Identification and immunological characterization of the domain of Actinobacillus actinomycetemcomitans leukotoxin that determines its specificity for human target cells.

Although extensive amino acid homology exists among the various Repeats in ToXin (RTX) family of bacterial cytolysins, the cellular and species specificities remain unique for individual toxins (i.e. Actinobacillus actinomycetemcomitans leukotoxin (LtxA) kills human monomyelocytes while a related toxin, Pasteurella hemeolytica leukotoxin (LktA) kills bovine lymphoid cells). To determine the Ltx domain responsible for species specificity, ltxA/lktA chimeric toxin genes were expressed in tandem with the ltxC gene under control of the P lambda promoter. The ability of lysates to kill either HL-60 (human) or BL-3 (bovine) cells was assessed by trypan blue exclusion. The critical area required for the chimeric toxins to recognize human target cells is a 253-amino acid fragment (residue 688-941) that contains the GGXGXDX(L[I[V[W[Y[F)X repeats. A panel of 12 neutralizing anti-LtxA monoclonal antibodies also recognized specificities within the 253-amino acid fragment. Epitope mapping of the monoclonal antibody panel showed that all antibodies bound to one of three sites on the LtxA molecule. One monoclonal recognized epitope A which was composed of LtxA residues 698-709 (KLDYYYTNKGFK), six antibodies recognized epitope B, a peptide composed of residues 746-757 (LIYGYDGDDRLY), whereas the remaining five monoclonals recognized epitope C, which is composed of residues 926-937 (DRARLKRQFELQ).

individual toxins of this group exhibit unique species and cell type specificity for killing. These findings suggest that although the mechanism of RTX toxin-mediated lysis may be similar, differences in host cell specificity indicate that certain regions of the RTX have undergone significant divergence from each other. Some members of the RTX toxin family are cytolytic against a narrow range of target cell types. For example, the leukotoxins from Pasteurella hemeolytica and A. actinomycetemcomitans are lytic only to leukocytes from ruminants (1,(15)(16)(17) and granulocytes from certain primates (16,181, respectively. In contrast, the E. coli a-hemolysin is toxic to a wide range of cell types from humans and other animal species (19). The extensive amino acid sequence similarity that is shared by R'IX provides an approach for locating the species-specific functional domains of the leukotoxin molecule.
Mutational analysis has been an important tool for ascribing characteristics and functions to specific regions of the RTX molecule. Truncation mutants were useful in demonstrating RTX carboxyl-terminal secretion functions (20, 211, whereas deleting regions of the various toxins has helped t o define domains responsible for Ca2+ and erythrocyte binding (22,23) and pore formation (21, 22,24,25). However, in many instances it is difficult to determine if an observed loss of cytotoxicity encountered with specific mutants is due to the deletion of the functional element or simply the inability of the mutant toxin t o achieve the proper tertiary structure. Chimeric toxins, on the other hand, would be expected to contain all of the functional domains of the native toxin, exhibit normal or near normal protein folding, and maintain their ability to kill either bovine or human target cells as long as the respective "species recognition unit" is present. Toward this end, investigation of the differences in target cell specificity among RTX toxins has revealed discrete regions of the broadly reactive hemolysins which can confer specificity for new cell types and host species to the target cell-limited leukotoxins (26,27). Studies of the structure of the regions responsible for species specificity will not only provide insight into the phylogenetic evolution of these molecules but will also urovide imuortant ureliminarv data on consists of a site for N recognition (Nut R ) and the rho-independent termination site (tR,). The ribosome binding site is indicated, and an NdeI site has been inserted at the initiation codon. The ZtxC gene was cloned into the NdeI site. Either the ZtxA or ZktA genes or chimeric toxins were cloned in at the KpnI and SstI sites.

MATERIALS AND METHODS
Bacterial Strains-A. actinomycetemcomitans, strain JP2, used in this study was obtained from Dr. N. Taichman (Department of Pathology, University of Pennsylvania, Philadelphia). The strain was grown in PYG medium (5 g of Bacto-peptone, 5 g of Trypticase-peptone, 10 g of yeast extract, 10 g of dextrose, 8 mg of CaCl,, 8 mg of MgSO,, 40 mg of &HPO,, 40 mg of KH,PO,, 400 mg of NaHCO,, 80 mg of NaCl in 1 liter of dH,O) for 24 h at 37 "C in an atmosphere of 5% CO,. E. coli strain AR120 was grown in LB medium (1% Tryptone, 0.5% yeast extract, 1.0% NaCl) with aeration at 37 "C. P. hemeolytica Al, strain 112, was obtained from Dr. Gordon Gheilen, University of California (Davis, CA) grown in PYG media and frozen in PYG, 20% glycerol as stocks.
Construction of ltxCllktA a n d ltxCiZtxA Gene Cassettes-Oligonucleotide primers used in this study were synthesized in our laboratory by a modification of the phosphite method (28). Initially these primers were utilized to amplify the ZtxC and the ZtxA genes from genomic DNA in the following manner. Primers a and b were utilized to amplify a 534-bp fragment which contained the ltxC gene, the ZtxClZtxA intragenic space, and 2 residues of the ZtxA gene. NdeI and KpnI restriction sites were engineered into the 5' and 3' ends of this fragment respectively. NdeI contains ATG in its restriction site (primer a, underlined) which was used as a translation start signal for ZtxC. Primers c and d were used to amplify a 3183-bp piece of DNA which included the ZtxA gene (from bp lo), termination codon, and 12 bp of the ZtxAlZtxB intragenic space. Placement of the KpnI site in the ZtxA gene was achieved by insertion of a glycine residue between ZtxA residues 2 and 3 and the placement of a silent mutation in residue 3 of the ZtxA gene. Addition of glycine residues had no effect on leukotoxicity (data not shown).

Primer a 5'-CATATGGAGAAAAATAATAATTTTGAAGT-3' Primer b 5'-GGTACCTGCCATAATCTATTCT-3' Primer c 5'-GGTACCACTACACTGCCAAAT-3' Primer d 5'-GAGCTCGCCAATAACTCATTAAGCAGT-3'
Fragments containing the I? hemeolytica ZktA gene were amplified from genomic DNA by PCR using oligonucleotide primers "e" and "f." A KpnI site (primer e, underlined) was engineered into the coding sequences of residues 2 and 3 of the lktA gene. Primers e and f amplify a 2865-bp product that contains coding sequences for residues 4-953 of the ZktA gene and a transcript termination codon (primer f, italics). An SstI site (primer f, underlined) was added to the 3' end of the PCR fragment.

Primer f 5'-GAGCTCTTAAGCTGCTCTAGCAAATT-3'
Amplification was carried out on a Perkin-Elmer 9600 Thermocycler using TaqI polymerase according to the protocol supplied by the manufacturer.
Expression of Ztx and Zkt Leukotoxin Genes-PCR-generated Ztx and Zkt genes were cloned into the pMGl(29) expression vector (Fig. 1). This plasmid cloning vehicle is derived from pKC30 (30-32) (a gift from Drs. Allan Shatzman and Martin Rosenberg, SmithKline Beckman Laboratories, Swedeland, PA) and utilizes the bacteriophage h promoter P, and carried in the lysogenic host AR120. In addition to providing a strong tightly controlled promoter, the system also ensures that P,-directed transcription efficiently traverses either ItxllktC or ItxlZktA genes by providing the phage A antitermination function, N, and a site for N utilization (Nut site) within the P, transcription unit. E. coli transitional regulatory eEcient ribosome recognition and translation sites were also engineered into the P, transcription unit. An NdeI site was placed at the ATG initiation codon and a ZtxC PCR product with NdeI restriction sites at the 5' end was cloned into this site. ItxlZktA PCR products were cloned in tandem utilizing the KpnI site described above.
Protein synthesis was induced by addition of nalidixic acid (60 pg/ml) to the culture medium for 4 h. Nalidixic acid induces the endogenous E. coli SOS response, resulting in cleavage of the CI repressor by the RecA protein and induction of expression. Cultures containing induced bacteria were sonicated (2 min, standard tapered Microtip, Heat Systems Ultrasonics model W385), centrifuged, and passed through a Detoxi-G e P column (Pierce). Cytotoxicity of sonicates processed in this manner was >98% inhibitable by either heat (60 "C, 1 h) or anti-A. actinomycetemcomitans leukotoxin antibody.
Determination of Leukotoxic Activity-In cytotoxicity assays, the HL-60 human promyelocytic leukemia cell line is killed when incubated with A. actinomycetemcomitans leukotoxin (331, but not with P. hemeo-Zytica toxin. This cell line served as the cytotoxicity paradigm for human cells. BL-3, a bovine leukemia cell line, is readily killed by P. hemeo-Zytica leukotoxin. However, incubation of these cells with A. actinomycetemcomitans leukotoxin has no effect on viability and consequently this cell line was the model for bovine species specificity. Leukotoxic activity of recombinant ltxA and lktA gene products and chimeras was determined essentially as described previously (341, except that lysis was visualized by staining the cells with trypan blue. Briefly, E. coli containing ZtxClltxA and ZktA plasmids were grown to late log phase, induced, and prepared as described above. HL-60 cells were cultured in RPMI 1640 containing 10% fetal calf serum, 1% glutamine, 1% minimal essential medium vitamin solution (Life Technologies, Inc.), 1% minimal essential medium nonessential amino acid solution (Life Technologies, Inc.), and 50 pg/ml of gentamycin at 37 "C under 7% GO,. BL-3 cells (CRL 8037, American Type Culture Collection, Bethesda, MD) were grown in 50% Leibovitz L-15, 50% Dulbecco's modified Eagle's medium with 10% fetal calf serum, 1% glutamine, 1% sodium pyruvate. Prior to use, target cells were washed twice in RPMI 1640 to remove gentamycin and suspended in RPMI 1640 at 4 x lo6 cells/ml. Then 50 pl of cells were added to sonicated bacterial supernatant (50 pl), and the suspensions were incubated a t 37 "C for 60 min. Negative controls consisted of incubating target cells in RPMI 1640 and sonicates from uninduced cells and induced cells that contained the pMGl plasmid without an insert. The cells were placed on ice, 100 pl of trypan blue (0.4%) added, and surviving cells counted in a hemocytometer. At least four fields were counted and averaged for each dilution assayed. Percent lysis was calculated by dividing the number of surviving cells by the number of cells in the negative controls.
SDS-Polyacrylamide G e l Electrophoresis a n d ImmunobZotting-Protein lysates of cells containing recombinant leukotoxin genes were subjected to analysis by SDS-polyacrylamide gel electrophoresis on 12.5% slab gels and immunoblotting. Blots were developed with a rabbit anti-leukotoxin antibody as the first antibody and a horseradish peroxidase-conjugated goat anti-rabbit immunoglobulin reporter antibody as described previously (35).
and Sal1 (2821 bp) in ZtxA and CZaI (2302 bp) in IktA. Chimeric toxins were constructed by aligning ltxA and ZktA sequences (Fig. 21, identifying the location of the unique restriction sites and amplifying a portion of the A gene from either A. actinomycetemcomitans or P. hemeolytica DNA. Primer e (vide supra) was used as the forward primer. Reverse primers were synthesized using the aligned sequence of the lktA gene with a EcoRI, BamHI, BstEII, or a SaZI site added at the 3' end. PCR products were then ligated into a plasmid containing the ZtrC and a corresponding sequence of the ZtxA genes. The technique was modified for construction of chimera CH57 when CH43 was cut with Sal1 and SstI and a PCR fragment from the P. hemeolytica A gene which contained Sal1 on the 5' end and SstI on the 3' was ligated into place. Construction of chimeras that utilizes a ClaI site from the P. hemeo-Zytica A gene was achieved by amplifying genomic DNA from A. actinomycetemcomitans with a forward primer that contained a ClaI restric-
actinomycetemcomitans and I? hemeolytica toxins are compared in the splice site region (residues that are encoded at the restriction site are underlined).
tion site added to the 5' end, and primer d was used as the reverse primer. After PCR amplification these fragment were cloned into pMGl plasmid that contained ltxC and lktA plasmid by restricting with ClaI and SstI.
Eppitope Mapping ofthe 1t.d Repeat Region-Forty-six noncleavable 12-mer peptides consisted of the LtxA-deduced amino acid sequence (36,37) beginning with residue 403 and extending through residue 955 were synthesized on plastic "gears" (peptides  according to manufacturer's instructions (Cambridge Research Biochemicals, Wilmington, DE). Additionally, a second series of 46 peptides that covered the same amino acid sequence, but with a 6 residue offset (residues 409-961) (peptides , was synthesized to permit mapping antibodies whose combining site might be at the juncture of two peptides in the first set. The panel of anti-leukotoxin hybridomas (35) was grown to stationary phase in serum-free medium (50% Dulbecco's modified Eagle's mediumhigh glucose:50% Ham's F-12 supplemented with 5 pg/ml insulin, 5 pg/ml transferrin, 1.2 pg/ml ethanolamine, 0.44 ng/ml sodium selenite, 2 mM 1.-glutamine, and 1 mM sodium pyruvate), then harvested by centrifugation and the supernatant frozen until assayed.
Between assays the pins were cleaned by sonicating for 10 min in 0.1 M phosphate buffer with 1% SDS and 0.1% P-mercaptoethanol, pH 7.2, at 60 "C, washing with H, O a t 60 "C, and rinsing in boiling methanol. Antibody removal was confirmed by repeating the blocking and second antibody and substrate incubations. were utilized as splice sites for chimeric toxins (Fig. 2). Oligonucleotide primers were synthesized and DNA fragments amplified from either A. actinomycetemcomitans or I? hemeolytica genomic DNA.

Construction and Immunological Characterization
Native ltxA and lktA genes and chimeric toxin genes were cloned into pMGl and expressed in tandem with the 1txC gene under the transcriptional control of the bacteriophage A P,, promoter. Sonicates of induced and uninduced cells were subjected to SDS-polyacrylamide gel electrophoresis and Western blot analysis (Fig. 3). Some cross-reactivity was noted with our polyclonal antisera, whereas none of our mAbs reacted with I? hemeolytica toxin. Furthermore, when sonicates were adjusted for protein concentrations and loaded on a gel, the bands developed on Western blotting were approximately the same den- sity, indicating that expression levels of the various chimeric Ltx were equal. Immunoreactivity was not observed in any sonicate when the P, promoter was not induced. Migration patterns of the LktA/LtxA chimeras ranged between M, values of the native toxins (LktA, 102,000; LtxA, 116,000). Differences in migration patterns of chimeric toxins on our Western blot were minimal due to the fact that most of the sequence gapping occurs in the carboxyl terminus of the LktA protein, whereas the chimeric toxins we used in this study all contained LtxA residues in this region.
Promiscuous Activation of LktA by the 1txC Gene Product-Previous studies (27,38) have reported differing results of the ability of RTX C gene products to activate toxins from a heterologous bacterial species. Moreover, it is not known if the respective gene products of the ltxC and lktC genes are functionally equivalent in their ability to activate either structural toxin gene. Since the present study requires that the ltxC gene product activate LktA, several studies were done to determine if this activation could be successfully achieved. The results from one of these experiments are shown in Fig. 4a. Homologous constructs containing either ItxClltxA or IktCllktA genes produced products that killed only HL-60 or BL-3, respectively, reflecting the same specificity exhibited by the native toxins. A construct containing ItxCllktA also produced a toxin which lysed target cells (BL-3). Expression of either the LtxA or LktA toxin without an RTX C gene product failed to produce active toxin. The relative ability of ItxC and lktC gene products to activate lktA was then assessed by titrating ItxCllktA and 1ktCl lktA containing E. coli lysates of equal protein concentrations and comparing their ability to lyse BL-3 cells (Fig. 4b). As shown in the figure, the ItxCllktAconstruct was approximately 20 times more potent than that induced by the homologous construct (IktCllktA). Thus the ltxC gene product is fully competent to activate both the LtxA and LktA toxins. Treatment of LtxAtoxin with heat (60 "C, 1 h ) or preincubation with anti-Ltx antibody resulted in a >98% inhibition of toxicity.
Species . These chimeras demonstrated that replacement of up to 75% of the 1txA gene by lktA did not alter the specificity of the chimeric gene product. All three toxins acted like LtxA, since they killed human (HL-60) target cells but not bovine (BL-3) target cells. Only CH64 (SalI, 2821 bp) was able to kill bovine cells. No killing of either target cell was observed with CH41 (ClaI, 2482 bp). This chimera has a splice site that is located in the repeat region. An additional construct, CH57, which contained the BstEIUSalI from CH1 that had been cloned into lktA was found to be capable of lysing HL-60 cells but not BL-3 cells.
The results of these experiments indicate that the region from residue 688 to residue 941 is required for recognition of the HL-60 target cells. The region contains the characteristic GGXGXDX(L I I I V I W I Y I F)X repeats as well as 34 residues before and 95 residues after the repeats.
Inhibition of Chimeric Leukotoxin Cytotoxic Activity with Monoclonal Antibodies-An important feature in the characterization ofA. actinomycetemcomitans leukotoxin has been the inhibition of toxin activity by either polyclonal (39) or monoclonal (35) anti-A. actinomycetemcomitans leukotoxin antibody or serum from patients with A. actinomycetemcomitans infections such as localized juvenile periodontitis (40). Consequently, a series of experiments were done to determine which members of a panel of neutralizing anti-A. actinomycetemcomitans leukotoxin monoclonal antibodies (35) could also inhibit the lytic activity of our chimeric toxins. In these experiments, CH57 chimeric toxin was incubated with one of 12 different monoclonal antibodies for 30 min and then the mixture was added to HL-60 target cells. The result of this experiment is shown in Table I . All of the antibodies tested were capable of neutralizing CH57 toxin and the native CH1 toxin. No neutralizing effect was observed with any antibody when it was incubated with a native I? hemeolytica toxin (PP4). These results indicate that all of the neutralizing monoclonal antibodies tested recognize amino acid sequences between residue 688 and residue 941. The same region of the ltxA gene must be present in chimeric toxin genes for the toxins to kill human target cells.
Epitope Mapping of Anti-leukotoxin Antibodies-The region from residue 688 to residue 941 contains the characteristic GGXGXDX(L I I I V I W I Y I F)X repeats as well as 34 residues before and 95 residues after the repeats. Since our panel of monoclonal antibodies recognizes this region and was initially identified by antibody inhibition of leukotoxin-mediated killing (351, delineation of the antibody epitope(s) might provide important preliminary indications of critical Ltx amino acid sequences within this region that are necessary for species recognition by the A. actinomycetemcomitans leukotoxin.
In these studies 12 mAbs were isotyped ( To relate the structure of these epitopes to the unique functions of LtxA, the deduced amino acid sequence of the toxin was aligned with the sequences of three homologous toxins: the E, coli a-hemolysin, P. hemeolytica leukotoxin, and the A. pleuropneumoniae hemolysin. These alignments (Fig. 6) revealed a high level of sequence similarity for epitope B (six exact matches and three conservative substitutions out of 12 resides), whereas the sequences of epitopes A and C were more divergent. For example, LtxA epitope A contains 3 consecutive tyrosines which are not found in the other RTX and for which there is no homologous structure in the other proteins. In the epitope C region, only the E. coli a-hemolysin has a corresponding sequence, and this dodecapeptide only contains three exact matches and three conservative substitutions with respect to the LtxA sequence. It therefore seems likely that the unique functional specificity of the LtxA gene product resides, at least in part, in the unique sequences of epitopes A and C. Moreover, despite the high level of sequence similarity among the four toxins in epitope B, antibodies directed against LtxA epitope B react only with the homologous RTX, suggesting either that the epitope B region is not accessible to antibody in the other toxins or that even these similar sequences can assume alternate conformations. Taken together, these analyses of homologous sequences to the LtxA epitopes A, B, and C show that the unique specificity of the LtxA toxin for human leukocytes probably results from both sequential and conformational elements.

DISCUSSION
Actinobacillus is a member of the Family Pasteurellaecae, a group of nonenteric, fermenting, Gram-negative rods, which are of considerable importance in both human and veterinary medicine. Although various Actinobacillus species are found in animals, only A. actinomycetemcomitans is routinely cultured

CH57, by anti-LtxA monoclonal antibodies
The ability of a panel of monoclonal antibodies (30) to inhibit the cytolytic effects of native r! haemolytica (PP4) and A. actinomycetemcomztans (CH1) toxins and Lt&ktA chimeric toxin (CH57). In these experiments an LD,, dose of toxin was incubated with anti-leukotoxin monoclonal antibody (30 min, 4 "C) after incubation, the mixture was added to either 2 x lo5 HL-60 cells (CHI and CH57) or BL-3 (PP4) cells in 100 pl and incubated (45 min, 37 "C). Cell viability was assessed by Trypan Blue exclusion, and results are reported as the mean and standard deviation of four assays. Results are expressed as percent kill of an untreated control.  from humans. This organism has been associated with a variety of infectious disease processes in man including endocarditis, brain abscesses, osteomyelitis, subcutaneous abscesses, and periodontal disease (13,17,41,42). A. actinomycetemcomitans leukotoxin is of considerable interest, since it represents a potential virulence factor in infections caused by this organism and possesses a unique biological specificity, killing only cells of the monomyelocytic lineage of man and some higher non-human primates. Cloning and analysis of the leukotoxin operon represented an important first step toward explaining these unique properties (36). In the present study, we continued our investigations of this interesting molecule by constructing a series of A. actinornycetemcomitansIP hemeolytica chimeric toxins to determine the region of the toxin that was responsible for the specificity of this molecule for human monomyelocytes.
Our studies provide novel information on the location and nature of the LtxA region that is necessary for recognition of the human target cells. The experimental data we have presented indicate that the region from residue 688 to residue 941 is required for recognition of the HL-60 target cells. The principal feature of the region is a series of 14 tandemly repeated nonapeptides that have the consensus sequence GGXGXD- Recently, the three-dimensional structure of another protein that contains these glycine-rich repeats, alkaline protease from Pseudomonas aerugenosa, has been solved (431, and analysis of the "repeats" has indicated that they form a novel folded structure which has been named the "parallel p-roll." In the p-roll, the GGXGXD motif forms a series of successive p-turns which are wound in continuous right-handed spirals. In addition to forming the p-turns, the GGXGXD motif forms a series of Ca2+ binding sites. Individual X(L I I I V I W I Y I F X sequences of the motif form p-strands that interconnect the individual turns. The exact function of the p-roll in determining the specificity of RTX is not understood at the present time. However, searches of both the protein and DNA sequence data bases revealed several important features of the GGXGXDX-(L I I I V I W I Y I FK repeats. First, the motif is unique to a group prokaryotic proteins comprised of RTX and other molecules, the most notable of which are a series of five metalloproteinases (13,(44)(45)(46). Second, the proteins which contain GGX-GXDX(L I I I V I W I Y I F)X repeats also appear to utilize secretory proteins, such as the RTX B and D gene products for translocation to the bacterial cell surface rather than rely on the canonical signal sequence used by most bacterial proteins. The current studies clearly indicate that the repeat region of LtxA is essential in forming a unique species recognition unit which is required for the P. hemeolyticalA. actinomycetemconitans chimeric toxins to kill a human target cell line such as HL-60. The mechanism by which this is achieved is not yet clear; however, when one chimera was spliced within the repeat region (CH41), it failed to kill either target cell line. CH41 produced a recombinant toxin that contained nine GGXG-XDX(L I I I V I W I Y I F)X repeats instead of the 14 repeats normally found in native A. actinonycetemcomitans leukotoxin. Far less drastic deletions have resulted in toxins that have different specificities or an increased requirement for Ca2+ in order to complete the lytic process (27).
Our epitope mapping studies have shown that a panel of 12 monoclonal antibodies has defined three distinct linear epitopes within residues 688-941. The three regions have been designated epitope A, epitope B, and epitope C. Epitope A (698KLDYYYTNKGFK70g) was recognized by one monoclonal antibody (mAb 28) from the panel and is located 12 residues from the initiation of the first GGXGXDX(L I I 1 V I W I Y I F K repeat. Studies with other RTX indicate that epitope A could be the putative acylation site for LtxA. Rowe et al. (47) have described an erythrolysin-neutralizing monoclonal antibody to HylA (mAb D12) which recognizes an epitope within residues 673-726 of HylA. Since mAb Dl2 recognizes biologically active HylA, but not HylA, produced in the absence of HylC, these observations have led to the conclusion that the mAb Dl2 epitope contains the HylA acylation site. Analysis of the Dl2 epitope with deletion analysis mutants suggests that it may be linear and formed by residues 684-701. Utilizing a panel of noncleavable leukotoxin peptides, we have shown that LtxA mAb 28 clearly is linear and overlaps with corresponding residues described by the Dl2 epitope. Studies are currently being done to determine the relationship of epitope A and the LtxA acylation site.
Epitope B (746LIYGYDGDDRLY57) is located within the GGXGXDX(L I I I V I W I Y I F)X repeat region and is recognized by six monoclonals from our panel (mAb 3, 13, 16, 46, 53, 83). All of these antibodies bind an epitope that is composed of the p-strand (746LIY) from the third repeat and the entire fourth repeat (GYDGDDRLY57). It is not clear at the present time as to why this was the only repeat of the 14 repeats recognized by our monoclonal antibodies. An obvious explanation is that the fourth repeat is playing a crucial role in target cell recognition and for this reason we were able to identify six neutralizing monoclonal antibodies which reacted with it. An alternative explanation is that the fourth repeat is located on the surface of the leukotoxin molecule, whereas other repeats are found internally. We are currently examining this point by constructing a series of chimeric toxins where LtxA repeats will have been replaced with repeats from another RTX. If the chimeric toxins continue to kill HL-60 cells we can conclude that the fourth repeat is simply located on the surface of the molecule. On the other hand, loss of toxicity in chimeras with repeat substitutions could mean that this area of the repeats could be playing a role in target cell recognition. Pellett et al. (48) have reported an epitope recognized by anti-HylA mAb A10 which also lies within the GGXGX-DX(L I I I V I W I Y I F K repeat region of HylA. Utilizing HlyA deletion mutants, the epitope of this mutant is reported to be somewhere between residues 745 and 829 (repeats 4 and 11) of HylA. However, several differences exist between mAb A10 and the six LtxA mAb presented above. First, the HylA antibody is a "pan-reactive"antibody which recognizes several additional members of the RTX family. Anti-Ltx mAbs 3, 13, 16, 46, 53, and 83 do not recognize the P. hemeolytica leukotoxin, nor do they recognize any of the other 13 repeats within LtxA in spite of a similarity in sequence. Furthermore, the HylA antibody is non-neutralizing, whereas the anti-Ltx antibodies all neutralize the effects of the A. actinornycetemconitans leukotoxin. Until the epitope of E. coli a-hemolysin is more clearly defined, it is impossible to determine if these antibodies are recognizing the same or different epitopes. If in fact mAb A10 does recognize the fourth glycine-rich repeat of HylA as its anti-Ltx counterparts do, it would indicate that this repeat does play a critical role in target cell recognition by A. actinonycetemcomitans leukotoxin, but not E. coli a-hemolysin.
The final five monoclonal antibodies recognized a sequence mapped to 926DRARLKRQFELQ937. Epitope C begins 78 residues after the terminal (14th) repeat. Comparison of amino acid sequences from this epitope with corresponding sequences from other RTX with LtxA indicates that considerable diversity of structure exists in this region for various members of the RTX family. LktA and AppA are gapped when compared with LtxA. Although the HylA and LtxA sequences are homologous in 3 of 12 residues, deletion mutants of LtxA which terminate before this epitope (residue 910) will not kill target cells, whereas those which terminate at a Sal1 site in the gene (residue 941) are toxic for HL-60 target cells. This indicates that the region encompassing this epitope is critical for the maintenance of cytolysis. 2 In this study, we have successfully utilized chimeric toxins to show that insertion of the A. actinomycetencomitans leukotoxin gene into the P. hemeolytica leukotoxin gene will result in a switch of the toxin from killing bovine to human target cells. The portion of the gene which is critical for this determination of target cell specificity is a series of glycine-rich repeats and residues which flank the repeats on both the amino-and carboxyl-terminal ends. The mechanism by which the repeats affect the species specificity of RTX is not clear at the present time; however, some form of a species recognition unit must exist, as chimeras with splice sites within the repeat region do not kill either bovine or human target cells. Our epitope mapping studies have defined three epitopes which are necessary for toxigenic functions. The combination of our chimeric toxins and the characterization of our mAb panel have provided important preliminary data and critical reagents for further study of toxin function.