TIM-3 and TIM-1 Could Regulate Decidual γδTCR Bright T Cells during Murine Pregnancy

Pregnancy is an immunological enigma where paternal antigens are present at the fetomaternal interface. What regulates the local immunotolerance, which is necessary to prevent rejection of the conceptus, is still under strong investigation. Gamma/delta T cells are believed to play a role in the local regulation of this immunotolerance towards the semiallogenic fetus. Gamma/delta T cells from the uterus and spleen of pregnant and nonpregnant mice were analyzed by flow cytometry. We confirmed that the rate of γδT cells in the decidua increases during murine pregnancy and half of decidual γδT cells are CD4+. Furthermore, we found a unique association of CD4 or CD8 coreceptor expression with their γδTCR intensity, where in all investigated groups CD4- or CD8-positive γδT cells seemed principally to be γδTCRdim. In addition, compared to peripheral γδT lymphocytes, a greater proportion of decidual γδT cells expressed the cytotoxic marker CD107a and markers of Th1 or Th2 polarization (TIM-3, TIM-1), where decidual γδTCRbright cells were characterized by high TIM-3 and TIM-1 receptor expression. On the other hand, no difference in the expression of CD160, a receptor with dual function affecting cytotoxicity and T cell inhibition, was detected. Within lymphocytes expressing CD107a, TIM-1, or CD160, the rate of γδT cells was significantly higher in the decidua. According to our results, cytotoxic potential of decidual γδTCRbright cells could be regulated by TIM-3 ligation, while the TIM-1 receptor seems to be able to influence the Th1-Th2 balance at the fetomaternal interface. These mechanisms could play a part in the active maternal immunotolerance towards the fetus, allowing an efficient protection against pathogens during healthy murine pregnancy.


Introduction
In pregnancy, a distinct balance of T helper-(Th-) mediated immunity is crucial to provide protection against pathogens, while mediating tolerance towards the semiallograft [1][2][3]. It is still unclear how this balance is adjusted exactly. T cell immunoglobulin domain and mucin domain-containing molecule-(TIM-) 3 is expressed on Th1 cells [4]. Engaging galectin-9 (Gal-9) probably modulates the balance of Th1 and Th2 cytokines, by downregulating the Th1 response, which is crucial for allograft tolerance [5,6]. Together with programmed cell death protein-(PD-) 1 and other inhibitory receptors, TIM-3 mediates CD8+ T cell exhaustion [7]. TIM-3 can be found on activated human natural killer (NK) cells, where it suppresses the NK cytotoxicity [8]. Over 60% of decidual NK cells are TIM-3+ and produce high levels of Th2 cytokines [9], which seem to be essential for a successful pregnancy [10]. In contrast, TIM-1, which is a receptor for phosphatidylserine and TIM-4, is mainly expressed on Th2 cells and has an activating function [11][12][13][14]. However, the results of this activation appear to be different according to the type of ligation [12]. Xiao et al. described that proinflammatory as well as antiinflammatory cytokine production can be triggered by distinct types of anti-TIM-1 monoclonal antibodies [15]. In addition, it has been also shown that the engagement of agonistic anti-TIM-1 monoclonal antibodies in CD4+/forkhead box protein 3 (FOXP3)+ cells leads to downregulation of FOXP3 [16], which is a commonly used marker for regulatory T cells [17]. Despite the evidences about the influence of TIM-1 in allograft tolerance [18], the presence and role of TIM-1 receptor at the fetomaternal interface are poorly investigated.
Immune cells must recognize maternal and paternal antigens of the trophoblast, which leads to the regulation of Th1-Th2 cytokine balance. CD160 could participate in this immunological signaling, because it is capable of binding to classical and nonclassical major histocompatibility complex (MHC) class I molecules, like human leukocyte antigen-(HLA-) A, HLA-B, HLA-C, HLA-E, and HLA-G [19][20][21]. Beside γδT cells, all intestinal intraepithelial T lymphocytes and some γδT cells and CD56dim NK cells express CD160 [19,[22][23][24], which is a marker for cytotoxicity on NK cells [19] and could also enhance NK cell-derived Th1 cytokine production [21]. According to our present knowledge, herpes virus entry mediator (HVEM) is a second ligand for CD160, which could deliver cell survival promoting inhibitory signaling [24][25][26]. Similarly to HLA-C, HLA-E, and HLA-G molecules, HVEM is also expressed in the placenta [27][28][29][30]. Therefore, CD160 could be able to regulate cell survival and function during implantation or maternal tolerance.
Gamma/delta T cells, in general, have been the scope of intense investigations over the last decades. They are able to react on a MHC-non-restricted manner and are strong in cytotoxic potential and cytokine production, and they expand during pregnancy [31]. Thus, γδT cells presumably play an important role in shaping the pro-/anti-inflammatory balance of the fetomaternal interface. Indeed, human decidual γδTCR+/CD56dim cells show TGFβ and IL-10 immunosuppressive cytokine profile on the mRNA level [32]. Afterwards, Fan et al. did confirm that around half of decidual γδT cells release these Th2 cytokine proteins and that decidual γδT cell-derived IL-10 enhance human trophoblast proliferation and invasion [33].
However, γδT cells seem to play a significant role in the maintenance of healthy pregnancy; the exact mechanisms are still poorly understood. In this study, we focused on the phenotypical aspects of γδT cells at the fetomaternal interface. Since healthy human decidual samples are hardly available, a mouse model was necessary to acquire in vivo data. Therefore, we implemented a mouse model with the objective of investigating the expression of CD4, CD8, CD107a, TIM-3, TIM-1, and CD160 within γδT cell populations of the uterus and the spleen in nonpregnant and pregnant mice. We aimed to examine the effect of pregnancy on the ratio of positive cells and the cell surface density of the above markers.

Materials and Methods
2.1. Animal Model. In our mouse model, the same experimental setting was used as in our previous work [34], which is described briefly in each chapter below.
Twenty-four pathogen-free BALB/c mice (2 months old) were kept on a 12 h light/dark cycle at 20-22°C and 40-60% humidity and were fed with standard food pellets and tap water. For fertilization, potential mates were paired up every evening and examined for the presence of the copulatory plug in the next morning. As soon as a copulatory plug was detected, it was considered as gestation day 0.5. Gestating female mice (n = 14) were killed on gestation day 14.5 by cervical dislocation, and the spleen and the uterine horns were harvested aseptically afterwards. Nonpregnant female mice (n = 10) were used in this experiment as well and were killed at the age of 3 months. The mice were purchased from Pécs Experimental Central Animal Laboratory. Animal housing, care, and application of experimental procedures were in accordance with institutional guidelines under approved protocols (No. BA02/ 2000-6/2012, National Food Chain Safety and Animal Health Control Office of the Government Office of County Baranya). Concerning animal welfare, all efforts were made to minimize suffering.

Isolation of Mononuclear
Cells from the Decidua. Following our protocol, described previously [34], after the mice were killed, the abdomen was carefully opened and access to the uterus was gained by pushing intestinal tissue to the side. The uterus was then removed by surgical cuts at the cervix and the ovaries. Then, the uteri were fixed to a clamp at the cervix, which gave enough stability and allowed carefully cutting along the uterine horns. Then, the decidua was separated from the placenta disc under a dissecting microscope. The average number of deciduae per mouse was 5.5. Isolated deciduae were pooled, sliced with scissors, and digested with type IV collagenase (Sigma-Aldrich) at 37°C for 30 minutes. Thereafter, the isolated cells were collected in a fresh tube through a 70 μm nylon cell strainer (BD Biosciences). Subsequently, cells were washed in RPMI 1640 medium (Lonza) supplemented with penicillin (1 × 10 5 U/l, Lonza) and streptomycin (0.05 g/l, Lonza). The supernatant was aspirated, and the pellet was resuspended in PBS and filtered via a 40 μm nylon cell strainer (BD Biosciences). Finally, the isolated cells were resuspended in RPMI 1640+10% fetal bovine serum (FBS) (Gibco).

Isolation of Mononuclear Cells from the Endometrium.
Endometrial mononuclear cells from the nonpregnant endometrium were isolated as described previously [34]. Here, the uterus was removed from the abdominal cavity, sliced with scissors, and digested with type IV collagenase at 37°C for 30 minutes. The isolated cells were collected in a fresh tube, filtered through a 70 μm nylon cell strainer, and washed in RPMI 1640 medium supplemented with penicillin and streptomycin. The supernatant was aspirated, and the pellet was resuspended in PBS and filtered via a 40 μm nylon cell strainer. Finally, the isolated cells were resuspended in RPMI 1640+10% FBS.

Isolation of Mononuclear
Cells from the Spleen. According to our previous protocol [34], the spleens were cut into small pieces by a sharp sterile surgical knife, immersed in 2 ml PBS, and pushed gently through a 70 μm nylon cell strainer with a syringe plunger. After that, cells were washed in PBS. Supernatant was aspirated, and the pellet was resuspended in PBS and filtered again via a 40 μm nylon cell strainer. Then, mononuclear cells were separated by Ficoll-Paque gradient (GE Healthcare). Isolated cells were collected and resuspended in RPMI 1640 medium supplemented with penicillin, streptomycin, and 10% FBS.

Fluorochrome Labelling and Flow Cytometric Analysis.
The isolated mononuclear cells (10 6 cells in 100 μl PBS/tube) were incubated with the fluorochrome-labelled monoclonal antibodies at room temperature for 30 min. Hereafter, the cells were washed with PBS and resuspended in a 1% paraformaldehyde (PFA) solution in PBS and stored darkly at 4°C until fluorescence-activated cell sorting (FACS) analysis, performed using a CyFlow® Space flow cytometer (Sysmex Partec GmbH, Görlitz, Germany) equipped with FloMax 2.60 for data acquisition. 2.6. Activation for CD107a Functional Assay. The cells were incubated for 4 h (37°C, 5% CO 2 ) with the fluorochromelabelled anti-CD107a antibodies in RPMI 1640 medium, ?supplemented with penicillin, streptomycin, and 10% FBS; furthermore, ionomycin (1 μg/ml) (Sigma-Aldrich) and phorbol myristate acetate (PMA) (25 ng/ml) (Sigma-Aldrich) were used to activate the cells. Before labelling the cells with the other fluorochrome-labelled monoclonal antibodies, the cells were washed and resuspended in PBS. Finally, cells were fixed in 1% PFA and evaluated by FACS as described in the previous paragraph.
Despite the possibility that the expression of CD160 might change upon activation [35], in our protocol, as described previously, a short stimulation of 4 hours with PMA/ionomycin did not influence the surface expression of CD160 significantly [34].

Statistics.
For comparison of data, we performed multiple types of statistical analysis performed using Microsoft Excel version 15.15. We used paired-samples t-test to evaluate the relation of two corresponding data from the same mouse and independent-samples t-test to compare pregnant and nonpregnant mice. The results presented in the text represent the mean ± SEM of the corresponding set of data. Differences were determined as significant, if the p value was equal to or less than 0.05.

Expression of γδTCR on Lymphocytes in the Uterine and Splenic Samples in Pregnant and Nonpregnant
Mice. This study is focused on different subsets of γδT cells in pregnant and nonpregnant mice. The ratio of γδT cells to other lymphocytes depending on the sampled tissue was foremost investigated. Therefore, mononuclear cells, gated on CD45+, of the decidual tissue in pregnant and endometrial tissue in nonpregnant mice were analyzed. Lymphocytes from pregnant and nonpregnant splenic samples were used as control groups. The ratio of γδT cells in uterine samples is significantly higher compared to that in the peripheral control (15 25 ± 0 03 in decidual and 4.88 ± 0 01 in endometrial vs. 2 01 ± 0 00 in pregnant and 1 58 ± 0 00 in nonpregnant splenic samples, p ≤ 0 02). Furthermore, decidual lymphocytes express γδTCR significantly more often than lymphocytes of endometrial samples (p ≤ 0 02) (Figure 1).

CD107a and TIM-1, TIM-3, and CD160 Expression
Profile of γδT Cells Depending on the Isolated Tissue's Origin in Murine Pregnancy. To get a clearer idea of the function of γδT cells at the fetomaternal interface, we investigated their potential cytotoxicity by assessing their CD107a expression upon stimulation [36]. This marker of degranulation is significantly more often expressed by γδT cells of the decidua than the spleen (40 75 ± 3 66 in the decidua vs. 33 65 ± 3 27 in the spleen, p ≤ 0 05). We also wanted to examine the cells for their immunoregulatory function. Thus, we analyzed the expression of TIM-3, TIM-1, and CD160 on γδT cells. The rate of TIM-3-expressing cells among γδT cells in the decidua is slightly but significantly higher than that in the spleen (19 22 ± 1 92 in the decidua vs. 15 36 ± 1 62 in the spleen, p ≤ 0 01). One-third of decidual γδT cells show TIM-1 positivity, which is six times more often than the splenic γδT cells do (28 45 ± 4 57 in the decidua vs. 4 25 ± 0 58 in the spleen, p ≤ 0 01). CD160 is rarely expressed on γδT cells, and we did not find any significant alterations in the rate of positive cells for this receptor depending on the location of the cells (1 27 ± 0 38 in the decidua vs. 0 81 ± 0 35 in the spleen) (Figure 4(a)).
Furthermore, the expression of functional markers is also connected to the intensity of γδTCR expression on the respective cells. When gating the γδTCRdim and γδTCRbright subsets in pregnant mice, we found a significantly higher rate of CD107a+ cells among decidual γδTCRdim lymphocytes compared to the splenic γδTCRbright subset (48 0 ± 3 5 in decidual γδTCRdim cells vs. 23 0 ± 5 3 in splenic γδTCRbright cells, p ≤ 0 05), while there was no significant difference detected between the organ-related γδTCRdim or γδTCRbright subsets.
Interestingly, in the decidua, the rate of TIM-1+ cells was almost three times higher among the γδTCRbright subset (56 0 ± 3 9) than among the γδTCRdim one (18 3 ± 1 6, p ≤ 0 01). Although a similar tendency could be observed in the splenic samples, it did not reach the level of significance.
Finally, we could not detect any significant difference in the rate of CD160+ cells between the investigated groups ( Figure 4(c)).

The Role of γδT Cells in Specific Functional Phenotypes.
In this part of this study, we investigated the ratio of γδT cells in different functional subsets of lymphocytes. The portion of γδT cells among CD107a-expressing lymphocytes is six times greater in the decidua than in the spleen (59 05 ± 5 31 in the decidua vs.

Discussion
Gamma/delta T cells seem to play a part in the maintenance of a healthy pregnancy. Their effect appears to be primarily localized to the fetomaternal interface. In this study, we focused on the prevalence and the possible functions of γδT cells in murine pregnancy.
In terms of CD4 expression, decidual γδT cells are outstanding: approximately half of them are CD4+, which is twice as much as in the endometrium, pregnant spleen, or nonpregnant spleen. In contrast, CD8 expression is similar in all groups. Due to the design of the study, a possible double positivity could not be ruled out. Here, we demonstrate for the first time that CD4 and CD8 receptor coexpression with γδTCR depends on the intensity of the γδTCR expression. We were able to describe two distinct populations: a γδTCRdim one, which was mainly CD4+ or CD8+ and a γδTCRbright one, which was predominantly negative for CD4 and CD8. The presence of these γδT subpopulations is particularly characteristic among resident γδT cells. Hence, we assume that subsets within the γδT cell population with altering functional characteristics should exist in the decidua. Those subsets might show similar variations like CD56dim and CD56bright NK cells. Our findings are in contrast with the results of Mincheva-Nilsson et al., who described human decidual γδT cells as a CD4/CD8 double-negative population, although this result was based on a single human donor flow cytometric experiment [37]. Further investigations are necessary to clarify this contradiction. However, this difference could also be traced back to the discrepancy in the investigated species, although CD4+ γδT cells have been described in other studies both in mice and in humans [38]. Murine CD4+ γδT cells secrete high levels of Th2 cytokines [39]. Furthermore, after TGFβ stimulation, naïve circulating γδT cells mature into CD25+/FOXP3+ γδT cells [40,41]. Due to their Treg-like nature, it is presumable that those decidual γδT cells express CD4, too. Additionally, it is possible that a Th2 subpopulation exists in addition to a Treg subpopulation within the γδT cells at the fetomaternal interface.
Although γδT cells with their significant cytokine production are able to form the local Th1-Th2 balance, which is an indispensable part of the immunotolerance towards the semiallogenic fetus, the precise regulation of this γδT cell function at the fetomaternal interface is still not well understood. Here, we described that the prevalence of γδT cells expressing TIM-3, commonly expressed by Th1 cells, is slightly higher in the decidua than in the spleen. It suggests that upon binding its ligand Gal-9, which is expressed in the placenta, TIM-3 could inhibit the Th1 function on γδT cells [42]. Interestingly, the prevalence of TIM-1+ γδT cells is six times higher in the decidua than in the spleen. TIM-1 is mainly found on Th2 cells and about half of the TIM-1positive lymphocytes are γδT cells in the decidua, which is twice as many as in the spleen. When investigating the expression of functional markers within the newly described γδTCRbright and γδTCRdim subsets, it appears that the above immune checkpoint molecules are mainly expressed on the γδTCRbright subset. We assume that those cells could be regulated by TIM-3 and TIM-1, while the impact of these markers seems to be less pronounced in the γδTCRdim subset, which might be rather controlled by CD4 or CD8. The fact that the rate of TIM-1+ cells in the decidual γδTCRbright subset is so outstanding might be correlated to their activity at the fetomaternal interface, since TIM-1 is known to be upregulated upon activation [43]. These findings indicate a shift in the Th1-Th2 profile of γδT cells towards the anti-inflammatory Th2 side in the decidua compared to the spleen. This is in line with the previously described Th2 bias in the decidua and the findings of McGrath et al., who proved that TIM-4, a natural ligand of TIM-1, promotes tolerance at the fetomaternal interface [44]. Approximately one-third of decidual γδT cells express CD107a, and the density of its surface expression is significantly higher on decidual γδT cells than on splenic γδT   cells. This phenomenon could be due to the character of the γδTCRbright subset, which had significantly higher CD107a expression intensity than the γδTCRdim one (data not shown). The cytotoxic activity of the decidual γδTCRdim and γδTCRbright cells could be regulated by CD8 or TIM-3 molecules, respectively. Among CD107aexpressing lymphocytes, over 50% are γδT cells in the decidua, whereas in the spleen, only 10% of these lymphocytes are γδT cells. Our results indicate that decidual γδT cells might conserve their cytotoxic potential in the microenvironment of the fetomaternal interface, which could be useful in the protection against pathogens, as well as in the process of remodeling of the uterine tissue during placentation. In summary, it is easy to assume that diverse phenotype-related functions can be fulfilled by decidual γδT cells. The activity of each decidual γδT cell subsets is largely determined by their coreceptors. CD160 plays a special role as a coreceptor, since both activation and inhibition can be mediated upon its binding. In CD4+ T cells, CD160 acts as an inhibitory coreceptor [24], whereas its signaling activates CD8+ T cells [45]. In our model, only a very small percentage of γδT cells expressed CD160. Thus, we assume that CD160 plays a minor role in the regulation of γδT cell function. Whether the activation upon CD160 binding influences γδT cells at the fetomaternal interface or γδT cells can upregulate CD160 under certain conditions at different time points during pregnancy is so far unknown.

Conclusions
Gamma/delta T cells have been suggested to promote tolerance towards the semiallogenic fetus in pregnancy. This is in line with our findings, and due to their TIM-1 positivity, decidual γδT cells could be involved in the formation of local pregnancy-specific Th2 bias. Based on our results relating the different decidual γδT cell subsets, we suppose that each phenotypic subpopulation, CD4-/CD8-positive γδTCRdim or TIM-3-/TIM-1-positive γδTCRbright, could correspond to a distinct function, which presumption needs further verification. In summary, beside cytotoxic potency, decidual γδT cells appear to have an additional, pronounced regulatory capacity in the placental microenvironment by showing a ?tolerance promoting Th2-like phenotype. Further investigations are necessary to explore the exact functional consequences of our findings.

Data Availability
The flow cytometric data used to support the findings of this study are available from the corresponding author upon request.