Importance of the Carboxyl Terminus in the Folding and Function of α-Hemolysin of Staphylococcus aureus *

The physical state of two model mutants of α-hemolysin (αHL), αHL(1–289), a carboxyl-terminal deletion mutant (CDM), and αHL(1–331), a carboxyl-terminal extension mutant (CEM), were examined in detail to identify the role of the carboxyl terminus in the folding and function of native αHL. Denatured αHL can be refolded efficiently with nearly total recovery of its activity upon restoration of nondenaturing conditions. Various biophysical and biochemical studies on the three proteins have revealed the importance of an intact carboxyl terminus in the folding of αHL. The CDM exhibits a marked increase in susceptibility to proteases as compared with αHL. αHL and CEM exhibit similar fluorescence emission maxima, and that of the CDM is red-shifted by 9 nm, which indicates a greater solvent exposure of the tryptophan residues of the CDM. In addition, the CDM binds 8-anilino-1-naphthalene sulfonic acid (ANS) and increases its fluorescence intensity significantly unlike αHL and CEM, which show marginal binding. The circular dichroism studies point that the CDM possesses significant secondary structure, but its tertiary structure is greatly diminished as compared with αHL. These data show that the CDM has several of the features that characterize a molten globule state. Experiments with freshly translated mutants, using coupled in vitro transcription and translation, have further supported our observations that deletion at the carboxyl terminus leads to major structural perturbations in the water-soluble form of αHL. The studies demonstrate a critical role of the carboxyl terminus of αHL in attaining the native folded state.

␣-Hemolysin (␣HL) 1 of Staphylococcus aureus has attracted lot of attention from structural biologists and biotechnologists for its potential applications in biotechnology and therapeutics (1). It is a 293-amino acid polypeptide that binds target cells as a monomer, and the cell-bound monomers undergo lateral diffusion to form a transmembrane heptameric pore. The heptamer is a rigid mushroom-like structure, resistant to SDS and temperatures up to 80°C (2). The water-soluble monomer undergoes a series of conformational changes to form the heptameric pore on the membrane. The amino acid residues 110 -148, termed the stem domain, penetrate the membrane bilayer to access the interior of a target cell for the functional pore formation.
The activity of ␣HL was earlier shown to be critically dependent on an intact amino and carboxyl termini (3). Recent studies from our laboratory have shown that deletion of four amino acids at the amino terminus of ␣HL leads to delayed pore opening. Although this mutant (␣HL(5-293)) could undergo the oligomerization process as fast as native ␣HL, the conformational changes that lead to the opening of the pore were retarded (4). Fluorescence spectroscopic studies carried out by Valeva et al. (5) have also arrived at similar conclusions regarding the role of the amino terminus. Previous studies have shown that deletion of three, five, or eight amino acids at the carboxyl-terminal end of ␣HL impairs its oligomerization and pore formation abilities. The carboxyl-terminal deletion mutants, however, have been shown to bind to rabbit red blood cells (rRBCs), where they remain in a cell-bound monomer form (3). In addition, the carboxyl-terminal deletion mutants are very inefficient in forming hetrooligomers with full-length ␣HL. All of the studies conducted so far have attributed the role of the amino terminus for pore opening and the carboxyl terminus for the initial oligomerization process of ␣HL. However, the reasons for the inefficient oligomerization of the carboxylterminal deletion mutants are not yet clear.
It has been observed that the carboxyl-terminal end of ␣HL becomes more exposed to the solvent in the oligomeric state than in the monomer form in solution, as revealed by IASD modification of single cysteine mutants (IASD is a membraneimpermeant reagent that covalently modifies surface-accessible cysteine residues in proteins (6,7)). This observation was supported by the crystal structure of the fully assembled ␣HL pore, which shows that the carboxyl terminus is well exposed to solvent. Thus, the carboxyl terminus does not appear to be critically involved in interprotomer interactions (3). Hence, a deletion of as few as three carboxyl-terminal residues ought not to have any drastic effect on the oligomerization process. Another possibility is that the carboxyl-terminal deletion is hampering a process prior to the oligomerization step, which might occur in the water-soluble monomer or at the membrane-bound monomer stage. Therefore, in order to have a better understanding of the role of the carboxyl terminus in the structure and function of ␣HL, we have aimed to examine the properties of a carboxyl-terminal deletion and an extension mutant.
The extension mutant was constructed with an aim to see whether or not ␣HL can accept a polypeptide at the carboxyl terminus and still carry out its folding and function. For this purpose, we have added a nearly neutral, nonaromatic amino acid-rich sequence to the carboxyl terminus of ␣HL. In this paper, the results from biochemical and biophysical studies on the above proteins have revealed that the carboxyl terminus of ␣HL stabilizes the native structure of ␣HL. In the absence of the carboxyl terminus, ␣HL is unable to acquire its watersoluble, fully folded, native form. This defective folding caused by truncation is responsible for the loss of function of ␣HL.

MATERIALS AND METHODS
All bacterial strains employed in this study were obtained from commercial sources. Ultrapure bovine serum albumin, S-Sepharose, and 8-anilino-1-naphthalene-sulfonate (ANS) were obtained from Sigma. Protein estimations were carried out with the Bradford reagent (Bio-Rad) using ultrapure lipid-free bovine serum albumin as the standard. Hemolysis assays were carried out with freshly drawn blood from New Zealand White rabbits of the local animal facility. All other chemicals used were of analytical grade.
Cloning of ␣HL, CDM, and CEM-The cloning of ␣HL was described in detail by Vandana et al. (4). The PCR amplification of CDM was achieved by advancing the stop codon present in ␣HL at 294 to Glu 290 by a downstream primer that contains a HindIII site. (The upstream primer is the same as that used for ␣HL.) The resultant PCR product was digested with NcoI and HindIII and ligated to the pT7 vector described earlier (4). The CEM was constructed in two stages by PCR amplification of the ␣HL gene using an upstream primer that contains an EcoRI site at Asn 293 and the downstream primer (with HindIII site) of ␣HL reported earlier (4). The upstream primer eliminates the stop codon present in the ␣HL gene. The primers containing the EcoRI and HindIII sites were first joined by PCR using ␣HL template. The CEM in final form was obtained by re-PCR of ␣HL template using a T7 promoter primer and the EcoRI and HindIII joined product obtained in the first stage. The resultant PCR product was digested with NcoI and EcoRI to remove the 3Ј-untranslated region of ␣HL. This double-digested PCR product was ligated to the parent pT7 vector between NcoI and EcoRI. As a result of removal of the stop codon of ␣HL, the translation proceeds beyond the EcoRI site and incorporates the following 38-amino acid stretch from the vector backbone: 294 SSSVDKLEY-SIVSPKSELDPAANKARKEAELAAATAEQ 331 .
Purification of ␣HL-␣HL was purified from S. aureus wood 46 (ATCC 10832) as reported earlier (8). Briefly, a 2% mid-log phase inoculum of S. aureus was added to 1 liter of tryptic soy broth, and the culture was grown for 18 h at 37°C. The cells were removed by centrifugation, and the supernatant was brought to 80% saturation with ammonium sulfate. The mixture was left overnight at 4°C with mild stirring. The precipitate was collected by centrifugation and dialyzed against 10 mM sodium acetate, pH 5.2, buffer at 4°C for 48 h with at least eight changes. The dialysate was clarified and loaded on an S-Sepharose column pre-equilibrated with 10 mM sodium acetate buffer, pH 5.2. The bound ␣HL was eluted with a step gradient, and the yield of ␣HL was typically 3-4 mg/liter. The protein was Ͼ95% pure as judged by SDS-PAGE (9).
Purification of CDM-E. coli BL21(DE3) cells harboring the CDM plasmid were grown at 37°C and induced at 0.4 A 600 with 0.2 mM isopropyl-1-thio-␤-D-galactopyranoside. After 4 h of induction, the cells were harvested by centrifugation at 4000 ϫ g for 20 min, resuspended in buffer A (50 mM Tris-HCl, pH 8.0, containing 5 mM EDTA), and treated with 0.2 mg/ml lysozyme at 20°C for 25 min. The cells were passed once at 1000 p.s.i. through a French press followed by 5-min sonication. The resultant cell lysate was centrifuged at 6000 ϫ g for 10 min. The inclusion body pellet was washed three times with 1 M urea and 0.5% Triton X-100 in buffer A, rinsed with buffer A, and stored at Ϫ4°C until further use. The inclusion bodies were suspended in 8 M urea in buffer B (10 mM sodium acetate, pH 5.2), diluted 2-fold with buffer B, and centrifuged at 13,000 ϫ g. The supernatant was passed through an S-Sepharose column pre-equilibrated with 4 M urea in buffer B. The mutant protein eluted with 200 mM NaCl and 4 M urea in buffer B. The protein was renatured either by dialysis against buffer B at 4°C or by simple dilution as desired for specific experiments, and hereafter this protein is referred to as renatured. The renatured protein was found to be Ͼ95% pure as judged by SDS-PAGE.
Purification of CEM-Purification of the soluble protein was carried out employing similar protocol reported by us earlier for ␣HL(5-293) (4).
Limited Proteolysis-The proteins (native and refolded as applicable) were subjected to digestion with Proteinase K at 25°C by keeping the substrate:enzyme ratio at 50:1. At appropriate time points, the enzyme was inactivated by the addition of 5ϫ Laemmli sample buffer and boiled at 100°C for 5 min. The samples were analyzed by 14% SDS-PAGE.
Hemolysis Assays-The lysis of rRBCs was measured by adding 100 l (7 g/ml) of protein to 100 l of K-PBSA (150 mM NaCl, 20 mM KH 2 PO 4 , pH 7.4, containing 1 mg/ml bovine serum albumin) in well number 1, and a 100-l aliquot was taken for 2-fold serial dilution to 12 wells. An equal volume of 2% rRBCs was added to the wells and left at 25°C for 1 h. At the end of the incubation period, the well number exhibiting 50% lysis was recorded visually. For CDM, the lysis was recorded after 24 h. Unless specifically mentioned, all hemolytic assays were performed in 96-well microtiter plates.
Quantitative Hemolysis Assay-The toxins (2 g) were added to 1 ml of K-PBSA and a 500-l aliquot was subjected to 2-fold serial dilutions with the same buffer. To the serially diluted samples an equal volume (500 l) of 1% rRBCs was added and incubated at 25°C for 30 min. The samples were then centrifuged, and the absorbance at 545 nm (due to hemoglobin release) was plotted against toxin concentration. For the mixed hemolysis assays with ␣HL and CEM, known quantities of toxins were mixed, and the assay was carried out as outlined above.
␣HL Unfolding and Refolding-A stock solution of ␣HL (1.75 mg/ml) was diluted with freshly prepared 10 M urea or 8 M guanidine HCl to a final concentration of 0.350 mg/ml ␣HL in 8 M urea and 6 M guanidine HCl, respectively. For acid denaturation studies, ␣HL was incubated in 50 mM citrate-phosphate buffer at pH 2.5, 3.5, and 7.0 at a concentration of 350 g/ml. After incubation at 25°C for different periods of time in the above described conditions, samples were withdrawn and diluted 50 times with various buffers (given in Tables I and II), and a hemolysis assay was performed immediately.
Fluorescence Measurements-Fluorescence measurements were carried out in a Perkin-Elmer LS-50B spectrofluorometer. The protein samples (30 g/ml in 10 mM MOPS, pH 7.0) were excited at 295 nm, and the slit widths were 5 nm for both excitation and emission. All fluorescence spectra are corrected for buffer contribution and are an average of at least eight scans. For the pH-induced unfolding studies, ␣HL and CDM were incubated in 50 mM citric acid/Na 2 HPO 4 buffer of pH values ranging from 2.5 to 7.0 for 20 min before measuring the emission spectrum. The buffer solutions were prepared by suitably mixing solutions of 50 mM citric acid and 50 mM Na 2 HPO 4 in order to obtain the desired pH.
For ANS binding studies, a stock solution of ANS (5 mM) was prepared in methanol. Binding of ANS to ␣HL, CDM, and CEM was carried out in cuvettes containing 30 g/ml protein in 10 mM MOPS, pH 7.00, and 50 M ANS. ANS fluorescence was obtained in the range of 410 -580 nm with the excitation wavelength fixed at 390 nm using the same slit widths mentioned above.
CD Studies-CD spectra were recorded on a Jobin Yvon spectropolarimeter, which was calibrated with D-10-camphorsulfonic acid. The proteins were diluted to a concentration of 0.1 mg/ml in 10 mM sodium acetate, pH 5.2, buffer. The far (190 -250 nm) and near (250 -320 nm) UV spectra were recorded in 1-mm and 5-cm path length quartz cuvettes, respectively. The spectra are an average of four scans, with subtraction of appropriate blanks.
Coupled In Vitro Transcription and Translation-Supercoiled plasmid DNA was used for in vitro transcription and translation in an Escherichia coli T7-S30 extract in the presence of rifampicin and complete amino acid mix as reported earlier (4). A 5-l aliquot was withdrawn at various time intervals from the initiation of translation. The 5-l aliquot was added to a 96-well plate containing 200 l of 0.5% rRBCs. The decrease in absorbance at 595 nm due to hemolysis was monitored at regular time intervals.

RESULTS
Purification and Characterization of ␣HL, CDM, and CEM-All of the toxins employed for the present study were constructed under the control of the T7 promoter. The relative sizes of the polypeptides are in agreement with the cloning strategy (Fig. 1A). Unlike recombinant ␣HL and CEM, the soluble form of CDM could not be obtained due to extensive inclusion body formation (Ͼ95%). Since inclusion body formation is often due to temperature-sensitive denaturation of the protein and can be avoided at lower culture growth temperatures (10), the CDM culture was grown at 30°C. However, this had no effect on the extent of inclusion body formation. Hence, we have purified CDM by solubilizing its inclusion bodies as described under "Materials and Methods." The purity of all three proteins was routinely assessed by SDS-PAGE and was found to be Ͼ98% as shown in Fig. 1A.
We have carried out limited proteolysis of the three proteins in solution with Proteinase K. Limited proteolysis is a sensitive probe for analyzing the conformation of proteins (11). Polypeptides that are devoid of tertiary structure have been observed to be very sensitive to proteolysis (12,13). Proteinase K cleaves ␣HL monomer in solution between residues 131 and 136 of the polypeptide chain, yielding an approximate two halves of ␣HL. Such a cleavage does not occur for the membrane-bound forms of monomer and oligomer because these residues get occluded in the membrane (14). As seen in Fig. 1B, the refolded CDM was completely digested by Proteinase K in minutes, whereas ␣HL gave the expected two halves, which were resistant toward further protease attack. This enhanced susceptibility of CDM towards protease suggests that its structure is not as rigid as that of ␣HL.
Hemolytic Activity-The hemolytic activity of the three proteins was compared by a quantitative assay. As seen from Fig.  2, the CDM exhibited no lysis at all in the time course of the experiment, whereas the CEM showed efficient lysis. However, it is about 5-fold weaker than ␣HL and increasing the amount of CEM 5 times in the assay gave an identical curve like that of ␣HL.
Unfolding and Refolding of ␣HL-The denaturation and renaturation of ␣HL were carried out by employing a wide spectrum of denaturing conditions as described under "Materials and Methods." It is clear from Table I that Ͼ95% of hemolytic activity of ␣HL can be recovered upon restoration of nondenaturing conditions. However, when ␣HL was incubated at pH 3.5 at room temperature, the recovery of hemolytic activity decreases with longer incubation times. At pH 3.5, the carboxyl-terminal portion of ␣HL is said to undergo a transition to a molten globule-like state with exposed hydrophobic regions (15). The apparent reason for the loss of activity at pH 3.5 could be due to aggregation of the partially unfolded intermediate. However, total recovery of activity was achieved when the incubation was carried out at 4°C, which could be either due to slow denaturation or due to a significant decrease in hydrophobic interactions at the lower temperature.
Fluorescence Studies-The normalized fluorescence emission spectra of native and denatured states of ␣HL, CDM, and CEM are shown in Fig. 3, and the emission maxima obtained are 336, 345, and 336 nm, respectively. It is interesting to note that the emission maximum of CDM exhibited a 9-nm red shift with respect to native ␣HL, indicating a change in the polarity of environment of tryptophans due to solvent exposure (16). The emission maxima for all three toxins were further red-shifted in presence of 8 M urea to 352.5 nm. In contrast to CDM, the fluorescence spectrum of CEM was identical to that of ␣HL, indicating that the extra 38 residues at the C-terminal end of ␣HL did not have any influence on its folding and function. The residues that were deleted in the case of CDM are Glu 290 -Met 291 -Thr 292 -Asn 293 , and the residues extended in the case of CEM do not contain any tryptophans. Hence, ␣HL, CDM, and CEM contain an equal number of aromatic residues in their primary sequence, and the differences in their fluorescence spectra clearly reflect the degree of compactness of the individual toxins.
The fluorescence emission maximum as a function of pH for ␣HL and CDM is depicted in Fig. 4. The curve obtained for ␣HL The toxins were added to 1 ml of K-PBSA and subjected to 2-fold serial dilutions. An equal volume of rRBCs was added to get a final concentration of 0.5% and 1 g/ml of toxin in the first dilution. After incubation at 25°C for 30 min, the absorbance at 545 nm of the centrifuged sample was measured. , OE, q, and f, ␣HL, CDM, CEM, and CEM (5 times the amount of other toxins in each tube), respectively. is in agreement with previous results (15). The CDM in 8 M urea was either 1) diluted with an appropriate buffer to the desired pH (final urea concentration was kept at about 80 mM) or 2) renatured by dialysis, and the pH of the dialysate was adjusted to the desired value. The curves obtained by both of the approaches do not overlap (as one would expect) even after incubation for 48 h, in that the case 2 CDM appears to be more resistant to acid denaturation compared with the case 1 CDM. This hysteresis could be due to formation of soluble aggregates among the renatured CDM molecules, because it has been observed that aggregation commonly interferes with the correct equilibrium, giving rise to such hysteresis (17). This possibility was examined by glutaraldehyde cross-linking of the renatured CDM, and high molecular weight forms of CDM were observed in contrast to ␣HL (data not shown).
Probing the Hydrophobic Regions of ␣HL, CDM, and CEM-ANS has been extensively used to characterize the hydrophobic pockets of proteins and enzymes for understanding their folding and function (18). The fluorescence emission of ANS is known to increase when the dye binds to hydrophobic regions of proteins that are normally absent in totally unfolded states and rarely present in native states (19). While the fluorescence of ANS marginally increased in the presence of ␣HL and CEM, the increase in case of CDM was rather dramatic, as shown in Fig. 5. In addition, the emission maximum of ANS had blueshifted from 513 nm in buffer to 483.5 nm upon binding to CDM. On the other hand, the denatured states of the three proteins did not bind any ANS (data not shown). This result indicates that CDM has hydrophobic regions exposed to the solvent, unlike ␣HL and CEM. It is interesting to note that the CEM did not show any concomitant increase in fluorescence intensity of ANS. These observations reflect the compactness of native ␣HL and the role of the carboxyl terminus in maintaining such a compact structure.
CD Studies-The secondary structure of ␣HL and CDM was examined by far UV-CD spectroscopy. The CD spectrum of CDM shows a significant content of secondary structure and is characteristic of a predominantly ␤-sheet protein, as is the case for ␣HL. However, comparison with the ␣HL spectrum suggests some minor conformational differences between the two species (Fig. 6A). The spectrum of ␣HL shows a minimum at 214.5 nm and is consistent with earlier reports (14,15,20). In case of CDM, the minimum was red-shifted to 218.5 nm. This might result from a change in the polarity of the environment of ␤-sheets, since it is well known that the ␤-sheet is very sensitive to a change in environment conditions (21).
The near UV-CD spectra of ␣HL and CDM are shown in Fig.  6B. The ␣HL spectrum is consistent with the previously published reports. Comparison of the two spectra shows a drastic difference in the tertiary structure of the two proteins. In case  of CDM, the negative peaks at 265 and 295 nm, which correspond to the vibrionic regions of phenylalanine and tryptophan, respectively (22), are totally absent. This indicates that most of the phenylalanine and tryptophan are in a more mobile environment. However, CDM still possesses the positive peak at 280 nm, whose ellipticity is about 30% of the corresponding peak in the ␣HL spectrum. This reveals that some rigid tertiary contacts are present in CDM, particularly around a fraction of the tyrosine residues, but the overall spectrum reflects a significant disorder in the tertiary structure of CDM. DISCUSSION In the present study, we have designed a carboxyl-terminal deletion and a carboxyl-terminal extension mutant of ␣HL in order to investigate the contribution of the carboxyl-terminal residues of the protein to its structural organization and function.
Our efforts to isolate the CDM have met with partial success, since it appears to be unstable. The protein is almost exclusively found in inclusion bodies, unlike recombinant ␣HL and CEM, which can be isolated in soluble, active form. A variety of attempts to purify the CDM by ion exchange and gel filtration techniques have led to precipitation of the protein in the column. Hence, a wide spectrum of conditions was employed in an attempt to stabilize the CDM, and the conditions suitable for spectroscopic studies are low ionic strength and pH 5.0 -7.0. The CEM, on the other hand, can be purified like ␣HL. The hemolytic data presented in Table I show that ␣HL can be unfolded and refolded to its native form in a variety of conditions. Refolding is extremely efficient, with almost total recovery of activity. Hemolysis studies carried out with the mutants showed that the CDM is very weakly lytic, which is in agreement with prior studies with other C-terminal deletion mutants (3). On the other hand, the CEM, which has 38 extra amino acids, was able to lyse the rRBCs efficiently (Fig. 2).
The CDM has a tendency to aggregate both in vitro and in vivo (as seen by the extensive inclusion body formation). Studies have shown that inclusion bodies form due to aggregation of partially folded intermediates (23). Hence, the occurrence of CDM in inclusion bodies suggests that the mutant protein was unable to achieve the final folded conformation in vivo. Limited proteolysis experiments have revealed that the CDM gets completely digested unlike ␣HL, which exhibited its typical "two halves" pattern. This pronounced susceptibility of CDM to proteolytic digestion suggests a more relaxed structure in which many proteolytic sites that are otherwise hidden in ␣HL are getting exposed. All these observations suggest that the CDM possesses a non-␣HL like structure.
The fluorescence emission of the CDM lies in between the native and denatured states of ␣HL. This red shift observed for the CDM indicates that its tryptophan residues are more exposed to the solvent as compared with ␣HL. This was further corroborated with binding studies with ANS, a dye widely used to probe the molten globule states of proteins. The fluorescence intensity of ANS increased by about 10-fold upon binding to CDM but showed negligible increase in the presence of ␣HL and the denatured states of the two proteins. The increase is accompanied by a blue shift of the emission maximum by 29.5 nm, which is an indication of specific binding. These results indicate the presence of exposed hydrophobic regions in CDM, which are sequestered in native ␣HL.
The far UV-CD spectrum of CDM shows that it possesses nearly native-like secondary structure. However, its tertiary structure as analyzed by near UV-CD is greatly diminished. The near UV-CD spectrum indicates that most of the phenylalanine and tryptophan of CDM are in a mobile, symmetric environment but some tertiary contacts are present in the environment of some of the tyrosines. A cursory glance at the distribution of the aromatic residues along the ␣HL polypeptide chain shows that seven of the eight tryptophan residues are in the carboxyl-terminal half of ␣HL, and eight of the 14 tyrosine residues are located on the N-terminal half. This gives rise to the possibility that the tertiary contacts may be localized on the first half of the polypeptide chain.
All of these observations strongly suggest that significant perturbations to the ␣HL structure had occurred in the absence of the carboxyl terminus. Interestingly, the additional amino acids at the carboxyl terminus (as in CEM) did not have any influence either on the structure or the function of native ␣HL. The CEM exhibited all of the characteristics of ␣HL. Proteolytic digestion revealed an ␣HL-like pattern, and the protein was able to oligomerize and cause the lysis of rRBCs. It has the same fluorescence emission maximum as ␣HL, which suggests that the tryptophans of both of the proteins are in a similar environment. Like ␣HL, it does not have any of its hydrophobic regions exposed to the solvent, as demonstrated by the negligible ANS binding. All of these points illustrate that it is possible to build residues at the carboxyl terminus of ␣HL without inhibiting its folding or function.
The non-␣HL like structure of the CDM can be interpreted as its being unable to "attain/maintain" the native, ␣HL-like conformation, and for that the carboxyl terminus is very crucial. We have tested for such a possibility by in vitro transcription and translation, whereby the activity of the CDM was assessed as it emerged from the ribosome during in vitro translation. The hemolytic activity of ␣HL, CDM, and CEM was examined as a function of initiation of translation. It is clear from Fig. 7 that at constant rate of transcription and translation, ␣HL begins to show lysis within minutes after initiation of translation, whereas the CDM fails to show any lysis. On the other hand, the CEM lyses rRBCs efficiently but marginally slower when compared with ␣HL. This shows that freshly synthesized ␣HL folds rapidly to attain its native, active structure, but the CDM is unable to do so although it lacks just four amino acids. This suggests that the CDM is probably stuck in a partly folded inactive state.
The physical state of the carboxyl-terminal deletion mutants of ␣HL is intriguing, since these mutants cannot participate in interprotomer interactions to form oligomers on rRBCs. They can however, bind the rRBCs and remain in the form of a cell-bound monomer (3). This is not surprising, because partially folded states of proteins are shown to bind membranes with high affinity (24). The crystal structure of ␣HL reveals that the N-terminal latch makes extensive contacts with the inner ␤-sheet of an adjacent protomer. Yet, when as many as 22 residues are deleted from the N terminus, ␣HL is still able to form oligomers (although yet to be fully characterized). This is because in a heptamer, each protomer participates in about 120 salt bridges and hydrogen bonds and 850 van der Waals contacts (2). In the absence of a few of these interactions, as in the case of an N-terminal deletion mutant, the net force of the remaining interactions is enough to drive the process to the oligomeric state. However, this does not seem to happen in the case of carboxyl-terminal deletion mutants, and they are unable to carry out such fruitful interactions either among themselves (i.e. formation of homooligomer) or with ␣HL (i.e. formation of hetrooligomer). Our data are clearly able to reveal the reason behind the inefficient oligomerization of carboxyl-terminal deletion mutants. The present data unequivocally prove that the CDM possess a partially folded nonnative conformation with greatly diminished tertiary structure. Such a species could still bind rRBCs through nonspecific hydrophobic interactions but does not possess the requisite motifs for carrying out interprotomer interactions. As a result, it cannot form either homooligomers with itself or hetrooligomers with ␣HL. We have also examined the lysis efficiency of the hetrooligomers of ␣HL and CEM by mixing a fixed amount of ␣HL with increasing amounts of CEM. As seen in Table II, the percentage of lysis increases with increasing amounts of the CEM, which shows that the extended carboxyl terminus does not interfere with the assembly of ␣HL (i.e. during formation of functional pore). This is in contrast to the amino-terminal deletion mutant ␣HL(5-293), which retards the lysis of ␣HL as reported by us earlier (4). This again supports our conclusion that the carboxyl terminus is not very critical for the oligomerization process but instead is involved in the folding of the monomer form.  (3). Typically, an aliquot of translation mix was withdrawn and added to 0.5% rRBCs, and the decrease in light scattering of rRBCs at 595 nm due to lysis was monitored at regular time intervals. E, Ⅺ, and ‚ represent ␣HL, CDM, and CEM, respectively at 3.5 min after initiation of translation. q, f, and OE represent ␣HL, CDM, and CEM, respectively, at 6.5 min after initiation of translation. No detectable lysis was observed before 3 min. The molten globule states of proteins have been attributed with four distinct features, viz. compactness, a near native secondary structure, loss of tertiary structure, and exposure of hydrophobic regions (25). High affinity of CDM for ANS reveals that it has hydrophobic regions exposed to the solvent. In addition, the CDM exhibited the presence of substantial secondary structure accompanied by a significant loss of tertiary structure. We could not, however, determine the compactness of the CDM due to its tendency to interact with gel filtration matrices. All these observations suggest that in the absence of the carboxyl terminus, the physical state of ␣HL may lie close to a molten globule-like state. Furthermore, the observations presented here strongly suggest that ␣HL is able to fold to its native form only after its complete synthesis, and/or the CDM is not able to "latch on" to native ␣HL-like structure because of the absence of the carboxyl terminus.
In summary, for the first time, the role of the carboxyl terminus of ␣HL has been delineated in greater detail. While the amino terminus is important for the functional pore formation step, the carboxyl terminus is crucial for correct folding and maintenance of the water-soluble monomer form of ␣HL. Removal of just four residues from the carboxyl terminus makes the protein unable to proceed beyond a molten globulelike state. This information would be valuable to understand how membrane binding toxins like ␣HL carry out their folding and function. Since single amino acid substitutions at the carboxyl terminus do not affect the function, the length (i.e. backbone) of the protein appears to be vital instead of the actual sequence at the carboxyl terminus. In addition, we have shown that the addition of an extra sequence at the carboxyl terminus does not affect the folding or function of ␣HL. Such molecules could form the basis for construction of new molecules in the future.