We have recently characterized an MHC class II-deficient human cell line, SW480, that supports the proliferation of purified human T cells in the presence of the staphylococcal enterotoxin and superantigen SEC1, but not the closely related isotypes SEC2 or SEC3. We now investigate the structural basis of this dichotomy and explore possible mechanisms that may account for it. Differences in activity between SEC1 and SEC2 were not attributable to differences in biochemical modification, to differences in Vβ specificity, or to the potential to induce anergy. SEC2 inhibited SEC1-mediated T cell activation in the presence of SW480 cells, suggesting that SEC2 could compete with SEC1 for binding to the TCR but was unable to productively signal through the TCR. Utilizing a panel of hybrid enterotoxins we identified specific amino acids near the NH2-terminus of SEC1 that abrogated MHC class II-independent T cell activation, yet did not alter potency in the presence of class II+ APC. These residues mapped to the putative TCR binding domain of SEC1, and suggest that subtle differences in TCR binding affinity or the topology of the SEC1-TCR interaction can compensate for the lack of MHC class II and hence promote T cell proliferation.
The staphylococcal enterotoxins are potent T cell mitogens, coined superantigens (1), that act by binding MHC class II molecules (2, 3) as unprocessed proteins (2, 3, 4), and cross-linking TCRs bearing particular β-chain variable elements (1, 5, 6). In this way, the enterotoxins are able to activate large numbers of T cells, enhancing the secretion of lymphokines such as IL-2 and IFN-γ (7, 8), and potentially leading to the deletion or inactivation of reactive T cell subsets (1, 9).
Staphylococcal enterotoxin B (SEB)4 (10) and the three isotypes of SEC (11, 12, 13) comprise a related group of enterotoxins that are nearly 70% identical at the amino acid level, and share a common three-dimensional folding pattern (14, 15, 16). These highly homologous enterotoxins have served as models to define regions on these proteins that promote MHC class II binding and presentation, TCR recognition, and T cell activation. Kappler et al. (17) utilized random mutagenesis to identify specific residues in the NH2-terminus of SEB that mediated interactions with TCR and MHC class II molecules. One residue in particular, Asn23, is conserved among all the enterotoxins and played a critical role in T cell activation, while two additional residues, Asn60 and Tyr61, altered only a subset of Vβ-specific responses. By testing intergenic SEC1-SEC2 hybrids, Deringer et al. (18) determined that residues at position 26 similarly controlled differences in Vβ-specific T cell activation. Recently, SEC2 and SEC3 have been cocrystallized with a murine TCR β-chain (19) and have provided a detailed map of enterotoxin-TCR interactions. Thirteen residues of SEC2 or SEC3, including Thr20, Asn23, and Tyr26, interact with elements of the TCR. Interestingly, all hydrogen bonds that contribute to this interaction exist between the amino acid side groups of the enterotoxin and main chain atoms of the TCR, suggesting that recognition is dependent on the overall conformation of the TCR and hence is relatively independent of the exact amino acid sequence.
Although enterotoxins presented by MHC class II molecules expressed on either professional APC (5) or MHC class II-transfected cell lines (20) are extremely potent T cell mitogens, several studies have indicated that MHC class II molecules are not absolutely required to activate T cells. Avery et al. (21) demonstrated that SEE and the three isotypes of SEC activated T cells obtained from MHC class II-deficient mice in an APC-dependent manner and promoted CTL-mediated lysis of class II-negative targets. Dohlsten et al. (22) and Herrmann et al. (23) similarly reported that SEB and SEC1 promoted the lysis of several MHC class II-negative human cell lines by CTL expressing a reactive TCR. Recently, Lando et al. (24) described a system whereby an SEA-Fab fusion protein presented on the surface of an MHC class II-deficient cell resulted in T cell proliferation and CTL-mediated lysis of the presenting cell. Utilizing this same system, the structural elements of SEA that conferred the ability to activate T cells in the absence of MHC class II were identified as amino acids contained within the NH2-terminal region spanning residues 20 to 27 (25), analogous to the region in SEE previously shown by our laboratory to be involved in the Vβ-specific recognition of TCR (26, 27). These studies indicate that enterotoxins may differentially activate T cells in an APC-dependent but MHC class II-independent manner, and suggest that this response is regulated through essential enterotoxin-TCR interactions.
We have recently characterized an experimental system in which certain enterotoxins are capable of inducing T cell proliferation in the presence of the MHC class II-deficient adenocarcinoma accessory cell line, SW480 (28). Specific binding of SEB to SW480 cells was not detected under a variety of circumstances, while SEB immobilized on beads activated T cells in the presence of SW480 cells, suggesting that SEB may bind and signal through TCR as molecules free in solution. SW480 cells were found to express several putative costimulatory molecules, and blocking studies revealed that ICAM-1/LFA-1 and LFA-3/CD2 interactions played a significant role in the activation of T cells. We interpreted these findings to suggest that enterotoxins can bind and signal through the TCR in the absence of a specific presenting molecule, and stimulate T cell proliferation with the addition of costimulation provided by SW480 cells.
While surveying a panel of staphylococcal enterotoxins for the ability to activate T cells in this manner, we noted that SEC1 acted as a potent mitogen, while SEC2 was unable to elicit T cell proliferation, representing an interesting functional difference between these two isotypes. Further analyses revealed that SEC2 competitively inhibited SEC1-mediated T cell activation but did not induce T cell anergy, suggesting that SEC2 was able to interact with the TCR but not signal in the absence of MHC class II. This functional difference existed despite the fact that SEC1 and SEC2 share over 95% identity at the amino acid level (29), activate overlapping subsets of Vβ-specific T cells (18, 30), and are equivalently potent when presented by a variety of MHC class II molecules (20). Utilizing a panel of hybrid enterotoxins incorporating nonconserved residues of SEC2 into the analogous positions of SEC1 (31), we identified specific amino acids in the NH2-terminus of SEC1 that abrogated class II-independent T cell activation, yet did not alter potency in the presence of class II. These residues map to the putative TCR binding domain of SEC1, and suggest that differences in the SEC1-TCR interactions promote T cell activation in the absence of MHC class II.
Materials and Methods
Cell lines and Abs
The SW480 cell line was obtained from the American Type Culture Collection (ATCC) (Rockville, MD) and maintained in RPMI 1640/10% FCS. mAbs against HLA-DR (L243), HLA-DR/DQ (L227), and CD11b (OKM1 and LM2/1.6.11) used for the negative selection of T cell populations were purified from the culture supernatants of B cell hybridomas obtained from ATCC. mAbs against Vβ3.1 (8F10), Vβ12 (S511), and Vβ13.1,3 (BAM13) were purchased from Endogen (Cambridge, MA), and Vβ17 (E17.5F3.15.13) and Vβ18 (BA62) from Immunotech, Inc. (Westbrook, ME).
Construction and purification of hybrid enterotoxins
The sec genes secMNDON and secFRI361, isolated from the genomes of Staphylococcus aureus strains MNDON and FRI361, were used to express native SEC1 and SEC2, respectively (12, 32). secMNDON was the origin of all the hybrid enterotoxins used in this study. Production of the SEC1 hybrids was described previously (31). Briefly, site-directed mutagenesis was employed to construct SEC variants in which one or more of the seven nonconserved residues in SEC1 was substituted with the corresponding residue from SEC2. Hybrid constructs were sequenced to confirm the presence of the correct nucleotide alterations and to insure that mutations at secondary sites in the gene had not occurred. The sec structural genes encoding native and hybrid proteins were subcloned into the chimeric expression vector (pMIN164) (13) and introduced into a nontoxigenic strain of S. aureus (RN4220) using protoplast transformation techniques (33). For large-scale production, clones expressing the native or hybrid enterotoxins were grown with aeration under erythromycin selection to stationary phase in dialyzable beef heart medium as previously described (13).
Native and hybrid enterotoxins were purified using preparative isoelectric focusing as previously described (34). Briefly, culture supernatants were precipitated with 4 vol of ice-cold ethanol and stored for several days at 4°C to facilitate recovery of the insoluble material. The precipitated material was recovered, redissolved, and dialyzed overnight against pyrogen-free distilled water to generate a crude toxin concentrate. The proteins in the retentate were resolved by preparative flatbed isoelectric focusing to obtain purified toxins. The proteins were initially separated in a pH gradient of 3.5 to 10. Fractions containing the toxins were identified by immunodiffusion, pooled, and refocused in a narrow pH gradient of either 6.0 to 8.0 or 7.0 to 9.0 (depending on the isoelectric point of the protein) to achieve maximum purification. Fractions containing homogenous enterotoxins, as determined by SDS-PAGE analysis, were pooled, aliquoted, and stored in a lyophilized state.
Purification of human T lymphocytes
PBMC, obtained from buffy coats of healthy donors (Gulf Coast Blood Center, Houston, TX) by density gradient centrifugation, were stained with a mixture of anti-class II (L243, L227) and anti-monocyte (LM2/1.6.11, OKM1) mAbs, and the cells were separated on goat anti-mouse-Ig-conjugated magnetic beads (Advanced Magnetics, Cambridge, MA). Two rounds of negative selection were employed to remove any contaminating APC, and typically yielded >98% CD3+ cells that were judged to be naive/resting based upon light scatter properties and cell surface expression of markers of activation/maturation (CD25−, CD45RA+).
Measurement of APC-dependent T cell proliferation
SW480 cells (1 × 107) were treated with 100 μg/ml of mitomycin C (Sigma, St. Louis, MO) for 1 h at 37°C, and washed extensively with HBSS/2% FCS. Purified human T cells (1.2 × 105) and either mitomycin C-treated SW480 cells (6 × 104) or autologous, irradiated (1500 rad) PBMC (2.4 × 105) were cultured in 200 μl of assay medium (RPMI 1640/10% FCS/100 mg/ml gentamicin/1% antibiotic-antimycotic mixture/2 mM l-glutamine/5 mM HEPES; all components from Life Technologies, Grand Island, NY) in a 96-well flat-bottom plate (Costar, Cambridge, MA) for 3 days. The cells were labeled for an additional 18 h with 1 μCi of [3H]thymidine (DuPont NEN, Boston, MA), harvested, and counted by liquid scintillation spectroscopy.
Flow cytometric analysis of T lymphocyte blasts
Purified T cells (5 × 106) and either mitomycin C-treated SW480 cells (2.5 × 106) or autologous, irradiated (1500 rad) PBMC (12.5 × 106) were cultured in 5 ml of assay medium in six-well plates (Costar) for 3 days. Viable cells were isolated from culture by density gradient centrifugation and recultured for an additional 24 h in the presence of 18 ng/ml IL-2 (R&D Systems, Minneapolis, MN) to expand the total number of proliferating cells and to promote the restoration of TCR expressed at the cell surface. Cells were stained with one of several TCR Vβ-specific mAbs and analyzed by flow cytometry on an Epics Profile (Coulter Corp., Hialeah, FL). Forward angle and 90° light-scatter patterns were used to restrict the analysis to blast-transformed T cells, as initially characterized by flow cytometric measurements of total DNA content and incorporated bromodeoxyuridine as a function of proliferation (data not shown).
SEC1, but not SEC2, activates human T cells in the absence of MHC class II
We have recently characterized an MHC class II-deficient human adenocarcinoma cell line, SW480, that supports proliferation of rigorously purified human T cells in the presence of SEB, SED, and SEC1, but not SEC2, SEC3, or SEE. SEA was weakly mitogenic under these conditions. Because of the extensive amino acid similarity between SEC1 and SEC2, we sought to determine the functional differences between these two enterotoxins. As seen in Figure 1, SEC1 stimulated robust T cell proliferation in the presence of SW480 cells after 4 days in culture with an ED50 of approximately 10 ng/ml. SEC2, however, was not mitogenic over a broad range of protein concentrations (Fig. 1), or at any time throughout the 4-day culture period (data not shown). In contrast, SEC1 and SEC2 were equivalently potent when presented in the context of MHC class II+ autologous PBMC with an approximate ED50 of 20 pg/ml. Importantly, SEC2 revealed no inhibitory effects over the concentration range of 0.001 to 1 μg/ml that might account for the lack of activity in the presence of SW480 cells. These observations illustrate an interesting function of SEC1, namely the ability to activate T cells without the need for professional or other MHC class II+ APC, that is not shared by the closely related isotype SEC2. This functional dichotomy is presumably due to structural differences encoded by the limited number of nonconserved amino acids that exist between SEC1 and SEC2 (Fig. 2). These residues may alter activity by changing the physical characteristics of the enterotoxin as it exists in solution or the way it binds and signals through the TCR.
SEC1 and SEC2 do not undergo differential posttranslational modification or multimerization
The amino acid differences between SEC1 and SEC2 may promote differential posttranslational modification of the mature protein or multimerization of the enterotoxins in solution that in turn alters the T cell signaling capacity. This is supported by the findings of Stevens et al. (35), who recently reported that species-dependent posttranslational modifications of the streptococcal superantigen SSA altered Vβ specificity and toxin susceptibility to protease digestion. In addition, both SEB and SED have been observed to form dimers or multimers in solution (36, 37), raising the possibility that SEC1 might selectively form multimeric complexes capable of cross-linking TCR similar in nature to anti-CD3 mAbs. Both SEC1 and SEC2 resolved as discrete 27.5-kDa molecules by SDS-PAGE under reducing conditions (Fig. 3,A) or by electrospray mass spectrometry (Table I), in agreement with their predicted m.w.s, suggesting that the enterotoxins lacked any covalently linked amino acid modifications, such as glycosylation (38), phosphorylation (39), or acylation (40). The 16-Da difference noted between the observed and predicted molecular masses of SEC1 and SEC2 are consistent with the oxidation of methionine to methionine sulfoxide. Interestingly, preparations of both SEC1 and SEC2 revealed at least three distinct bands by native PAGE (Fig. 3,B), suggesting the presence of protein multimers. The migration patterns of the bands from SEC1 and SEC2 differed under these conditions due to a difference in isoelectric points. SEC1 and SEC2 were further analyzed by gel filtration chromatography to confirm the presence of protein multimers in the absence of any possible gel artifacts. As seen in Figure 4, the chromatograms of both SEC1 and SEC2 revealed major peaks of monomeric protein that eluted at approximately 28 kDa, as well as minor 56-kDa peaks. The 56-kDa fractions when analyzed by SDS-PAGE resolved as 28-kDa bands identical to monomeric SEC1 (data not shown) and corresponded to 3.7% of the monomeric SEC1 peak and 1.0% of the monomeric SEC2 peak, respectively. When rerun over the gel filtration column, the 56-kDa fraction eluted again as a 56-kDa peak with no traces of a 28-kDa peak. Likewise, when the 28-kDa fraction was rerun over the column, no traces of a 56-kDa peak were detected, despite the addition of Zn2+ or Mg2+, suggesting that these two species are not in equilibrium (data not shown). Moreover, the 28-kDa, 56-kDa, and unseparated fractions of SEC1 were equivalently potent T cell mitogens in the presence or absence of MHC class II molecules (data not shown). Thus, a small amount of SEC1 and SEC2 can exist as dimers in solution. However, no significant differences in activity were noted between these proteins that might account for the differences in the ability to activate T cells in the absence of MHC class II.
|.||Molecular mass (Da) .||.||.|
|.||Predicteda .||Observedb .||Δc .|
|.||Molecular mass (Da) .||.||.|
|.||Predicteda .||Observedb .||Δc .|
Molecular mass predicted by the deduced amino acid sequence of the mature proteins.
Molecular mass observed by electrospray mass spectrometry, performed as previously described (35).
Difference in molecular mass calculated as follows: observed mass − predicted mass.
SEC1 and SEC2 stimulate overlapping Vβ-specific T cell subsets
Kotzin et al. (30) previously demonstrated that SEC1 and SEC2 stimulate overlapping subsets of T cells as defined by TCR Vβ usage. Utilizing reverse transcriptase-PCR analysis, however, Deringer et al. (18) reported slight quantitative differences in the ability of these enterotoxins to activate Vβ3 and Vβ13.1 T cells that is mediated by the nonconserved amino acid at position 26 in SEC1. A difference in Vβ utilization by SEC1 and SEC2 might alter the potency of a T cell proliferative response due to differences in the frequency of reactive T cells. We therefore examined Vβ utilization by SEC1 and SEC2 in the presence or absence of class II to determine whether Vβ specificity could account for the functional differences in class II dependency. As seen in Table II, SEC1 and SEC2 stimulated overlapping Vβ subsets in the presence of MHC class II+ PBMC, as denoted by robust proliferation of Vβ17 T cells and intermediate levels of proliferation of Vβ3, Vβ12, and Vβ13 T cells. Vβ18 is not recognized by either SEC1 or SEC2 and thus served as a measure of specificity. SEC1 in the presence of SW480 cells activates similar Vβ-specific subsets as in the presence of class II molecules, while SEC2 plus SW480 cells yielded no detectable T cell blasts. The sum of the proportion of Vβ3, Vβ12, Vβ13, and Vβ17 T cell blasts resulting from stimulation by SEC1 plus SW480 constituted 33.8% of the total blast population, while the sum of the proportion of these same subsets stimulated by SEC1 plus PBMC constituted 25.2% of the total blasts. This difference indicates an expansion of fewer Vβ subsets in the presence of SW480 cells, and may reflect a need for higher affinity interactions with a more limited subset of TCR to activate T cells in the absence of MHC class II. Newton et al. (41) similarly noted that when the low-affinity binding site for MHC class II was disrupted in SEA, the mutant lost the ability to activate four of seven Vβ families normally stimulated by native SEA. These findings thus demonstrate that SEC1 is capable of activating equivalent Vβ subsets in the presence of either SW480 cells or MHC class II+ PBMC, and moreover, that these subsets are comparable to the Vβ subsets activated by SEC2 in the presence of class II. These observations suggest then that the inability of SEC2 to activate T cells in the absence of MHC class II is not simply due to a difference in Vβ specificity.
|Toxin .||APC .||% Positive T Cell Blasts .||.||.||.||.|
|.||.||Vβ3 .||Vβ12 .||Vβ13 .||Vβ17 .||Vβ18 .|
|Toxin .||APC .||% Positive T Cell Blasts .||.||.||.||.|
|.||.||Vβ3 .||Vβ12 .||Vβ13 .||Vβ17 .||Vβ18 .|
T cells were activated with SW480 plus SEC (200 ng/ml) or PBMC plus SEC (2 ng/ml) and stained with a panel of Vβ-specific mAbs. Analysis was restricted to the blasting T cell population based on flow cytometric light scatter properties. Values represent the percentage of the T cell blast population that stained positive for the indicated mAb.
ND indicates that no blasts were detected.
SEC2 inhibits T cell activation induced by SEC1 in the absence of MHC class II
SEC2 may fail to stimulate T cells in the absence of MHC class II for several reasons. SEC2 may simply not bind TCR with sufficient affinity to elicit signaling. This hypothesis is in contrast to the observations of Malchiodi et al. (42) that SEC2 is able to bind soluble TCR β-chain in the absence of class II with a slightly higher affinity than SEC1. Alternatively, SEC2 may engage the TCR and fail to signal, or deliver an aberrant signal that subsequently makes the T cell refractory to proliferation. Since SEC1 and SEC2 stimulate overlapping T cell subsets, we attempted to differentiate among these possibilities by testing the ability of SEC2 to compete with SEC1 for TCR-mediated signaling in the absence of class II. As seen in Figure 5, the level of proliferation induced by a submaximal concentration of SEC1 in the presence of SW480 cells was inhibited in a dose-dependent manner by SEC2. In the presence of PBMC, however, both SEC1 and SEC2 were stimulatory and the level of proliferation induced by the combination of SEC1 and SEC2 was greater than either enterotoxin alone. These findings indicate that SEC2 is able to engage the TCR and compete with SEC1 for interactions that result in T cell activation in the absence of class II. Since SEC2 is able to bind the TCR, we addressed whether SEC2 was subsequently anergizing the T cells as a mechanism of nonresponsiveness. We pretreated purified T cells with 2 μg/ml of SEC2, a concentration observed to inhibit T cell proliferation induced by SEC1 in the presence of SW480 cells, for 24 h and examined the T cells for markers of anergy induction and for the ability to proliferate upon restimulation in the presence of MHC class II. Pretreatment of the T cells failed to result in CD2 or CD25 up-regulation, or CD3 down-regulation (data not shown) as had been previously observed during the induction of anergy by SEB (43). In addition, the levels of proliferation resulting from restimulation of these pretreated T cells by SEC2 in the presence of PBMC were not significantly inhibited compared with untreated T cells (Table III). These data suggest that anergy does not account for the inability of SEC2 to stimulate T cell proliferation in the absence of class II. Taken together, these findings indicate that SEC2 is able to engage the TCR in the absence of class II, but fails to deliver a productive signal that leads to proliferation, while SEC1 is able to bind and signal through the TCR on its own.
|Pretreatment .||cpm (mean ± SD) .|
|SEC2||106,873 ± 5,849|
|None||111,886 ± 3,994|
|Pretreatment .||cpm (mean ± SD) .|
|SEC2||106,873 ± 5,849|
|None||111,886 ± 3,994|
Purified T cells were cultured with or without SEC2 (2 μg/ml) for 24 h at 37°C, washed extensively, and recultured in the presence of PBMC plus SEC2 (50 pg/ml). [3H]Thymidine incorporation was measured in a standard 4-day proliferation assay. Data points represent the mean cpm of duplicate determinations ± SD.
Residues in SEC1 that mediate class II-independent activation reside in the putative TCR binding domain
As noted in Figure 2, the amino acid differences between SEC1 and SEC2 are interspersed throughout the NH2-termini among residues that comprise the putative TCR and MHC class II binding domains. We sought to define the amino acids in SEC1 that contributed to MHC class II-independent T cell activation to examine in detail the mechanisms that underlie this mode of T cell activation. We utilized a panel of hybrid enterotoxins (31) incorporating the nonconserved residues of SEC2 into analogous positions of SEC1 and tested these hybrids for the ability to activate T cells in the presence of SW480 cells. We limited our analysis to positions 16, 20, 22, and 26, which are identical among SEB and SEC1, but differ in both SEC2 and SEC3, and hence correlate with the ability to induce class II-independent T cell proliferation. The results obtained from several experiments are summarized in Figure 6. The hybrid molecule containing the single amino acid change K16E was similar to SEC1 in potency, suggesting that this residue does not play a role in T cell activation in the absence of class II. The hybrid molecule containing the single amino acid substitution V26Y was approximately 100-fold less potent than SEC1, as noted by a shift in the dose-response curve, suggesting that the Tyr26 substitution may alter TCR binding interactions for certain β-chains, as predicted by mutational and structural studies (16, 18). However, the hybrid containing the two amino acid changes E22G and V26Y possessed a potency similar to SEC1, suggesting that perhaps Tyr 26 itself deforms the TCR binding pocket slightly, and this perturbation is relieved by the removal of Glu at position 22. Taken together, these data suggest that position 26, although important in SEC2 for mediating Vβ-specific interactions with the TCR, is not critical for T cell activation in the absence of class II. The hybrid containing the single amino acid substitution L20T, however, was severely impaired in the ability to stimulate T cell proliferation, as noted by its shallow dose-response curve. When the L20T substitution was introduced into SEB, the T cell mitogenicity of this mutant was reduced >75% compared with recombinant SEB (data not shown), suggesting that position 20 is a critical residue in the SEB and SEC families of enterotoxins for the induction of class II-independent T cell proliferation. When the L20T substitution was combined with the amino acid change E22G in SEC1, the resulting hybrid completely lost the capacity to stimulate T cell proliferation over a wide range of protein concentrations, similar to the hybrid containing substitutions at positions 20, 22, and 26, and native SEC2 (Fig. 6). The hybrid containing the amino acid changes L20T and V26Y was also weakly mitogenic, comparable to the hybrid containing Thr at position 20. In contrast, the hybrid enterotoxins, when presented to T cells in the context of PBMC, were all equivalently potent at a concentration of 20 pg/ml corresponding to the ED50 of SEC1 and SEC2 in the presence of MHC class II+ APC (Fig. 7). This suggests that the hybrids contain no major conformational defects, only minor changes in the local topology of the TCR binding domain, and further suggests that the functional deficiencies of the hybrids containing the substitution L20T can be compensated by class II presentation. These findings indicate that the amino acids at positions 20 and 22 are necessary in SEC1 to facilitate T cell activation in the absence of class II, while SEC2 requires a component of MHC class II for effective T cell stimulation in the presence of SW480 cells.
We have characterized the activation of T cells by SEC1 in the absence of MHC class II molecules and have shown that the closely related enterotoxin SEC2 fails to induce T cell proliferation under similar circumstances. The native and hybrid enterotoxins utilized in these studies were produced in a nontoxigenic strain of S. aureus to insure freedom from any contaminating superantigens, and were subsequently shown to lack any differential posttranslational modifications or to undergo differences in protein multimerization that might secondarily alter toxin activity. Several of the nonconserved amino acids of SEC1 and SEC2 reside in the putative TCR binding domains, suggesting that differences in the TCR interactions directly influence mitogenic activity in the absence of class II. The difference in function, however, could not be attributed solely to differences in Vβ specificity since SEC1 and SEC2 activated equivalent subsets of T cells in the presence of class II, which overlap with the T cell subsets stimulated by SEC1 in the absence of class II. Site-directed alterations of residues in SEC1 to the corresponding residues in SEC2 revealed that two amino acid changes within the TCR binding domain of SEC1 abrogated activity, suggesting that slight alterations of enterotoxin-TCR interactions could account for differences in the ability to activate T cells in the absence of class II.
The amino acid substitutions L20T and E22G in SEC1 resulted in a complete loss of T cell mitogenicity in the absence of MHC class II. Interestingly, these alterations did not affect T cell potency in the presence of class II, suggesting that these residues play a specialized role in promoting class II-independent TCR interactions. These two amino acid modifications may have altered the function of the native SEC1 molecule in a number of ways. The crystal structure of SEC2 complexed with a murine Vβ8 β-chain demonstrated that the side group of Thr20 in SEC2 contributes to a hydrogen bond with a main chain atom of Thr55 in the TCR. In addition, Thr20 is also involved in contacting Lys57 of the TCR. The alteration L20T in SEC1 may therefore alter the topology of the β-chain binding pocket and hence disrupt critical TCR interactions that mediate TCR binding affinity, off-rate or signal transduction. This is supported by structural differences revealed by the comparison of the crystal structures of SEB and SEC2 (16). In SEC2, Nδ2 of Asn23 is engaged in a hydrogen bond with Oγ1 of Thr20. In SEB, and possibly SEC1, this hydrogen bond is absent due to the presence of Leu at position 20, leaving Nδ2 of Asn23 available for hydrogen bonding. In this state, perhaps N23 of SEB and SEC1 is able to hydrogen bond with a critical residue of the TCR, thereby stabilizing enterotoxin-TCR binding and signaling. Alternatively, the introduction of Thr at position 20 may reconstitute the formation of a hydrogen bond with main chain atoms of the TCR, as observed with SEC2, and consequently force the hybrid into a nonproductive interaction with the TCR. Gly22 in SEC2, on the other hand, was not shown to be involved in any direct TCR interactions. The presence of Glu at position 22 in SEC1 may thus introduce another contact with the TCR, perhaps through an ionic interaction with its negatively charged side group, as suggested by the topology of the TCR binding site of SEB (14). Alternatively, Glu22 itself may simply effect a change in the local topology of SEC1 that coordinates the orientation of other TCR binding residues. This latter hypothesis is supported by the findings that the combined amino acid changes E22G and V26Y in SEC1 did not lead to a loss of T cell mitogenicity in the absence of class II, but rather restored potency to the hybrid containing the single amino acid change V26Y. It remains to be seen whether conserved residues other than those identified in SEC2 are utilized by SEC1 to bind TCR either in the presence or absence of class II.
These observations suggest that residues within SEC1 are able to compensate for a lack of MHC class II presentation and enable the enterotoxin to bind and signal directly through the TCR, while under these same conditions, it is necessary for SEC2 to be presented by class II+ APC to deliver a productive signal. This requirement for MHC class II may reflect the need by SEC2 to undergo an allosteric change upon class II binding to expose critical TCR binding residues. This seems unlikely in light of crystallographic data that suggest that there are no large conformational changes in either SEB or the class II molecule upon formation of the SEB-HLA-DR1 complex (44). Moreover, SEC1 and SEC2 both bind to soluble β-chain in the absence of MHC class II (42). Alternatively, SEC2 may require residues of MHC class II to contact the TCR to fully activate T cells. Modeling of the SEC2-TCR β-chain structure with HLA-DR1 predicted contact between the α helix of MHC class II β1 region and both CDR1 and CDR2 of the TCR Vα-chain (19). Furthermore, Deckhut et al. (45) have provided evidence that TCR recognition of SEB in the context of an I-E molecule utilizes functional interactions between the β-chain of class II and the TCR α-chain. The observation that the ED50 of SEC1 in the presence of class II+ APC is 500-fold lower than in the absence of class II suggests that SEC1 may similarly utilize components of the MHC class II molecule when available to increase T cell potency and perhaps expand Vβ reactivity. The seeming requirement for class II may also arise because SEC2 requires additional costimulation ordinarily provided by an MHC class II+ professional APC to fully activate T cells. This possibility is less likely, however, since SEC1 and SEC2 are both potent mitogens when presented by HLA-DR1-transfected murine fibroblasts lacking conventional costimulatory molecules for human T cells (20).
Lando et al. (24) recently described a system in which SEA, but not SEE, fused to the Fab portion of a tumor-specific Ab and presented on the surface of an MHC class II-deficient tumor cell line stimulated T cell proliferation and CTL activity. Antonsson et al. (25) further demonstrated that structural elements contained in the NH2-terminus of SEA spanning residues 20 to 27 supported class II-independent T cell activation. These investigators thus speculated that residues in this region might serve a more general function in the activation of T cells, perhaps by enhancing the affinity of the enterotoxin-TCR interaction. Our findings confirm that residues contained in this region do indeed enhance T cell activation. In the context of SW480 cells, the enterotoxins that contain the amino acid motif Leu20-X-Glu22 (Fig. 2), including SEB and SEC1, are potent mitogens, whereas enterotoxins containing the amino acid motif Thr21-X-Gly22, including SEC2 and SEC3, are not mitogenic. The mitogenic potential of SEA in the absence of class II, despite possessing an NH2-terminal TCR binding motif similar to SEC2 and SEC3 (25), may result from additional amino acid differences between these two families of enterotoxins that contribute to TCR binding and signaling. In contrast to this model, Avery et al. (21) reported that SEC1, SEC2, SEC3, and SEE were all equivalently potent in activating T cells obtained from MHC class II-knockout mice, while SEA and SEB were not mitogenic. One obvious explanation for this striking difference is that of critical differences in contact residues between enterotoxins and the responsive human (our data) and mouse (21) TCR Vβ. In support of this hypothesis, and consistent with the observations of Avery et al. (21), we have examined a library of SEA-SEE hybrid molecules (26, 27) for the ability to activate T cells obtained from class II-knockout mice. Interestingly, an SEA hybrid incorporating the four nonconserved amino acids spanning residues 20 to 27 of SEE was not mitogenic, while an SEA hybrid containing the 11 nonconserved amino acids within the first 70 residues of SEE was equally potent to recombinant SEE (data not shown). These observations indicate that the recognition of murine T cells is not fully dependent on a motif contained within the NH2-terminal region spanning residues 20 to 27 of SEE, suggesting that other structural motifs of the enterotoxins are required to contact murine TCR in the absence of MHC class II molecules. Definitive resolution of such observations, however, will likely require comparative crystallographic analyses. Alternatively, different structural motifs within SEE may be utilized to contact murine TCR in the context of an alternate enterotoxin receptor. That alternative receptors exist for enterotoxins is supported by the findings of Beharka et al. (46), who described a low-affinity enterotoxin receptor expressed on macrophages obtained from class II-knockout mice, suggesting that this receptor may be active in presenting selected enterotoxins to murine T cells.
In conclusion, we have identified a motif in the SEB and SEC families of enterotoxins that promotes T cell activation in the absence of MHC class II molecules. We previously showed that an analogous motif in the SEA and SEE families of enterotoxins mediates differences in Vβ specificity (27), suggesting that this domain of the staphylococcal enterotoxins in general serves to engage certain TCR and either alter the binding capacity through changes in the affinity or avidity of the toxin-TCR interaction, or alter the nature of signaling through the TCR. These modifications of the toxin-TCR interaction by SEB and SEC1 thus support a mechanism to initiate T cell activation that effectively eliminates the need for MHC class II+ APC and may alter the subsequent immune response in vivo by perhaps selectively stimulating a subtype of T cell or by promoting activation-induced cell death. Thus, staphylococcal enterotoxins may utilize multiple mechanisms to activate T cells and manipulate the resulting effector response.
The authors thank Claudia Beck Deobald and Mai Van for their expert technical assistance.
This work was supported by United States Public Health Service Grant RO1-AI30036 (R.R.R.), and United States Public Health Service Grant RO1-AI28401 and United States Department of Agriculture Grant NRI94-02399 (G.A.B.).
Abbreviations used in this paper: SEA through SEE, staphylococcal enterotoxin A through E.