Local versus systemic control of numbers of endometrial T cells during pregnancy in sheep

Authors


Dr P. J. Hansen, PO Box 110920, University of Florida, Gainesville, FL 32611–0920, USA.

Summary

Pregnancy in sheep is associated with changes in numbers of specific T-lymphocyte populations in the uterine endometrium. These changes probably contribute to evasion by the conceptus of maternal immunological rejection and indicate a possible role for T cells in placental growth, parturition and post-parturient uterine defence against infection. The purpose of the present experiment was to evaluate the relative importance of systemic signals (i.e. those present throughout the uterus or from the circulation, including conceptus hormones secreted into the maternal blood) versus locally acting conceptus signals for regulating changes in numbers of endometrial lymphocytes during pregnancy. The approach taken was to surgically confine pregnancy to one uterine horn and compare differences in lymphocyte numbers between the two uterine horns as well as between both horns of pregnant ewes with those of ovariectomized ewes. As compared with ovariectomized ewes, there was a decline in numbers of CD45R+ lymphocytes within glandular epithelium and an increase in γδ T-cell number within the luminal epithelium. These changes occurred in both the pregnant and non-pregnant uterine horns of unilaterally pregnant ewes. Moreover, there were no significant differences in lymphocyte numbers between the two uterine horns of unilaterally pregnant ewes. Expression of CD25 was absent in tissues from both uterine horns. In conclusion, changes in numbers of endometrial lymphocytes during pregnancy, rather than due to locally acting signals of conceptus origin, are the result of hormonal signals of maternal or conceptus origin that either act directly on endometrial lymphocytes or stimulate the uterine endometrium to induce synthesis of regulatory molecules that affect lymphocyte dynamics.

Abbreviations
NK

natural killer cells

NPH

non-pregnant horn

OvUS

ovine uterine serpin

PH

pregnant horn

TGF-β

transforming growth factor-β.

Introduction

The uterus contains populations of T cells sufficient to induce graft rejection.1,2 Proper adjustments in uterine lymphocyte function to inhibit anti-fetal responses are therefore critical for pregnancy success; disruption of this regulation can lead to pregnancy loss.3 That certain lymphocyte populations are involved in anti-fetal responses is indicated by passive immunization experiments. For example, administration of antibody to asialoGM1 reduces abortion in the CBA/J × DBA/2 mouse abortion model4 and following activation of natural killer (NK) cells by injection of poly(I)·poly(C).5 Some lymphocyte populations, in contrast, may play a role that favours conceptus survival. In mice, for example, abortion can be increased by administration of anti-CD8 early in pregnancy6 and γδ T-cell subsets that produce transforming growth factor-β (TGF-β) have been identified in the mouse uterus that can suppress other lymphocytes.7,8 Also, NK cells have been implicated in the regulation of murine placenta growth.9

In sheep, there are pregnancy-associated changes in endometrial lymphocyte numbers that suggest differential regulation of specific lymphocyte populations by conceptus and maternal factors. The endometrial portion of the placentome, which is formed by the fusion of maternal caruncles and the fetal cotyledons, is nearly devoid of lymphocytes. Rather, lymphocytes are confined primarily to epithelial areas of the interplacentomal endometrium.10 Approximately half of all lymphocytes from the luminal epithelium of non-pregnant ewes are CD8+ CD45R γδ. The other lymphocytes are roughly equally divided into a population of CD8+ CD45R+ γδ cells and a population of CD8+ CD45R+ γδ+ T cells.11 During pregnancy, the number of non-granulated T cells decreases within endometrial glands, while numbers in the epithelium lining the lumen of the uterus initially decline and then become similar to values for non-pregnant ewes.12 Beginning in mid-pregnancy, in contrast, there is a large increase in the number and degree of granulation of a population of granulated lymphocytes in the lumen.11,12 These granulated T cells have been identified as γδ T cells.11 Given the roles of γδ T cells in regulation of cell growth13 and immunoglobulin A (IgA) secretion,14 these cells may promote fetal or placental development or contribute to anti-microbial defence postpartum. Recently, it has been reported that endometrial γδ T cells express perforin;15 perhaps they are cytotoxic and contribute to placental detachment from the endometrium during parturition.

Evidence suggests that these changes in lymphocyte numbers are regulated by progesterone and local conceptus signals. Treatment of ovariectomized ewes with progesterone decreased numbers of CD45R+ lymphocytes in glandular and luminal epithelium.16 In addition, progesterone induces endometrial secretion of ovine uterine serpin (OvUS),17 a protein that can inhibit lymphocyte proliferation and natural killer activity and can block immune responses in vivo.18 Other immunosuppressants produced in sheep include prostaglandin E2 (PGE2) from uterus and placenta,19 partially identified glycoproteins from placenta20 and TGF-β from the endometrium.21

The fact that γδ T cells increase in number during pregnancy despite the presence of several molecules with anti-proliferative activity may reflect their resistance to inhibition by these regulators. For example, OvUS inhibited concanavalin A-induced up-regulation of CD25 expression on γδ T cells but not on γδ+ T cells.22 In addition, endometrial γδ T cells may be activated by some locally acting signal from the conceptus. Using a unilaterally pregnant model, in which the conceptus was restricted to one uterine horn, there was an increased expression of activation antigens such as CD25 on γδ T cells from the endometrium ipsilateral to the conceptus.23 This finding, which is consistent with data in humans that the placenta can activate γδ T cells,24 has been called into question by recent findings that CD25 expression was absent from endometrial lymphocytes.25

The purpose of the present experiment was to evaluate the relative importance of systemic signals (i.e. those present throughout the uterus or from the circulation, including conceptus hormones secreted into the maternal blood) versus locally acting conceptus signals for regulating changes in lymphocyte numbers in the endometrium during pregnancy. The approach taken was to confine pregnancy surgically to one uterine horn and compare differences in lymphocyte number between the two uterine horns (where only the horn ipsilateral to the conceptus would be exposed to locally acting conceptus factors) and between both horns of the unilaterally pregnant ewes with uteri of ovariectomized ewes (where both horns of pregnant animals would be exposed to changes in circulating hormones of pregnancy as well as uterine-derived factors induced by those hormones while the uterus of ovariectomized animals would be exposed to neither).

Materials and methods

Materials

Hybridoma cells producing antibodies to CD8 (7C2), CD45R (20–96) and γδ T-cell receptor (TCR) (86D) were obtained from the European Collection of Animal Cell Culture (Salisbury, UK) and were maintained at the Hybridoma Core Laboratory, University of Florida Interdisciplinary Center for Biotechnology Research. Antibody to CD8 was purified using a Hi-Trap Protein G Sepharose column (Pharmacia & Upjohn, Kalamazoo, MI). The antibodies towards CD45R and γδ T cells were used as ascites fluid. Monoclonal antibody towards CD25 (clone 155.2) was obtained as an ascites fluid from the Centre for Animal Biotechnology, University of Melbourne, Parkville, Victoria, Australia. The immunohistochemistry kit (HistoScan Universal Monoclonal Detector Kit) was obtained from Biomeda (Foster City, CA), Tissue-Tek OCT Compound was from Miles Diagnostic (Elkhart, IN) and mouse IgG1, control mouse ascites fluid (clone NS-1) and normal goat serum were from Sigma (St Louis, MO). Other reagents were from Sigma or Fisher Scientific (Pittsburgh, PA).

Animals

Rambouillet-type ewes, maintained on a diet of Bermuda hay ad libitum, were used for the experiment. Six ewes were unilaterally ovariectomized via midventral laparotomy and a single uterine ligation was made on the side ipsilateral to the side of the ovariectomy.26 After approximately 3 weeks, ewes were bred to a fertile ram at the next oestrus. Pregnant ewes were slaughtered via captive-bolt stunning and exsanguination at day 140 of pregnancy and reproductive tracts were removed. Four additional ewes were bilaterally ovariectomized and a uterine ligation placed around each uterine horn. Ewes were slaughtered 8 months after ovariectomy. These animals served as controls that were not exposed to any ovarian steroids that might contribute to systemic effects of pregnancy on endometrial lymphocyte number. Ligations were placed around the uterus of ovariectomized animals to parallel the presence of a ligation in pregnant animals.

Tissue collection and embedding

Tissue samples were excised from randomly chosen regions of the intercaruncular endometrium of all ewes. For pregnant ewes, tissues were collected from the ligated, non-pregnant horn (NPH) and from the pregnant horn (PH). Tissue samples (∼3 mm3) were frozen in aluminium boats (1 cm3) containing OCT freezing medium using isopentane cooled with liquid nitrogen. Tissue blocks were stored at −70° until sectioning was performed.

Histochemistry

Immunohistochemistry was performed on tissues from two separate blocks from each uterine horn (pregnant ewes) or uterus (ovariectomized ewes). Frozen blocks of tissue were sectioned on a cryostat at 5-µm thickness. Sections were placed on polylysine-coated slides and fixed with 95% ethanol. Sections were air dried onto the slide and then rehydrated with phosphate-buffered saline (PBS) containing 1% (w/v) normal goat serum (PBSG) and kept at 4° until processed. All histochemical steps were performed at room temperature and slides were rinsed with PBSG and blotted dry between each step. Slides were treated with PBSG, pH 7·4 containing 0·6% (w/v) H2O2 for 3 min to eliminate endogenous peroxides. Tissues were then treated with conditioner from the Biomeda kit for 5 min. Sections were incubated for 6 hr with antibody to CD8 (6 µg/ml), CD45R (1 : 800), γδ T cells (1 : 300) or CD25 (1 : 500) diluted in PBS. Mouse IgG1 (5 µg/ml) and control mouse ascites fluid (1 : 300) were used as negative controls. Other steps, including incubation with anti-mouse IgG linked to biotin (30 min), streptavidin-peroxidase (30 min) and 3-amino-9-ethyl-carbazole (10 min) were conducted using reagents supplied in the kit. Slides were rinsed with deionized water for several minutes and carefully blotted before applying mounting medium and coverslips.

Quantification of positive cells

Slides were observed at ×400 magnification in a bright-field microscope containing a graticule in one eyepiece. The number of positive cells was counted in three tissue types: luminal epithelium, glandular epithelium and the stroma. For each tissue type, three randomly selected sites were chosen for morphometric analysis. The number of positive cells in 30 squares of the graticule (each square having a surface area of 625 µm2) were counted. For epithelium, squares were chosen for counting so as to ensure that the entire area of the squares was within the epithelium. For stroma, squares were chosen for counting so as to evaluate a cross-sectional area extending from the lumen to the deep stroma.

Statistical analysis

Data were analysed by least square analysis of variance using the General Linear Models procedure of SAS.27 For the analysis, each uterine horn of the pregnant ewes was considered as an independent experimental unit. Data were analysed separately for each of three sites (luminal epithelium, glandular epithelium and stroma). The mathematical model included effects of treatment (ovariectomized; NPH of pregnant ewes; PH of pregnant ewes) and error. Orthogonal contrasts were used to partition variance due to the overall effect of pregnancy (ovariectomized versus NPH + PH) and the effect of local presence of the conceptus (NPH vs. PH).

Results

Distribution and appearance of lymphocytes

For CD45R, γδ T cells and CD8, positive cells were located predominantly in the luminal and glandular epithelium (Fig. 1) and only a few scattered lymphocytes were present in the stroma. There were no cells positive for CD25 detected in endometrium although the antibody did react with some cells in the spleen and a proportion of mitogen-activated peripheral blood lymphocytes (results not shown).

Figure 1.

Immunohistochemical localization of T cells in the sheep endometrium. The first three panels are low magnification images that illustrate the distribution of cells staining for CD45R (a), γδ (b) and CD8 (c). A section of endometrium in which antibody was replaced by control mouse IgG1 is shown in (d). Panels (e)–(h) are higher magnification images to demonstrate effects of pregnancy on numbers of CD45R+ cells in the glandular epithelium: (e) is from an ovariectomized ewe and (f) is from a pregnant ewe; and on numbers of γδ T cells in luminal epithelium: (g) is from an ovariectomized ewe and (h) is from a pregnant ewe.

Pregnancy was associated with an increase in the thickening of the epithelial lining of the endometrial glands and lumen (compare Fig. 1e with Fig. 1f and Fig. 1g with Fig. 1h) but the distribution of lymphocytes was unaffected by pregnancy status. In general, the appearance of positive-staining cells was similar for all tissues except that intraepithelial CD45R+ cells from ovariectomized ewes stained less intensely with the reaction product than did cells from either the PH or NPH of pregnant ewes (Fig. 1e versus Figure 1f). Except for CD45R+ cells from ovariectomized ewes, staining intensity of CD8+ cells was lower than for the other two markers.

Local and systemic effects on numbers of lymphocytes

Data are shown in Fig. 2. The number of CD45R+ lymphocytes in luminal epithelium did not differ between the three uterine types. In contrast, there was a reduction (P < 0·0001) in numbers of CD45R+ cells in the glandular epithelium of both uterine horns from pregnant ewes as compared to ovariectomized ewes. Values were 1·69 ± 0·15 cells/625 µm2 for ovariectomized ewes versus 0·26 ± 0·12 cells/625 µm2 for PH and 0·46 ± 0·12 cells/625 µm2 for NPH. The difference between PH and NPH was not significant.

Figure 2.

Numbers of endometrial lymphocytes. Data are least-squares means ± SEM for endometrial tissue collected from ovariectomized ewes (OVX) and from the non-pregnant horn (NPH) and pregnant horn (PH) of unilaterally pregnant ewes at day 140 of pregnancy.

Numbers of γδ T cells in the luminal epithelium were higher (P = 0·05) for both horns of pregnant ewes as compared to the ovariectomized ewes and there was no difference between PH and NPH. Numbers of cells were 0·28 ± 0·15 cells/625 µm2 for ovariectomized ewes, 0·70 ± 0·13 cells/625 µm2 for PH and 0·62 ± 0·13 cells/625 µm2 for NPH. Numbers of γδ T cells in the glandular epithelium were unaffected by uterine type.

Although there was a tendency for numbers of CD8+ cells in luminal epithelium to be higher in pregnant ewes, there was no significant effect of uterine type on numbers of CD8+ cells in either luminal or glandular epithelium.

Discussion

As compared to ovariectomized ewes, the pregnancy-associated decrease in CD45R+ cells in glandular epithelium and increase in luminal γδ T cells mirror earlier observations in pregnant ewes.12 Present results indicate that these changes, rather than due to locally acting conceptus factors, are instead the result of hormonal signals of maternal or fetal origin that either act directly on endometrial lymphocytes or stimulate the uterine endometrium to induce synthesis of regulatory molecules that affect lymphocyte dynamics. This is shown by the fact that pregnancy-associated changes in lymphocyte populations occurred in both the pregnant and non-pregnant uterine horn of pregnant ewes and there were no differences in numbers of lymphocytes between the pregnant and non-pregnant horns.

The pregnancy-associated decrease in intraepithelial CD45R+ cells that occurred in both uterine horns may be largely the result of the immunosuppressive actions of progesterone. Uterine fluid is rich in progesterone-induced immunosuppressive substances28 and progesterone treatment of ovariectomized ewes decreased numbers of CD45R+ cells in glandular and luminal epithelium.16 Chief among these immunosuppressive molecules is OvUS, which can inhibit a variety of T-cell- and NK-cell-mediated phenomena.18 In addition, PGE2 and TGF-β, both of which can inhibit lymphocyte proliferation in sheep,29,30 are produced by the endometrium of pregnant ewes.19,21 The failure of a decrease in CD45R+ cells in the luminal epithelium, despite the fact that progesterone treatment reduced numbers of CD45R+ cells in both glandular and luminal epithelium,16 probably reflects the increase in γδ T cells in the lumen, since these cells are CD45R+.11 It was also observed that the staining intensity of CD45R+ cells was greater for pregnant ewes than ovariectomized ewes, suggesting that there is up-regulation of CD45R expression on at least some populations of CD45R+ cells.

It was hypothesized that the increase in numbers of γδ T cells seen in the luminal epithelium duing mid- and late-pregnancy12 represents stimulatory effects of local signals from the conceptus because trophoblast can activate γδ T cells in humans24 and expression of CD25 and other activation markers on intraepithelial γδ T cells in luminal epithelium of unilaterally pregnant ewes was higher for cells from the uterine horn ipsilateral to the conceptus.23 This hypothesis is not supported by the data in two respects. First, immunohistochemistry failed to identify cells in the endometrium that stained for CD25. This result is similar to another recent report of lack of expression of CD25 on endometrial γδ T cells.25 Thus, the detection of CD25 on endometrial γδ T cells by Liu et al.23 may have represented up-regulation of expression after cells had been harvested. The present data also failed to support the hypothesis of local stimulation of endometrial lymphocyte populations because there were more luminal γδ T cells in both the non-pregnant and pregnant uterine horns of unilaterally pregnant ewes as compared to endometrium from ovariectomized ewes. Moreover, the number of γδ T cells in the luminal epithelium was similar for the pregnant and non-pregnant uterine horns.

Thus, as for CD45R+ cells, the signal for the increase in luminal γδ T-cell number is likely to be hormonally regulated. Maternal or conceptus hormones could act directly on luminal γδ T cells or induce synthesis of locally acting endometrial signals involved in γδ T-cell growth such as interleukin-7 and interleukin-15.31,32 Interestingly, the increase in γδ T-cell numbers seen in the luminal epithelium was not observed in glandular epithelium. This indicates that either the stimulatory signal is absent in endometrial glands, the population of γδ T cells in the glandular epithelium are functionally distinct from those in the lumen, or that other regulatory signals in the glands block activation of γδ T cells.

The absence of a difference between pregnant and ovariectomized ewes in numbers of CD8+ cells may reflect the fact that CD8 is expressed by both γδ T cells and αβ T cells11 and increases in some CD8+ subpopulations may be offset by decreases in other subpopulations. In addition, some pregnancy-induced changes in CD8+ populations in endometrium may not be apparent using immunohistochemistry because only a portion of endometrial CD8+ cells appear to be detected by immunohistochemistry. Indeed, although analysis by flow cytometry suggests that all CD45R+ and γδ T cells express CD8,11 the numbers of CD8+ cells determined by immunohistochemistry were less than for CD45R+ cells (present study; see also refs. 10, 16). Although not significant, there was a tendency for CD8+ cells to increase in the luminal epithelium in unilaterally pregnant ewes, particularly in the pregnant horn. These increases may reflect changes in numbers of γδ T cells and, perhaps, activation of a subpopulation of CD8+ cells by some local conceptus factor.

Changes in lymphocyte number in the endometrium could represent regulation of lymphocyte proliferation, differentiation, migration, or apoptosis. Several uterine-derived molecules that can inhibit lymphocyte proliferation have been described in the sheep including OvUS,18 PGE219 and TGF-β.21 In addition, suppressor cells in the ovine uterus that have characteristics distinct from T cells33 are increased in number by both oestradiol-17β and progesterone.34 The possibility that endometrial lymphocyte dynamics involve changes in the rate of differentiation in situ exists because, at least in the human, lymphocyte differentiation markers (recombinase-activating genes 1 and 2) are expressed in the endometrium.35 A recent paper has demonstrated migration of lymphocytes into the ovine uterine endometrium.36 The degree to which migration occurs may be low, however, since there are few lymphocytes resident in the endometrial stroma through which lymphocytes migrating to and from the vasculature must traverse. Yet another potential explanation for the pregnancy-associated decrease in lymphocyte numbers (in particular the decrease in CD45R+ cells in glandular epithelium) may be the hypertrophy of endometrial glands during pregnancy. The increase in volume of the glands would result in a corresponding reduction in the density of lymphocytes within the epithelium if the rate of lymphocyte growth was lower than the rate of growth of the endometrial epithelium.

In conclusion, the present experiment points out the importance of hormonally derived signals for regulating lymphocyte numbers during pregnancy. Further work should focus on identification of these endocrine signals, their origin (fetal versus maternal) and of the local, endometrial cytokine and growth factor networks through which they regulate endometrial lymphocytes.

Acknowledgments

The authors thank Fabiola Paula-Lopes, for technical advice on histochemistry; the entire laboratory for assistance during surgery; Gary Barcham, Els Meeusen and the Centre for Animal Biotechnology, University of Melbourne, Parkville, Victoria, for donation of anti-CD25; Dean Glicco for animal care; and the University of Florida Interdisciplinary Center for Biotechnology Research for technical support including Scott Whittaker and the Electron Microscopy Laboratory for help with cryostat sectioning and computer imaging and the Hybridoma Core Laboratory for production of antibodies. Research was supported in part by USDA NRI 96–35203–3304 and a grant from the Florida Milk Checkoff Program. This is Journal Series Number R-07512 of the Florida Agricultural Experiment Station.

Ancillary