmembrane domain contribute substantially to pentameric stability, whereas the polar or chemical properties of the sulfhydryl group play only a minor role; 3) the functional potency of these PLB variants does not correlate with oligomeric stability; and 4) acetylation of the N-terminal methionine has neither a functional nor a structural effect in full-length PLB.

of PLB is primarily responsible for inhibition of the Ca-ATPase (3,4,(7)(8)(9). The residues presumed responsible for stabilizing the pentameric structure of PLB are located in the hydrophobic transmembrane domain (10 -12). These transmembrane domain residues are largely composed of Leu and Ile, but this arrangement is punctuated with three cysteines in a five-residue repeat (Fig. 1). Mutation of these cysteines (at positions 36,41, and 46 to Ser, Ala, or Phe, respectively) induces changes in the oligomeric stability of PLB (13). The mutation Cys-41 to Phe shows the strongest effect, decreasing the apparent pentameric stability. We have found that Cys-41 is unreactive and is located at a crucial site for the maintenance of the pentameric structure (5). Based on these results, a structural model for the PLB pentamer has been proposed, in which each pair of subunits is stabilized by interhelical interactions between leucines 37, 44, and 51 with isoleucines 40 and 47 to form a Leu/Ile zipper (5,11,12). The transmembrane Cys residues do not appear to be involved in intermolecular disulfide bonding (2,15). To evaluate the role of the chemical and steric packing properties of the cysteines in the pentameric structure of PLB as well as their function, we have used Fmoc solid-phase peptide synthesis to design a sterically identical PLB derivative. In the present study, we replaced all cysteine residues in PLB with ␣-amino-n-butyric acid (Abu), which is isosteric to cysteine (16). Like many eukaryotic proteins, PLB is "capped" at the N-terminal methionine by posttranslational acetylation (17). The N-terminal cytoplasmic portion of WT-PLB has a net charge of ϩ3 with the posttranslational acetylation, but this charge increases to ϩ4 in the absence of the acetyl group. It is not known whether this acetyl group in the full-length PLB is necessary for pentameric stabilization and interaction with the Ca-ATPase. No effect of nonacetylated PLB  was reported, and only acetylated peptide Ac-PLB  showed inhibition of the Ca-ATPase (18). To clarify this subject, we determined the oligomeric states of acetylated and nonacetylated Abu-PLB in detergent solution by SDS-PAGE. The synthetic peptides were then co-reconstituted with the Ca-ATPase in lipid vesicles, and Ca-ATPase inhibition assays were performed in comparison to WT-PLB.

EXPERIMENTAL PROCEDURES
Peptide Synthesis and Purification of Abu-PLB-Materials, solvents, instrumentation, and general methods of solid-phase peptide synthesis were essentially as described in our previous publications (15, 16, 20 -22). We used acetic anhydride for the acetylation of the N-terminal amino group (23). First, an Fmoc removal step was carried out on 200 mg of peptide resin (15), followed by treatment with 0.5 M acetic anhydride in 10 ml of N,N-dimethylformamide. After 2 h, the acetylated peptide resin was filtered and then used for cleavage and purification (15). Fractions containing peptides were lyophilized to yield 26 mg of Abu-PLB ( Fig. 1)  Expressed Wild-type PLB-WT-PLB was expressed in Sf21/baculovirus insect cell system and purified by monoclonal antibody affinity chromatography as described previously (7,12).
SR Vesicles-SR vesicles were prepared from the fast-twitch skeletal muscle of New Zealand White rabbits (24). The Ca-ATPase from the SR vesicles was purified using a reactive-red affinity column (25).
Analysis of Peptide Size and Composition-SDS-PAGE was performed using 16.5% Tris/Tricine gel (Bio-Rad) (15). The peptide samples from the stock methanol/chloroform 2:1 solution were dried overnight. 20 l of 1% SDS was added to the samples that contained 5 g of Abu-PLB and WT-PLB. For SDS-PAGE, samples contained 20 l of Tricine sample buffer (26) with a final SDS concentration of 1.5%. The temperature was controlled during electrophoresis by using a recirculating water bath. For the quantitation of Abu-PLB and WT-PLB monomers, the gels were scanned by a densitometer using the transmittance mode, and then the bands were quantitated using the volume (area ϫ density) analysis method (27).
Mass spectral data was acquired with a Bruker Biflex III matrixassisted laser desorption/ionization time of flight system, which is equipped with an N 2 -laser (337 nm, 3-ns pulse length) and a microchannel plate detector. The data was collected in the linear mode, positive polarity, with an accelerating potential of 19 kV. Each spectrum is the accumulation of 100 -400 laser shots. The samples were co-crystallized with the matrix 3,5-dimethoxy-4-hydroxycinnamic acid (sinapinic acid).
Ca-ATPase/PLB Co-reconstitution-The method used for the functional reconstitution of Ca-ATPase with PLB has been described (28). In short, 33 g of Abu-PLB or its analogues was dried and solubilized in 240 l of chloroform containing 2.4 mg of lipids (DOPC/DOPE, 4:1). The dried film of lipid and PLB was hydrated with 120 l of 25 mM imidazole, pH 7.0, by vortexing followed by a brief sonication. The resulting vesicles were diluted to 20 mM imidazole, pH 7.0, 0.1 M KCl, 5 mM MgCl 2 , 10% glycerol. Then, 4.8 mg of ␤-octyl glucoside was added, followed by 60 g of purified Ca-ATPase. The final volume was adjusted to 300 l with buffer. The detergent was then removed by incubation with 120 mg of hydrated Biobeads for 3 h at room temperature. The Ca-ATPase/PLB lipid vesicles were separated from Biobeads and assayed immediately. All Ca-ATPase/PLB co-reconstitution in the present study used a fixed molar ratio of 10 PLB/Ca-ATPase. As shown previously (27), this ratio gives substantial effects, comparable to those in cardiac SR.
ATPase Activity Measurements-Ca-ATPase activity was measured by an enzyme-linked assay performed in microtiter plates (200 l total volume in each well) as described previously (15,29). Each well contained 0.2-0.6 g of Ca-ATPase (1-3 l of vesicles) and was added to a buffer containing 50 mM imidazole, pH 7.0, 0.1 M KCl, 5 mM MgCl 2 , 0.5 mM EGTA, 0.5 mM phosphoenolpyruvate, 2.5 mM ATP, 0.2 mM NADH, 2 IU of pyruvate kinase, 2 IU of lactate dehydrogenase, and 1-2 g of calcium ionophore (A23187). Each assay was done in triplicate at each of 12 different free calcium concentrations. The absorbance of NADH was monitored at 340 nm to determine the rate of ATP hydrolysis. The assays were performed at 15 and 25°C in a Thermomax microplate reader (Molecular Devices). Each data point represents average Ϯ S.E. (n Ն 6). A t test was used to determine the statistical significance of the differences between peptides and the effects of temperature.
Computational Modeling-We started with our previously constructed, experimentally verified model for the PLB transmembrane domain (residues 35-52) (5). This model was derived from that of Adams et al. (30) (PDB entry 1PSL) by capping the N and C termini with N-acetyl and NЈ-methylamide groups, respectively, and then rotating each helix counter-clockwise around its axis by about 50°to produce a structure conforming to a leucine-zipper motif (12), as supported by cysteine reactivity and site-directed spin labeling (5). This structure was then used as the starting point for simulations carried out using AMBER 5.0 (31,32), as recently updated for the peptide backbone parameters (parm96.dat). Before energy minimization, side chain rotamers were generated using the program SQWRL (33) to relieve unfavorable interactions between side chains. Side chains were found to be in the most favorable conformation for an ␣-helix (33).
The template was energy-minimized using a 12-Å nonbonded cutoff, a distance-dependent dielectric constant of 4 r, and a converge criterion of 5 ϫ 10 Ϫ5 kcal/mol Å in the root mean square of the Cartesian elements of the energy gradient. Positional constraints were imposed on the backbone heavy atoms using a harmonic potential with a force constant of 5 kcal/mol Å 2 . Minimum energy conformations of the mutants were generated by replacing the side chains while maintaining the same backbone templates as in the wild-type protein.
Interhelical energy differences were obtained as ⌬E h ϭ E h -E h 0 , where E h is the nonbonded interchain energy per helix (34), and E h 0 is the value for WT-PLB.
The buried surface area was defined as ⌬A ϭ (A i ϩ A j ) -A ij , where A i and A j are the surface areas of the individual helices i and j, and A ij is the surface area of the dimer complex. Values were averaged over the five helix interfaces of the pentamer. Surface areas were generated using GRASP (35) with a probe size of 1.4 Å. Free energies of disassociation were estimated from the buried surface area by considering a penalty of 20 cal mol Ϫ1 Å -2 upon loss of interfacing surface (36). ily as a 30-kDa pentamer, with a faint band at 6 kDa (monomer). This shows that in SDS solution at 25°C, the pentamer is just as stable for Abu-PLB as for WT-PLB.

Characterization of Abu-PLB-Mass
Inhibitory Function of WT-PLB and Abu-PLB- Fig. 4 shows the effects of synthetic WT-PLB and Abu-PLB on Ca-ATPase activity as a function of Ca 2ϩ concentration, measured in reconstituted membranes.
Both WT-PLB and Abu-PLB decrease the activity of the Ca-ATPase at pCa below 5.5, resulting in an increase in pK Ca (the calcium concentration, in pCa units, required for 50% calcium activation). Abu-PLB shifted pK Ca by -0.27 (control, 6.23 Ϯ 0.02; Abu-PLB, 5.96 Ϯ 0.02), whereas the shift by WT-PLB was -0.20 (6.03 Ϯ 0.03) for this co-reconstitution system. There is a significant difference between the control and Abu-PLB (p ϭ 2.5 E-06) but not between Abu-PLB and WT-PLB (p ϭ 0.2).
Thus, Abu-PLB shows similar inhibitory activity to WT-PLB, and Cys residues are not required for inhibition of the Ca-ATPase.
Inhibitory Function of Acetylated and Nonacetylated Abu-PLB-To study the function of the acetyl group capping the N terminus, acetylated and nonacetylated Abu-PLB were reconstituted in membranes. Both peptides showed the same increase in pK Ca (Fig. 5).
There is no significant difference between acetylated and nonacetylated Abu-PLB (p ϭ 0.60). This result shows that the acetyl group plays no role in the inhibition of the Ca-ATPase.
Inhibitory Function of WT-PLB and Ala-PLB- Fig. 6 shows the inhibitory effects of Ala-PLB and WT-PLB on Ca-ATPase activity.
Ala-PLB decreases pK Ca by 0.29 (control, 6.31 Ϯ 0.02; Ala-PLB, 6.02 Ϯ 0.02), whereas the decrease from WT-PLB is 0.20   (6.03 Ϯ 0.03). There is a significant difference between the control and Ala-PLB (p ϭ 1.7 E-09) but not between Ala-PLB and WT-PLB (p ϭ 0.6). Table I summarizes the inhibitory potencies, measured as in Figs. 4 and 6, and oligomeric stabilities, measured from densitometry of SDS-PAGE, of the three PLB derivatives at three different temperatures. There was no significant difference in inhibitory potency among the three peptides, nor was there a significant temperature dependence. In contrast, there were substantial differences in oligomeric stability. At all three temperatures, WT-PLB was predominantly pentameric, and Ala-PLB was completely monomeric. At low temperature, Abu-PLB exhibited high pentameric stability, comparable to that of WT-PLB, but the Abu-PLB pentamer was much less stable at 37°C. Fig. 7 shows a comparison of the mobilities of the WT-PLB, Abu-PLB, and Ala-PLB on SDS-PAGE at room temperature. WT-PLB and Abu-PLB showed mobility characteristic of pentamers, whereas Ala-PLB was completely monomeric. DISCUSSION Previous studies have shown that mutations of the three cysteines in PLB to serine, alanine, or phenylalanine disrupts the pentameric structure of PLB, suggesting that the cysteine side chains are crucial for the oligomeric stability of PLB (12,13). The principal goal of the present study was to determine whether the steric properties of Cys are sufficient for pentameric stability or whether the specific chemical properties of the thiol group are important. Another goal was to clarify the role of the N-terminal acetyl group for pentameric stability and interaction with the Ca-ATPase. We synthesized acetylated and nonacetylated Abu-PLB, an analogue of PLB in which the three Cys residues were replaced with Abu ( Fig. 1), an amino acid analogue that is known to be isosteric with cysteine. SDS-PAGE of Abu-PLB indicated that it is primarily pentameric, as is WT-PLB (Fig. 3), indicating clearly that the apparent requirement of the three Cys residues for pentameric stability of PLB is based primarily on steric packing, not on the chemical properties of the thiol groups.
N-terminal acetylation of Abu-PLB indicated no difference in the oligomeric stability and inhibitory function. Fig. 8 illustrates a structural model for helix packing in the transmembrane domain of the pentameric WT-PLB (left), Abu-PLB (middle), and Ala-PLB (right), focusing on the role of one of the three Cys residues, Cys-36. Note that the steric structure of the helix interface is indistinguishable for WT-PLB and Abu-PLB, showing excellent matching of the two helical surfaces. In this steric view, Abu-36 substitutes perfectly for Cys-36. However, in Ala-PLB, there is clearly a cavity created due to the smaller side chain of Ala-36, which should result in a loss of interhelical van der Waals stabilization energy. Table II summarizes calculations of predicted energetic properties of these three structures, focusing on the interhelical potential energy. Unlike total energies, interhelical potential energies can be compared between peptides with different sequences and can provide a direct comparison between mutants to estimate relative stability (34). We calculated interhelical energies, E h , as described under "Experimental Procedures," for the structural models of WT-PLB, Abu-PLB, and Ala-PLB (Fig. 8) by placing the side chains of residues at positions 36, 41, and 46 in the gauche(ϩ) ( 1 ϭ -60°) conformation.
As shown in Table II, the decreased pentameric stability (increased percentage of monomer) of Ala-PLB correlates with a calculated loss of energetically favorable interhelical van der Waals interactions, due to the smaller Ala side chain compared with Cys (⌬E h ). In contrast, the isosteric Abu and Cys side chains result in isoenergetic interactions. These results mirror the calculated loss in free energy of association (⌬⌬G in Table  II), as obtained from a loss in the buried interhelical surface area, which is significantly less for Ala-PLB than for either WT-PLB or Abu-PLB. The estimated ⌬⌬G ϭ 0.72 kcal/mol for an Ala-PLB dimer interface is about 50% of that calculated for the glycophorin mutants L75A and I76A (36). The smaller values for Ala-PLB reflect the smaller volume change for a Cys to Ala mutation, as compared with a Leu (Ile) to Ala mutation. Whereas the PLB-Ala peptide is mostly monomeric, as measured by SDS-PAGE, the truncated peptide (residues 26 -52) is mostly oligomeric (15). The different behavior between fulllength and truncated peptide suggest that a decrease in van der Waals interactions upon Cys-to-Ala mutation is not sufficient for depolymerization but that the N-terminal region also contributes to the energetics of the PLB pentamer (15).
Although the pentameric stabilities of WT-PLB and Abu-PLB are far more similar than that of Ala-PLB, SDS-PAGE at higher temperatures showed clearly that the WT-PLB pentamer is significantly more stable than Abu-PLB (Table I). This difference must be explained by interactions other than van der Waals, probably involving the specific chemical (polar) properties of the thiol group. Examination of the model structure reveals a possible hydrogen bond between Cys-36 and the backbone oxygen of Leu-37 and between Cys-36 and Cys-41 (Fig. 9).
These two interactions are absent in the model previously proposed by Arkin et al. (37), where Cys-36 faces the interior of the bundle, and Cys-41 and Cys-46 are exposed to the lipid. In that model, only intrachain hydrogen bonds were proposed between Cys at position i and the carbonyl of residue i-4 (37). Although such interactions can also occur in the current model, Cys-36 and Cys-41 may also be involved in interhelical hydrogen bonds.
The structural model of PLB, based on a "leucine zipper" template, places Cys-36 and Cys-41 at positions g and e, respectively, of the helical heptad repeat, whereas Cys-46, at position c, faces the lipid environment (Fig. 9A) (3). According to this model, side chain-side chain interactions can occur between Cys-36 and Cys-41 and between these two residues and the backbone atoms of Leu-37 (Fig. 9B). The lower oligomeric stability of Abu-PLB and Ala-PLB compared with WT-PLB, could, therefore, depend on differences between interhelical interactions at positions 36 and 41 of these peptides.
A close inspection of the interfacial region between Cys-36 and Cys-41 shows that additional types of interactions are also possible. Whereas the 1 ϭ gauche(ϩ) conformer of Cys-36 is preferred in ␣-helices (33), wild-type PLB has an additional state ( 1 ϭ trans) available for interaction. In this orientation, the interhelical energies are similar to those with 1 gauche(ϩ) (Table II) (38). Furthermore, side chain rotation of Cys-36 results in a S . . . S distance of 4.0 Ϯ 1.0 Å, which is similar to that observed for hydrogen-bonded S . . . S atoms in cysteine crystals (S . . . S ϭ 3.854 Å) (39).
Inhibitory Function-Abu-PLB and Ala-PLB have the same inhibitory function as WT-PLB (Table I), showing clearly that neither the steric properties of the Cys residues nor their chemical and polar properties are important for their inhibition of the Ca-ATPase. The similar potency of these two PLB analogues does not correlate with their significant differences in oligomeric stability (Table I). At first glance, this appears to contradict the proposal that only the monomeric form of PLB binds to the Ca-ATPase and inhibits it (3,4), and it raises the possibility that the SDS-PAGE assay used in the present study does not accurately reflect PLB oligomeric stability in the membrane. However, previous analyses of PLB oligomeric stability in lipid bilayers, using spectroscopic probes, have shown that SDS-PAGE is remarkably accurate in determining the distribution of oligomers for PLB and its mutants in lipid bilayers (3,4,6). Spectroscopic analysis also shows that the Ca-ATPase depolymerizes PLB, so that even WT-PLB, which is only 10 -20% monomer in SDS or in lipid bilayers (6), is 40% monomeric in the presence of Ca-ATPase (7). Thus, it is quite likely that each of the PLB variants in the present study is sufficiently monomeric in the presence of the Ca-ATPase to exert full inhibitory function.
These results are consistent with previous findings that there is sometimes a lack of quantitative correlation between oligomeric stability and inhibitory potency, suggesting that other structural factors are important (27). Although some PLB mutants that have greatly decreased pentameric stability, as shown by SDS-PAGE, have greater inhibitory activity than WT-PLB, others have negligible inhibitory activity (9). A few mutants have been shown to be more potent inhibitors than WT-PLB, despite retaining full pentameric stability on SDS-PAGE (25). Of course, one possible explanation is that the oligomeric state of a PLB mutant in SDS-PAGE does not correspond to its state in the lipid bilayer and does not reflect the effects of the Ca-ATPase on this oligomeric state. This possibility has been addressed by spectroscopic probes, which have been used to measure oligomeric interactions of PLB in lipid bilayers (4). These studies showed that there is a dynamic equilibrium between monomeric and pentameric PLB in a lipid bilayer and that this equilibrium agrees with SDS-PAGE in the absence of Ca-ATPase (3, 6), but the equilibrium shifts in favor of the monomer in the presence of the Ca-ATPase (4, 7). It has been suggested that inhibitory activity of PLB or its mutants  depends not only on the oligomeric state but also on the affinity for the Ca-ATPase (7,9,27). The present study underscores the complexity of this functional interaction.
We found that the N-terminal acetyl group of Abu-PLB has no effect on oligomeric stability or functional significance on Ca-ATPase inhibition (18). N-terminal acetylation, in addition to its role in targeting proteins for distribution within the cell, has been shown to be important in the function of actin (40,41), and the proteolytic stability of peptides (42). In addition to an increase in steric bulk of the N-terminal end of PLB, the acetyl capping reduces the net positive charge, which could be critical for the interaction with the Ca-ATPase if electrostatic interactions were vitally important. It can be argued that the addition of extra charge would disrupt any interaction surface that existed, even if the forces involved in the interaction were largely hydrophobic. Because no change in PLB inhibitory function occurs when the acetyl cap is omitted, we speculate that the extreme N-terminal end of PLB does not interact at all with the Ca-ATPase. This result is in agreement with the literature (15,43), and it suggests that the N terminus of PLB can be labeled without affecting the interaction with the Ca-ATPase (44). With the recent publication of a high resolution x-ray crystal structure of SERCA1 (19), modeling of the PLB/ SERCA interaction surface should be possible in the near future.
In summary, substitution of the three cysteine residues in PLB with alanine (Ala-PLB) completely destabilizes the PLB pentamer, but substitution by ␣-amino-n-butyric acid (Abu-PLB) causes only slight destabilization. Thus, the thiol groups of the Cys residues of PLB are not required for pentamer formation. In particular, intermolecular disulfide bonds are not necessary. Because Abu is isosteric to Cys, these results indicate that steric packing is the principal factor determining pentameric stability. However, the Abu-PLB pentamer is significantly less stable than WT-PLB at temperatures above 25°C, indicating that specific chemical properties of the thiol groups do play a significant role. Examination of the structure reveals a possible hydrogen bond between Cys-36 and the backbone oxygen of Leu-37 and between Cys-36 and Cys-41. WT-PLB, Abu-PLB, and Ala-PLB have no significant differences in their inhibition of the Ca-ATPase, indicating that neither Cys residues nor oligomeric stability is essential for inhibitory function of PLB. This study illustrates that chemical synthesis offers unique advantages in the analysis of membrane protein structure and function.