The nonpolymorphic soluble HLA-G1 (sHLA-G1) isoform has been reported to be secreted by trophoblast cells at the materno-fetal interface, suggesting that it may act as immunomodulator during pregnancy. In this paper, we report that affinity-purified β2-microglobulin-associated sHLA-G1 triggered apoptosis in activated, but not resting CD8+ peripheral blood cells. We demonstrate by Western blotting that sHLA-G1 enhanced CD95 ligand expression in activated CD8+ cells. Cytotoxicity was inhibited by preincubation of the cells with a CD95 antagonist mAb (ZB4) or a soluble recombinant CD95-Fc, indicating that apoptosis is mediated through the CD95/CD95 ligand pathway. Finally, we show that such sHLA-G1-induced apoptosis depends on the interaction with CD8 molecules, with cell death being blocked by various CD8 mAbs.

During pregnancy, the fetus can be considered as a semiallogenic allograft. However, the fetus is tolerated by the mother, suggesting an induction of tolerance against paternally derived Ags. The mechanisms that account for such a tolerance are still not clearly understood. The fetus is never in direct contact with maternal tissues; therefore, many of the tolerance mechanisms may occur only at the materno-fetal interface in the placenta. One of these interfaces is the site of implantation where fetally derived extravillous cytotrophoblast cells invade the decidua basalis and encounter maternal immune cells such as NK cells, T cells, and macrophages (1). The lack of expression of HLA-A and -B class Ia molecules on these trophoblast cells and, in contrast, the expression of HLA-G class Ib may contribute to the proper outcome of pregnancy (2). One of the major structural characteristics of the HLA-G gene is its potentiality to be transcribed in alternatively spliced isoforms, including membrane-bound and soluble proteins (3, 4). The HLA-G1 soluble form (sHLA-G1)3 is encoded by a spliced transcript containing part of intron 4, which generates a stop codon that results in the deletion of the transmembrane and cytoplasmic domains, yielding a 37-kDa soluble protein that associates to β2-microglobulin (β2m) (4). Such intron-retaining soluble isoform appears to be a unique structural feature of HLA-G that is not observed with other soluble HLA class I molecules. sHLA-G1 was shown to be secreted during pregnancy (5, 6, 7). Some placental cells, including extravillous cytotrophoblast, macrophages in chorionic villi, and endothelial cells lining fetal placental blood vessels, have been identified as sHLA-G secretors (8, 9, 10, 11).

In this report, we show that purified sHLA-G1 triggers in vitro apoptosis of activated CD8+ cells by interacting with the CD8 molecules. We found that sHLA-G1 enhanced CD95 ligand (CD95-L) expression in activated CD8+ cells and that apoptosis was the consequence of a CD95/CD95-L interaction.

Jurkat cell line was provided by Dr. B. Rubin (Centre National de la Recherche Scientifique, Toulouse, France). PG CD8 is a HLA-B*2705-restricted CD8+ synovial cytotoxic T cell line (12). These cells were cultured in IL-2 (150 IU/ml) and activated by PHA (1 μg/ml) every 2 wk and cultured in IL-2 (150 IU/ml) the rest of the time. The following mAbs were used: 7C11 and ZB4 (Immunotech, Marseille, France); CD95 agonist and antagonist, respectively; BL4 CD4 (gift of Prof. J. P. Revillard, Institut National de la Santé et de la Recherche Médicale U503, Lyon, France); W6/32, HLA class I heavy chains associated with β2m; IP48 and CD8×8; CD8α-chain, produced locally; and CD95-L (clone 33; Transduction Laboratories, Lexington, KY). Soluble recombinant CD95-Fc was obtained by fusion of the extracellular domain of CD95 to the Fc fragment of human IgG1, transfection in Chinese hamster ovary cells, and recovery from the cell culture supernatant.

β2m-associated sHLA-G1 and sHLA-B7 were purified from culture supernatants of JAR cell line transfected with sHLA-G1 encoding cDNA (13) or of C1R cell line transfected with cDNA encoding sHLA-B7 (14), respectively, using immunoaffinity column, as has been described (13). Briefly, N-hydroxysuccimide-activated Sepharose 4 fast flow (Pharmacia Biotech, Uppsala, Sweden) was incubated with W6/32 HLA class I mAb (5 mg protein/ml Sepharose) overnight at room temperature. Before use, the column was blocked with 100 mM ethanolamine (pH 9.0) overnight. After washes with PBS, 100–200 ml culture supernatants were applied onto the columns overnight at 4°C. After washes with PBS, bound Ag was eluted with 0.1 M glycine buffer (pH 11.5) and neutralized with 1 M Tris buffer (pH 7.5). Ag purity was confirmed by SDS-PAGE and Western blotting, as previously described (13).

PBMC from healthy donors were isolated by centrifugation of heparinized blood on a layer of Ficoll-Hypaque (Lymphoprep; Pharmacia Biotech). Cells were washed and resuspended in RPMI 1640 with Glutamax (Life Technologies, Cergy-Pontoise, France), 10% FCS, sodium pyruvate (1 mM), and antibiotics. Cultures were maintained for 3 days in the presence or not of PHA (5 μg/ml; Sigma, St. Louis, MO). After PHA activation, viable cells were depleted of CD4+ cells by adherence onto plastic flasks coated with CD4 mAb (BL4). Such PBMC suspensions contained an average of 67% (±15%) of CD8+ cells and less than 5% of CD4+ cells as measured by flow cytometry CD8-FITC or CD4-PE mAb binding, respectively. Viable CD8+ cells (106/ml) were incubated in 96-well microplates with sHLA molecules or mAbs at indicated concentrations. Cell death was then evaluated by fluorescence microscopy measurement of nuclear condensation and fragmentation after Hoechst 33342 staining (Sigma) (15). Results were expressed as percentage of specific apoptosis, according to the following formula: % specific apoptosis = [(% of apoptotic treated cells − % of apoptotic control cells) × 100]/(100 − % of apoptotic control cells). Apoptosis was also evaluated by the detection of phosphatidylserine expression by flow cytometry after addition of FITC-labeled human Annexin V (Bender MedSystems, Vienna, Austria) (16).

The CD95-L-dependent cytotoxicity was measured by [3H]DNA released from Jurkat cells induced by CD95-L-producing cells, as previously described (17, 18). Jurkat cells were pulsed for 12 h with 20 μCi/ml of [3H]TdR (Amersham, les Ulis, France). After washes, [3H]-labeled Jurkat cells (0.2 × 106 cells/ml) were incubated with resting or PHA-activated CD8+ cells previously treated for 15 h with sHLA-G1 (1 μg/ml) or sHLA-B7 (1 μg/ml) at a ratio of one Jurkat cell for three CD8+ cells, with or without the antagonist CD95 mAb ZB4 (2 μg/ml). After 12 h of culture, [3H]DNA release induced by apoptosis of Jurkat cells was measured using a Packard direct beta counter (Packard, Meriden, CT). Results were expressed as percentage of cytotoxicity, according to the following formula: % cytotoxicity = [(cpm spontaneous − cpm sample) × 100]/cpm spontaneous. Cytotoxicity inhibited by ZB4 was considered as specific to CD95-L-dependent apoptosis.

Cells were stained with CD8-FITC or CD4-PE (Immunotech) or CD25-FITC (PharmMingen, San Diego, CA) mAbs for 30 min at 4°C, washed, and analyzed on a Coulter (Margency, France) Epics Elite flow cytometer gated to exclude nonviable cells.

Western blotting was performed as described (13). Briefly, cell lysates were separated onto a 12% SDS-PAGE and transferred on nitrocellulose membranes. Membranes were blocked with PBS/0.1% Tween 20/5% nonfat dry milk and incubated with CD95-L mAb and then with peroxidase-labeled rabbit anti-mouse IgG and enhanced chemiluminescence Western blotting detection reagent (Amersham).

To investigate the effect of sHLA-G1 on CD8+ cells, PBMC were activated for 3 days by PHA, depleted of CD4+ cells, and incubated with sHLA-G1, sHLA-B7, or an agonist CD95 mAb (7C11), used as positive control. We first analyzed induction of apoptosis by Hoechst staining. sHLA-G1 reproducibly induced specific apo- ptosis of PHA-activated CD8+ cells (Fig. 1,A). By comparison, sHLA-B7 induced only a slight or no apoptosis, whereas CD95 mAb 7C11 triggered a stronger specific killing. Twenty eight ± 10% of CD8+ cells were estimated to be activated by PHA, as measured by expression of the IL-2R α-chain CD25 (data not shown). This suggested that about 50% of activated CD8+ cells were killed by sHLA-G1, compared with 100% of killing by the CD95 mAb. Almost no apoptotic effect of sHLA-G1 or sHLA-B7 was detected by this technique on resting CD8+ cells (Fig. 1,A). Apoptosis was then evaluated by Annexin V staining (Fig. 1 B). When added to PHA-activated CD8+ cells, sHLA-G1, like CD95 mAb, triggered externalization of phosphatidylserine measured by the binding of Annexin V as a typical feature of apoptosis (11.8 and 17.7% specific apoptosis, respectively), whereas both sHLA-G1 and CD95 mAb had little effect on resting CD8+ cells (5.5 and 8.1% of specific apoptosis, respectively). As control, sHLA-B7 had no specific effect on resting and activated CD8+ cells (no specific apoptosis).

FIGURE 1.

sHLA-G1 induces apoptosis of activated CD8+ cells. A, PBMC were activated for 3 days with PHA (5 μg/ml). After removal of dead cells and depletion of CD4+ cells, PHA-activated CD8+ cells (106 cells/ml) were incubated for 20 h with sHLA-G1s (1 μg/ml), sHLA-B7 (1 μg/ml), or CD95 agonist mAb 7C11 (1 μg/ml). Control resting CD8+ cells were similarly incubated with sHLA-G1, sHLA-B7, or CD95 mAb. Apoptosis was determined by fluorescence microscopy measurement of nuclear condensation and fragmentation after Hoechst 33342 staining. Results are expressed as percentage of specific apoptosis, as described in Materials and Methods. Spontaneous apoptosis in medium alone did not exceed 15%. Values are the mean ± SD of ten experiments. B, Flow cytometry analysis after Annexin V-FITC staining. PHA-activated CD8+ cells or control resting CD8+ cells (106 cells/ml) were incubated with culture medium or with sHLA-G1s (1 μg/ml), sHLA-B7 (1 μg/ml), or the CD95 agonist mAb 7C11 (1 μg/ml). After 24 h, cells were stained with Annexin V-FITC and analyzed by flow cytometry. Percentages of Annexin-V-positive cells are indicated in each panel. Spontaneous apoptosis in medium alone did not exceed 20%. These results are representative of six independent experiments.

FIGURE 1.

sHLA-G1 induces apoptosis of activated CD8+ cells. A, PBMC were activated for 3 days with PHA (5 μg/ml). After removal of dead cells and depletion of CD4+ cells, PHA-activated CD8+ cells (106 cells/ml) were incubated for 20 h with sHLA-G1s (1 μg/ml), sHLA-B7 (1 μg/ml), or CD95 agonist mAb 7C11 (1 μg/ml). Control resting CD8+ cells were similarly incubated with sHLA-G1, sHLA-B7, or CD95 mAb. Apoptosis was determined by fluorescence microscopy measurement of nuclear condensation and fragmentation after Hoechst 33342 staining. Results are expressed as percentage of specific apoptosis, as described in Materials and Methods. Spontaneous apoptosis in medium alone did not exceed 15%. Values are the mean ± SD of ten experiments. B, Flow cytometry analysis after Annexin V-FITC staining. PHA-activated CD8+ cells or control resting CD8+ cells (106 cells/ml) were incubated with culture medium or with sHLA-G1s (1 μg/ml), sHLA-B7 (1 μg/ml), or the CD95 agonist mAb 7C11 (1 μg/ml). After 24 h, cells were stained with Annexin V-FITC and analyzed by flow cytometry. Percentages of Annexin-V-positive cells are indicated in each panel. Spontaneous apoptosis in medium alone did not exceed 20%. These results are representative of six independent experiments.

Close modal

sHLA-G1 induced specific apoptosis of PHA-activated CD8+ cells at a concentration as low as 0.25 μg/ml with a maximum effect at 1 μg/ml (Fig. 2,A). Maximum sHLA-G1-induced apoptosis was observed after 24 h of incubation with a kinetics comparable to the CD95 mAb-induced apoptosis (Fig. 2 B).

FIGURE 2.

Dose response and kinetics of sHLA-G1-induced apoptosis. A, PHA-activated CD8+ cells were incubated for 20 h with sHLA-G1s at indicated concentrations. Apoptosis was determined by fluorescence microscopy after Hoechst 33342 staining, and results are expressed as specific apoptosis as in Fig. 1 A. Values are the mean ± SD of five separate experiments. B, PHA-activated CD8+ cells were incubated either with sHLA-G1s (1 μg/ml; •) or with the 7C11 agonist CD95 mAb (1 μg/ml; □). At indicated times, apoptosis was similarly determined. Values are expressed as the mean ± SD of three experiments.

FIGURE 2.

Dose response and kinetics of sHLA-G1-induced apoptosis. A, PHA-activated CD8+ cells were incubated for 20 h with sHLA-G1s at indicated concentrations. Apoptosis was determined by fluorescence microscopy after Hoechst 33342 staining, and results are expressed as specific apoptosis as in Fig. 1 A. Values are the mean ± SD of five separate experiments. B, PHA-activated CD8+ cells were incubated either with sHLA-G1s (1 μg/ml; •) or with the 7C11 agonist CD95 mAb (1 μg/ml; □). At indicated times, apoptosis was similarly determined. Values are expressed as the mean ± SD of three experiments.

Close modal

Knowing that activated T cells express CD95 receptors (19) and are sensitive to CD95-L-mediated apoptosis (20), we examined the contribution of CD95/CD95-L interaction in sHLA-G1-induced apoptosis. We first investigated by Western blotting whether sHLA-G1 increased expression of CD95-L in PHA-activated CD8+ cells (Fig. 3 A). Using a CD95-L-specific mAb, one specific band in the range of 37 kDa was clearly detectable in PHA-activated CD8+ cells incubated with sHLA-G1 for 24 h (lanes 1 and 2) or 12 h (lanes 3 and 4). Depending on cell concentration, no signal or a faint band was revealed in control untreated PHA-activated CD8+ cells (lanes 5 and 6, respectively), whereas no signal was detectable in resting CD8+ cells (lanes 7 and 8). These results demonstrated that sHLA-G1 enhanced CD95-L expression in PHA-activated CD8+ cells.

FIGURE 3.

sHLA-G1-induced apoptosis of activated CD8+ cells is dependent on CD95/CD95-L interaction. A, sHLA-G1s induces expression of CD95-L in PHA-activated CD8+ cells: Western blotting analysis. The following cell lysates were separated onto a 12% SDS-PAGE and probed with the CD95-L-specific mAb clone 33 at a dilution of 1:1000; PHA-activated CD8+ cells incubated with sHLA-G1s (1 μg/ml) for 24 h (PHA + sHLA-G1/24h) or 12 h (PHA + sHLA-G1/12h) and PHA-activated CD8+ cells incubated for 12 h in medium alone (PHA+) or in resting CD8+ cells (PHA−). Lanes 1, 3, 5, and 7: 2 × 105 cells; lanes 2, 4, 6, and 8: 4 × 105 cells. Red ponceau staining of membranes assessed that equivalent amount of proteins levels was loaded in each slot of the same series. B, Functional CD95-L interaction was assessed by Jurkat cell DNA fragmentation. PHA-activated CD8+ cells were treated for 15 h with sHLA-G1 (1 μg/ml) or sHLA-B7 (1 μg/ml) and then were incubated with [3H]-labeled Jurkat cells (0.2 × 106 cells/ml) at a ratio of one Jurkat cell for three CD8+ cells, with (□) or without (▪) the antagonist CD95 mAb ZB4 (2 μg/ml). After 12 h of culture, [3H]DNA release induced by apoptosis of Jurkat cells was measured. Results are expressed as percentage of specific DNA fragmentation (mean ± SD of three separate experiments). C, Inhibition of sHLA-G1-induced apoptosis in PHA-acivated CD8+ cells by the antagonist CD95 mAb ZB4 and the CD95-Fc molecule (which interacts with CD95-L). PHA-activated CD8+ cells were incubated for 1 h with medium alone (□), with the antagonist CD95 mAb ZB4 (2 μg/ml; ▪), or with CD95-Fc (▦). Then, sHLA-G1s (1 μg/ml), sHLA-B7 (1 μg/ml), or the agonist CD95 mAb 7C11 (1 μg/ml) was added for 20 h, and apoptosis was determined by fluorescence microscopy after Hoechst 33342 staining. Results are expressed as specific apoptosis. Spontaneous apoptosis did not exceed 15% with all the agents tested. Values are the mean ± SD of seven independent experiments.

FIGURE 3.

sHLA-G1-induced apoptosis of activated CD8+ cells is dependent on CD95/CD95-L interaction. A, sHLA-G1s induces expression of CD95-L in PHA-activated CD8+ cells: Western blotting analysis. The following cell lysates were separated onto a 12% SDS-PAGE and probed with the CD95-L-specific mAb clone 33 at a dilution of 1:1000; PHA-activated CD8+ cells incubated with sHLA-G1s (1 μg/ml) for 24 h (PHA + sHLA-G1/24h) or 12 h (PHA + sHLA-G1/12h) and PHA-activated CD8+ cells incubated for 12 h in medium alone (PHA+) or in resting CD8+ cells (PHA−). Lanes 1, 3, 5, and 7: 2 × 105 cells; lanes 2, 4, 6, and 8: 4 × 105 cells. Red ponceau staining of membranes assessed that equivalent amount of proteins levels was loaded in each slot of the same series. B, Functional CD95-L interaction was assessed by Jurkat cell DNA fragmentation. PHA-activated CD8+ cells were treated for 15 h with sHLA-G1 (1 μg/ml) or sHLA-B7 (1 μg/ml) and then were incubated with [3H]-labeled Jurkat cells (0.2 × 106 cells/ml) at a ratio of one Jurkat cell for three CD8+ cells, with (□) or without (▪) the antagonist CD95 mAb ZB4 (2 μg/ml). After 12 h of culture, [3H]DNA release induced by apoptosis of Jurkat cells was measured. Results are expressed as percentage of specific DNA fragmentation (mean ± SD of three separate experiments). C, Inhibition of sHLA-G1-induced apoptosis in PHA-acivated CD8+ cells by the antagonist CD95 mAb ZB4 and the CD95-Fc molecule (which interacts with CD95-L). PHA-activated CD8+ cells were incubated for 1 h with medium alone (□), with the antagonist CD95 mAb ZB4 (2 μg/ml; ▪), or with CD95-Fc (▦). Then, sHLA-G1s (1 μg/ml), sHLA-B7 (1 μg/ml), or the agonist CD95 mAb 7C11 (1 μg/ml) was added for 20 h, and apoptosis was determined by fluorescence microscopy after Hoechst 33342 staining. Results are expressed as specific apoptosis. Spontaneous apoptosis did not exceed 15% with all the agents tested. Values are the mean ± SD of seven independent experiments.

Close modal

To investigate the functionality of such a CD95-L expression in PHA-activated CD8+ cells treated with sHLA-G1, we then used the CD95+ Jurkat cells, which are susceptible to CD95-L-mediated apoptosis (17, 18). PHA-activated CD8+ cells were incubated either with sHLA-G1 or sHLA-B7 for 15 h and then cocultured for a further 12 h with [3H]thymidine-labeled Jurkat target cells in the presence or absence of the antagonist CD95 mAb ZB4. PHA-activated CD8+ cells treated with sHLA-G1 induced [3H]thymidine release from Jurkat cell (Fig. 3,B). Preincubation with the antagonist CD95 mAb ZB4 completely abolished this effect, suggesting that sHLA-G1 induced expression of functional CD95-L. In contrast, sHLA-B7 did not induce specific [3H]thymidine release (Fig. 3 B).

Consistently with the previous data, we next demonstrated that a preincubation with the same antagonist CD95 mAb ZB4 prevented not only 7C11- but also sHLA-G1s-induced apoptosis of PHA-activated CD8+ T cells (Fig. 3 C), whereas no lysis was observed in nonactivated cells (data not shown). Similar results were obtained when CD95-Fc, which blocked interaction between CD95 and CD95-L, was used instead of ZB4. These results further confirmed that sHLA-G1-induced apoptosis is mediated by the CD95/CD95-L pathway.

sHLA-G1 may interact with CD8+ T cells through either HLA-G-restricted TCR (21) or the CD8 molecules (22, 23). We investigated the contribution of CD8. PHA-activated CD8+ cells were preincubated for 1 h with CD8 mAbs before the addition of sHLA-G1, sHLA-B7, or agonist CD95 mAb 7C11. Both CD8 mAbs used in this study strongly inhibited sHLA-G1-induced apoptosis, whereas they had no significant effect on 7C11- or sHLA-B7-induced apoptosis (Fig. 4). The IgG control had no effect. These data indicated that sHLA-G1-induced apoptosis was CD8-dependent. This was further confirmed by sHLA-G1 induction of apoptosis of an alloreative CD8+ T cell line exhibiting a HLA-B27-restricted TCR (Ref. 12 and data not shown). These latter results show that sHLA-G1-induced apoptosis was not associated with a single way of activation (i.e., PHA). In addition, they exclude the possibility that sHLA-G1 did interact with the TCR to induce such apoptosis. It was recently shown that CD4 cross-linking by HIV gp 120 triggered CD4+ T cell apoptosis (24). Our data confirm that apoptosis could be induced through accessory molecules.

FIGURE 4.

CD8 mAbs inhibit sHLA-G1-induced apoptosis of activated CD8+ cells. PHA-activated CD8+ cells were incubated for 1 h with medium alone, with CD8×8 (ascite 1/1000) or IP48 (ascite 1/200) CD8 mAbs, or with IgG control (10 μg/ml). Then, sHLA-G1s (1 μg/ml), sHLA-B7 (1 μg/ml), or the agonist CD95 mAb 7C11 (1 μg/ml) was added, and after 20 h, the percentage of apoptotic cells was determined by fluorescence microscopy after Hoechst 33342 staining. Results are expressed as specific apoptosis as in Fig. 1 A. Spontaneous apoptosis did not exceed 15% with any of the agents tested. Values are the mean ± SD in three independent experiments.

FIGURE 4.

CD8 mAbs inhibit sHLA-G1-induced apoptosis of activated CD8+ cells. PHA-activated CD8+ cells were incubated for 1 h with medium alone, with CD8×8 (ascite 1/1000) or IP48 (ascite 1/200) CD8 mAbs, or with IgG control (10 μg/ml). Then, sHLA-G1s (1 μg/ml), sHLA-B7 (1 μg/ml), or the agonist CD95 mAb 7C11 (1 μg/ml) was added, and after 20 h, the percentage of apoptotic cells was determined by fluorescence microscopy after Hoechst 33342 staining. Results are expressed as specific apoptosis as in Fig. 1 A. Spontaneous apoptosis did not exceed 15% with any of the agents tested. Values are the mean ± SD in three independent experiments.

Close modal

Although unlikely, our results cannot completely exclude that CD4+ T cells were also susceptible to sHLA-G1-mediated apoptosis. Because an interaction between sHLA-G1 and CD4 is unusual for a class I molecule, a binding to other unknown receptors present on CD4+ cells therefore would be implied. It has been demonstrated that HLA-G binds to several killer cell Ig-like receptors, including Ig-like transcripts 2 and 4, expressed in particular on myelomonocytic cells (reviewed in Ref. 1). Therefore, we cannot rule out that sHLA-G1 has bound to such receptors expressed on some of these cells present in our cultures, modulating one or several of their Ag-presenting functions and thereby having indirect effects on CD8+ cells.

sHLA class Ia molecules have been reported to induce apoptosis in alloreactive CTL (25). The mechanisms underlying this phenomenon appear to be distinct from what we observed with sHLA-G1, because sHLA class Ia triggered cell death by interaction with TCR, whereas sHLA-G1 induced apoptosis by interaction with CD8. Interaction of HLA-G with CD8 was already reported to involve a conserved negatively charged loop localized in the α3 domain of the HLA-G heavy chain (22). Two main structural differences between sHLA-G1 and sHLA-B7 could explain the differential effect of these two molecules on apoptosis of CD8+. First, at position 228 within α3 binding domain to CD8, HLA-G has a valine instead of the conserved threonine residue found in all other HLA class Ia (26). Second, sHLA-G1 has a translated part of intron 4 (4), which could modify the general conformation of the CD8 binding sites. Based on these structural differences, one can predict that the CD8 binding affinity would differ between sHLA-G1 and sHLA-B7. Experiments using HLA-G/HLA-B7 chimera molecules remain to be done to investigate these possibilities. A very recent report mentioned that sHLA class I molecules, isolated from human sera, induced apoptosis in activated CD8+ but not in CD4+ T cells (27). However, these results did not discriminate between classical and nonclassical soluble HLA class I products. Therefore, it cannot be excluded that part of these soluble MHC class I molecules were HLA-G, possibly secreted by activated macrophages (10).

Because sHLA-G1 is mainly expressed in the placenta (1), we hypothesize that sHLA-G1 may, among other functions (1, 2), contribute to the elimination of CD8+ alloreactive maternal immune T cells in vivo at the materno-fetal interface through the CD95/CD95-L pathway. Several observations favor such a hypothesis. First, sHLA-G was shown to be secreted by invading extravillous cytotrophoblasts present in the decidua (10, 11). Second, a level of sHLA-G1 significantly higher than that in early abortion recently has been reported in early intact pregnancy (28). Third, few CD8+ T cells have been found in the human decidua by immunohistochemistry or flow cytometry analysis (29), whereas apoptotic nuclei mainly arising from CD45+ leukocytes recently have been detected at the same materno-fetal interface (30). These latter two observations suggest that CD8+ T cells could have been eliminated from the decidua. In addition to T cells, it has been found that CD8 was also present on the surface of some decidual CD56+ NK cells (29). Furthermore, some reports concluded that decidual NK cells had the potential to kill trophoblast when activated by IL-2 or other cytokines (29). We speculate that such potentially harmful maternal CD8+ activated NK cells could also be bound by sHLA-G and induced to apoptosis. Such induction of apoptosis by sHLA-G1 in maternal decidual lymphocytes should contribute to deter trafficking of activated maternal CD95-expressing CD8+ T cells among the decidua, the trophoblast, and the fetus and therefore could be an important mechanism implicated in the immune tolerance of the fetal allograft. Due to its very low polymorphism (2), sHLA-G1 recombinant molecule may be especially relevant to organ transplantation by regulating the elimination of alloimmune CD8+ T cells.

We thank Drs. B. Rubin and F. C. Grumet for cells and Prof. J. P. Revillard for mAbs. We thank N. Bonnefoy-Berard, L. Genestier, and J. P. Revillard for critical review of the manuscript.

1

This work was supported by Institut National de la Santé et de la Recherche Médicale, Association pour la Recherche sur le Cancer (9489), Etablissement Français des Greffes, Agence Nationale pour la Recherche sur le Sida, and Université Paul Sabatier.

3

Abbreviations used in this paper: s, soluble; β2m, β2-microglobulin; CD95-L, CD95 ligand.

1
Le Bouteiller, P., C. Solier, J. Proll, M. Aguerre-Girr, S. Fournel, F. Lenfant.
1999
. Placental HLA-G protein expression in vivo: where and what for?.
Hum. Reprod. Update
5
:
223
2
Le Bouteiller, P., A. Blaschitz.
1999
. The functionality of HLA-G is emerging.
Immunol. Rev.
167
:
233
3
Ishitani, A., D. E. Geraghty.
1992
. Alternative splicing of HLA-G transcripts yields proteins with primary structures resembling both class I and class II antigens.
Proc. Natl. Acad. Sci. USA
89
:
3947
4
Fujii, T., A. Ishitani, D. E. Geraghty.
1994
. A soluble form of the HLA-G antigen is encoded by a messenger ribonucleic acid containing intron 4.
J. Immunol.
153
:
5516
5
Hiby, S., A. King, A. Sharkey, Y. Loke.
1999
. Molecular studies of trophoblast HLA-G: polymorphism, isoforms, imprinting and expression in preimplantation embryo.
Tissue Antigens
53
:
1
6
Rebmann, V., K. Pfeiffer, M. Passler, S. Ferrone, S. Maier, E. Weiss, H. Grosse-Wilde.
1999
. Detection of soluble HLA-G molecules in plasma and amniotic fluid.
Tissue Antigens
53
:
14
7
Puppo, F., M. Costa, P. Contini, S. Brenci, E. Cevasco, M. Ghio, R. Norelli, A. Bensussan, G. Capitano, F. Indiveri.
1999
. Determination of soluble HLA-G and HLA-A, -B, and -C molecules in pregnancy.
Transplant. Proc.
31
:
1841
8
Blaschitz, A., F. Lenfant, V. Mallet, M. Hartmann, A. Bensussan, D. E. Geraghty, P. Le Bouteiller, G. Dohr.
1997
. Endothelial cells in chorionic fetal vessels of first trimester placenta express HLA-G.
Eur. J. Immunol.
27
:
3380
9
Hunt, J. S., D. Vassmer, T. A. Ferguson, L. Miller.
1997
. Fas ligand is positioned in mouse uterus and placenta to prevent trafficking of activated leukocytes between the mother and the conceptus.
J. Immunol.
158
:
4122
10
Chu, W., M. E. Fant, D. E. Geraghty, J. S. Hunt.
1998
. Soluble HLA-G in human placentas: synthesis in trophoblasts and interferon-γ-activated macrophages but not placental fibroblasts.
Hum. Immunol.
59
:
435
11
McMaster, M., Y. Zhou, S. Shorter, K. Kapasi, D. Geraghty, K. H. Lim, S. Fisher.
1998
. HLA-G isoforms produced by placental cytotrophoblasts and found in amniotic fluid are due to unusual glycosylation.
J. Immunol.
160
:
5922
12
Dulphy, N., M. A. Peyrat, V. Tieng, C. Douay, C. Rabian, R. Tamouza, S. Laoussadi, F. Berenbaum, A. Chabot, M. Bonneville, et al
1999
. Common intra-articular T cell expansions in patients with reactive arthritis: identical β-chain junctional sequences and cytotoxicity toward HLA-B27.
J. Immunol.
162
:
3830
13
Fournel, S., M. Aguerre-Girr, A. Campan, L. Salauze, A. Berrebi, Y. Lone, F. Lenfant, P. Le Bouteiller.
1999
. Soluble HLA-G: purification from eukaryotic transfected cells and detection by a specific ELISA.
Am. J. Reprod. Immunol.
42
:
22
14
Grumet, F. C., S. Krishnaswamy, K. See-Tho, E. Filvaroff, D. D. Hiraki.
1994
. Soluble form of an HLA-B7 class I antigen specifically suppresses humoral alloimmunization.
Hum. Immunol.
40
:
228
15
Fournel, S., L. Genestier, J. P. Rouault, G. Lizard, M. Flacher, O. Assossou, J. P. Revillard.
1995
. Apoptosis without decrease of cell DNA content.
FEBS Lett.
367
:
188
16
Koopman, G., C. P. Reutelingsperger, G. A. Kuijten, R. M. Keehnen, S. T. Pals, M. H. van Oers.
1994
. Annexin V for flow cytometric detection of phosphatidylserine expression on B cells undergoing apoptosis.
Blood
84
:
1415
17
Matzinger, P..
1991
. The JAM test: a simple assay for DNA fragmentation and cell death.
J. Immunol. Methods
145
:
185
18
Fournel, S., E. Robinet, N. Bonnefoy-Berard, O. Assossou, M. Flacher, H. Waldmann, G. Bismuth, J. P. Revillard.
1998
. A noncomitogenic CD2R monoclonal antibody induces apoptosis of activated T cells by a CD95/CD95-L-dependent pathway.
J. Immunol.
160
:
4313
19
Klas, C., K. M. Debatin, R. R. Jonker, P. H. Krammer.
1993
. Activation interferes with the APO-1 pathway in mature human T cells.
Int. Immunol.
5
:
625
20
Fournel, S., L. Genestier, E. Robinet, M. Flacher, J. P. Revillard.
1996
. Human T cells require IL-2 but not G1/S transition to acquire susceptibility to Fas-mediated apoptosis.
J. Immunol.
157
:
4309
21
Horuzsko, A., J. Antoniou, P. Tomlinson, V. Portik-Dobos, A. L. Mellor.
1997
. HLA-G functions as a restriction element and a transplantation antigen in mice.
Int. Immunol.
9
:
645
22
Sanders, S. K., P. A. Giblin, P. Kavathas.
1991
. Cell-cell adhesion mediated by CD8 and human histocompatibility leukocyte antigen G, a nonclassical major histocompatibility complex class 1 molecule on cytotrophoblasts.
J. Exp. Med.
174
:
737
23
Schmidt, C. M., E. Garrett, H. T. Orr.
1997
. Cytotoxic T lymphocyte recognition of HLA-G in mice.
Hum. Immunol.
55
:
127
24
Bottarel, F., M. Feito, M. Bragardo, S. Bonissoni, D. Buonfiglio, S. DeFranco, F. Malavasi, T. Bensi, U. Ramenghi, U. Dianzani.
1999
. The cell death-inducing ability of glycoprotein 120 from different HIV strains correlates with their ability to induce CD4 lateral association with CD95 on CD4+ T cells.
AIDS Res. Hum. Retroviruses
15
:
1255
25
Zavazava, N., M. Kronke.
1996
. Soluble HLA class I molecules induce apoptosis in alloreactive cytotoxic T lymphocytes. [Published erratum appears in 1996 Nat. Med. 2:1267.].
Nat. Med.
2
:
1005
26
Geraghty, D. E., B. H. Koller, H. T. Orr.
1987
. A human major histocompatibility complex class I gene that encodes a protein with a shortened cytoplasmic segment.
Proc. Natl. Acad. Sci. USA
84
:
9145
27
Puppo, F., P. Contini, M. Ghio, S. Brenci, M. Scudeletti, G. Filaci, S. Ferrone, F. Indiveri.
2000
. Soluble human MHC class I molecules induce soluble Fas ligand secretion and trigger apoptosis in activated CD8+ Fas (CD95)+ T lymphocytes.
Int. Immunol.
12
:
195
28
Pfeiffer, K., V. Rebmann, M. Pässler, H. van der Ven, K. van der Ven, H. Grosse-Wilde.
1999
. Early pregnancy induces significant changes in soluble HLA levels.
Placenta
20
:
A52
29
Loke, Y. W., A. King.
1995
.
Human Implantation: Cell Biology and Immunology
Cambridge Univ. Press, Cambridge, U.K.
30
Hammer, A., G. Dohr.
1999
. Apoptotic nuclei within the uterine decidua of first trimester pregnancy arise from CD45 positive leukocytes.
Am. J. Reprod. Immunol.
42
:
88