A leucine zipper stabilizes the pentameric membrane domain of phospholamban and forms a coiled-coil pore structure.

Phospholamban is a phosphoprotein regulator of cardiac sarcoplasmic reticulum which is phosphorylated in response to β-adrenergic stimulation. Previous results have shown that phospholamban forms Ca-selective channels in lipid bilayers. The channel-forming domain has been localized to amino acid residues 26-52, which form a stable pentameric, helical structure. The specific residues responsible for stabilizing the pentameric membrane domain of phospholamban have been identified by mutational analysis. Residues 26-52 were individually mutated to Ala or Phe, and the ability of the resulting mutant to form a pentamer or other oligomer was assessed by SDS-polyacrylamide gel electrophoresis analysis. Replacement of Leu, Ile, Leu, Ile, or Leu by Ala prevented pentamer formation, indicating their essential involvement in the oligomeric assembly. The heptad repeats, and 3-4-residue spacing of the essential amino acids suggest that residues 37-52 adopt a pentameric coiled-coil structure stabilized by a leucine zipper motif formed by the close packing of Leu, Ile, Leu, Ile, and Leu. The resulting symmetric structure contains a central pore defined by the hydrophobic surface of the five stabilizing leucine zippers, which are oriented to the interior and form the backbone of the pentamer.

PLB 1 is a small oligomeric phosphoprotein of cardiac SR that regulates Ca 2ϩ transport across this intracellular membrane organelle (Tada and Inui, 1983;Sham et al., 1991). Phosphorylation of PLB activates the Ca 2ϩ pump of SR and increases Ca 2ϩ uptake (Lindemann et al., 1983) by a mechanism that remains to be completely understood (James et al., 1989;Cantilina et al., 1993;Colyer, 1993;Toyofuku et al., 1994;Voss et al., 1994). Structural analysis of PLB has been instrumental in the discovery of its activity as a channel protein (Kovacs et al., 1988), but the physiological relevance of this activity to Ca 2ϩ sequestration is not yet clear .
Following the development of a method to purify PLB to homogeneity from SR vesicles, analysis by SDS-PAGE suggested that PLB was a pentamer of identical 5-6-kDa subunits (Wegener and Jones, 1984;Jones et al., 1985) and further indicated that each subunit was dually phosphorylated by cAMP-and Ca 2ϩ /calmodulin-dependent protein kinases (Sim-merman et al., 1986;Wegener et al., 1986). Peptide mapping studies showed that each subunit contained two domains: a cytosolic, hydrophilic domain incorporating the two phosphorylation sites and a hydrophobic domain responsible for the oligomeric subunit interactions . Sequence analysis of PLB confirmed that it is a noncovalent oligomer of identical subunits , each containing 52 amino acids (Fujii et al., 1987). The amino-terminal hydrophilic domain contains the residues serine 16 and threonine 17 phosphorylated by cAMP-and Ca 2ϩ /calmodulindependent kinases, respectively , whereas the hydrophobic domain is located within residues 26 -52. Empirical analysis of the carboxyl-terminal residues 26 -52 suggested that they could form an amphipathic helix sufficiently long to traverse the SR membrane and that five such helices could assemble to a pentameric pore-forming structure . Single channel recording experiments have demonstrated that PLB in planar bilayers does in fact exhibit voltage-regulated Ca 2ϩ channel activity (Kovacs et al., 1988). Although the primary structure of PLB suggests that the membrane-embedded region of the protein is localized to residues 26 -52, the role of specific residues in this domain to the oligomeric stability and channel activity is largely unknown. Solubilized preparations of residues 26 -52 have been shown by circular dichroism spectroscopy to adopt a predominantly helical configuration (Simmerman et al., 1989). Residues 26 -52 incorporated as pentamers into phospholipid bilayers are also mainly helical . Further questions remain regarding the arrangement, orientation, and topography of the native membrane-spanning domain of PLB in SR vesicles.
To probe the structure and stabilizing interactions of the membrane-spanning channel domain of PLB, mutational analysis has been implemented. Mutagenesis has been used successfully to determine the sequence specificity of interacting helices in glycophorin A (Lemmon et al., 1992a(Lemmon et al., , 1992b, heat shock factor (Rabindran et al., 1993), and the GCN4 DNAbinding domain (van Heeckeren et al., 1992;Harbury et al., 1993). It is assumed that replacement of a residue with Ala, which results in disruption of the PLB pentameric assembly, indicates the essential role of that residue in the stabilization of the quaternary structure. Such has been found the case for a de novo synthetic approach to examining the stability of twostranded coiled-coils (Zhou et al., 1992). The results of PLB mutagenesis indicate that a leucine zipper stabilizes the pentameric membrane domain of PLB and forms a coiled-coil pore structure. A preliminary account of this work appeared earlier in abstract form (Simmerman et al., 1994).

EXPERIMENTAL PROCEDURES
Constructs for Mutagenesis and in Vitro Transcription-For efficient in vitro transcription and translation of PLB we inserted the protein coding region of PLB into the BglII site of the cloning vector pSP64T (Krieg and Melton, 1984). Base pairs Ϫ3 through ϩ159 of canine cardiac PLB cDNA  were amplified by polymerase chain reaction using primers with flanking BglII sites and four extra nucleotides at the 5Ј ends. The 176 base pair product was digested with BglII, electroeluted from a 1% agarose gel, and ligated into the BglII site of pSP64T. For mutagenesis, the HindIII-EcoRI fragment of pSP64T containing frog hemoglobin untranslated regions and the PLB protein coding region was subcloned into the HindIII-EcoRI sites of pAlter-1™. Site-directed mutagenesis from pAlter-1™ was conducted with an Altered Sites II Mutagenesis System™ (Promega) according to the manufacturer's instructions. 39-Mer antisense oligonucleotides carrying mismatch bases specific for each mutated residue were used (Kunkel, 1985). The nucleotide sequences of all of the PLB constructs used were confirmed by dideoxy sequencing (Sanger et al., 1977).
In Vitro Transcription and Translation-Plasmids encoding wildtype PLB and PLB mutants were linearized with SalI to make run-off transcripts. In vitro transcription was performed with 2 g of each linearized cDNA using an SP6 MEGAscript™ kit (Ambion). Capping nucleotide concentration was 2.5 mM. In vitro translation was conducted using the Red Nova™ Lysate system without methionine (Novagen). 5 g of cRNA was translated in 25 l of rabbit reticulocyte lysate using [ 35 S]methionine as the label. Translations were conducted for 60 min at 30°C. Reactions were stopped by adding 10 l of 1 mg/ml RNase A and then incubated at ambient room temperature for 24 h for optimal pentamer formation. Labeling efficiency was determined by trichloroacetic acid precipitation of 2-l aliquots from the reaction mixtures. Only samples with equivalent amounts of [ 35 S]methionine incorporation were analyzed further. Pentamer stability of the translated products was assessed by SDS-PAGE in 10% polyacrylamide slab gels using the buffer system of Porzio and Pearson (1977). Generally, 3 l of in vitro translate was added to 7 l of SDS sample buffer (Laemmli, 1970) containing 3% SDS. In order to analyze all of the alanine or phenylalanine mutations in the same slab gel, we used 0.35-mm-thick polyacrylamide gels cast in a nucleic acid sequencing apparatus (Hoefer). For autoradiography, the gels were fixed, dried, and exposed to x-ray film overnight at room temperature. Quantification of percentages of PLB pentamer and monomer formation was performed on the dried gels using a GS-250 Molecular Imager (Bio-Rad).
Expression and Purification of Recombinant PLB-Wild-type PLB and the mutants L37A and C41L were expressed and purified from Sf21 insect cells. Expression and purification of wild-type PLB was recently described . For expression of the PLB mutants L37A and C41L, the BglII inserts encoding these proteins were excised from pAlter-1™ and inserted into the BglII site of the baculovirus transfer vector pVL1393. Sf21 cells were then co-transfected with pVL1393 and linearized baculovirus DNA using the Baculogold TM System (Pharmingen). Purification of the mutated PLB proteins was conducted as recently described, using monoclonal antibody affinity chromatography . Canine cardiac PLB was isolated from SR vesicles as described previously by Jones et al. (1985). Protein concentrations were determined by the method of Schaffner and Weissman (1973).

RESULTS
In Vitro Translation of PLB-In agreement with the earlier work of Cook et al. (1989), we observed that PLB monomers synthesized in a rabbit reticulocyte lysate spontaneously associated to form pentamers, as assessed by SDS-PAGE analysis ( Fig. 1, WT). The percentage pentamer formation was approximately 50% of the total protein synthesized, and the characteristic dissociation of pentamers by boiling in SDS (Wegener and Jones, 1984) was retained. After pentamers, monomers were the next most stable species detected, followed by dimers. The same oligomeric forms of PLB were observed previously in cardiac SR vesicles (Jones et al., 1985), suggesting that recombinant PLB behaved in similar fashion to the natural protein.
Interestingly, we observed that when translates were analyzed by SDS-PAGE immediately after reactions were terminated, only approximately 25% of the PLB monomers had associated into pentamers. Incubating the recombinant monomers for 24 h at room temperature prior to SDS-PAGE increased the proportion of pentamers to approximately 50%, the maximal value obtained. This observation suggested that pentamer formation occurred prior to addition of SDS for PAGE. Inclusion of pancreatic microsomes in the translation system did not affect the extent of pentamer formation (data not shown). However, as pointed out by Cook et al. (1989), rabbit reticulocyte lysates contain some membranes, so the data do not distinguish whether membranes are required for pentamer formation.
Alanine Mutations-Residues 26 through 52 of PLB were individually changed to Ala by in vitro translation of the mutated cRNAs and each mutated protein was then analyzed by SDS-PAGE ( Fig. 1 and Table I). Most of the mutated proteins retained the ability to form pentamers, with the pentameric mobility form contributing 25% or greater of the total protein synthesized. The notable exceptions occurred at Leu residues 37, 44, and 51 and Ile residues 40 and 47, occupying positions a and d of a heptad repeat pattern of residues a-g beginning at Leu 37 . Changing any of these amino acids to Ala prevented pentamer formation (Table I). Replacement of Phe 32 with Ala also prevented pentameric assembly, although this residue does not belong to the repeating heptad pattern of Leu and Ile After 24 h at room temperature, SDS sample buffer was added and electrophoresis was conducted using a 10% polyacrylamide gel (Porzio and Pearson, 1977). The resulting autoradiograph is shown. Pentamer and Monomer denote the pentameric and monomeric mobility forms of PLB, respectively. The diffuse area of radioactivity visible above the PLB monomer band contained PLB dimers, which were poorly resolved in this gel, but were identified clearly with use of a 7-18% gradient gel (data not shown). Ϯ Boil indicates whether WT samples were boiled in SDS prior to PAGE. No RNA is a control showing that negligible 35 S incorporation occurred when PLB cRNA was omitted from the translation mix. After autoradiography, the dried gel was placed in a GS-250 Molecular Imager (Bio-Rad), and the amount of recombinant PLB in each lane was quantified. Background radioactivity in the No RNA lane was deducted for each determination. Eight replicate experiments of this type were performed, and the averaged results for the percentages of PLB pentamer formation for each mutation are listed in Table I. Molecular weight standards (ϫ 10 Ϫ3 ) are shown on the left.
residues. The expression level of recombinant proteins was similar for all the mutants analyzed.
Phenylalanine Mutations-In order to further assess sitespecific contributions to pentamer stability and to map surface orientations, we changed each amino acid in the transmembrane region to Phe ( Fig. 2 and Table I). Leu residues 37 and 44 and Ile residues 40 and 47 were again observed to play key structural roles. Changing any of these amino acids to Phe eliminated pentamer formation. The pentameric stability of several other Phe mutations was also sharply reduced. This included Phe substitutions at Gln 29 , Ile 33 , Asn 34 , Cys 41 , Met 50 , and Leu 51 . It is notable that all residues occupying the b, c, and f positions of the heptad pattern beginning at residue Leu 37 tolerated replacement with Phe, viz. residues Ile 38 , Leu 39 , Leu 42 , Ile 45 , Cys 46 , Val 49 , and Leu 52 .
Changing Cys 41 to Phe was previously reported to destabilize the pentamer (Fujii et al., 1989), as is confirmed in Fig. 2 and Table I. We also tested if substitution with another hydrophobic residue at this position would disrupt the pentamer. Unexpectedly, changing Cys 41 to Leu led to a new oligomeric species running with a mobility expected for a tetramer (Fig. 2). Formation of tetramers by the C41L mutation is characterized below.
Leucine/Isoleucine Mutations-To further analyze the roles of the critical Leu and Ile residues at positions 37, 40, 44, 47, and 51 in maintaining pentamer stability, we mutated each of these residues to its structural isomer ( Fig. 3 and Table II). The requirement for Ile at position 40 appeared to be stringent, as replacement of this amino acid with Leu completely prevented pentamer formation. The stability of the pentamer was also substantially reduced, but not entirely eliminated, when Leu 37 was changed to Ile. In contrast, changing Leu 44 to Ile, Ile 47 to Leu, or Leu 51 to Ile did not have a major effect on pentamer formation.
Mutant Proteins Purified from Sf21 Cells-In order to confirm the reliability of the in vitro translation system for identifying the amino acid residues contributing to PLB pentamer stability, and to provide additional information on the different PLB mobility forms, we expressed and purified PLB containing two of the more revealing mutations from Sf21 insect cells . The mutations chosen for further analysis, L37A and C41L, preferentially formed monomers and tetramers, respectively, when analyzed by the in vitro translation method. PLB mutated at these two positions, as well as recombinant wild-type PLB and natural PLB purified from SR vesicles, were subjected to gradient gel electrophoresis followed by Coomassie Blue staining, in order to permit direct visualization of all five mobility forms of the protein. Fig. 4 demonstrates that wild-type PLB, purified from either SR vesicles or Sf21 cells, migrated predominately as a pentamer, with some monomers and dimers also present. Boiling the wild-type proteins in SDS prior to PAGE dissociated the pentamers and allowed the visualization of tetramers and trimers, as well as the other two mobility forms. Most importantly, the C41L mutation migrated principally as a tetramer, aligning exactly with tetramers formed from wild-type PLB boiled in SDS. A very weak pentamer band could be detected with the C41L mutation, which eliminated the trivial possibility that the tetrameric form of C41L was actually a pentamer but with an aberrant mobility. As observed by in vitro translation, the L37A mutation migrated principally as a monomer. Some dimers were also detected, which were also observed with in vitro translates when gradient gels were used. In other experiments we were able to detect a minor amount of pentamer formation by L37A, but only when electrophoresis through the stacking gel was conducted at a very slow rate with a very high concentration of L37A (Ն20 g of protein/gel lane) (data not shown). Thus, unique association constants for each of the mutant proteins analyzed may exist, the determination of which is beyond the scope of this study.

DISCUSSION
In the present site-directed mutagenesis study, residues responsible for stabilizing the pentameric structure of PLB were identified by mutation to Ala and were found to be Leu 37 , Ile 40 , Leu 44 , Ile 47 , and Leu 51 (Fig. 1). The mechanism by which substitution of Ala for the critical stabilizing residues prevents the pentameric structure is not by disrupting the native alpha helical structure, since Ala is a strong helix-forming residue (Chou and Fasman, 1978). Furthermore, residues neighboring the stabilizing amino acids may be changed to Ala with no effect on quaternary structure and, therefore, presumably no effect on secondary structure either. We conclude that specific characteristics of the aliphatic side chains of Leu 37 , Ile 40 , Leu 44 , Ile 47 , and Leu 51 stabilize the oligomeric structure of PLB.
The most striking observation from the results is the spacing between critically sensitive residues: an alternating series of three leucines and two isoleucines with each isomeric position separated from the next by seven residues. The heptad repeat pattern is a diagnostic feature of the leucine zipper structural motif (Landschulz et al., 1988), in which the region containing the leucine heptad repeats forms a helix, and the leucines line up along one face of the helix (at a pitch of 3.5 residues/turn) to promote oligomerization of the helices in a parallel orientation. In PLB the repeating isoleucines are offset at 3-4-residue intervals from the leucine heptad repeat (Fig. 5A), conforming to the model of a coiled-coil in which hydrophobic residues occupy positions a and d of a repeating heptad of a-g residues (O'Shea et al., 1989;Zhou et al., 1992;Zhu et al., 1993). Thus Leu 37 , Leu 44 , and Leu 51 occupy the a position, whereas Ile 40 and Ile 47 occupy the d position in the motif. The present mutagenesis results are thus consistent with a model of PLB in which residues 37-52 form a 3.5 residue/turn helix, creating a leucine zipper of three helical turns and a complimentary pair of aligned isoleucines with the appropriate axial spacing of two helical turns to interdigitate with the leucines, forming a symmetric, coiled-coil pentamer (Fig. 5B). Five identical zippers formed by interaction between the three leucines of one helix with the two isoleucines of the adjacent helix thus stabilize the PLB quaternary structure. Mixed Leu/Ile zippers have been observed previously (Cohen and Parry, 1990;Atkinson et al., 1991), and the coiled-coil model for the parallel PLB helices is consistent with the observation that parallel orientation of helices is very unusual except in a coiled-coil structure (Oas et al., 1990). The length of the region containing sites important for PLB pentamer formation further suggests extensive interhelical interactions that are more consistent with a coiled-coil model (Zhou et al., 1992) and are not consistent with a single closest approach crossover point between adjacent rigid helices. The proposed model of PLB residues 37-52 (Fig. 5B) suggests that residues occupying the b, c, and f axial positions are oriented to the exterior surface of the structure and that the PLB pentamer would be insensitive to a bulky residue substitution at these sites. Our observation that all residues predicted to occupy the b, c, and f positions of the heptad repeat, viz. Ile 38 , Leu 39 , Leu 42 , Ile 45 , Cys 46 , Val 49 , and Leu 52 , accepted Phe without loss of pentamer formation supports our model of this domain as a coiled-coil bundle of leucine zipper helices with these residues oriented to the exterior.

FIG. 2. In vitro translation of phenylalanine mutations.
Amino acid residues 26 -31, 33-34, and 36 -52 of PLB were individually changed to phenylalanine, and the mutant proteins were in vitro translated and analyzed as described in the legend to Fig. 1. In addition, a cysteine 41 to leucine (C41L) mutation was also analyzed. The percentages of pentamer formation for each of the phenylalanine mutations are listed in Table I. FIG. 3. In vitro translation of leucine/isoleucine mutations. Residues 37,40,44,47, and 51 of PLB were mutated to leucine or isoleucine (listed at top of figure) by in vitro translation and analyzed as described in Fig. 1. The percentages of pentamer formation from four separate experiments of this type are listed in Table II.  . 4. SDS-PAGE of purified wild-type (WT) PLB and C41L and L37A mutations. 11 g of PLB purified from canine cardiac SR vesicles (Cardiac) and insect cells (Sf21 Expressed) were electrophoresed in an 8 -17% polyacrylamide gradient gel according to Laemmli (1970), and the gel was stained with Coomassie Blue. WT and mutant proteins are designated at the top of the figure. Ϯ Boil indicates whether samples were boiled in 5% SDS immediately prior to electrophoresis. The numbers 5 to 1 in the left margin indicate the pentameric through monomeric mobility forms of PLB.
In contrast to the absence of steric hindrance inherent to the b, c, and f axial positions of the leucine zipper helical domain model, residues occupying the e and g positions are in closer proximity in the cleft between adjacent helices (Fig. 5B). Although not specifying the primary stabilizing interactions between helices, these sites may contribute secondarily to oligomeric stability through interactions between their side chains (Hu et al., 1993), and these sites are more likely to exhibit a restricted range of acceptable substitutions based on the volume occluded by adjacent structure. Residues Cys 41 and Ile 48 in the e position and residues Leu 43 and Met 50 in the g position all accepted replacement with Ala without preventing pentamer formation. However, replacement with Phe at positions 41 and 50 prevented pentamer formation, but the bulky side chain did not prohibit pentamer formation at positions 43 and 48. The results suggest that the cleft positions within the PLB leucine zipper helical structure are not equivalent, with positions 41 and 50 more restricted than positions 43 and 48. The formation of a PLB tetramer upon substitution of Cys 41 with Leu ( Figs. 2 and 4) further indicates a unique role for this site in the higher order structure of PLB. A site-directed mutagenesis study of the role of PLB Cys residues in pentamer stability (Fujii et al., 1989) has shown previously that PLB quaternary structure is most intolerant of changes in Cys 41 . Pentamer formation and stability decreased as the size of the substituted side chain increased. Confirmed by the present work, these results are best interpreted as the effect of steric hindrance on pentamer formation and the disruption of close packing on stability, consistent with the model of PLB presented here in which Cys 41 is confined to an interfacial cleft between adjacent helices.
It has been proposed that replacement of the Leu at positions a or d in a leucine zipper with the ␤-branched Ile may influence the stoichiometry of the oligomer formed through the specificity of packing interactions (DeGrado et al., 1989;Zhu et al., 1993). This has been demonstrated using mutants of the GCN4 leucine zipper (Harbury et al., 1993). However, that study did not address structural determinants for oligomers larger than a tetramer. Our results indicate that the PLB pentamer is primarily stabilized by interactions between leucines in the heptad position a (residues 37, 44, and 51) with isoleucines in position d (residues 40 and 47) (Fig. 4). According to the algorithm derived from GCN4 mutants, occupancy of the a position with Leu and the d position with Ile would be expected to specify a tetramer. As this is not observed with PLB, other residues in the PLB heptad pattern must therefore be involved in specifying the number of subunits in the oligomer. Our observation that substitution of Cys 41 for Leu results in a PLB tetramer argues that this site, in the e position of the heptad, is important in determining the aggregation number of the PLB oligomer. It thus appears that whereas occupancy of positions a and d with either Leu and Ile is necessary for the coiled-coil motif and primarily responsible for the stability of the resulting oligomer, it is not sufficient to specify the number of helices in the coiled-coil quaternary structure.
To probe the role of ␤-branched residues at positions a and d in determining the oligomeric state of PLB, we independently mutated Leu 37 , Leu 44 , and Leu 51 to Ile and changed Ile 40 and Ile 47 to Leu. Mutations of Leu 44 , Ile 47 , and Leu 51 had no effect on the formation of the native PLB pentamer, whereas mutation of Leu 37 to Ile nearly completely eliminated pentameric assembly of PLB and mutation of Ile 40 to Leu completely abolished the ability of the PLB monomers to associate (Fig. 3). Intermediate sized oligomers were not observed for any of these mutations. These results confirm the concept that in PLB the a and d heptad positions stabilize the pentameric assembly but do not specify the number of PLB helices in the oligomer. The tolerance for substitution of Leu/Ile isomers at the three Cterminal positions 44, 47, and 51 may indicate a greater structural flexibility in this region of the oligomer. This suggests that the contributions of the leucine zipper residues to the PLB pentamer are not equivalent, with residues Leu 37 and Ile 40 contributing more importantly than residues Leu 44 , Ile 47 , or Leu 51 . Taken together with the special sensitivity of Cys 41 to mutation, the results suggest that specific contacts in the region of residues 37-41 are primarily responsible for stabilizing the pentameric PLB assembly, with contacts in other regions of the coiled-coil serving as secondary sources of stabilizing energy. The location of these critical residues at the N terminus of the leucine zipper motif suggests that correct initiation and propagation of the nascent leucine zipper helix during translation is important for native PLB quaternary structure.
The prevention of pentamer formation observed by mutation of Ile 33 to Phe and also by mutations at Gln 29 , Phe 32 , Ile 33 , and Asn 34 indicates that residues 26 -36 must contribute in some way to the PLB pentameric structure, as suggested previously . Although Ile 33 fits the PLB pattern of alternating Ile and Leu residues at 3-4 residue spacing, the zipper must not extend N-terminally from Leu 37 to Ile 33 , as Ala is accepted at the latter site without loss of pentamer formation. Therefore, the side chain of Ile 33 does not contribute to the essential hydrophobic interactions stabilizing the pentamer. The aromatic side chain of Phe 32 appears obligatory at this position, as its replacement with Ala prevents formation of the PLB pentamer. The essentiality of a specific, outward facing orientation at this structural site to enable pentamer formation suggests that Phe 32 may play a key function in correct folding of secondary structural elements during translation, prevent- FIG. 5. Heptad repeat cartoon model of PLB monomer and pentamer. A, residues 37-52 of monomeric PLB are configured as a 3.5 residues/turn helix with positions a through g of the heptad repeat circled. Leucine and isoleucine residues constituting the zipper are localized to positions a and d, respectively (filled circles). Changing any amino acid (circled) here to alanine destabilizes the pentamer. B, PLB pentamer. The five monomers are arranged to allow intermolecular contact at positions a and d (shaded), such that the leucine and isoleucine residues of one monomer align with leucine and isoleucine residues of adjacent monomers. Alanine substitutions here disrupt the pentamer. Alanine substitutions at positions e and g do not affect stability, but some phenylalanine mutations here are disruptive.
ing premature initiation of the zipper-type helix at Ile 33 . Thus we propose that although residues 26 -36 appear uninvolved in the close-packed hydrophobic interactions stabilizing the pentamer, they are critical for appropriate initiation and folding of PLB leucine zippers and their post-translational orientation and assembly into the pentamer. Interestingly, the belt of Phe 32 and Phe 35 aromatic side chains encircling the pentameric structure defines the interface between the polar and nonpolar regions of PLB. PLB shares this structural motif with membrane channel porins (Cowan et al., 1992;Weiss and Schulz, 1992), in which the band of aromatic residues may protect polar structural elements at the membrane interface from conformational fluctuations during vertical displacements of the protein within the membrane (Kreusch et al., 1994). It may be that PLB phenylalanines also serve to correctly position PLB within the bilayer and to protect polar interfacial residues.
The PLB oligomeric structure is stabilized only by contacts between hydrophobic residues without contribution from side chain electrostatic or hydrogen bonding interactions, commonly observed in other natural coiled-coils (Talbot and Hodges, 1982;Cohen and Parry, 1990). Synthetic two-stranded coiled-coils having only Leu at the a and d heptad positions are much more stable than naturally occurring two-stranded coiled-coils containing a significant proportion of Val and Ala at the critical a and d heptad positions (Hodges et al., 1990). It was thus suggested that natural coiled-coil proteins have not evolved for maximum stability. Considering that the PLB coiled-coil pentamer is stabilized only by Leu and Ile interactions, it appears that the PLB oligomeric structure has uniquely evolved very specifically for great stability.
The PLB membrane domain does not conform to the model of integral membrane proteins as "inside-out" with respect to the relative polarity of surface and buried residues (Engelman et al., 1986). Instead, it is a cylinder of close-packed hydrophobic helices (Fig. 5), which may fulfill predictions that the coiled-coil leucine zipper motif may serve as the defining structure for certain channels (Lear et al., 1988;Karle et al., 1988;DeGrado et al., 1989;Popot and Engelman, 1990;Harbury et al., 1993). Modeling suggests the pore formed by a pentameric coiled-coil would have a diameter of 5 Å (Lear et al., 1988), sufficient to accommodate a water molecule (2.8-Å diameter) or a hydrated small ion. The association of channel activity with the pentameric domain of PLB residues 26 -52 emphasizes the utility of PLB as a natural model system for elucidating channel structure/function relationships and mechanisms of ion conductance, as suggested previously (Kovacs et al., 1988).
While this work was in progress (Simmerman et al., 1994), a study was published using chimeric protein analysis to examine PLB transmembrane domain structure (Arkin et al., 1994). The two approaches concur that amino acids stabilizing the pentamer line up on faces of a 3.5 residues/turn helix, and the resulting pore structure is defined by hydrophobic residues. However, the wild-type chimeric protein studied by Arkin et al. (1994) did not associate with the same oligomeric distribution as native PLB. The random substitutions created by saturation mutagenesis were confined to residues 35-52, and the enabling role of residues 26 -36 in PLB pentameric assembly was not reported. The use of an indirect, qualitative detection method also limited identification of some structurally important residues. For example, the PLB fusion protein containing C41L was expressed as a monomer by Arkin et al. (1994), whereas we clearly show that when native PLB is substituted with C41L, the protein is a tetramer. Our results provide evidence that the leucine zipper helical motif can indeed stabilize an oligomeric membrane domain to form a coiled-coil structure with potential relevance to ion channels.