Ofer Mandelboim, ^* Daniel M. Davis, ^* Hugh T. Reyburn, ^* Mar Valés-Gómez, Eric G. Sheu, Laszlo Pazmany, Jack L. Strominger

An important feature of the human immune system is the ability of T cells to respond to small quantities of antigen. Class II major histocompatibility complex (MHC)-restricted T cells that expressed a costimulatory natural killer (NK) cell receptor for class I MHC proteins were cloned. In the presence of low doses of superantigen, the proliferative response of these T cell clones was three- to ninefold greater when the T cells were costimulated by way of the NK receptor. Thus, the action of costimulatory NK receptors on T cells may play a significant role in initiating and sustaining immune responses.

Department of Molecular and Cellular Biology, Harvard University, 7 Divinity Avenue, Cambridge, MA 02138, USA.
^*   These authors contributed equally to this work.

   To whom correspondence should be addressed. E-mail: jlstrom{at}fas.harvard.edu

Recently, an isoform of NK inhibitory receptors was described. These p50 NK receptors share similar extracellular sequences with the p58 receptors but have altered transmembrane regions, including the addition of a charged lysine residue, and truncated cytoplasmic tails (3, 11), so that they lack the YXXL sequences (L, Leu; Y, Tyr; X, any amino acid) to which protein tyrosine phosphatases can bind. Such short-tailed NK receptors do not mediate inhibition of NK cell-mediated lysis but instead activate or coactivate NK clones (11). To investigate the possible effect of NK activating receptors on T cell function, we studied T cell clones isolated as by-products in NK cell cloning (12).

Two of these clones, TANK-1 and TANK-9, were prepared from a donor whose HLA type is HLA-A1, -A2, -B7, -B8, -Cw7, -DR1, -DR17, -DQw2, and -DQw5. Clone TANK-1 was positive for TCRBV17 (V17) and expressed neither NK1, NK2, nor NK3 receptors by flow cytometry (13). Cloning by polymerase chain reaction (PCR) and sequencing of cDNA derived from TANK-1 (14) revealed the expression of the short-tailed activating NK receptor, clone 39 (3). Clone TANK-9 was positive for TCRBV2 (V2) and expressed both NK1 and NK2 receptors by flow cytometry (13). PCR typing of cDNA derived from TANK-9 (15) revealed that TANK-9 expresses the activating, short-tailed form of the NK1 receptor (NKAR1) and the inhibitory, long-tailed form of the NK2 receptor (NKIR2), as well as TCRVA24 (V24). These T cells with activating NK receptors (TANK cells) were positive for TCR, CD4, and CD3 and negative for TCR, CD8, CD16, CD56, and CD94 by flow cytometry (13, 16).

To determine the effect of NK activating receptors on T cell responses, we examined the proliferative response of TANK-1 and TANK-9 in the presence of superantigen-coated 721.221 cells, human HLA-DR1⁺ B cells deficient in cell-surface expression of class I MHC proteins (17), and 721.221 transfectants (18). As expected from the TCRVB gene segment expressed by each clone, TANK-9 responded to toxic shock syndrome toxin 1 (TSST1) (Fig. 1A) and not to staphylococcal enterotoxin A (SEA) or SEB (19), whereas TANK-1 responded to SEB (Fig. 1B) and not to SEA (19). At all the tested concentrations of TSST1, the proliferation of TANK-9 was greater when the target cell was transfected with either HLA-Cw4 or -Cw6 and reduced when the target cell was transfected with either HLA-Cw3 or -Cw7 (Fig. 1A).

Target cells that enhanced the proliferative response expressed class I MHC proteins that were recognized by NK1 receptors, whereas target cells that inhibited the response expressed alleles that were recognized by NK2 receptors (9, 11). This result implies that the short-tailed NK1 receptor present on the cell surface of TANK-9 facilitates the enhancement of the T cell proliferative response, whereas the long-tailed NK2 receptor mediates inhibition of such a response. The decrease in proliferation of TANK-9 in response to TSST1 at 50 ng/ml compared to the proliferative response at 0.5 ng/ml is likely to be due to the induction of anergy (20).

Proliferation of TANK-1 in response to SEB presentation was enhanced by target cells transfected with either HLA-Cw4 or HLA-Cw7 (Fig. 1B). Two other TANK cell clones behaved similarly (19). Because the only NK receptor expressed on TANK-1 was the activating receptor, clone 39, it can be inferred that clone 39 facilitates an enhanced T cell response in the presence of the class I MHC alleles HLA-Cw4 and HLA-Cw7. These two class I MHC proteins belong to the two distinct groups of HLA-C proteins that are recognized by either NK1 or NK2 receptors. The molecular basis for the recognition of both HLA-Cw4 and -Cw7 by clone 39 is unknown, but the sequence of clone 39 is chimeric between NKAR1 and NKAR2.

The costimulatory effect of HLA-Cw4 or -Cw7 on the proliferative response of TANK-1 is greatest at lower concentrations of superantigen, causing a 300 to 900% increase in stimulation compared with untransfected target cells (Figs. 1, B and C, and 2A). Thus, the function of costimulatory NK receptors on T cells in vivo might be to allow a subset of T cells to be particularly responsive to small quantities of antigen. In addition, the enhanced T cell response of TANK-1 costimulated by HLA-Cw4 or -Cw7 peaks on the fourth day after initial stimulation and is maintained for a further 5 days (Fig. 1C). This suggests that the function of the costimulatory NK receptor on T cells might also be to sustain the immune response for an extended period. Possible reasons for the sharp decrease in the proliferation of TANK-1 at about 5 days after initial stimulation (Fig. 1C) include the up- or down-regulation of a particular protein on either the T cells or target cells, depletion of cytokines, or an exhaustion of nutrients within the media.

To determine directly whether the activating NK receptor on T cells facilitates the enhancement in proliferation, we used the monoclonal antibodies (mAbs) HP3E4 and GL183 to NK1 and NK2, respectively, in assays of TANK-9 proliferation in the presence of TSST1 at 0.05 ng/ml. Monoclonal antibody HP3E4 blocked the enhancement of TANK-9 proliferation mediated by 721.221 cells transfected with HLA-Cw4 or -Cw6, and GL183 blocked the inhibition of TANK-9 mediated by HLA-Cw3- or HLA-Cw7-expressing target cells (Fig. 2A). Analogous data were obtained with superantigen at 0.5 ng/ml. These data confirm that the NK1 and NK2 receptors on TANK-9 do indeed mediate the alterations in the proliferative response to TSST1.

Because none of the available mAbs binds specifically to clone 39, the role of this molecule in enhancing the proliferative response of TANK-1 could not be directly assessed by antibody blocking. However, the increase in proliferation of TANK-1 in the presence of SEB at 0.5 ng/ml, mediated by 721.221 cells transfected with HLA-Cw4 or -Cw7, was blocked by mAb PA2.6 to class I MHC proteins (Fig. 2B). Analogous data were obtained with SEB at 0.05 ng/ml. These data confirm that the increased proliferation of TANK-1 is mediated by class I MHC receptors.

The nonclassical class I MHC protein, HLA-G, has recently been shown to inhibit both NK1 and NK2 clones (21). Transfection of HLA-G into 721.221 cells enhanced the proliferation of TANK-1 in the presence of SEB, compared to untransfected cells (Fig. 3A). In contrast, the carcinoma cell line JEG-3, which expresses HLA-G but is deficient in class II MHC expression, could not stimulate TANK-1 in the presence of SEB (Fig. 3A). This result implies that superantigen cannot be presented by the class I MHC protein to cause significant proliferation of TANK-1, and that stimulation must be mediated primarily by superantigen associated with class II MHC molecules. In support of this conclusion, the proliferative response of TANK-1 to SEB was diminished in the presence of mAb LB3.1 to class II MHC (19).

SEB-coated 721.221 cells that had been transfected with HLA-G but that had lost most cell-surface expression of the protein after an extended period in culture (HLA-G[2]) caused the same extent of TANK cell proliferation as untransfected cells (Fig. 3A). Thus, it is the expression of HLA-G that facilitates the enhanced proliferation of TANK-1, rather than any artefactual alteration in the phenotype of transfected 721.221 cells. Although HLA-G can facilitate action through both NK1 and NK2 receptors, HLA-G-transfected 721.221 cells inhibited the proliferative response of TANK-9 to TSST1 (Fig. 3B) (22).

Thus, both NK activating receptors and NK inhibitory receptors are present on T cell clones, where they can alter the proliferative response of T cells stimulated by superantigen. NK inhibitory receptors can also inhibit the proliferative response of T cells stimulated by peptide antigen (19) as well as by superantigen (10), and it is reasonable to infer that NK activating receptors costimulate T cells activated by peptide antigens. Therefore, distinct combinations of activating and inhibitory receptors on each T cell could adjust the response of the human immune system to specific antigens.

Because the presence of an activating NK receptor on a T cell potentially facilitates a response to small amounts of antigen, it is possible that TANK cells are important in initiating immune responses. In particular, because, by flow cytometry, NK receptors occur primarily on T cells of a memory phenotype (19, 23), costimulatory NK receptors may especially facilitate the rapid induction of secondary T cell-mediated immune responses. In addition, the expression and function of activating class I MHC receptors may allow activation of a T cell whose TCR may interact weakly with self peptide. Thus, expression of such receptors could also be significant in triggering the onset of autoimmune disease.

REFERENCES AND NOTES

1. V. A. Boussiotis, J. G. Gribben, G. J. Freeman, L. M. Nadler, Curr. Opin. Immunol. 6, 797 (1994) [Medline].
2. E. Ciccone et al., Proc. Natl. Acad. Sci. U.S.A. 87, 9794 (1990) [Medline]; E. Ciccone et al., J. Exp. Med. 175, 709 (1992) [Medline]; V. Litwin et al., ibid. 178, 1321 (1993) [Medline]; M. Colonna and J. Samaridis, Science 268, 405 (1995) [Medline]; A. D'Andrea et al., J. Immunol. 155, 2306 (1995) ; J. E. Gumperz and P. Parham, Nature 378, 245 (1995) [Medline]; A. Moretta et al., Annu. Rev. Immunol. 14, 619 (1996) [Medline].
3. N. Wagtmann et al., Immunity 2, 439 (1995) [Medline].
4. M. Colonna, E. G. Brooks, M. Falco, G. B. Ferrara, J. L. Strominger, Science 260, 1121 (1993) [Medline]; M. Colonna, G. Borsellino, M. Falco, G. B. Ferrara, J. L. Strominger, Proc. Natl. Acad. Sci. U.S.A. 90, 12000 (1993) [Medline].
5. O. Mandelboim et al., J. Exp. Med. 184, 913 (1996) .
6. M. Cella et al., ibid. 180, 1235 (1994); J. E. Gumperz et al., ibid. 181, 1133 (1995) [Medline].
7. L. Olcese et al., J. Immunol. 156, 4531 (1996) [Medline]; D. N. Burshtyn et al., Immunity 4, 77 (1996) [Medline]; A. M. Fry et al., J. Exp. Med. 184, 295 (1996) [Medline]; K. S. Campbell et al., ibid., p. 93.
8. H. Spits, L. L. Lanier, J. H. Phillips, Blood 85, 10 (1995) .
9. M. C. Mingari et al., Int. Immunol. 7, 697 (1995) [Medline]; C. S. Falk, A. Steinle, D. J. Schendel, J. Exp. Med. 182, 1005 (1995) [Medline]; S. Ferrini et al., Eur. J. Immunol. 24, 2294 (1994) [Medline]; H. Nakajima, H. Tomiyama, M. Takiguchi, J. Immunol. 155, 4139 (1995) [Medline].
10. J. H. Phillips, J. E. Gumperz, P. Parham, L. L. Lanier, Science 268, 403 (1995) [Medline].
11. A. Moretta et al., J. Exp. Med. 182, 875 (1995) [Medline]; R. Biassoni et al., ibid. 184, 645 (1996) .
12. TANK cell clones were prepared in the same manner as the NK clones used in (5), except that the CD4⁺ T cells were not depleted.
13. Cells were typed by flow cytometry for the presence of CD3, CD4, CD8, CD16, CD56, TCR, and TCR (with antibodies from Becton-Dickinson), the type of TCR (with antibodies from Coulter Immunology), CD94 (with mAb HP3B1), and NK receptors [with mAbs specific for NK1 receptor (HP3E4), NK2 receptor (GL183, Coulter Immunology), and NK3 receptor (DX9)].
14. Oligonucleotides complementary to nonpolymorphic regions of the known NK receptor sequences were chosen as GATGGTACATGTCATAGGAGCTCC (at the 3 end) and GAAAACCTTCCCTCCTGGCCC (at the 5 end). The resultant PCR product that was amplified from cDNA derived from the T cell was cloned into pCRII (Invitrogen). The sequence of the insert was determined by automated sequencing at the Molecular Biology Core Facilities at the Dana-Farber Cancer Institute, Boston, MA. This sequence was then aligned against sequences currently held in the GenBank database with the program BLAST. The insert obtained from the cDNA of clone TANK-1 was 100% identical to that of the NK receptor, clone 39.
15. Reverse transcriptase-PCR typing of NK receptors was performed with oligonucleotides complementary to polymorphic regions of the extracellular portions of the known NK receptors and with oligonucleotides complementary to the long and short cytoplasmic tail sequences of NK receptors (H. T. Reyburn et al., in preparation).
16. TANK cells are not exclusively CD4⁺ TCR⁺ because TCR⁺ and CD8⁺ TCR⁺ TANK cells have also been obtained (O. Mandelboim et al., in preparation).
17. Y. Shimizu and R. DeMars, J. Immunol. 142, 3320 (1989).
18. Proliferation assays were performed as follows: The target cells were irradiated on a cesium source for 30 min (30 Gy). Thereafter, 50,000 T cells, 25,000 target cells, various amounts of superantigen, and any other appropriate reagents were mixed in a total volume of 200 µl of RPMI-10% fetal calf serum in each well of a 96-well plate. After incubation at 37°C and 5% CO₂ for 2 days (or the time indicated), 1 µCi of [³H]thymidine was added to each well, and the cells were further incubated at 37°C overnight. The cells were then harvested (Harvester 96 Mach III M, Tomtec) and counted on a liquid scintillation counter (1450 Microbeta Plus, Wallac). In analysis of the counts per minute (cpm) from each well, the background cpm from a well in which identical reagents and target cells were placed in the absence of any T cells was subtracted.
19. O. Mandelboim, D. M. Davis, H. T. Reyburn, E. G. Sheu, unpublished data.
20. J. F. A. P. Miller and G. Morahan, Annu. Rev. Immunol. 10, 51 (1992) .
21. L. Pazmany et al., Science 274, 792 (1996) [Medline].
22. This may be because HLA-G has a greater binding affinity for NK2 than for NK1 receptors or because HLA-G can interact with other inhibitory receptors on lymphocytes yet to be determined. Alternatively, the signals for inhibition of the proliferative response could be dominant over activating signals when both are triggered in this clone.
23. A. D'Andrea, J. Exp. Med. 194, 784 (1996) .
24. We thank M. Lopéz-Botet (mAbs HP3B1 and HP3E4) and L. Lanier (mAb DX9). Supported by EMBO and the Fullbright Commission (O.M.), The Wellcome Trust (H.T.R.), the Arthritis Foundation (L.P.), and NIH grant CA 47554.

3 July 1996; accepted 12 November 1996