Pivotal role of CEACAM1 protein in the inhibition of activated decidual lymphocyte functions (original) (raw)
CEACAM1 is expressed on different decidual lymphocytes after activation. To test the possible role of CEACAM1 in controlling decidual lymphocyte functions, we isolated decidual lymphocytes from first-trimester elective pregnancy terminations as described in Methods. Obtained tissues were identified as decidua by histologic analysis. Lymphocytes were isolated from nine different deciduae and quadruple-stained using flow cytometry for the expression of CD3, CD16, CD56, and CEACAM. In agreement with previous observations (2), the total decidual lymphocyte population contained mainly CD16– NK cells (70–80%, characterized by CD3– CD56bright; data not shown), but T (characterized by CD3+ CD56–) and NKT (characterized by CD3+ CD56+) cells were also identified (5.3% and 3.2%, respectively; data not shown). Little or no staining for the CEACAM1 protein was observed among all decidual lymphocyte populations tested (Figure 1, a–c).
CEACAM1 staining of decidual lymphocytes. Decidual lymphocytes were isolated and quadruple-stained as described in Methods. (a–c) CEACAM1 staining on nonactivated decidual NK cells (a), T cells (b), and NKT cells (c). One representative experiment is shown out of three performed. Decidual lymphocytes were cultured in the presence of IL-2 as described (20) and then screened for CEACAM1 expression with the 5F4 mAb. (d–f) CEACAM1 staining for activated decidual NK clone (d), T clone (e), and NKT clone (f). Similar results were obtained when other lymphocyte clones were used. (g and h) Staining of EVTs for HLA-G and CEACAM1, respectively. Bold lines represent mAb staining and thin lines show background staining.
Various lymphocytes were cloned (see Methods) and cultured for 3 weeks in the presence of IL-2 (50 U/ml). Remarkably, staining with the 5F4 anti-CEACAM1 mAb (see Figure 2) revealed a dramatic increase in the CEACAM1 protein expression on the surface of the vast majority of NK, T, and NKT cell clones tested (85%, 86%, and 95%, respectively; surface expression of CEACAM1 on representative clones is shown in Figure 1, d–f, and data not shown). This is in marked contrast to NK cells derived from peripheral blood, in which surface CEACAM1 expression could be detected on only 2–3% of IL-2–activated CD16+ NK clones and on about 45% of the IL-2–activated CD16– clones (15, 21). Notably, the expression levels of the CEACAM1 on the surface of all tested clones were more than threefold above background (data not shown). This level of expression was reported to be sufficient for effective inhibition of NK cytotoxicity (15).
Staining of .221 cells expressing various members of the CEACAM family using specific anti-CEACAM antibodies. We generated .221 transfectants as described in Methods. Each row shows the staining performed on a particular transfectant (indicated at left), and each column shows the staining with a particular antibody (indicated at top). Bold lines represent antibody staining and thin lines show background staining on .221 cells. One representative experiment is shown out of three performed.
As the CEACAM1 protein interacts homotypically with other CEACAM1 proteins (15, 22, 23) (see Figures 6 and 7) and decidual lymphocytes are in direct contact with embryonic EVT cells in vivo (3), it was important to test whether EVT cells express the CEACAM1 protein. EVT cells were obtained from the same elective pregnancy terminations from which decidual lymphocytes were isolated and were tested for the expression of HLA-G and CEACAM1. As the expression of HLA-G is restricted to EVT cells only (6), isolated cells were identified as EVT cells by using specific staining with the anti–HLA-G specific mAb MEM-G/13B. The mAb MEM-G/13B specifically stains the class I MHC–negative .221 cells transfected with HLA-G; it did not stain .221 cells transfected with other class I MHC cDNA (data not shown). FACS staining analysis of isolated EVT cells showed that these cells express the HLA-G (Figure 1g) and the CEACAM1 (Figure 1h) proteins. These findings suggest that CEACAM1 might mediate direct interactions between activated decidual lymphocytes and EVTs and thus might display a novel control mechanism protecting the embryo from sustaining damage.
CEACAM1-Ig specifically binds to CMV-infected fibroblasts. (a) Binding of CEACAM1-Ig to .221/CEACAM1 cells (bold line) but not to parental .221 (thin line). The figure shows a representative experiment out of three performed. (b) Day-by-day staining of uninfected and CMV-infected HFF cells in the presence or absence of 300 μg/ml of the antiviral agent PFA. Cells were stained with CEACAM1-Ig and with the control CD99-Ig fusion protein as described in Methods. Data are presented as fold increase above the staining of uninfected cells. The average of two independent experiments is shown.
The functional interactions between BW/CEACAM1ζ and CMV-infected HFFs elicit IL-2 secretion. (a) Spontaneous IL-2 secretion by BW and various BW transfectants after 48 hours of incubation. The average of 20 independent experiments is shown. (b) IL-2 secretion by BW/CEACAM1ζ cells coincubated for 24 hours with irradiated .221 or with .221/CEACAM1 cells. The average of six independent experiments is shown. (c) IL-2 secretion after coincubation of BW or BW/CEACAM1ζ cells with uninfected or CMV-infected HFF cells for 48 hours. No IL-2 secretion above background levels was observed when PFA was included in the assay (only day 6 is shown). Experiments were performed concomitantly with the flow cytometry binding assays of CEACAM1-Ig shown in Figure 5. The average of two independent experiments is shown.
CEACAM1 interactions inhibit decidual NK cytotoxicity. We have previously demonstrated that the CEACAM1-mediated inhibition of NK cells can be blocked by using rabbit polyclonal anti-CEACAM antibodies and not by the mAb 5F4 or the mAb Kat4c (15). It was previously reported that the CEACAM1 protein interacts with other CEACAM proteins, such as CEACAM5 and CEACAM6, and that the binding site of CEACAM1 was located at the N-terminal Ig-V–type domain of the CEACAM1 protein (23). The N-terminal Ig-V–type domain of the CEACAM family reveals 70–90% sequence similarity among the different variants. It was therefore important to determine the specificity of all anti-CEACAM1 antibodies used in this work. We transfected .221 cells with CEACAM1 (15), CEACAM6, and CEACAM5 and stained them for surface expression using the various anti-CEACAM antibodies.
Figure 2 shows that all anti-CEACAM antibodies specifically recognized members of the CEACAM family. This is because no staining was observed on either nontransfected .221 cells or the control HLA-B27–transfected .221 cells. The 5F4 mAb recognized the CEACAM1 protein only, whereas the Kat4c mAb and the rabbit polyclonal antibodies directed against CEACAM recognized CEACAM1, CEACAM6, and CEACAM5 proteins (Figure 2).
To investigate whether the CEACAM1 protein is functional, IL-2–activated decidual NK clones, expressing the CEACAM1 protein (a representative clone is shown in Figure 3a), were tested in killing assays against .221 cells and .221 cells transfected with CEACAM1 (.221/CEACAM1). The generation of these transfectants was described previously (15). The CEACAM1+ NK clones effectively killed .221 cells, whereas inhibition of lysis was observed when .221/CEACAM1 cells were used (Figure 3b). The inhibition of NK killing by .221/CEACAM1 cells was the result of the CEACAM1 homotypic interactions, as lysis of .221/CEACAM1 cells was restored when rabbit anti-human CEACAM antibodies were included in the assay. The addition of a control rabbit serum derived from ubiquitin-immunized rabbit had no effect (data not shown). No difference in the lysis of .221 or .221/CEACAM1 cells was observed when CEACAM1– NK clones were used (data not shown). In agreement with a previous report (12), most decidual NK clones displayed only limited cytotoxicity against the .221 target cells (10–20% lysis; data not shown). Importantly, the killing of .221/CEACAM1 cells by “low killer” decidual NK clones was also decreased because of the homotypic CEACAM1 interactions, and the addition of anti-CEACAM polyclonal antibodies restored lysis (data not shown).
CEACAM1-mediated inhibition of decidual NK cytotoxicity. Decidual NK clones were stained for CEACAM1 expression. (a) CEACAM1 staining of decidual NK clone 17 using the anti-CEACAM1 mAb 5F4 (bold line). The thin line shows the control staining. (b) Killing and inhibition of NK clone 17 by .221 cells and by .221 cells transfected with CEACAM1 (.221/CEACAM1). Blocking experiments were performed using 40 μl/ml of anti-CEACAM antibodies. Average of three independent experiments is shown. Similar results were obtained when other CEACAM1+ NK clones were used.
CEACAM1 interactions inhibit staphylococcal enterotoxin B–induced decidual T cell proliferation. As the expression of CEACAM1 protein was also demonstrated on the vast majority of T lymphocytes activated by IL-2, we also tested the effect of CEACAM1 interactions on T cell proliferation. Superantigens can induce T cell proliferation by binding to class II MHC proteins and specific TCR Vβ chains. The staphylococcal enterotoxin B (SEB) superantigen interacts with various TCR Vβ chains, including Vβ3 and Vβ17. Decidual T cell clones were obtained as described in Methods and screened by flow cytometry for the expression of CD4, Vβ3, and Vβ17 by using specific mAb’s. A representative T cell clone, no. 1, stained brightly for both CD4 and Vβ17 (Figure 4, a and b), and moderately for CEACAM1 (Figure 4c). The SEB-induced proliferation of this T cell clone was assayed as described in Methods. A dramatic increase in the T cell proliferation was observed when cells were incubated with .221 cells in the presence of 250 ng/ml of SEB (50-fold above the background proliferation without SEB; data not shown). Efficient inhibition of the T cell proliferation (around 50%) was observed in all SEB concentrations tested when cloned T cells were incubated with .221/CEACAM1 cells (Figure 4d). The expression levels of the class II MHC proteins were similar on both .221 and .221/CEACAM1 cells (data not shown).
CEACAM1-mediated interactions inhibit SEB-induced T cell proliferation. Decidual T cell clones were tested for expression of CD4 (a), Vβ17 (b), and CEACAM1 (c) by flow cytometry. Bold lines indicate mAb staining and thin lines indicate control staining. (d) Fifty thousand cells of the presented T cell clone were incubated for 2 days with 25,000 irradiated .221 cells or with .221 cells transfected with CEACAM1 (.221/CEACAM1), in the presence of decreasing SEB concentrations as indicated in the figure. Proliferation was measured with 3H-thymidine incorporation. The figure represents the average of ten independent experiments. Similar results were obtained when other T cell clones were used.
CEACAM1 interactions inhibit secretion of cytokines from decidual NKT cells. Cytokines might play an important role in fetus development (24). NKT cells that are present among the decidual lymphocyte population (see Figure 1) are able to produce large amount of cytokines. The functional effect of CEACAM1 interactions on cytokine secretion of decidual NKT cells has never been investigated. Decidual NKT clones were cultured as described in Methods and screened for CEACAM1 expression by flow cytometry, using the anti-CEACAM1 5F4 mAb (a representative NKT cell clone, no. 3, is shown in Figure 5a). NKT clone 3 spontaneously secreted IFN-γ into the media, as measured by ELISA (Figure 5b). Other cytokines such as IL-4, IL-5, IL-13, TNF, and macrophage inflammatory protein-1α could not be detected in culture supernatant of this clone (data not shown). Cross-linking of CEACAM1 for 24 hours with the Kat4c mAb dramatically decreased the amount of IFN-γ detected in the medium of this NKT cell clone (Figure 5b). In order to determine whether the inhibitory effect observed after cross-linking of CEACAM1 on NKT cells is the result of decreased secretion or decreased production of IFN-γ, we stained for the presence of intracellular IFN-γ, before and after cross-linking of CEACAM1, as described in Methods. Untreated NKT cells showed little staining for intracellular IFN-γ (median fluorescence intensity twofold above background; Figure 5c). After cross-linking with the Kat4c mAb, the staining for intracellular IFN-γ increased significantly (median fluorescence intensity 4.5-fold above background; Figure 5d). These findings therefore suggest that CEACAM1 engagement on NKT cells suppresses the cytokine secretion machinery and not de novo synthesis.
CEACAM1-mediated inhibition of IFN-γ secretion from NKT cells. (a) CEACAM1 expression on isolated activated NKT clone. The bold line shows the staining with 5F4 mAb, and the thin line shows the control staining. (b) The amount of IFN-γ in culture supernatant of mAb-treated and untreated NKT clone cells measured by ELISA. The average of two independent experiments is shown. Cross-linking of surface CEACAM1 was performed without (c) or with (d) the Kat4c mAb, and intracellular staining for IFN-γ was performed. One representative experiment is shown out of two performed. Similar results were obtained when other NKT cell clones were used.
In vivo upregulation of CEACAM1 on decidual lymphocytes. The above observations suggest a major role for the CEACAM1 protein in the regulation of decidual lymphocyte functions after IL-2 activation. In vivo activation of decidual lymphocytes might occur as a result of viral infection. CMV is the leading cause of congenital viral infections in Western countries (25, 26). We therefore tested whether CEACAM1 expression could be detected on the surface of lymphocytes obtained from deciduae of women who had primary CMV infection during gestation with documented intrauterine manifestations. Second and third-trimester pregnancy terminations of women diagnosed with primary CMV infection necessitate the administration of labor-promoting agents that might have some immunological effects. To control the experiment, we analyzed the expression of CEACAM1 protein on the surface of lymphocytes obtained either from third-trimester caesarian sections with labor (Table 1) or from the deciduae taken from caesarian sections without labor (Table 1). Decidual lymphocytes were obtained and stained for the presence of CEACAM1 on NK, NKT, and T cells as above. Only very limited numbers of NKT cells were isolated, and therefore the expression of CEACAM1 on NKT cells could not be determined. Remarkably, a significant elevation of CEACAM1 expression was observed in NK and T cells obtained from deciduae of CMV-infected women, whereas little or no expression of CEACAM1 was observed in the two control groups (Table 1). CEACAM1 expression can vary significantly between different CMV-infected deciduae. In one patient, 90% and 95% of the NK and T cells, respectively, expressed the CEACAM1 protein, whereas in the second patient the expression of the CEACAM1 protein was limited to 10% and 10.2% of NK and T cells, respectively. However, the expression of the CEACAM1 protein, even in the second patient, was still very significant compared with that of the control groups, and it was similar to the expression level of other class I MHC inhibitory receptors, which vary between 5% and 20% (9). There are several possible reasons for the differences in the level of CEACAM1 expression, such as subjective local immune response, course of CMV infection, and the time of pregnancy termination after the initiation of infection. The mild expression of the CEACAM1 protein on trophoblasts obtained from normal decidua (Figure 1h) was still maintained on trophoblasts obtained from infected decidua (data not shown).
CEACAM1 expression is upregulated on decidual lymphocytes from women with primary CMV infection
CMV-infected fibroblasts express a novel ligand for CEACAM1. The results presented above demonstrate that CEACAM1 expression is upregulated in vivo in lymphocytes obtained from CMV-infected deciduae. Expression of CEACAM1 was observed on EVT cells obtained from either normal or CMV-infected deciduae (Figure 1 and data not shown). Thus, CEACAM1 homotypic interactions might be expected to occur in vivo, leading to lymphocyte inhibition.
Only two cases of CMV-infected deciduae are presented here, as studies in vivo are limited for several reasons. In addition to the fact that primary CMV infection during pregnancy is quite rare (26), the detection and diagnosis are quite difficult. Furthermore, deciduae from CMV-infected women were used only if they spontaneously detached, to avoid unnecessary additional procedures. It is clear, however, that in both presented cases, a significant CEACAM1 upregulation was observed. To further establish the effect of CMV infection with regard to CEACAM1 inhibition and to test whether the CMV uses the CEACAM1 inhibitory mechanism to avoid attack by the immune system, we used the CMV-infected HFFs. HFF cells were infected with CMV strain AD169 with moi 2–3. Importantly, no staining of either infected or uninfected HFF cells with anti-CEACAM1, -5, and -6 Kat4c mAb was observed at any time point before or after the infection. Infected cells were harvested at different time points at 24-hour intervals after the infection and stained for the presence of CEACAM1 ligand using CEACAM1-Ig fusion protein, as described in Methods.
The CEACAM1-Ig fusion protein specifically stained the .221/CEACAM1 cells and did not stain the .221 cells (Figure 6a), indicating that CEACAM1 homotypic interactions are strong enough to be detected by this method. No staining of CEACAM1-Ig was observed in the first 4 days after the infection (Figure 6b and data not shown). Importantly, CEACAM1-Ig staining was observed starting on day 5 and reaching maximum on day 6 after the infection. All infected cells were positively stained with anti-pp65 mAb (data not shown). The CEACAM1-Ig binding observed was only to the HFF-infected cells, not to the uninfected cells (data not shown). No changes in the level of the control CD99-Ig fusion protein staining were observed at any time point (Figure 6b). As CEACAM1 can interact only with the CEACAM1, -5, and -6 variants (14), and as it was also reported that CEACAM variants cannot be detected on the surface of human fibroblasts (27), these results strongly suggest the existence of a novel ligand for CEACAM1 on the surface of CMV-infected HFF cells. This novel ligand appears late after the infection. To further test this hypothesis, we performed similar experiments in the presence of the antiviral agent PFA, which is known to block viral DNA synthesis and early-late-phase transition. Progeny virus titers in culture supernatants were determined on day 4 after infection by a standard plaque titration assay on HFFs. In the absence of PFA, virus titer was 3 × 106 plaque-forming units/ml, whereas in the presence of PFA no virus could be detected. In agreement with the above observations demonstrating the appearance of CEACAM1 ligand on the surface of CMV-infected HFFs, the addition of PFA completely abolished the binding of CEACAM1-Ig to the infected HFF cells (Figure 6b).
We next tested whether the CEACAM1 interactions with the CMV-infected HFFs are functional. Mouse BW cells were stably transfected with a chimeric molecule composed of the extracellular portion of CEACAM1 fused to mouse ζ chain(as described in Methods). Engagement of CEACAM1 leads to the secretion of mouse IL-2, mediated by the ζ chain. The IL-2 amounts in the cell supernatants can be measured by ELISA. Secretion of IL-2 could be detected in the culture supernatants of the BW cells transfected with CEACAM1ζ, but not in the culture supernatants of the BW cells or BW cells transfected with CD16ζ (Figure 7a). Moreover, IL-2 secretion was also detected in the supernatants of BW/CEACAM1ζ cells when cells were incubated with .221/CEACAM1 cells, but not with they were incubated with .221 cells (Figure 7b). Thus, homotypic CEACAM1 interactions are strong enough to induce IL-2 secretion in this system. In agreement with the CEACAM1-Ig staining data, efficient secretion of IL-2 was observed (on days 5 and 6 after the infection) in the supernatants of BW/CEACAM1ζ cells cultured with infected HFFs. This IL-2 secretion was blocked by the addition of PFA (Figure 7c). No IL-2 secretion was observed in the culture supernatants of BW/CD16ζ cells incubated with uninfected or infected HFF cells (data not shown).
To further substantiate the above results, we cultured the clinical CMV strain isolated from the infected decidua (patient 6; Table 1) with infected HFF cells. The propagation of the virus was much slower than that of the laboratory AD169 strain (data not shown). Consistent microscopic monitoring of infected HFF cells revealed that even after prolonged propagation time, only partial infection could be achieved. One month after initiation of infection, infected HFF cells were analyzed for recognition by CEACAM1. HFF cells were stained with Ig-fused proteins, including CEACAM1-Ig and the control CD99-Ig. No staining of uninfected HFF cells was observed (data not shown). Specific staining of the infected HFF cells could be observed with the CEACAM1-Ig but not with the CD99-Ig (20% and 2% staining, respectively; Figure 8a). No staining was observed when anti-CEACAM antibodies were used (data not shown), indicating that CEACAM1-Ig recognizes a novel CMV-induced ligand on infected HFFs. We also tested whether this recognition is capable of eliciting IL-2 secretion from BW/CEACAM1ζ cells. IL-2 levels were measured in the supernatants of BW or BW/CEACAM1ζ cells cocultured with HFF cells infected with the clinical CMV strain isolated from patient 6. In agreement with the CEACAM1-Ig staining, increased IL-2 secretion could be detected only in the supernatants of the BW/CEACAM1ζ cells coincubated with infected HFF cells (Figure 8b). The moderate elevation of IL-2 secretion and the partial staining of CEACAM1-Ig (Figure 8, a and b) are correlated with the low infection levels of this clinical CMV strain observed in vitro, and with the moderate percentages of CEACAM1+ lymphocytes isolated from infected decidua no. 2 (patient 6; Table 1). Similar results were obtained with another clinical CMV strain, isolated from a neonate’s urine (data not shown).
CMV isolated from infected decidua induces a ligand for the CEACAM1 on infected HFF cells. (a) Staining of HFF cells infected with clinical CMV strain with CD99-Ig or with CEACAM1-Ig. No staining was observed when proteins were omitted, indicated by the horizontal line. FSC, forward scatter. (b) IL-2 secretion from BW or BW/CEACAM1ζ cells coincubated with HFF-infected cells for 48 hours. The average of two experiments is shown.
These combined results show a major role for the CEACAM1 protein in the inhibition of decidual lymphocyte functions after activation and suggest that the CMV virus might have developed a unique mechanism, mediated via the CEACAM1 protein interactions with an unknown ligand, to avoid decidual immune cell attack.