Assembly of the Oligomeric Membrane Pore Formed by Staphylococcal a-Hemolysin Examined by Truncation Mutagenesis*

The a-hemolysin (aHL) from Staphylococcus aureus causes the lysis of susceptible cells such as rabbit erythrocytes (rRBCs). Lysis is associated with the formation of a hexameric pore in the plasma membrane. Here we show that truncation mutants of aHL missing 2 to 22 N-terminal amino acids form oligomers on the surfaces of rRBCs but fail to lyse the cells. By contrast, mutants missing 3 or 5 amino acids at the C terminus are very inefficient at oligomerization but do lyse rRBCs, albeit extremely slowly. The C-terminal truncation mutants, retarded as monomers on the cell surface, undergo a conformational change in which the protease-sensitive loop located near the midpoint of the polypeptide chain becomes occluded. Judged by this criterion, polypeptides truncated at the N terminus, frozen as nonlytic oligomers, are in a similar conformation. A second proteolytic site near the N terminus of the polypeptide becomes inaccessible in the lytic pore formed by the wild-type polypeptide, supporting the idea that a second conformational change occurs upon pore formation. These findings suggest a pathway for assembly of the lytic pore in which aHL first binds to target cells as a monomer, which is converted to a nonlytic oligomeric

The a-hemolysin (aHL) from Staphylococcus aureus causes the lysis of susceptible cells such as rabbit erythrocytes (rRBCs). Lysis is associated with the formation of a hexameric pore in the plasma membrane. Here we show that truncation mutants of aHL missing 2 to 22 N-terminal amino acids form oligomers on the surfaces of rRBCs but fail to lyse the cells. By contrast, mutants missing 3 or 5 amino acids at the C terminus are very inefficient at oligomerization but do lyse rRBCs, albeit extremely slowly. The C-terminal truncation mutants, retarded as monomers on the cell surface, undergo a conformational change in which the protease-sensitive loop located near the midpoint of the polypeptide chain becomes occluded. Judged by this criterion, polypeptides truncated at the N terminus, frozen as nonlytic oligomers, are in a similar conformation. A second proteolytic site near the N terminus of the polypeptide becomes inaccessible in the lytic pore formed by the wild-type polypeptide, supporting the idea that a second conformational change occurs upon pore formation. These findings suggest a pathway for assembly of the lytic pore in which aHL first binds to target cells as a monomer, which is converted to a nonlytic oligomeric intermediate before formation of the pore. In keeping with this model, an N-terminal truncation mutant blocks the slow lysis induced by a C-terminal truncation mutant, presumably by diverting the weakly lytic subunits into inactive oligomers.
Information about the assembly of oligomeric membrane proteins is important for understanding the biosynthesis of integral membrane proteins, the actions of cytolytic toxins and immune proteins, and how receptors aggregate with each other or with regulatory proteins. While considerable knowledge is accumulating about the integration into membranes of individual protein subunits (1)(2)(3)(4), including large polypeptides with multiple domains (5), far less is known about how membrane protein subunits come together. Recent findings concerning Na/K-ATPase (6), acetylcholine receptor (7), complement C9, and perforin (8) have demonstrated intermediates in assembly in complex cellular systems. A more complete picture would be available if such work were combined with i n vitro biochemical and biophysical studies of assembly intermediates (4).
* This study was supported by the Department of Energy, Divisions of Energy Biosciences and Materials Sciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Because of its simplicity, a-hemolysin (aHL)' is a model system for studying the assembly of an oligomeric membrane pore (9). The aHL polypeptide of 33,200 Da (293 amino acids) is secreted by Staphylococcus aureus as a water-soluble monomer, which assembles into cylindrical pores in susceptible cell membranes (10,11). The secondary structure of both the monomer and hexamer is predominantly @-sheet (12,13). It has been proposed that aHL undergoes a conformational change during assembly, involving the separation of two rigid domains connected by a loop near the midpoint of the polypeptide chain, converting the molecule into an amphipathic rod, six of which form a pore (12). A glycine-rich sequence (residues 119-143) (14), thought to be the loop, is susceptible to proteases in the monomer but not in the hexamer (12). Here, we use deletion mutagenesis to demonstrate two potential intermediates in assembly: a membrane-bound monomer in which the loop is occluded and a nonlytic oligomeric pore precursor of variable stoichiometry.
Deletion Mutagenesis-Fragments of the aHL gene were amplified by the polymerase chain reaction from the plasmid pT7NPH-8 in which the internal NdeI site in the gene has been removed by mutagenesis (17). This copy of the aHL gene contains a G + A substitution at position 1058 resulting in the replacement of Ser-217 by a Gln residue, which has no effect on the activity of the polypeptide. . The EcoRI, PstI, and ribosome binding sites at the 5'-ends of these primers were not used in the work described here. Each primer contained an NdeI site at the initiation codon followed by an alanine codon. The latter ensures efficient cotranslational removal of the N-terminal methionine (17) and limits proteolytic degradation (18). Residue 1 in staphylococcal aHL (s-aHL) is also alanine. The 5"primer for the C- The abbreviations used are: aHL, a-hemolysin of S. aureus; DOC, deoxycholate; r-aHL, recombinant a-hemolysin isolated from E. coli; RBC, red blood cell; rRBC, rabbit RBC; SDS, sodium dodecyl sulfate; s-aHL, wild-type a-hemolysin isolated from S. aureus.
terminal truncations was 5' CGGGATCCTAATACGACTCACTA-TAGGG, which hybridizes to the T7 promoter region upstream from the NdeI site in pT7NPH-8. The 3"primers were 5' GCGCAAGCT-TTCATTATTCTTCTTTTTCCCAATCG, aHL(1-290); 5' GCG-CAAGCTTTCATTATTTTTCCCAATCGATTTTATATC,aHL(l-288); 5' GCGCAAGCTTTCATTAATCGATTTTATATCTTTCTG, trHL(1-285). These primers encode a HindIII site followed by two termination anticodons and the desired anticodons of the aHL sequence. The amplification products were cut with NdeI and HindIII and inserted into the pT7 vector pT7SflA, a derivative of pT7flA (19) containing a synthetic transcription stop signal inserted at the HglII site upstream from the T7 promoter to prevent read-through from the ampicillin promoter (17). To obviate possible problems with polymerase chain reaction errors the number of cycles was minimized (usually lo), and each experiment described herein was done at least three times with each of three independent clones. The following additional mutants were constructed by the same method aHL ( Coupled in Vitro Transcription and Translution-Supercoiled plasmid DNA was used for in vitro transcription/translation in an Escherichia coli S30 extract (Promega L464) in the presence of T7 RNA polymerase, rifampicin, and [%]methionine (17). For hemolysis assays, a complete amino acid mix was used to increase the concentration of methionine, ensuring the synthesis of concentrations of mutant polypeptides in the range 10-50 pg/ml. For the remaining experiments described here, in vitro transcription/translation was carried out in the presence of [3sS]methionine of high specific radioactivity (17). All aHL constructs described here were transcribed and translated with efficiencies closely similar to that found for wild-type r-trHL.
Hemolysis Assay"S30 extracts were serially diluted in microtiter wells with 20 mM KH2P0.,, 150 mM NaCI, pH 7.4, containing 1 mg/ ml bovine serum albumin (K-PBSA) before the addition of an equal volume of 1% rabbit red blood cells (rRBCs). The extent of hemolysis in each well was estimated after 1,4, 18, and 24 h a t 20 "C (20). The concentrations of the polypeptides were subsequently quantitated by SDS-polyacrylamide gel electrophoresis and autoradiography, taking into account the number of methionine residues in each mutant. Where necessary small corrections were made to the hemolytic titers.
Oligomer Formation on rRBCs-To investigate oligomer formation as shown in Fig. lA and Table I (entry B), 5.0 pl of the S30 mix was incubated with 10% rRBCs (50 pl) for 30 min on ice in K-PBSA. The cells were recovered by centrifugation at 4 "C, resuspended in 50 pl of K-PBSA, and brought to 37 "C for 2 min. s-aHL (1 pl, 1.8 mg/ml; prepared as described by Walker et al. (17)) was then added, and the cells were incubated at 37 "C for 15 min, during which lysis occurred. Two variations of this experiment were also performed. In the first (Table I, entry A), no chase with unlabeled s-aHL was done. In the second (Table I, entry C), the mutant aHL was premixed with unlabeled s-aHL (1 pl, 1.8 mg/ml) and the final incubation was carried out at 20 "C rather than 37 "C. Membranes or cells were recovered by centrifugation, dissolved in 2 X Fairbanks loading buffer, warmed at 45 "C for 5 min, and subjected to electrophoresis in a 4.5% SDS-polyacrylamide gel (17). The stoichiometries of the oligomers, as recorded in Table I, were assigned by assuming that the major species formed by ["SJr-aHL is a hexamer (12,17) and a linear relationship between the mass of an oligomer and the negative log of the migration distance. For entries A and B in the Table, it was assumed that homo-oligomers were formed. For entry C and the experiments with DOC (see below), it was assumed that hetero-oligomers containing a single mutant subunit were formed.
Oligomer Formation in DOC-To investigate oligomer formation in detergent, mutant "S-labeled polypeptides in S30 mix (7.5 pl) were mixed with 10 mM methionine, 100 mM Tris-HCI, pH 8.0, containing 1 mM 0-mercaptoethanol, followed by the addition of 1 mM leupeptin (1 pl) and s-aHL (1.8 mg/ml; 30 pl). Sodium DOC (32 mM, pH 8.2, containing 10 mM Tris-HCI) was then added over 30 min to 6.4 mM (4 X 2.5 pl), followed by a 20-min incubation, all at room temperature. After each addition the mixture was stirred for a few seconds. Finally, one volume of 2 X Fairbanks loading buffer was added at room temperature, followed by analysis in a 4.5% SDS-polyacrylamide gel (17). The addition of cold methionine greatly reduced the intensity of faint nonpertinent bands apparently caused by, among other things, the incorporation of ["S]methionine into s-aHL. Nevertheless, after treatment with DOC extremely faint bands corresponding to oligomers were noted with several mutants that did not oligomerize at all on rRBCs (e.g. see Table I).
Limited Proteolysis-Mutant "S-labeled polypeptides in S30 mix (10 pl) were allowed to bind to 5% rRBCs in K-PBSA (90 pl) for 1 h at 20 "C. In the experiment displayed in Fig. 2, the cells were washed twice with K-PBSA, resuspended in the same buffer (20 pl), and treated with proteinase K (0.1 pg) for 0, 2, 15, and 60 min at 20 "C. "S-Labeled polypeptides, diluted 80-fold in K-PBSA without the prior addition of rRBCs, were also treated with proteinase K under the same conditions. After the addition of phenylmethylsulfonyl fluoride (0.4 pl, 50 mM) to each sample, it was heated in Laemmli loading buffer for 5 min at 95 "C and subjected to electrophoresis in a 12% SDS-polyacrylamide gel (17). In the experiment displayed in Fig. 3, proteolysis was with 1 pg of proteinase K for 5 min at 20 "C.

RESULTS AND DISCUSSION
Hemolytic Actiuity of aHL Truncation Mutants-a-Hemolysin expressed in E. coli cells or lysates (r-aHL) is essentially identical to staphylococcal aHL (s-aHL) as judged by criteria based on protein chemistry, the ability to bind to erythrocytes and to form oligomers on cells or in detergent, hemolytic potency, and the properties of single channels formed in lipid bilayers (17). In the present work, truncation mutants of aHL were made by ligating segments of the hemolysin gene, generated by polymerase chain reaction amplification, into a T7 transcription vector followed by coupled in uitro transcrip-

Properties of aHL deletion mutants
Hemolytic activity is recorded as the well in a 2-fold serial dilution in which 50% hemolysis of rRBCs occurred. The final dilution in well 1 was 1:4. In parentheses are the relative concentrations at which hemolysis took place (r-aHL = 1). The ability of a mutant polypeptide to bind to rRBCs was determined by experiments such as that shown in Fig. L4. Oligomer formation on rRBCs was determined using three separate procedures: A, no chase with unlabeled s-aHL; B, chase with unlabeled s-aHL (for an example see Fig. LA); C, mutant polypeptide and unlabeled s-aHL premixed. Oligomer formation in DOC was determined as described in Fig. 1B. Key: s, more than 50% conversion to oligomer; w, weak oligomer band; vw, very weak oligomer band; ?, extremely weak oligomer band of uncertain origin in DOC experiment.
Protease resistance was determined as shown in Fig. 2 (12)). The open arrows indicate fragments generated after cleavage a t sites near the N terminus (molecular mass of -32 kDa based on I4C-labeled markers). tion/translation in an E. coli ,330 extract supplemented with T 7 RNA polymerase and rifampicin. All seven of the mutants described here in detail (Table I) bound tightly to rabbit erythrocytes (rRBCs; see below). Mutants with N-terminal truncations as small as two amino acids (aHL(A3-293)) exhibited no hemolytic activity at all toward rRBCs (Table I).
Oligomerization of N-terminal Truncation Mutants-The mutant polypeptides were tested for the ability to form SDSstable oligomers (21) on rRBC membranes (Fig. a). When 2 or 11 amino acids were deleted at the N terminus (aHL(A3-293), aHL(A12-293)), aHL formed mainly pentamer, traces of hexamer, and a lower mass species, probably a tetramer. An N-terminal truncation mutant lacking 22 amino acids (aHL(A23-293)) formed a hexamer, albeit inefficiently. Removal of 38 amino acids or more (aHL(A39-293), and data not shown) resulted in complete loss of the ability to form stable oligomers on rRBCs. Closely similar results were obtained whether or not treatments of rRBCs with mutant polypeptides were followed with a chase with unlabeled sa H L (Table I, entries A and B), showing that homo-oligomers formed by the N-terminal truncation mutants were locked in states that prevented their conversion to s-aHL-containing hetero-oligomers, which differ in size distribution (see below).
While oligomerization of a H L is presumably triggered by receptors on susceptible cells (20,22), hexamers are generated in vitro by treatment of monomeric a H L with heat (23), DOC (24), or lipids (13). aHL(A3-293), aHL(A12-293), and aHL(A23-293), but not aHL(A39-293), were efficiently incorporated into oligomers in DOC when premixed with excess s-aHL (Fig. 1B). The size distributions of these structures differed from those formed by the mutant subunits on rRBCs (Fig. 1A) suggesting that these N-terminal mutants, defective in lysis but capable of self-assembly, can form hetero-oligomers containing s-aHL subunits. This finding was confirmed by experiments with rRBCs using mutant subunits premixed with s-aHL in which the size distributions of the oligomers were very similar to those in DOC ( Table I, entry C).
Oligomerization of C-terminal Truncation Mutants-Deletions of 3 or 5 amino acids at the C terminus (aHL(1-290), aHL(1-288)) greatly inhibited the ability of aHL to form oligomers on rRBCs. Traces of hexamer were observed, which were no longer detectable after the deletion of eight or more C-terminal amino acids (aHL(1-285)) (Fig. 1A). Hexamer formation by aHL(1-290) and aHL  was even less efficient when the s-aHL chase was omitted, suggesting that these mutants are better incorporated into hexamers that include the full-length polypeptide. Traces of hexamer were also observed after treatment of aHL(1-290) and aHL  with DOC in the presence of s-aHL. It is presumably hexamer that causes the slow lysis of rRBCs by aHL(1-290) and aHL(1-288) ( Table 1).
Conformations of Mutant aHL Polypeptides Probed by Limited Proteolysis-The conformations of aHL(A3-293) and aHL(1-290), the mutant polypeptides with the smallest Nand C-terminal truncations, were probed in solution and on the surface of the red cell by limited proteolysis with proteinase K (12). After binding to rRBCs, both resisted cleavage in the loop region (Fig. 2). Proteinase K has previously been shown to cleave aHL in solution before Val-140 and Ile-136 (major sites) and before Asn-139 and Gly-134 (minor sites) (25). Even though aHL(1-290) is almost entirely monomeric under these conditions (Fig. 1A), quantitative autoradiography indicated that the glycine-rich loops of 30 f 15% of the bound polypeptide were resistant to proteolysis. A plausible interpretation is that this fraction of aHL(1-290) undergoes a conformational change on the rRBC surface (see below). The resistant fraction of bound aHL( 1-290) was nonetheless cleaved at additional sites that were deduced to lie within 11 amino acids of the N terminus by comparison of the proteolytic patterns generated from several truncation mutants bound to rRBCs (Fig. 3). The N-terminal site of cleavage was further confirmed by using the labeled mutants [35S-Cys] aHL-S3C and ["S-Cys]aHL-T292C. Wild-type a H L contains no cysteine residues so the radiolabel acts as a marker for an intact N or C terminus. Radiolabel was present in membrane- (see "Experimental Procedures") showed that the concentrations of the aHL mutants in the S30 extracts prior to dilution were almost identical. The final dilution of the S30 mixes in well 1 was 1 in 8. After the addition of an equal volume of 1% rRBCs, the extent of hemolysis was estimated a t various time points ( t ) . T h e results shown were averaged from four experiments. Inhibition of hemolysis is far greater than expected were cuHL(1-290) simply diluted by the aHL(A3-293) preparation.
bound, -32-kDa fragments of aHL-T292C, but not in fragments of aHL-S3C.* The N-terminal truncation mutant (aHL(A3-293)) was also cleaved near the N terminus (Figs. 2B and 3). The cleavage proceded to completion but more slowly than with aHL(1-290), perhaps because of steric hindrance to proteolysis in the nonlytic oligomeric state. By contrast, a large fraction of the full-length polypeptide, r-aHL, was completely resistant to cleavage (Fig. 2 A ) . While this may indicate a second conformational change upon formation of the aHL pore, it must be recognized that the N-terminal truncation could itself affect the rate of proteolysis. The small fraction of full-length r-aHL bound to rRBCs that is cleaved at the N-terminal sites is likely to be residual monomer (Fig. 2 4 ) .
Interpretation of the Mutagenesis Experiments-One drawback of any study using mutagenesis is the possibility that the mutant polypeptides are incorrectly folded (26). Several arguments suggest that the present study is less vulnerable than many to this criticism. First, while several of the mutants B. J. Walker and H. Bayley, unpublished results were more sensitive than r-aHL to proteinase K in solution, they did exhibit initial cleavage in the loop region. When bound to rRBCs the loop was protected, but a site near the N terminus was cleaved. This conservation of the proteolytic sites found in wild-type aHL suggests that the folding of the mutant polypeptides approximates the native structure. Second, the mutant polypeptides that have been described in most detail here are not inactive but defective in specific and varied aspects of assembly. Third, although modifications that alter the phenotype of aHL need not be drastic (e.g. in aHL(A3-283) the N-terminal Ala-Asp-Ser.. . becomes Ala. . ., and in aHL(1-290) just 3 residues . , .Met-Thr-Asn are removed from the C terminus), we are not advocating the involvement of deleted amino acids in specific functions. Rather, we are suggesting that selective modification of the Nor C-terminal domains leads to the accumulation of interesting assembly intermediates. This reserved interpretation remains valid if the mutations produce regional disturbances in folding rather than simple voids in the protein structure. The recent finding that the N-and C-terminal halves of aHL can be separately synthesized and reconstituted into a functional hemolysin3 strengthens the assumptions that the polypeptide can be considered as a two-domain structure (12) and that at least the unaltered half of each truncation mutant is properly folded. In summary the main conclusions of the study seem incontrovertible: 1) N-terminal lesions in aHL yield polypeptides that form oligomers without hemolytic activity; 2) C-terminal lesions yield polypeptides that bind to rRBCs as monomers in which the loop connecting the N-and C-terminal domains is occluded.
Model for Assembly of the aHL Pore-The data are consistent with a working model for the assembly of aHL (Fig. 4), which provides a succinct summary of the biochemical findings. At this point, other possibilities cannot be strictly excluded, including those in which the monomer is lytic (for discussions see Refs. 11,[27][28][29]. The N-terminal truncation mutants, aHL(A3-293), cuHL(A12-293), and aHL(A23-293), are trapped as nonlytic oligomers (3), suggesting that the step 3 + 4 is blocked. By contrast, the C-terminal truncation mutants, aHL( 1-290) and aHL( 1-288), appear to be retarded a t 2, membrane-bound monomers with an occluded loop region. These mutant polypeptides are very slowly converted to hexamers with concomitant cell lysis (4). In keeping with the suggested sequence of events, aHL(A3-293) blocks the slow hemolysis induced by aHL(1-290) (Fig. 5), without affecting the binding of the latter to rRBCs (data not shown). The model is consistent with earlier findings (12,27,30,31). phobic reagents (27), while a nonlytic oligomeric pore precursor (3) has been proposed, because proteolytically cleaved s-aHL forms oligomers but does not lyse cells (30). Preliminary structural studies also suggest that oligomeric forms of s-aHL that do not fully penetrate the membrane can be trapped in two-dimensional crystals (31). Besides unifying previous mechanistic data and proposals, the present findings suggest that the C terminus of the polypeptide is involved in the initial aggregation of the aHL monomers, while an intact N terminus is required to complete pore formation (Fig. 4). As r-aHL can be obtained in milligram amounts (17), it should now be possible to isolate the putative intermediates for biophysical studies after treatment of aHL mutants with DOC (24) or by fractionation of aHL-treated rRBCs (32).