Highly Biased CDR3 Usage in Restricted Sets of β Chain Variable Regions During Viral Superantigen 9 Response

The nomenclature used for BV genes is according to Wei et al. (11). PBMCs were separated by Ficoll centrifugation (Pharmacia Biotech AB, Uppsala, Sweden) and stimulated for 24 h with 1 μg/ml of PHA (Murex Diagnostics, Guelph, Canada). Total RNA was extracted from 4 × 10⁷ cells with RNAzol B (CINNA/BIOTECX Laboratories, Houston, TX) and 10 μg were reverse transcribed with 5 μg of oligodT and 60 U of AMV-RT (Life Sciences, St. Petersburg, FL). The BV6S7 cDNAs were amplified by PCR using primers and conditions described elsewhere (12), digested with BamHI, and fractionated on 2% gels.

Human T cells were purified from four donors homozygous for either BV6S7 allele using rosetting with sheep red blood cells (Quélab, Montréal, Canada) and PBMCs separated by Ficoll centrifugation. Purity was assessed by staining with anti-CD3 (OKT3-FITC) and anti–HLA-DR (L-243) mAbs (Becton Dickinson, Mountain View, CA) and T cell purity was >99%, whereas the non–T cell fraction contained <15% of T cells. T cells were stimulated in the presence of irradiated autologous feeder cells with DAP-DR1 cells, which are murine fibroblasts transfected with DR1 α and β chain cDNAs, or DAP-DR1 expressing vSAG9 (13). DAP-DR1 and DR1-vSAG9 cells were treated for 1 h at 37°C with 100 μg/ml mitomycin C (Sigma Chemical Co., St. Louis, MO) and cocultured in 96-well plates in DMEM, 5% FCS, 2 mM l-glutamine, and 20 μg/ml gentamycin. 6 × 10⁵ human T cells were cultured with 2 × 10⁵ DAP-DR1 or DR1-SAG9 in the presence of 5 × 10⁵ autologous irradiated feeder cells. After 2 d, 10 U/ml of recombinant human IL-2 was added and the culture proceeded for an additional 7 d. For PHA stimulation, 10⁶ T cells were cocultured in 24-well plates with 2 × 10⁵ irradiated feeder cells and 1 μg/ml PHA in 1.5 ml. Proliferation was measured by [³H]thymidine incorporation (Dupont Co.–New England Nuclear, Boston, MA) after 3 d and determined with a β plate counter (Pharmacia Biotech AB).

FACS^® analysis was performed after T cell stimulation with DR1-SAG9 or PHA, using CD4, CD8, and BV-specific mAbs. The BV-specific mAbs used were anti-BV2, 3, 8, BV13S3 (JU74), 17, 19, 21 (Immunotech Coulter, St. Laurent, Canada), BV5 (MH3-2), BV5 (3D11), BV6S7*1 (OT145), BV9 (MKB1), BV12 (S511), BV13 (BAMB), BV13S1 (H131), BV13S2 (H132), and BV23 (HUT-78). Cells were stained for CD4 (OKT4-PE) and CD8 (OKT8-PerCP) (Becton Dickinson) and 3 × 10⁵ live cells were gated according to forward and side light scatter and analyzed on a FACScan^®. For quantitative PCR (qPCR), cDNAs were synthesized as described above and PCR was performed using conditions and primers described elsewhere (14). PCR products were fractionated on 12% polyacrylamide gels and exposed overnight on PhosphorImager screens (Molecular Dynamics, Sunnyvale, CA). Quantification of signals was performed on a PhosphorImager running the IMAGEQUANT software (Molecular Dynamics) and normalized against a TCR α chain constant region internal control (14).

BV6 members were amplified by PCR using DeepVent (New England Biolabs, Mississauga, Canada) using the constant region primer GGTGTGGGAGATGTCGACTTTTGATGGCTCAAAC and the pan-BV6 primer: CCTTTACTGGTACCGACAGAGCCTGG. PCR products were digested with KpnI and SalI, cloned into pBSKS⁺ (Stratagene, La Jolla, CA), and recombinants were screened by PCR using reverse and universal primers (15). BV6 members were subdivided by RFLP based on the presence of BamHI and ApaLI sites (BamHI⁺ApaLI⁺: BV6S7*1; BamHI⁻ ApaLI⁺: BV6S7*2; BamHI⁻ApaLI⁻: BV6S5; BamHI⁺ApaLI⁻: BV6S1, BV6S3, BV6S4, BV6S11, and BV6S14). 10 μl of PCR reactions were digested with BamHI and ApaLI and fractionated on 2% gels. Sequence determination was performed using the T7 Sequencing Kit (Pharmacia Biotech AB).

Since vSAG9 can stimulate BV6S7⁺ human T cells and polymorphism between BV6S7 alleles is located within the HV4 region (16), we assumed that responsiveness to vSAG9 might differ between individuals homozygous for either allele and that analysis of BVs expanded in donors bearing a less responsive allele might reveal the contribution of non-HV4 elements in vSAG9 response.

Since the murine cell line DAP, expressing both HLA-DR1 (DAP-DR1) and vSAG7, has been shown to stimulate human T cells in a BV-restricted manner (14), we used DAP-DR1 transfected with vSAG9 (DR1-SAG9; reference 13) to stimulate T cells from an individual homozygous for BV6S7*1 (donor J). Proliferation measured by thymidine incorporation peaked after 4 d of coculture and was reproducibly fourfold higher compared to control DR1 cells (data not shown). FACS^® analysis of BV usage in human CD4⁺ T cells in response to vSAG9 or PHA, a mitogen that stimulates T cells independently of BV usage (17 and data not shown), shows that vSAG9 stimulated BV6S7*1⁺ and BV21⁺ T cells (Fig. 1,A). Since the BV-specific mAbs currently available cover ∼65% of the repertoire, qPCR analysis was performed and this confirmed that only BV6 and BV21 responded to vSAG9 and were amplified four- and threefold, respectively (Fig. 1,B). This selective expansion was reproduced on a second BV6S7*1 homozygous individual (data not shown). T cells derived from two individuals homozygous for the BV6S7*2 allele (donors S and M) were stimulated with vSAG9 and proliferation was six- and fourfold higher compared to control DAP-DR1 cells (data not shown). Quantitative PCR analysis was performed and, as with the BV6S7*1 donors, only BV6⁺ and BV21⁺ T cells were amplified (Fig. 2, A and B). The responsiveness of BV6 and BV21 is not surprising given that they share significant homology, notably in the CDR1 and HV4 regions (18). HLA typing for MHC haplotypes was performed and showed that donor J was DR1/DR1, donor S was DR7/DR7, and donor M was DR2/DR7. Although donors S and M express MHC molecules different from DR1, T cell proliferation after 3 d of coculture with the control DR1 transfectant was similar between the DR1 donor (3.2 × 10⁴ cpm) and the non-DR1 donors (3.1 × 10⁴ and 3.0 × 10⁴ cpm), indicating that no significant allogeneic response occurred, which is not surprising since DAP cells are known to be poor at eliciting alloresponses (19). From qPCR analysis, it is apparent that the overwhelming majority of T cells that responded to vSAG9 belonged to the BV6 family (Figs. 1,B, 2, A and B). Since the human BV6 locus contains seven expressed genes (18) and our qPCR did not allow discrimination between subfamily members, BV6s were amplified using a pan-BV6 primer and cloned for analysis. Approximately 200 clones from PHA- and vSAG9-stimulated cells were analyzed by RFLP. For donor J (BV6S7*1 homozygous), 98% of the clones obtained after vSAG9 stimulation used BV6S7. With donors S and M (BV6S7*2 homozygous), RFLP analysis showed that 64 and 87% of the clones used BV6S7, although a significant number of other BV6 gene segments were also obtained after stimulation (Table 1). For donor S, 27% of the BV6s present after vSAG9 stimulation were BV6S5⁺, whereas in donor M, 5% of BV6 gene segments were BV6S5⁺. For all donors, BV6S7 was clearly the best responder to vSAG9, as evidenced by the low numbers of BV6⁺, non-BV6S7 cells present after stimulation. For donor M, the proportion of BV6S7*2-positive clones present in the PHA-stimulated population was twice that of the other donors (Table 1) and appear to have dominated the vSAG9 response, perhaps explaining the lower number of BV6S5 found after stimulation.

To evaluate the possible contribution of the β chain CDR3 in recognition of the vSAG9–MHC complex, TCR junctional regions from T cells stimulated by vSAG9 or PHA were sequenced. Sequence analysis of the BV6S7 clones revealed no particular BJ selection from either vSAG9- or PHA-stimulated cells, with a random BJ usage and CDR3 length distribution (data not shown, Table 2). In contrast, the BV6S5 clones obtained after vSAG9 stimulation in donor S revealed a striking bias; 89% of BV6S5 clones amplified used the BJ1S5 element (Table 2), whereas BJ1S5 usage in BV6S5 clones from PHA-stimulated T cells was not elevated (10%; data not shown). Two dominant clonotypes were found, with CDR3 regions bearing a P(Q/E)NSG motif (Table 2), created entirely by N-additions at the V–J junction. In donor M, a dominant BV6S5 clonotype, having the totally N-encoded CDR3 region, LEHSTRP, represented 89% of the BV6S5⁺ clones present after vSAG9 stimulation (Table 2), whereas this clonotype could not be detected in PHA-stimulated cells (data not shown). The number of BV6S1, S3, and S4 present after vSAG9 stimulation was low in all donors, and the total lack of CDR3 bias seen in these clones suggests that they have not been amplified. Thus, the bias observed in the CDR3 region of BV6S5 clones suggests that intimate MHC–TCR contacts might exist during vSAG response. Other studies have yielded similar results, as sequence analysis of TCR-β junctional regions in BV6⁺ CD4⁺ T cells that survived deletion mediated by vSAG7 showed a BJ usage and CDR3 length distribution that differed significantly from H2-matched mice lacking vSAG7 (8). Introduction of an Eα transgene in SJL mice, which do not express I-E and therefore can not present the endogenous vSAG9, restores deletion of BV17⁺ T cells; when compared to nontransgenic mice, T cells having survived negative selection showed increased BJ2S5 and decreased BJ1S1 usage, indicating that the nature of the BJ segment can have an impact on T cell deletion (7). Our results are in agreement with these studies and extend them by providing evidence that the structure of the CDR3 plays a positive role in vSAG responsiveness.

Comparison of HV4 sequences between BV6s and BV21s reveals that they are highly homologous except for the presence of a glutamic acid at the tip of the HV4 loop in all nonresponding BV6s. The allelic polymorphism between the BV6S7s is located next to that residue in which the glycine present in BV6S7*1 is replaced by a glutamic acid in BV6S7*2. These differences might partially explain the differential reactivity observed between BV6 subfamily members. Although HV4 is clearly the overriding determinant in vSAG recognition, it appears that a BV with a suboptimal HV4 would become more dependent on stabilization provided by TCR–MHC contacts. As mutagenesis of the CDR1 in murine BV6 was shown to abrogate both vSAG7 and conventional antigen recognition (20), this argues in favor of a similar MHC– TCR topology. In addition, mutagenesis in CDR1 (but not CDR2) affects vSAG responsiveness, and mutagenesis in CDR2 (but not CDR1) affects bSAG reactivity, indicating that vSAGs and bSAGs probably bind differently to the TCR (21). Mutagenesis studies have shown the importance of CDR2 in bSAG recognition (21, 22) and the structure of SEC3 bound to TCR shows that it contacts the CDR2 peptide backbone and should act like a wedge between TCR and MHC, allowing only part of the MHC to contact the TCR (10). In addition, the structure of SEB bound to DR1 predicts that the SAG-MHC-TCR orientation should differ by 45° compared to MHC-peptide-TCR orientation (23).

Our results and those from others (7, 8) indicate that BJ usage impacts vSAG recognition. Although this region might be a contact site for vSAGs, we feel this is unlikely given it is located opposite to HV4 (4, 9) and that the two BV6S7*2 homozygous individuals used different BJs. The BJ bias observed during vSAG9 response might be due to a TCR interaction with a peptide present in the MHC groove, since the BJ gene segment contributes for a significant portion of the CDR3 (4, 9). The different CDR3s found in the two donors could be due to recognition of different peptides or a dominant peptide being recognized by both TCRs, since TCRs having identical BV segments, but different CDR3 sequences, can recognize the same peptide– MHC complex (24). Since the TCR α and β chains CDR1 and CDR2 are involved in MHC contacts (4), this would explain α chain biases (5, 6) and the contribution of the β chain CDR1 (20, 21), whereas the skewed BJ usage observed (7, 8) could be explained by CDR3 contacts with peptide–MHC complexes. Thus, the data in the literature about the role of non-HV4 regions in vSAG responses could be readily reconciled using a model in which vSAGs cross-link MHC and TCR in a way that allows the interactions occurring during conventional antigen recognition to exist.

We would like to thank Walid Mourad for performing HLA typing of individuals and are grateful to Pascal Lavoie, Hugo Soudeyns, and Ursula Utz for critical reading of the manuscript.

This work was supported by grant No. 6605-4847-AIDS from the National Health Research Development Program to F. Denis and grant No. RG-544/95M from the Human Frontier Foundation to R.-P. Sékaly. R.-P. Sékaly holds a Medical Research Council Scientist Award and C. Ciurli has received a PhD fellowship from the National Health Research Development Program of Canada.