Murine T Cell Determination of Pregnancy Outcome

Cellular Immunology 196, 71–79 (1999) Article ID cimm.1999.1535, available online at http://www.idealibrary.com on

Murine T Cell Determination of Pregnancy Outcome II. Distinct Th1 and Th2/3 Populations of Vg1 1d6 1 T Cells Influence Success and Failure of Pregnancy in CBA/J 3 DBA/2J Matings Petra C. Arck,* ,† David A. Ferrick,† ,‡ Darlene Steele-Norwood,§ Paul J. Egan, ¶ Kenneth Croitoru,§ Simon R. Carding, ¶ Johannes Dietl,\ and David A. Clark§ ,1 *Universitats Klinikum, Medizinische Fakultat der Humbolt, Universitat zu Berlin, Augustenburger Platz, 13353 Berlin, Germany; †Amgen Institute, University of Toronto, Toronto, Canada; ‡University of California at Davis, Department of Microbiology and Pathology, Davis, California; §Department of Medicine and Department of Pathology, McMaster University, Hamilton, Ontario, Canada; ¶ Department of Microbiology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania; and \Department of Obstetrics and Gynaecology, Universita¨t Wu¨rzburg, Wu¨rzburg, Germany Received September 16, 1998; accepted July 4, 1999

Vd6.3 given on day 8.5 boosted abortions to the same level. These results suggest that two populations of Vg1.1 1d6.3 1 T cells may arise in the decidua: an early population that is Th1, abortogenic, and present during the time of implantation, and a Th2/3 cell subset that is present in the decidua later during pregnancy and which is pregnancy-protective. © 1999 Academic Press Key Words: murine pregnancy abortion; gdT cells; Th1/Th2/Th3 cytokines.

At the fetomaternal interface, maternal effector cells come in intimate contact with fetal trophoblast cells which express paternal antigens. Failure of fetal trophoblast cells to activate maternal Th1 immune responses has been attributed in part to the absence of classical Class I and Class II major histocompatibilty complex (MHC) antigen expression and elaboration of factors which reduce TcR expression and shift any immune responses which may occur to Th2. Classical TcR ab 1 T cells have not been found to be able to respond to trophoblasts. Recently, TcR gd 1 T cells have been characterized in the low-abortion-rate pregnant C57Bl/10 mouse decidua, and the Vg1 1 subset may be able to respond to trophoblasts in a non-MHCdependent manner. Trophoblast-recognizing T cells with Vg1 receptors are also present in the decidua of CBA/J mice pregnant by DBA/2, an abortion-prone mating combination. To test the role of the Vg1 subset of decidual gd T cells in abortion-prone pregnancies, we altered this subset by injecting monoclonal antiVg1.1 antibody on gestation day 5.5, 1 day after implantation. This reduced detectability of a Vgd subset producing TNF-a and reduced the abortion rate. AntiVg2, which reacts with a similar proportion of decidual gd T cells as anti-Vg1.1, failed to prevent abortions. Vd6.3 1 cells are prominent at the fetomaternal interface, and anti-Vd6 antibody injected on day 5.5 prevented abortions. TGF-b2 1 gd cells first appear on day 8.5 of pregnancy; anti-Vg1.1 antibody injection on day 8.5 depleted these cells and boosted abortions; anti-

INTRODUCTION In allogeneic matings, paternal genes are selectively expressed in fetal trophoblast via imprinting (1). In some trophoblast populations, paternal antigens recognizable by the maternal immune system may be expressed (2, 3). The success of pregnancy has been proposed to depend upon resistance of the fetal trophoblast to attack by both antigen-specific T cells or antibody and nonspecific innate effector cells (2). Nevertheless, abortions do occur, both spontaneously and in response to specific stimuli such as infection (endotoxin) or stress (3). It has been suggested that, in a successful pregnancy, a Th2/3 . Th1 pattern is present systemically and locally at the fetomaternal interface, but in pregnancies with high rates of abortion, a Th1 . Th2/3 pattern may occur systemically and at the fetomaternal interface (5– 8). In murine CBA/J 3 DBA/2J matings which have a high spontaneous abortion rate, there is an unusual accumulation of gd T cells in uterine decidua beginning 1 day after implanation (9). This infiltrate predisposes the pregnant females to abort if there is exposure to a second signal, such as endotoxin

1

To whom correspondence should be addressed at Rm. 3V39, McMaster University, 1200 Main St. West, Hamilton, Ontario, Canada L8N 3Z5. Fax: 1 (905) 521-4971. E-mail: [email protected]. McMaster.ca.

71

0008-8749/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.

72

ARCK ET AL.

or stress, which boosts TNF-a release in decidua (9, 10). In the CBA/J 3 DBA/2J system, from studies of the effect of injecting anti-d monoclonal antibodies at different time points during pregnancy, we have deduced that gd T cells present on day 5.5– 6.5 of pregnancy in the uterus are predominantly Th1 (TNF-a 1 IFN-g 1), and gd T cells appearing on day 8.5 are predominantly Th2/3 (IL-10 1 TGF-b2 1) (9, 11, 12). CD8 1 T cells, thought to be TcR ab 1, promote a Th1 3 Th2/3 shift in decidual gd cells, and alloimmunization to paternal MHC promotes this transition and decreases the abortion rate (9, 13, 14). However, TcR ab 1 T cells are not thought to be able to recognize the fetal trophoblast by contrast to gd 1 T cells (15). Until now, only the Vg1 subset of gd T cells has been shown to react with the trophoblast as assessed by stimulation of IL-2 production in vitro from decidual gd cells cloned by a hybridoma technique (16) but no IL-2 has been detected so far in decidua; the IL-2 produced by these hybridomas is derived from the fusion partner rather than from the gd T cell itself, so one can only draw conclusions concerning signaling via the gd TcR and no inferences can be made with respect to what the original gd T cell might be producing in situ. At least six different Vgd populations may occur in the uterus of pregnant mice in addition to the Vg6 1 subset that has cannonical receptors and which is present both in the nonpregnant endometrium and in the tongue (16 –18). The purpose of the present study was to test whether the gd T cell populations affecting pregnancy success in the CBA/J 3 DBA/2 model belonged to the Vg1 1 T cell class that can recognize fetal trophoblasts and which might be able to distinguish between trophoblasts arising from mating with different strains of male, such as BALB/c, where the abortion rate is quite low. MATERIALS AND METHODS Animals. Female CBA/J and male DBA/2 mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and maintained in a barrier facility with a 12-h light/dark cycle. Animal care and experimental procedures followed institutional ethics guidelines and conformed to requirements of the Canadian Council on Animal Care. After overnight cohabitation of CBA/J females with a DBA/2 male, females with vaginal plugs (day 0.5 of pregnancy) were segregated. Some of the mated mice were injected intraperitoneally with 100 mmg anti-Vg1.1 antibody or 100 mg anti-Vd6.3 antibody on day 5.5 or 8.5 of pregnancy. Some groups were exposed to ultrasonic stress for 24 h beginning on day 5.5 to boost abortion rates as described previously (9, 10). All animals were sacrificed on day 13.5, the number of normal and resorbing sites was counted, and cells were taken for further study.

Antibodies. Anti-Vg1.1 monoclonal antibody was isolated from the supernatant of 2.11 hybridoma clone kindly provided by Dr. Jeffrey A. Bluestone (19). AntiVg2 monoclonal antibody (Vg2 in the Bluestone nomenclature and Vg4 in the Tonegawa nomenclature) was generated similarly using the UC3-10A6 clone (20). Anti-Vd6.2/6.3 antibody (hereafter denoted as anti-Vd6.3) was generated from hybridoma clone C504.17C as described elsewhere (21). In general, the hybridoma cells were cultured in vitro in DMEM with 10% FBS, 1% l-glutamine, 1% pyruvate, 5 3 10 25/ml 2-mercaptoethanol, 100 IU/ml penicillin, and 100 mg/ml streptomycin. Antibody was purified from supernatant by standard affinity column chromatography. Some of the anti-Vg1.1 antibody was conjugated to FITC. TRICOLOUR-labeled anti-TcR gd (CL7201TC, Caltag, Burlingame, CA) or PE-labeled anti-d (GL3, Pharmingen) was used at a predetermined optimum concentration of 10 mg/ml (1 mg/10 6 cells). FITC-labeled rat IgG 1 anti-murine TNF-a was obtained from Pharmingen and used at 1.2 mg/ml. Anti-TGF-b2 (3C7) was purchased from Genzyme (Cambridge, MA) and conjugated with PE by Molecular Probes (Eugene, OR) and used at 1.0 mg/ml. This antibody may cross-react with TGF-b3, but we have not found either TGF-b3 or TGF-b1 in supernatants from decidua and, by in situ hybridization, TGF-b3 mRNA 1 cells were uncommon compared to TGF-b2 mRNA 1 cells (22, 23). Therefore, operationally this antibody immunostain is referred to as anti-TGF-b2 and this provides a specific marker for the cells of interest. Preparation of cell suspensions and flow cytometry. The standard method we have used for flow cytometry is depicted in Fig. 1. Decidua and spleen cells were prepared from groups of treated and untreated mice as described in (9, 10) and incubated for 4 h at 37°C in brefeldin A (10 mg/ml, Epicentre Technologies). Decidual mononuclear cells were purified by centrifugation on lympholyte M (Cedarlane Labs., Hornby, Ontario). To determine the percentage of Vg1.1 1 T cells among the gd 1 T cells, 1 3 10 6 decidual cells were labeled with PE-conjugated anti-d (GL3/Pharmingen) and FITC-labeled anti-Vg1.1; alternately, TRICOLOUR-labeled anti-gd was used with streptavidin–PE. For intracellular detection of cytokine production in cells stained with monoclonal antibody, our standard protocol was used (9). Briefly, 3– 4 3 10 6 cells were incubated in 100 ml cytoflow buffer (PBS, 1% BSA, 5 mM EDTA, and 0.1% sodium azide) for 15 min at 25°C in the dark and the cells were washed and fixed in 100 ml solution A, Cell Perm. & Fix kit (Caltag), for 15 min at 25°C and washed. The cells were gently resuspended in 100 ml of solution B, Cell Perm. & Fix kit, and divided, and aliquots of 1 3 10 6 cells incubated for 15 min at 25°C in the dark with anti-cytokine antibody at the following

REGULATION OF ABORTION BY Vg1 T CELLS

73

FIG. 1. Method of cell harvesting and staining protocol for aquisition and analysis of murine decidual cells using three-color flow cytometry staining and lymphocyte scatter gate.

concentrations: FITC-conjugated anti-mouse TNF-a (Pharmingen) at 0.12 mg/100 ml sample and PE-labeled anti-mouse TGF-b2,3 at 0.1 mg/100 ml sample. Cell fluorescence was measured using a Becton–Dickinson FACScan. The data were analyzed using CellQuest software. Forward and side scatter were used to set the gates to capture viable small lymphoid cells for assessment of intracellular staining, as illustrated in Fig. 1. In some experiments, scatter gates were enlarged in order to capture all viable cells. Negative control samples incubated with irrelevant fluorochrome-conjugated isotype-matched antibodies were performed in parallel and generated ,0.3% fluorescent gd cells (9). Statistics. Resorption rates were determined from (total number of resorbing implants)/(total number of implants, resorbing and nonresorbing) and the significance of differences between groups was determined using distribution-free nonparametric methods, such as x 2 or Fisher’s exact test, where appropriate. Significance was set at P , 0.05. Resorption rates were also determined by calculating the arithmetic mean and

SEM using data from individual mice for purposes of comparison to the standard method. Flow cytometry data were analyzed using Quad statistics. RESULTS Figure 2 shows the effect of intraperitoneal administration of 100 mg of purified anti-TcR Vg1.1 antibody on the abortion (resorption) rate of CBA/J 3 DBA/2J pregnancies. The noninjected unstressed mice had a resorption rate of 10.1% (e.g., 1003 (7/69)), and after injection of antibody on day 5.5 of pregnancy, the abortion rate decreased (as predicted by our hypothesis) to 6.1%, which was not significantly different from the control group. In contrast, antibody injected on day 8.5 boosted the abortion rate significantly (as predicted by our hypothesis) to 47.5%. This reproduced the result from a prior experiment in which fewer mice per group were used. A 24-h exposure to sonic stress (which stimulates release of TNF-a in decidua (10)) significantly increased the abortion rate (as expected) to 30.5% com-

74

ARCK ET AL.

FIG. 2. Influence of injection of mAB to Vg1.1 1 cells on gestation day 5.5 or 8.5 on the abortion rate (resorptions) in unstressed and stress-treated CBA 3 DBA/2 pregnancies, as assessed on gestation day 13.5. Mean abortion rates for different groups were determined from the ratio of resorptions/total number of implantations, as is given above the error bar on each column. The standard deviation bars were determined from the binomial. The number at the column base indicates number of mice per group. The significance of differences (P , 0.05) was determined using x 2 (or Fisher’s exact test where appropriate). (*) Significant increase in abortion rate compatred to uninjected control group; (**) significant reduction in abortion rate compared to stress-treated control, and not significantly different from unstressed control result; (***) significant increase in abortion rate compared to day 5.5 treatment, but not different from either the uninjected stressed group or the unstressed mice given antibody on day 8.5 of pregnancy. The significance of differences in stressed versus no-stress groups was confirmed using Dunnett’s means comparison test employed in our previous studies (9, 11). Arithmetic means 61 SEM for each column from left to right were 11.0 6 3.3, 7.7 6 3.6, 45.3 6 9.8, 32.8 6 7.1, 18.0 6 4.7, and 41.7 6 8.3%. Student’s t test confirmed that the changes occurring in the direction predicted were significant.

pared to the control mouse value of 10.1%. Anti-Vg1.1 injected at the start of the stress on day 5.5 abrogated the stress-induced increase in the abortion rate (17.7 vs 30.5%). This also reproduced the result from a prior experiment. When the antibody was administered on day 8.5, the abortion rate of 44.4% was not different from either the rate observed in the stressed uninjected mice or the rate in the unstressed mice that had been injected with anti-Vg1.1 on day 8.5 of pregnancy, similar to the results previously obtained using anti-d (11) and in agreement with our hypothesis that Vg1.1 1 cells were responsible to such effects. These data suggested that different Vg1.1 1 cell subpopulations were being affected on days 5.5 and 8.5; depletion on day 5.5 prevented abortions (and blocked the stress effect), depletion on day 8.5 enhanced losses to a ceiling (limit) of 44 – 48%. By flow cytometry of the small lympoid population gated in Fig. 1, 17.1 6 3.5% were Vgd 1 (m 6 1 SD, n 5 3). Of the gd population, an average of 27% were Vg1.1 1 on day 13.5 of pregnancy (mean of three independent experiments); from prior work, anti-Vg2 1 reacted with 24% of the gd small lymphoid population (11). Although the frequency of Vg1.1 1 and Vg2 1 cells was similar, in an independent experiment, 100 mg

anti-Vg2 antibody injected on day 5.5 had no statistically significant effect on the abortion rate in either unstressed control mice (8% if injected, 10% not injected) or stressed mice (24% if injected, 29% not injected). These data were consistent with previous results showing no significant effect of injection of inert control fluids on abortion rates in control and in stressed mice (2, 9 –11, 13) and suggested that the Vg1.1 1 subset was functionally important. Figure 3 illustrates the effect of an injection of antiVg1.1 antibody on the frequency of intracellular cytokine 1 decidual Vd 1 T lymphoid cells in unstressed and stressed mice corresponding to Fig. 2. In the group of unstressed mice injected on day 5.5 of gestation with anti-Vg1.1 antibody, we found a change in cytokine pattern. TNF-a 1 decidual d 1 T cells decreased from 9.6% of gd cells in the control group to 5.9% in the anti-Vg1.1 antibody-injected group. The effect was more dramatic in the stressed females. Stress increased TNF-a 1 d 1 cells to 14.9%, and there was a substantial decrease after anti-Vg1.1 antibody injection from 14.9 to 2.2%. By contrast, anti-Vg1.1 antibody injected on day 5.5 of pregnancy did not have the same reducing effect on frequency of TGF-b2 1 decidual d 1 T cells (0.7 vs 0.7% in unstressed mice, 0.3 vs 0.7%

REGULATION OF ABORTION BY Vg1 T CELLS

75

FIG. 3. Effect of anti-Vg1.1 antibody injection on gestation day 5.5 or 8.5 on the levels of intracellular cytokines TNF-a and TGF-b2 of decidual gd 1 T cells, assessed on gestation day 13.5. Frequency of TNF-a 1 and TGF-b2 1 decidual gd T cells are presented in the upper two plots; the upper left quadrant of each plot shows the percentage of TNF-a 1 gd T cells, the lower right quadrant represents the percentage of TGF-b2 1 gd T cells.

in stressed mice). Injection of anti-Vg1.1 on gestation day 8.5 into unstressed mice led to an increased frequency of TNF-a 1 decidual d 1 T cells (9.6% in the unstressed control group vs 12.5% in the anti-Vg1.1 antibody-injected group) and a markedly lower frequency of TGF-b2 1 d 1 cells (0.7 vs 0.1% if injected). In stressed mice, the percentage of TGF-b2 1 cells decreased (from 0.7 to 0.3%), and although anti-Vg1.1 antibody injection on day 8.5 decreased the frequency of TNF-a 1 gd cells (14.9 vs 9.5%), there was a more striking decrease in TGF-b2 1 decidual d 1 T cells (0.3 vs 0.1%). Although one can calculate Th1:Th2/3 ratios from these percentage figures in order to provide a quantitative test of the relationship of change in Th1:

Th2/3 balance to pregnancy outcome, it is more informative if one considers the actual number of cells present in decidual tissue. Table 1 presents absolute lymphoid cell numbers calculated from the observations shown in Fig. 3 based on actual cell numbers counted in the decidual preparations per mouse. In the absence of antibody treatment, the number of TNF-a 1 gd TcR 1 cells increased significantly in stressed animals, whereas the TGFb2 1 gd 1 cell number decreased. Thus stress caused a Th1 4 Th2/3 shift. In the mice injected with anti-Vg1.1 antibody on day 5.5 and randomized to no-stress treatment, we found lower numbers of TNF-a 1 gd 1 cells in the nonstressed mice and no change in TGF-b2 1 cells

TABLE 1 Effect of Anti-Vg1.1 Antibody Injection on Gestation Day 5.5 or 8.5 in CBA/J 3 DBA/2 Pregnancies on Number of Cytokine 1 Decidual gd T Cells No-stress-treatment mice

Stress-treated mice

Control

Anti-Vg1.1, day 5.5

Anti-Vg1.1, day 8.5

Nil

Anti-Vg1.1, day 5.5

Anti-Vg1.1, day 8.5

Phenotype

(10) a

(10)

(6)

(10)

(4)

(4)

Total cells gd 1 TNF-a 1 gd 1 TGF-b2 1 gd 1 TNF-a/TGF-b2 Ratio

6.9 3 10 6 1.2 3 10 6 11.5 3 10 4 0.8 3 10 4

6.1 3 10 6 1.2 3 10 6 7.1 3 10 4 0.8 3 10 4

3.8 3 10 6 0.6 3 10 6 7.5 3 10 4 0.1 3 10 4

7.5 3 10 6 1.5 3 10 6 22.4 3 10 4 0.4 3 10 4

5.1 3 10 6 1.0 3 10 6 2.2 3 10 4 0.7 3 10 4

4.9 3 10 6 1.0 3 10 6 7.5 3 10 4 0.1 3 10 4

14.4

8.8

75.0

56.0

3.1

75.0

a

Parentheses show number of mice.

76

ARCK ET AL.

compared to the noninjected control, so anti-Vg1.1 treatment caused a Th1 3 Th2/3 shift. In the stressed mice, the number of TNF-a 1 d 1 cells in stressed mice was greatly reduced by injection of anti-Vg1.1 on day 5.5, but a stress reduction in the number of TGF-b2 1 gd 1 cells was prevented by antibody injection. Thus, injection of anti-Vg1.1 on day 5.5 also caused a Th1 3 Th2/3 shift in the stressed mice. In contrast to the effect of antibody treatment on day 5.5 of gestation, injection of anti-Vg1.1 antibody on day 8.5 had an opposite effect. The number of TNF-a 1 gd 1 cells in unstressed mice injected with anti-Vg1.1 on day 8.5 was similar to that in mice injected on day 5.5, but injection on day 8.5 caused a striking decrease in TGFb2 1 d 1 cells and hence a Th1 4 Th2/3 shift. In the stressed mice, injection on day 8.5 caused less reduction in TNF-a 1 cells than the day 5.5 injection, but a striking decrease in TGF-b2 1 cells led to an even more striking Th1 4 Th2/3 shift. Changes in the ratio of TNF-a 1/TGF-b2 1 cell numbers did not show an absolute match to abortion rates, perhaps because all sources of TNF-a in stressed mice (24) were not measured and included in the calculation. These data reproduced similar findings obtained in a prior experiment in which fewer mice were available for study. Because the flow cytometry was done using cells pooled from individual mice (to obtain a sufficient number for the study), the precision of the estimates can only be determined from the Poisson error, i.e., the square root of the number of cells counted; for example, if 3% of 10,000 cells counted were positive, the precision would be =300 5 17.3 which represents 65.8%. These calculations have not been included in the table to avoid clutter. Previous work by Heyborne et al. (16, 18) has shown that Vg1 1 decidual lymphocytes are predominantly Vd6 1 (Vd6l12 and Vd6.3), and Vd6.3 1 cells were prominent at the decidua–trophoblast interface in pregnant mice (16, 18). We found that 100 mg anti-Vd6.3 antibody injected on day 5.5 of pregnancy in our stresstreated mice reduced the abortion rate to 0/14 (0%), which was significantly less than the loss rate in the uninjected stressed control group (29%) and not different from the abortion rate in stressed mice which had received anti-Vg1.1 on day 5.5 of gestation. In an independent experiment, we found that anti-Vd6.3 administered to unstressed control mice on day 5.5 also reduced the abortion rate, but the different was not large enough to achieve statistical significance. In contrast, anti-Vd6.3 injected on day 8.5 into unstressed control mice gave an abortion rate of 44% (12/27), similar to the result with anti-Vg1.1 antibody shown in Fig. 2. We concluded that both the Th1-type gd (TNFa 1) and the Th2/3 (TGF-b2 1) functional subgroups of decidual gd T cells interacting with CBA/JxDBA/2 F1

trophoblasts that we detected were likely to be Vg1.1 1d6.3 1. DISCUSSION The lining of the nonpregnant mouse uterus and vagina contains a population of Vg6d1 cells with minimally rearranged TcR of limited (i.e., canonical) specificity (17). In contrast, at least six to seven TcR subsets of uterine gd T cells have been noted in the decidua of synpregnant and allopregnant C57Bl/10 mice (16, 18). Here, the only population able to react to the trophoblast by activating cytokine production was Vg1 1 (16). The dominant phenotype in C57Bl/10 decidua was Vg6d1 (approximately 55– 60% of all gd cells, but Vg1 1 and Vd6.3 1 were also noted) (18). Abortionprone DBA/2-mated CBA/J matings show a striking infiltration of gd cells in their decidua by contrast to low-abortion-rate matings of CBA 3 BALB/c and C57Bl/6 3 DBA/2 or C57Bl/6 (9). This increased infiltration of gd cells is thought to be functionally important as treatment with anti-d on day 5.5 of pregnancy abrogates susceptibility to abort later in pregnancy. The number of different subsets present in the infiltrate in DBA/2-mated CBA/J mice is unknown, but there must be .2 as the sum of Vg1.1 1 lymphoid cells (27%) and Vg2 1 lymphoid cells (24%) is only 51% of the total. The 27% Vg1.1 1 cells in the CBA 3 DBA/2 system appears greater than the 5% frequency of Vg1 1 cells reported for C57Bl/10 females (18), and this observation suggests that Vg1.1 1 cells might have some functional role in the abortions in the CBA 3 DBA/2 model. Since Vg1 1 cells in the C57Bl/10 system seemed to associate primarily with Vd6 (16), we hypothesized that Vg1.1d6.3 1 cells would prove relevant to pregnancy outcome. The functional role of different subsets of decidual gd T cells during pregnancy is unknown, but from the in vivo effects of anti-Vg1.1 and antiVd6.3 reported in this paper, one can ascribe a major role of Vg1.1 1d6.3 1 cells in producing either proinflammatory Th1 cytokines (TNF-a, IFN-g) that promote abortions or Th2/3 cytokines (IL-10, TGF-b2) that prevent abortions in the CBA/J 3 DBA/2 model (7, 9, 11, 25). As the latter functional subset arises at a different time in pregnancy, one may speculate that there are two distinct major subsets of gd cells that impact pregnancy outcome. The function and significance of other subsets remain to be determined. Based on Table 1, Vg1.1 1 cells did not appear to account for all of the TNF-a 1 gd cells and other sources merit analysis. From previous data (11), we knew that only 0.5% of Vg2 1 lymphoid cells were TNF-a 1. From Table 1 one can calculate from the decrease in TNF-a 1 cells/decrease in total gd 1 lymphoid cells after antiVg1.1 on day 5.5 that 5.5– 8.4% of the Vg1.1 1 population should be TNF-a 1; based on an injection on day 8.5

REGULATION OF ABORTION BY Vg1 T CELLS

of gestation, 1.3–5.6% should be TNF-a 1. Both antiVg1.1 and anti-Vd6.3 injected on day 5.5 prevented abortions. It is therefore probable that the Th1 (TNFa 1) abortogenic cells accumulating in early pregnancy (day 5.5) are Vg1.1 1d6.3 1. In contrast, TGF-b2 1 cells were only depleted by an injection on day 8.5 of pregnancy. From the in vivo boosting of abortion rates following antibody administration on day 8.5, it appears that the TGF-b2 1 cells were also Vg1.1 1 and Vd6.3 1. Flow cytometry experiments showing both surface markers are present on the same cytokine 1 cell are necessary to confirm this hypothesis. It is conventional to think that an injection of antibody depletes its target cell population in vivo. It is possible that antibodies may render the cells invisible by capping their TcR or by sequestering the cells. It must also be considered that the effect of the antibodies on the abortion rate may have been caused by stimulation of the Vg1.1 1d6.3 1 cells rather than by their elimination or inactivation. The latter is difficult to test since, on binding to the target cell in vivo, antibody might initially stimulate and then eliminate cell function or might prevent detection of a functional cell in subsequent flow cytometry via capping of the TcR. If the tagged antibody were injected it might, in theory, be possible to detect and isolate labeled cells and study the kinetics of their cytokine response in order to directly answer the question. However, the effects on abortion rates seen with anti-Vg1.1 and anti-Vd6.3 were the same as those obtained using anti-d (9). Further, the predicted reduction in detection of Th1 and Th2/3 cytokine 1 Vd 1 lymphoid cell subsets corresponded to the known effects on abortion rates of directly antagonizing these cytokines in vivo and to the known pro- and anti-abortive effects of the Th1 and Th2 cytokines in vivo. These observations argue stongly that the effects we observed in the present study were due to depletion or inactivation of a specific subset of gd T cells. We have preliminary data using a monoclonal anti-gd which indeed appears to stimulate in vivo rather than deplete (Arck, Ferrick, and Clark, unpublished data), but the key experiments required to prove the point have not yet been done. We have taken TNF-a as representing a Th1 phenotype by virtue of its association with cellular immunity and inflammation (26), notwithstanding possible overlap in some Th2-type cells (27); enumeration of IFN-g 1 cells which contribute to abortion (9, 26) as well as IL-4 1 and IL-10 1 cells, which are also present and prevent abortions (5, 9), requires study and association with subsets of gd T cells. The proposed role of cytokines produced by decidual gd T cells is regulation of cytokine production by NK cells and macrophages (9) plus addition to the cytokine milieu generated by these and other cells in the uterus (5). A detailed quantitation of the cytokine contribution of different cell phe-

77

notypes will be required to create a complete picture of the events at the fetomaternal interface that lead to pregnancy success or abortion. Nevertheless, the data in this paper argue that Vg1.1d6.3 cells may represent a dominant force determining these events. What determines whether a Vg1 1 T cell will develop along a Th1 or Th2/3 pathway? For classical TcR ab T cells, IL-4, IL-10, progesterone, and possibly atypical oligomorphic major histocompatibility complex (MHC) antigens such as HLA-G, HLA-E, and HLA-F (analogous to MHC Class Ib antigens in the mouse) have been suspected to foster Th2 development, and the cytokines IL-12 and IFN-g favor Th1 development (27– 31). Previous studies have shown that CD8 1 immunosuppressor cells are present in the spleen and uterus prior to day 8.5 of pregnancy and are boosted by alloimmunization so as to produce either a 34-kDa (32) or 1.5-kDa (13) inhibitory factor; these alloantigen-activated CD8 1 cells appear to promote a Th1 3 Th2/3 cytokine shift in the decidua gd population and prevent abortions (9, 14, 32). About 2/3 of these CD8 1 cells express TcR ab (9), but whether the remainder express gd remains to be determined. As anti-abortive CD8 1 cells are spontaneously activated in CBA/J female mice pregnant by BALB/c, assuming only gd T cells can respond to the trophoblast, these CD8 1 cells would likely bear TcR gd (15, 16, 33, 34). They appear functionally distinct, however, from the Th1 Vg1.1d6.3 1 cells described in this paper because in vivo depletion of the CD8 1 cells in CBA/J 3 DBA/2 and in CBA/J 3 BALB/c increases the abortion rate whereas anti-d and anti-Vg1.1 given at the same time in pregnancy decrease the abortion rate (9, 11, 35, Fig. 2, Table 1). However, the decrease in abortion rate produced by anti-d (11), anti-Vg1.1, and anti-Vd6.3 does not exclude a CD8 1 cell with this TcR phenotype because eliminating the Th1 cytokine-producing Vg1.1 1 cell subset would make the presence or absence (depletion) of a regulatory CD8 1 cell subset irrelevant. The presence of V g1.1 on CD8 1 regulatory cells is an attractive hypothesis as such cells should be able to react to the trophoblast. The existence of interacting cell populations with different surface and cytokine phenotypes mentioned above has an added dimension of complexity which merits comment. AsialoGM1 1 NK cells and macrophages have been considered to be the main types of innate immune system effectors involved in causing abortions (2, 9, 26, 36). Their activity is modulated by the gd T cell populations we have described (9). But there can be overlap between NK and T cells—so-called NK-T or natural T cells (37). Usually the TcR on these cells has been ab, and activation induced IL-4 production (37). AsialoGM1 1 cells show a similar kinetic pattern of appearance in decidua in early pregnancy in the CBA/J 3 DBA/2 model as gd T cells, and we have

78

ARCK ET AL.

recently reported a significant overlap of these phenotypic subsets (using scatter gates set to capture all viable cells); gd-only, asialoGM1-only, and double-positive populations which express intracellular TNF-a and TGF-b2 have been identified (9, 38). Where there are both NK activation receptors and gd TcR on a cell, it becomes theoretically possible to activate a Th1 cytokine profile by engaging the NK cell receptor (NK1) and a Th2 reponse by a trophoblast ligand such as CD1b, MHC Class Ib (6peptide), or HSP (6peptide) interacting with the TcR gd (39 – 43). MHC interaction with killer inhibitory or killer activating receptors may also affect cytokine production (44). A Th1 . Th2/3 cytokine response by NK-gdT cells which has recently been demonstrated in CBA/J 3 DBA/2 pregnancy decidua (38) could bias subsequent ab and gd T cell development toward Th1 and away from Th2. This may explain in part the reduction in abortion rates following injection of anti-asialoGM1 antibody in the CBA 3 DBA/2 model (26, 36, 38). Cells with only gd, by contrast, make Th1 ! Th2/3 responses and bias toward Th2/3 development. The innate system, currently envisaged as being composed of NK cells, macrophages, polymorphonulcear leukocytes, and mucosal gd T cells, is thought to be “hard wired” so as to detect and to respond to “danger” via an appropriate cytokine response that determines what the classical immune system’s TcR ab T cells will do when activated by antigen (37). Danger is viewed as conserved surface molecules of pathogens (37, 45). We have added psychic stress to the list of what signals danger and have identified substance P release as the responsible neurotransmitter. One may speculate that in some mating combinations, the genetics are such that the trophoblast itself is seen as danger (42). Haig has cast reproduction in the metaphor of a genetic conflict between the mother and her offspring (46); the mother would not regard her own genetic material as undesirable, but rather, it must be the genetic material of the male (preferentially expressed in the trophoblast) which she must analyze to decide whether the invading parasite within her uterus should be treated more hospitably than other parasites and pathogens. Further studies using the CBA/J 3 DBA/2 mouse model should lead to major advances in understanding the ligands underlying a powerful and important process that has driven, and continues to drive, the evolution of species. ACKNOWLEDGMENTS

REFERENCES 1. 2. 3.

4.

5. 6. 7.

8. 9. 10. 11. 12.

13. 14. 15. 16. 17. 18. 19. 20. 21.

22.

23. 24.

25. 26.

This study was supported by grants from MRC Canada. Dr. Arck received a fellowship from the Deutsche Forschungsgemeinschaft.

27.

Kunz, H. W., Yuan, X-J, Salgay, S. K., Dixon-McCarthy, B., and Gill, T. J., 3rd., Am. J. Reprod. Immunol. 38, 158, 1997. Clark, D. A., Crit. Rev. Immunol. 11, 215, 1991. Clark, D. A., Arck, P. C., Jalali, R., Merali, F. S., Chaouat, G., Underwood, J., and Mowbray, J. F., Am. J. Reprod. Immunol. 35, 330, 1996. Mowbray, J. F., Jalali, R., Chaouat, G., Clark, D. A., Underwood, J., Allen, W. R., and Mathias, S., Am. J. Reprod. Immunol. 37, 421, 1997. Lin, H., Mosmann, T. R., Guilbert, L. J., Tutinpopitat, S., and Wegmann, T. G., J. Immunol. 151, 4562,1993. Wegmann, T. G., Lin, H., Guilbert, L., and Mosmann, T. R., Immunol. Today 14, 353, 1993. Chaouat, G., Meliani, A. A., Martal, J., Raghupathy, R., Elliot, J., Mosmann, T., and Wegmann, T. G., J. Immunol. 154, 4261, 1995. Krishnan, L., Guilbert, L. J., Russell, A. S., Wegmann, T. G., Mosmann, T. R., and Belosevic, M., J. Immunol. 156, 644, 1996. Arck, P. C., Ferrick, D. A., Steele-Norwood, D., Croitoru, K., and Clark, D. A., Am. J. Reprod. Immunol. 37, 492, 1997. Arck, P. C., Merali, F., Stanisz, A., Stead, R., Chaouat, G., Manuel, J., and Clark, D. A., Biol. Reprod. 53, 814, 1996. Arck, P. A., Ferrick, D. A., Steele-Norwood, D., Croitoru, K., and Clark, D. A., Am. J. Reprod. Immunol. 37, 87, 1997. Clark, D. A., Merali, F., Hoskin, D., Steele-Norwood, D., Arck, P. C., Murgita, R., Croitoru, K., and Hirte, H., Biol. Reprod. 56, 1351, 1997. Arck, P. C., Merali, F., Chaouat, G., and Clark, D. A., Cell. Immunol. 171, 226, 1996. Clark, D. A., Chaouat, G., Mogil, R., and Wegmann, T. G., Cell. Immunol. 154, 143, 1994. King, A., Gardner, L., and Loke, Y. W., J. Reprod. Immunol. 30, 67, 1996. Heyborne, K., Fu, Y. X., Nelson, A., Farr, A., O’Brien, R., and Born, W., J. Immunol. 153, 2918, 1994. Roark, C. E., Vollmer, M. K., Campbell, P. A., Born, W. K., and O’Brien, R. L., J. Immunol. 156, 2214, 1996. Heyborne, K. D., Cranfill, R. L., Carding, S. R., Born, W. K., and O’Brien, R. L., J. Immunol. 149, 2872, 1992. Pereira, P., Gerber, D., Regnault, A., Huang, S. Y., Hermitte, V., Couthino, A., and Tonegawa, S., Int. Immunol. 8, 83, 1996. Dent, A. L., Matis, L. A., Hooshmand, F., Widacki, S. M., Bluestone, J. A., and Hedrick, S. M., Nature 343, 714, 1990. Belles, C., Kuhl, A. K., Donoghue, A. J., Sano, Y., O’Brien, R. L., Born, W., Bottomly, K., and Carding, S. R., J. Immunol. 156, 4280, 1996. Clark, D. A., Flanders, K. C., Hirte, H., Dasch, J. R., Cocker, R., McAnulty, R. J., and Laurent, J. G., Biol. Reprod. 52, 1380, 1995. Lea, R. G., Flanders, K. C., Harley, C. B., Manuel, J., Banwatt, D., and Clark, D. A., J. Immunol. 148, 778, 1992. Markert, U., Arck, P. C., McBey, B. A., Manuel, J., Croy, B. A., Marshall, J. S., Chaouat, G., and Clark, D. A., Am. J. Reprod. Immunol. 37, 94, 1997. Krishnan, L., Guilbert, L. J., Wegmann, T. G., Belosevic, M., and Mosmann, T. R., J. Immunol. 156, 653, 1996. Clark, D. A., Chaouat, G., Arck, P. C., Mittruecker, H. W., and Levy, G. A., J. Immunol. 160, 545, 1998. Mosmann, T. R., and Sad, S., Immunol. Today 17, 138, 1996.

REGULATION OF ABORTION BY Vg1 T CELLS 28. 29. 30.

31. 32. 33. 34.

35. 36.

Parham, P., Am. J. Reprod. Immunol. 34, 10, 1995. Maeda, H., and Shiraishi, A., J. Immunol. 156, 73, 1996. Piccinni, M-P., Giudizi, M-G., Biogiotti, R., Beloni, L., Giannarini, L., Sampognaro, S., Parronichi, P., Manetti, F., Annunziato, F., Livi, C., Romagnani, S., and Maggi, E., J. Immunol. 155, 128, 1995. Magina, M., Fuji, T., Kozima, S., Okai, T., Shibathy, Y., and Taketani, Y., Am. J. Reprod. Immunol. 38, 79, 1997. Szerereday, L., Varga, L., and Szekeres-Bartho, J., Am. J. Reprod. Immunol. 38, 418, 1997. Chaouat, G., and Menu, E., Contemp. Topics Microbiol. Immunol. 222, 102, 1997. Chaouat, G., Clark, D. A., and Wegmann, T. G., in “Early Pregnancy Failure: Mechanisms and Treatment, Beard,” Royal College of Obstetrics and Gynecology, London, 1988. Athanassakis, I., Chaouat, G., and Wegmann, T. G., Cell. Immunol. 129, 13, 1990. Baines, M. G., Duclos, A. J., Antecka, E., and Haddad, E. K., Am. J. Reprod. Immunol. 37, 471, 1997.

79

37. Fearon, D. T., and Locksley, R. M., Science 272, 50, 1996. 38. Arck, P. C., Dietl, J., and Clark, D., Biol. Reprod. 60, 227, 1999. 39. Arase, H., Arase, N., and Saito, T., J. Exp. Med. 183, 2391, 1996. 40. Shawar., Vyas, J. M., Rodgers, J. R., and Rich, R. R., Annu. Rev. Immunol. 12, 839, 1994. 41. Blanchere, N. E., Li, Z., Chandawarkar, R. Y., Suto, R., Jarkaria, N. S., Basu, S., Udomo, H., and Srivastava, K., J. Exp. Med. 186, 1315, 1997. 42. Clark, D. A., Am. J. Reprod. Immunol. 41, 233, 1999. 43. Su, B. C., Strad, D., McDonough, P. G., and McDonald, J. F., Am. J. Obstet. Gynecol. 159, 1195, 1988. 44. Neiss, J. H., Steele-Norwood, D., Manuel, J., Markert, U., and Clark, D. A., Am. J. Reprod. Immunol. 41, 394, 1999. 45. Chien, Y. H., Jores, R., and Crowley, M. P., Annu. Rev. Immunol. 14, 511, 1996. 46. Haig, D., Am. J. Reprod. Immunol. 33, 226, 1966.