THE LINK BETWEEN PDL1 COSTIMULATORY PATHWAY AND TH17 IN FETOMATERNAL TOLERANCE

Abstract

Fetomaternal tolerance has been shown to depend both on regulatory T cells (Tregs) and negative signals from the PD1-PDL1 costimulatory pathway. More recently, IL-17-producing T cells (Th17) have been recognized as a barrier in inducing tolerance in transplantation. Herein we investigate the mechanisms of PDL1-mediated regulation of fetomaternal tolerance using an alloantigen-specific CD4⁺ TCR transgenic mouse model system (ABM-tg mouse). PDL1 blockade led to an increase in embryo resorption and a reduction in litter size. This was associated with a decrease in Tregs, leading to a lower Treg/Teff ratio. Moreover, PDL1 blockade inhibited antigen-specific alloreactive T cell apoptosis, induced Tregs' apoptosis and a shift towards higher frequency of Th17 cells, breaking fetomaternal tolerance. These Th17 cells arose predominantly from CD4⁺Foxp3⁻ cells, rather than from Tregs' conversion. Locally in the placenta, similar decrease in regulatory and apoptotic markers was observed by RT-PCR. Neutralization of IL-17 abrogated the anti PDL1 effect on fetal survival rate and restored Tregs' number. Finally, the adoptive transfer of Tregs was also able to improve fetal survival in the setting of PDL1 blockade. This is the first report utilizing an alloantigen-specific model that establishes a link between PDL1, Th17 cells and fetomaternal tolerance.

INTRODUCTION

Acceptance of the semiallogeneic fetus by the mother during pregnancy represents a physiologic model of in vivo immune tolerance (1–2). Several mechanisms have been reported to take part in this complex interplay. More recently the importance of regulation and trafficking of regulatory T cells at the fetomaternal interface has been investigated, even though its mechanisms remain a matter of debate (3). Since tolerance has now been reconsidered as a balance rather than an all-or-none phenomenon (4), a number of studies both in transplantation and in fetomaternal mouse models are focusing on the role of regulatory T cells (Tregs), effector T cells (Teff) and costimulatory pathways, as well as their reciprocal influence in balancing the immune response (5–7).

The PD1-PDL1 negative costimulatory pathway has been demonstrated to be critical for peripheral transplantation tolerance due to its ability to alter the balance between pathogenic T cells and regulatory T cells (8–9). Expression of PDL1 in donor tissue was crucial for the prevention of chronic rejection in a heart transplant model (10) as well as for the maintenance of tolerance at the utero-placental interface (11). In a bone marrow transplantation model, both in vivo and in vitro, PD1-PDL1 costimulation blockade resulted in an increased lethality of graft-versus-host disease (GVHD) by increasing donor IFN-γ production and enhancing T cell allo-immune responses (12). In a diabetes model, PDL1 expression on the surface of islets provided protection against immunopathological injury after transplantation (13).

Our group has previously shown that the PD1-PDL1 negative costimulatory pathway plays a key role in inducing and maintaining fetomaternal tolerance in a mouse model. In particular, it was shown by Habicht et al. that the expression of PDL1 on the surface of Tregs is essential to exert their suppressive effect and to control the maternal immune response (14). In fact, blocking PDL1 resulted in a loss of regulatory function and reduction in fetal survival rate. It has also been suggested that a deficiency of PDL1, in of itself, may influence effector T cells and drive the balance towards fetal rejection by favouring T helper 1 (Th1) and T helper 17 (Th17) development and expansion (14). Consistent with that, several authors have reported the failure of the Th1/Th2 paradigm in fully explaining fetal rejection in different murine models (15–17) as well as allograft rejection in transplantation (18). It has also been recently observed that Th17 expansion is critical to initiate an allo-immune response particularly in a Th1- deficient environment. Although no data is available in the allogeneic pregnancy model, Th17 expansion represents a barrier to establish tolerance (19–20).

The aim of the present study was to investigate the mechanisms by which PDL1 mediates the interaction between Tregs and Teff and thereby directly affects the Th1/Th2/Th17 balance, leading to regulation at the fetomaternal interface. For this purpose, we took advantage of an alloantigen-specific CD4⁺ TCR transgenic mouse model system (ABM-transgenic mouse) to track alloantigen-specific cells generated during the immune response in a single class II mismatch murine model of fetomaternal tolerance.

MATERIALS AND METHODS

Mice

B6.PL-Thy1^a/CyJ (Thy1.1 B6) and B6.C-H-2^Bm12/KhEg (Bm12) mice were purchased from The Jackson Laboratory. ABM-(anti-Bm12) mice were generated as previously reported (21) and maintained as a breeding colony in our animal facility. Foxp3 Knock In (Foxp3-GFP) mice were a kind gift of Dr. Rudensky (University of Washington). ABM Tg.FoxP3-GFP mice on B6 background were generated by crossing ABM mice with Foxp3 KI mice for 8 generations in our own facility. All mice were housed in accordance with institutional and National Institutes of Health guidelines.

Treatment Protocol

Pregnant female Thy1.1 B6 mice were injected intraperitoneally (i.p.) with the blocking anti-mouse PDL1 mAb (MIH6) as previously reported (11). Neutralizing IL-17 mAb (MAB421; R&D Systems, Minneapolis, MN) has been recently described (20) and was administered i.p. at a dose of 84ug on 4.5, 5.5, 6.5, 7.5, 8.5 and 11 dpc.

Timed matings and resorption rates

Females were inspected daily for vaginal plugs and sighting a vaginal plug was designated as day 0.5 of pregnancy. Plugged females were either sacrificed between 11.5 –13.5 dpc to examine the number of implantations and resorbing sites or were monitored until parturition and the number of pups born was recorded.

Adoptive transfer of TCR-tg T cells

Adoptive transfer of ABM TCR-tg T cells was performed as previously described (21). Briefly, spleens and lymph nodes were harvested from naïve ABM TCR-tg mice and pooled single-cell leukocyte suspensions were prepared. CD4⁺ T cells were purified by magnetic bead negative selection (Miltenyi Biotec, Auburn, CA) with >90 % purity. Typically, >90% of CD4⁺ T cells expressed the tg TCR as determined by anti-TCR Vα2.1, and anti-TCR Vβ8.1 antibody staining. 3 × 10⁶ ABM TCR-tg naïve CD4⁺ T cells were then injected i.p. into Thy 1.1 B6 females mated with Bm12 males (day 0). On day 1 and day 2 post transfer of cells, few mice were sacrificed, spleen and lymph nodes taken and their numbers counted to determine if similar numbers were present to begin with in the mice designated to anti PDL1 treatment group versus control group. Other mice received anti PDL1 mAb or control IgG on 11.5 – 13.5 dpc, lymph nodes and spleens were collected, single-cell leukocyte suspensions were prepared. ABM TCR-tg cells T cells were identified by staining for CD90 1.2 (Thy1.2) surface marker at flow cytometry. Placenta samples were collected as well for RT-PCR and intracellular cytokine staining by flow cytometry. Adoptive transferred cells used were naïve Thy 1.2⁺ ABM TCR-tg (Vα8.1⁺Vβ2.1⁺) CD4⁺ T cells.

Adoptive transfer of Thy1.2⁺Foxp3-GFP TCR-tg T cells

Adoptive transfer of Thy1.2⁺Foxp3-GFP TCR-tg T cells was performed as previously described in the ABM TCR-tg system (21). Briefly, spleens and lymph nodes were harvested from ABM Tg.FoxP3-GFP mice and pooled single-cell leukocyte suspensions were prepared. CD4⁺ T cells were purified by magnetic bead positive selection (Miltenyi) with > 90 % purity. ABM Tg.FoxP3-GFP CD4⁺ T cells express CD90.2 (Thy 1.2⁺) on their surface. More than 90 % of CD4⁺ T cells expressed the tg TCR, and approximately 2% of CD4⁺ T cells were GFP⁺. CD4⁺ isolated cells were then flow-sorted for Foxp3-GFP⁺ and GFP^– cells with > 98% purity. Subsequently, 0.3 × 10⁶ Thy1.2⁺Foxp3-GFP positive were injected i.p into Thy 1.1 B6 females mated with Bm12 males (day 0) as previously described. As a control, adoptive transfer of 0.3 × 10⁶ Thy1.2⁺Foxp3-GFP negative cells into Thy 1.1 B6 females mated with Bm12 males (day 0) was also performed.

To evaluate the fate of these transferred cells, we harvested lymphoid organs on day 3 after cell injection in pregnant females, in the setting of administration of either IgG or anti PDL1 Ab. Part of the collected cells were then stained for Thy 1.2 and Foxp3, while part was cultured with Bm12 irradiated splenocytes for 48 hours, followed by measurement of IL-17 production by Luminex as detailed below. To evaluate the effect of antigen-specific Treg transfer on litter size, a group of pregnant females treated with either IgG or anti PDL1 were followed long-term.

Flow cytometry

Cells (1 × 10⁶), were stained with biotinylated Thy1.2 (CD901.2) followed by APC or PerCP-conjugated Streptavidin to identify the adoptively transferred cells. Teff were determined by staining with anti-CD4 PerCP, anti-CD8 FITC, anti-CD62L APC, and anti-CD44 PE (all from BD Pharmingen) and analyzed for cells bearing the CD44^hiCD62L^low phenotype. Tregs were detected by staining with anti-CD4 FITC, anti-CD25 PE and intracellular staining with Foxp3 APC (eBioscience) as per manufacturer's instructions. In addition Thy1.2⁺Tregs were also stained with PE-conjugated PDL1 (MIH5 clone from BD Pharmingen). Apoptosis was evaluated by Annexin V and 7-amino-actinomycin D staining. Four-color flow cytometry was performed on a FACSCalibur (Becton Dickinson, San Jose, CA), and cells were analyzed using FlowJo software.

ELISPOT

Splenocytes from pregnant mice (female Thy1.1 B6 mated with Bm12 male) were obtained as single cell suspensions and used as responder cells (5 × 10⁵). Splenocytes from male Bm12 mice were irradiated and used as stimulator cells. The ELISPOT assay was adapted to detect IFN-γ, IL-4 and IL-17-secreting cells as described before (14).

Luminex cytokines assay

For cytokine analysis, splenocytes harvested at 11.5 and 13.5 dpc from pregnant Thy1.1 B6 mice were restimulated with irradiated Bm12 splenocytes. The cell-free supernatants of individual wells were removed after 48h and analyzed by a multiplexed cytokine bead-based immunoassay (20).

Placenta intracellular cytokine staining

Placentae were collected at 13.5 dpc and were cut into small pieces and digested with 1mg/ml collagenase at 37°C. Single cell suspension was prepared by passing through the wire mesh. The cells were washed and resuspended in two Percoll fractions at densities of 1.095 and 1.030. The cells in the 1.030 fraction were underlayered with cells in the 1.095 density fraction. The cells were spun at 2500rpm for 30 min. Cells were taken from the interface, washed and counted for subsequent flow staining. Pan leukocytes (CD45) expressing Foxp3, IFN-γ or IL-17 were stained on the surface with anti-CD45 FITC and intracellularly with Foxp3/IFN-γ/IL-17 APC (eBioscience) antibodies as per manufacturer's instructions, then detected by flow cytometry.

Treg conversion in vitro

Thy1.2⁺ FoxP3-GFP⁻ T cells from ABM Tg.FoxP3-GFP mice, generated in our laboratory by mating ABM Tg mice with Rudensky reporter mice (B6 FoxP3-GFP knock-in), were used for in vitro studies. Isolated Foxp3-GFP knock-in CD4⁺GFP⁻ T cells (1 × 10⁶) were cocultured with syngeneic CD11c⁺ cells (0.2× 10⁶), soluble anti-CD28 (1 μg/mL), soluble TGF-β (1 ng/mL) and plate-coated with anti-CD3 (1 μg/mL), in the presence or absence of anti PDL1 mAb (1, 10 or 50 μg/ml). After 4 days in culture, the percentage of CD4⁺Foxp3⁺ cells was assessed by flow cytometry. We also collected the supernatant of these cultures at 48 hours in order to evaluate the production of IL17 by Luminex. Finally, CD4⁺GFP⁺ cells (0.2 × 10⁶) from ABM Foxp3-GFP knock-in TCR-Tg mice were cocultured with syngeneic CD11c⁺ cells (0.05× 10⁶) and irradiated Bm12 cells (0.2 × 10⁶) in the presence or absence of anti PDL1. Production of IL17 was evaluated after 48 hours of culture. Cell isolation was performed by FACS-sorting CD4⁺GFP⁻, CD4⁺GFP⁺ cells and CD11c⁺ cells, resulting in greater than 99% purity.

RNA extraction and real-time PCR (RT-PCR)

RNA extraction was performed according to the manufacturer's instructions (Invitrogen). Expression was measured as copies of any given gene divided by copies of the housekeeping gene GAPDH as previously described (14).

Statistics

Kaplan-Meier survival graphs were constructed and a log rank comparison of the groups was used to calculate p values. Student's t test was used for comparison of means between experimental groups examined by FACS analysis or ELISPOT assay. Differences were considered to be significant at p values < 0.05.

RESULTS

Reduction in fetal survival with PDL1 blockade correlates with a decrease in Tregs and Tregs/Teff ratio in the periphery and a lowering of expression of regulatory markers at the uteroplacental interface

In order to investigate the mechanisms by which PD1-PDL1 costimulatory pathway affects fetomaternal tolerance, we took advantage of the ABM-tg mouse (21) to track alloantigen-specific CD4⁺ cells with either an effector or a regulatory phenotype development in a Thy1.1 B6 pregnant mouse previously mated with a Bm12 male (see methods). We have previously shown that PDL1 is expressed at the fetomaternal interface and blockade of this molecule resulted in significantly reduced fetal survival rate in a fully allogeneic (CBA × B6) mating set up (11). Here we report that PDL1 blockade remarkably increased fetal resorption (34±10% vs. 1.4±10% n=7 p<0.0001) and resulted in a reduced litter size compared to Control IgG (5.8±0.4 vs 8.5±1, n=7 p<0.0001) in the in the Bm12 × B6 mating as well (Figure 1).

Figure 1. PDL1 blockade significantly increases fetal resorption and results in reduced litter size.

Thy 1.1 B6 females were mated with BM12 males and at days 6.5, 8.5, 10.5 and 12.5, anti PDL1 mAb or control IgG was injected. Some mice were observed for fetal resorption at day 13.5 dpc and others were allowed to go to term and litter sizes were counted. A. Scattered dot plots depicting percentages of embryos resorbing in mice treated with anti PDL1 (n=7) or Control IgG (n=7) and B. Scattered dot plots depicting litter size in mice treated with anti PDL1 or Control IgG are shown.

Subsequently we adoptively transferred Thy1.2⁺ABM-tg cells at day 4 post detection of vaginal plug and treated the female mice with either anti PDL1 mAb or control IgG. We analyzed samples at 11.5 and 13.5 days post coitum (dpc). Flow cytometry analysis demonstrated an increase in the percentage of Thy1.2⁺Teff (Thy1.2⁺CD44^hiCD62L^low) in spleens of animals treated with anti PDL1 compared to Control IgG (72± 1.8 vs. 65 ± 1.7%, p<0.01), particularly at 11.5 dpc while no differences were found at day 13.5 (56± 2.7 vs. 52 ± 2.8%, p=0.2). Absolute numbers did not show any difference neither at day 11.5 (216000 ± 25 vs. 206500 ± 13, respectively, p=0.7) (Figure 2A) nor at day 13.5 (215000 ± 55 vs. 218400 ± 33, p=0.9). As shown in Figure 2B, Thy1.2⁺Tregs (Thy1.2⁺CD25⁺Foxp3⁺) were also significantly decreased after PDL1 blockade at the same timepoint in lymph nodes (data not shown) and spleen compartments in both percentage (13 ± 2.7 vs. 4 ± 0.7%, p= 0.01) and absolute numbers of cells (382200 ± 30 vs. 217500 ± 33, respectively, p<0.01), while a less evident reduction was detectable at 13.5 dpc (5.8± 0.3 vs. 3.9 ± 0.65%, p= 0.03 and 305300 ± 28 vs. 221595 ± 22, p=0.048), suggesting a role for this costimulatory pathway in actively regulating the balance between alloantigen-specific Tregs and Teff during the tolerance process in the periphery. In fact, analysis of Tregs/Teff ratio showed a reduction at 11.5 dpc both in lymph nodes (0.5 ± 0.03 vs. 0.2 ± 0.02, p<0.0002) and spleen (0.14 ± 0.03 vs. 0.05 ± 0.008, n=11 p<0.03) in percentage values that was confirmed by calculating absolute numbers of cells in spleens (1.7 ± 0.2 vs. 1 ± 0.1 on controls, respectively, p<0.04) (Figure 2C). These results suggest that a significant decrease in Thy1.2⁺Tregs rather than an increase in Thy1.2⁺Teff plays a major role in affecting the balance of the immune response.

Figure 2. Blocking PD1-PDL1 costimulation is associated with a decrease in Regulatory T cells (Tregs) in an antigen specific manner.

Spleens and lymph nodes were harvested from ABM TCR-tg mice, CD4⁺ T cells were purified and tested for expression of the tg TCR as determined by anti-TCR Vα2.1, and anti-TCR Vβ8.1 antibody staining. 3 × 10⁶ ABM TCR-tg CD4⁺ T cells were then injected i.p. into Thy 1.1 B6 females mated with Bm12 males (day 0) and mice received either anti PDL1 or control IgG. On 11.5 –13.5 dpc, lymph nodes and spleens were collected, single-cell leukocyte suspensions were prepared. ABM TCR-tg cells T cells were identified by staining for CD90 1.2 (Thy1.2) surface marker at flow cytometry. Placenta samples were collected as well for RT-PCR and intracellular cytokine staining by flow cytometry. A. Bar graphs depicting absolute numbers of Thy1.2⁺CD44^hiCD62L^low (Thy1.2⁺Teff) cells in spleen of adoptively transferred pregnant mice at 11.5 dpc treated with anti PDL1 or Control IgG. Representative flow plots of Thy1.2⁺Teff from pregnant mice in each treatment group are shown on the right side of respective bar graphs. Data are representative of three independent experiments using at least n = 4 mice per group. B. Bar graphs depicting absolute numbers of Thy1.2⁺CD25⁺Foxp3⁺ (Thy1.2⁺Tregs) in spleen of adoptively transferred pregnant mice at 11.5 dpc treated with anti PDL1 or Control IgG. Representative flow plots of Thy1.2⁺Tregs cells from pregnant mice in each treatment group are shown on the right side. Data are representative of three independent experiments using at least n = 4 mice per group. C. Scattered dot plots depicting Tregs/Teff ratio (absolute numbers) in anti PDL1 and Control IgG-treated mice.

Prior studies have also reported that PDL1 blockade can affect regulatory markers locally in the placenta (11, 22). In line with this, we observed that the expression of some of the regulatory genes detected by real time-PCR (RT-PCR) at the fetomaternal interface is diminished in animals treated with anti PDL1 compared to controls. In particular, Foxp3 (0.00002 ± 0.00002 vs. 0.5310 ± 0.2042 copies/copies GADPH, p<0.05), IDO (0.31 ± 0.13 vs.1.08 ± 0.24 copies/copies GADPH, p<0.04) and ICOS (0.0001 ± 0.000000000004 vs. 0.44 ± 0.15 copies/copies GADPH, p<0.03) were down regulated in anti PDL1-treated groups (Supplemental Figure 1).

This data confirms that blocking PD1-PDL1 negative costimulatory pathway interferes with the fine balance between Teff, Tregs and regulatory molecules in the periphery as well as at the uteroplacental interface.

PDL1 blockade is associated with a switch in Th balance towards Th17

In order to investigate whether PD1-PDL1 costimulation is able to directly affect Teff cell generation and/or tip the Th1/Th2/Th17 balance to promote tolerance during pregnancy, we first analyzed the cytokine profile both locally in the placenta with intracellular cytokine flow staining / RT-PCR and in the periphery with ELISPOT / Luminex-based assays. The most interesting finding was that blocking PDL1 resulted in a major inhibition of Th1 cytokine production both in the periphery (ELISPOT: IFN-γ 24.33 ± 4.717 for anti PDL1-treated mice vs. 141.8 ± 29.56 spots per 0.5×10⁶ splenocytes of Control IgG-treated mice, p<0.0002; Luminex: IP10 20.38 ± 4.8 vs. 42.5 ± 7.0 pg/ml, respectively, p<0.02; TNF-α 51.45 ± 25.67 vs. 190.3 ± 12.98 pg/ml, respectively, p<0.007) and at the uteroplacental interface (TNF-α 0.000002 ± 0.000002 vs. 1.17 ± 0.41 copies/copies GADPH, respectively, p<0.05) with an increase in the production of Th2 cytokines (IL-4 and IL-5) as reported in Figure 3. Similar results were obtained at the second timepoint 13.5 dpc for Th1 cytokines (ELISPOT: IFN-γ 91.2 ± 25 for Control IgG-treated mice vs. 39± 6.7 spots per 0.5×10⁶ splenocytes of anti PDL1-treated mice, p=0.012; Luminex: IP10 42.5±7 vs. 20.38 ± 4.8 pg/ml, respectively, p=0.019; TNF-α 29.9 ± 6.4 vs. 9.53 ± 1.9 pg/ml, respectively, p<0.001) as well as for Th2 cytokines (ELISPOT: IL-4 88.8±8.6 for Control IgG-treated mice vs. 114±3.6 spots per 0.5×10⁶ splenocytes of anti PDL1-treated mice, p=0.018; Luminex: IL-4 21.67±3.4 vs. 40.67 ± 6 pg/ml, respectively, p=0.02; IL-5 51.1 ± 25 vs. 153.4± 73 pg/ml, respectively, p=0.010). Interestingly, we detected a significant increase in IL-6 production at 11.5 dpc and later at 13.5 dpc in anti PDL1-treated animals compared to controls (11.5 dpc: 285.4 ± 51.7 vs. 27.13 ± 3.3, respectively, p<0.0002; 13.5 dpc:1579 ± 368.4 vs. 741.8 ± 121.7 pg/ml, respectively, p<0.03). As it has been reported that IL-6 represents a differentiation factor for Th17 and a target for IL-17-producing cells (23), we examined IL-17 production both locally and in the periphery. As anticipated, ELISPOT and Luminex showed a significant increase in the level of IL-17 in the periphery in anti PDL1-treated mice compared to controls (Figure 4A). This increase in IL-17 was also demonstrable locally at the uteroplacental interface as evidenced by intracellular flow cytometry staining (Figure 4B) and by RT-PCR of the placentae of anti PDL1-treated mice vs. Control IgG-treated mice (1.1 ± 0.1 vs. 0.3 ± 0.05%, p<0.005 and 10.36 ± 3.3 vs. 1.87 ± 1.2 copies/copies GADPH, p<0.03), as shown in Figure 4 B.

Figure 3. PDL1 blockade results in a suppression of Th1 cytokines and an increase in both Th2 cytokines and IL-6.

A. Bar graphs depicting production of allospecific cytokine IFN-γ (by ELISPOT), IP-10 and TNF-α (by Luminex) by splenocytes from Thy1.1 females previously mated with Bm12 males, 11.5 dpc treated with anti PDL1 or Control IgG. Data are representative of three independent experiments using at least n = 4 mice per group. B. Bar graphs depicting production of allospecific cytokine IL-4 (by ELISPOT), IL-4 and IL-5 (by Luminex) by splenocytes from Thy1.1 females previously mated with Bm12 males, 11.5 dpc treated with anti PDL1 or Control IgG. Data are representative of three independent experiments using at least n = 4 mice per group. C. Bar graphs depicting production of allospecific cytokine IL-6 (by Luminex) by splenocytes from Thy1.1 females previously mated with Bm12 males, 11.5 dpc treated with anti PDL1 or Control IgG. Data are representative of three independent experiments using at least n = 4 mice per group.

Figure 4. Anti PDL1 treatment results in higher expression of IL-17 both in the periphery and locally in the placenta.

Splenocytes from pregnant mice (female Thy1.1 B6 mated with Bm12 male) were obtained as single cell suspensions and used as responder cells (5 × 10⁵). Splenocytes from male Bm12 mice were irradiated and used as stimulator cells. The ELISPOT assay was performed to detect IL-17-secreting cells. Splenocytes from pregnant Thy1.1 B6 mice were also restimulated with irradiated Bm12 splenocytes and the cell-free supernatants of individual wells were removed after 48h and analyzed by a multiplexed cytokine bead-based immunoassay for IL-17 production. A. Bar graphs depicting production of allospecific cytokine IL-17 by ELISPOT assay (left panel) and Luminex (right panel) using splenocytes from Thy1.1 females previously mated with Bm12 males, 11.5 dpc treated with anti PDL1 or Control IgG. Data are representative of three independent experiments using at least n = 4 mice per group. In parallel, placentae were removed and IL-17 expression was determined in anti-PDL1 mAb or control IgG antibody group by real time PCR or by Flow cytometry. For placenta intracellular cytokine staining, placentae were collected at 13.5 dpc and were digested with collagenase at 37°C and processed to get infiltrating cells as described in methods. Pan leukocytes (CD45) expressing IL-17 were stained on the surface with anti-CD45 FITC and intracellularly with anti-IL-17 APC antibodies and detected by flow cytometry. B. Left panel: scatter plots depicting expression of IL-17 cytokine by RT-PCR at the fetomaternal interface of pregnant mice treated as above. Right panel: bar graphs depicting detection of IL-17 cytokine by intracellular cytokine staining at the fetomaternal interface of pregnant mice treated as above with representative flow plots in each treatment group shown on the right side. The data represent the mean ± SD of the results obtained from three separate experiments.

Neutralization of IL-17 abrogates the anti PDL1 effect on fetal survival rate

Having found differences in Tregs as well as Th17 in our ABM-tg pregnancy model, we focused on delineating the interplay, if any, between Th17 and down-modulation of Tregs in this model. To dissect this, we first set up matings between Bm12 males and B6 Thy1.1 females and then treated them with anti PDL1 alone or in combination with IL-17 neutralizing antibody. Neutralization of IL-17 in anti PDL1-treated animals showed a significant increase in litter size compared to anti PDL1 alone (8.6 ± 0.74 vs. 4.7 ± 0.14, n=5 and n=7, respectively, p<0.0001) and a significant decrease in fetal resorption (16.2± 3.4 vs. 37.4 ± 1.7, n=5 and n=7, respectively, p<0.0002), as represented in Figure 5A, 5B. We then tested the role of blocking PDL1 in IL-17^−/− mice mated with Bm12 males and observed no difference in litter size (8 ± 0.4 vs. 8.5 ± 0.8, n=5, respectively, p=ns) as well as in embryo resorption rate (14.8± 4 vs. 15.7 ± 2, n=5, respectively, p=ns) when animals were treated with anti PDL1 compared to Control IgG (Figure 5C, 5D). This provided evidence for an important role for IL-17-producing cells in interfering with tolerance induction in this experimental model. Furthermore, it highlights the fact that PDL1 blockade in our single class II-mismatch B6 × Bm12 mating model led to a reduction in fetal survival and an increase in the fetal resorption rate by switching the Th balance in favor of Th17.

Figure 5. Neutralization of IL-17 abrogates anti PDL1 effect in reducing fetal survival rate and increasing embryo resorption.

Pregnant female Thy1.1 B6 mice were injected with the blocking anti-mouse PDL1 mAb. Neutralizing IL-17 mAb was administered to some of the mice treated with anti PDL1 antibody. The degree of fetal resorption was determined at day 13.5 dpc and litter size was determined in a set of mice at term. Pregnant IL-17^−/− mouse were used for fetal resorption and litter size studies for some experiments.

A. Scattered dot plots depicting percentages of embryos resorbing in mice treated with Control IgG (n=10), anti PDL1 (n=7), anti PDL1 plus anti IL-17 (n=10) and anti IL-17 alone (n=9).

B. Scattered dot plots depicting litter size in mice treated with Control IgG (n=7), anti PDL1 (n=7), anti PDL1 plus anti IL-17 (n=5) and anti IL-17 alone (n=5).

C. Scattered dot plots depicting percentages of embryos resorbing in IL-17^−/− mice treated with Control IgG (n=4) and anti PDL1 (n=4).

D. Scattered dot plots depicting litter size in IL-17^−/− mice treated with Control IgG (n=4) and anti PDL1 (n=4).

Neutralization of IL-17 significantly reduces anti PDL1-mediated expansion of Teff cells and abrogates the anti PDL1- mediated effect in reducing Treg numbers

Our next step was to determine the effect of neutralization of IL-17 in anti PDL1-treated animals on adoptively transferred Thy1.2⁺ABM-tg cells. While the expansion of Thy1.2⁺Teff is slightly increased in anti PDL1-treated animals upon IL-17 neutralization, there is a more marked effect in maintaining Thy1.2⁺Tregs compared to administration of anti PDL1 alone (Figure 6A, 6B). This results in a relatively increased Tregs/Teff ratio compared to anti PDL1 alone treated group (Figure 6C).

Figure 6. Neutralization of IL-17 significantly reduces anti PDL1-mediated expansion of effector cells and abrogates anti PDL1-mediated reduction of Tregs.

Spleens and lymph nodes were harvested from ABM TCR-tg mice, CD4⁺ T cells were purified and tested for expression of the tg TCR as determined by anti-TCR Vα2.1, and anti-TCR Vβ8.1 antibody staining. 3 × 10⁶ ABM TCR-tg CD4⁺ T cells were then injected i.p. into Thy 1.1 B6 females mated with Bm12 males (day 0) and mice received either anti PDL1 or control IgG. Neutralizing IL-17 mAb was administered to some of the mice treated with anti PDL1 antibody. On 11.5 dpc, lymph nodes and spleens were collected, single-cell leukocyte suspensions were prepared and ABM TCR-tg cells T cells were identified by staining for CD90 1.2 (Thy1.2) surface marker at flow cytometry. A. Bar graphs depicting absolute numbers of Thy1.2⁺Teff in spleen of adoptively transferred pregnant mice at 11.5 dpc with various treatments. Data are representative of three independent experiments using at least n = 4 mice per group. B. Bar graphs depicting absolute numbers of Thy1.2⁺Tregs in spleen of adoptively transferred pregnant mice at 11.5 dpc with various treatments. Data are representative of three independent experiments using at least n = 4 mice per group. C. Scattered dot plots depicting Tregs/Teff ratio (absolute numbers of cells) in pregnant mice at 11.5 dpc with various treatments.

PDL1 blockade inhibits antigen-specific alloreactive T cell apoptosis and induces apoptosis of Tregs

Since one of the most important mechanisms shown to induce and/or maintain fetomaternal tolerance is the apoptosis of activated maternal cells (2, 24) we investigated its possible link with PD1-PDL1 costimulatory pathway. We first focused our attention on the apoptosis of alloantigen-specific ABM-tg cells. Flow cytometry analysis at 11.5 dpc showed that both percentage (5.1 ± 1.6 vs. 10.2 ± 1.3%, respectively, p<0.04) and absolute number (533546 ± 25 vs 744170 ± 85 , p<0.05) of Thy1.2⁺AnnexinV⁺ apoptotic cells decreased particularly in the spleen of mice treated with anti PDL1 compared to Control IgG (Figure 7A and 7B). This data was corroborated by expression of apoptotic markers locally in the placenta by real time PCR (RT-PCR). RT-PCR analysis showed a significant reduction in DR5, a receptor involved in the TRAIL pathway for apoptosis, in the placentae of anti PDL1-treated mice vs. control IgG-treated mice (0.60 ± 0.16 vs. 1.58 ± 0.14 copies/copies GADPH, p<0.001). These data indicate a role for PDL1 blockade in inhibiting the apoptosis of activated alloreactive T cells at the periphery and limiting the availability of molecules involved in the apoptotic process.

Figure 7. PDL1 blockade inhibits antigen-specific alloreactive T cell apoptosis.

Spleen and lymph nodes were harvested from ABM TCR-tg mice and 3 × 10⁶ ABM TCR-tg CD4⁺ T cells (Thy1.2+) were then injected i.p. into Thy 1.1 B6 females (mated with Bm12 males) treated with anti PDL1 mAb or control IgG. At d11.5 dpc, spleens and lymph nodes were collected and apoptosis was evaluated by Annexin V and 7-amino-actinomycin D staining. Four-color flow cytometry was performed on a FACSCalibur and cells were analyzed using FlowJo software. A. Bar graphs depicting percentage of Thy1.2⁺AnnexinV⁺ apoptotic cells in spleen of adoptively transferred pregnant mice at 11.5 dpc treated with anti PDL1 or Control IgG. Representative flow plots of gating on Thy1.2⁺, 7AAD⁻and AnnexinV⁺ cells from pregnant mice in each treatment group are shown below the bar graphs.

B. Bar graphs depicting absolute numbers of Thy1.2⁺AnnexinV⁺ apoptotic cells in spleen of adoptively transferred pregnant mice at 11.5 dpc treated with anti PDL1 or Control IgG. Data are representative of three independent experiments using at least n = 4 mice per group.

C. Scattered dot plots depicting Thy1.2⁺Tregs undergoing apoptosis in adoptively transferred pregnant mice at 11.5 dpc under different treatments. Data are representative of three independent experiments (n = 3 mice per group).

We next explored if PDL1 blockade is responsible for inducing apoptosis of Tregs and if anti IL-17 co-administration affects Treg apoptosis. We observed that apoptosis of Thy1.2⁺Tregs (Thy1.2⁺CD25^hi) is increased in anti PDL1-treated animals compared to controls (489084 ± 51 vs. 204971 ± 37, p<0.01) and compared to groups treated with either a combination of anti PDL1 and anti IL-17 or anti IL-17 alone (227399 ± 43 and 126638± 10, respectively, p<0.02 and p<0.002 respectively) as shown in Figure 7C.

PDL1 blockade reduces Tregs and increases the conversion of CD4⁺Foxp3⁻ into IL17-producing T cells

In order to determine if PDL1 blockade prevents induction of Tregs, shifts Tregs to Th17 producers or both, we performed in vitro cultures of CD4⁺Foxp3⁻ or CD4⁺Foxp3⁺ cells from Foxp3 KI mice in the presence or absence of anti PDL1 mAb. Our in vitro data shows that anti PDL1 decreases conversion of CD4⁺Foxp3⁻ cells into CD4⁺Foxp3⁺ Tregs. This decrease in conversion is small but significant (p<0.01) (Figure 8A). Moreover, PDL1 blockade increases the frequency of IL17-producing cells from CD4⁺Foxp3⁻ cells in a dose dependent manner (Figure 8B). However when CD4⁺Foxp3⁺ Tregs were cultured in the presence of anti PDL1 mAb, they failed to convert to Th17-producing cells (Figure 8C).

Figure 8. PDL1 blockade reduces induction of Tregs and increases the amount of IL17-producing cells in vitro.

Isolated CD4⁺GFP⁻ T cells (1 × 10⁶) from ABM Tg.FoxP3-GFP mice were cocultured with syngeneic CD11c⁺ cells, soluble anti-CD28, soluble TGF-β and plate-coated with anti-CD3, in the presence or absence of anti PDL1 mAb. After 4 days in culture, the percentage of CD4⁺Foxp3⁺ cells was assessed by flow cytometry. For the Treg conversion, CD4⁺GFP-Foxp3⁺ cells from ABM Tg.FoxP3-GFP were cultured for 48 hours with syngeneic CD11c⁺ cells and irradiated Bm12 cells in the presence or absence of anti PDL1. A. Bar graphs depicting in vitro conversion of CD4⁺Foxp3-GFP⁻ cells into CD4⁺Foxp3-GFP⁺ cells incubated with anti PDL1 antibody. Below the bar graphs, representative flow histograms of Foxp3GFP⁺ of CD4⁺ gated cells are shown. B. Bar graphs demonstrating IL-17 production from CD4⁺Foxp3GFP⁻ cells incubated in vitro with anti PDL1 antibody in a dose dependent manner. C. Bar graphs depicting IL-17 production from CD4⁺Foxp3GFP⁺ cells incubated in vitro with anti PDL1 antibody at multiple concentrations.

To strengthen our in vitro findings, we utilized ABM Tg.FoxP3-GFP mice in which antigen-specific Tregs (CD4⁺Thy1.2⁺Foxp3-GFP⁺) can be detected, isolated and adoptively transferred. Flow-sorted Thy1.2⁺CD4⁺Foxp3-GFP⁺ (Tregs) and GFP⁻ cells (0.3 × 10⁶) were injected intraperitoneally into Thy 1.1 B6 female mice mated with Bm12 males and treated with either anti PDL1 or Control IgG. We then harvested these cells 72 hours after injection to evaluate their fate (Supplemental Figure 2). About 15–20% of injected cells were present on secondary lymphoid organs three days after injection (Figure 9A). While both control and anti PDL1 treated groups lost Foxp3 expression after transfer of Thy1.2⁺CD4⁺Foxp3-GFP⁺ cells, the group treated with anti PDL1 showed a greater conversion to Foxp3-GFP⁻ cells (Figure 9B). Moreover, IL-17 production was very low on both groups, refuting the hypothesis of increased conversion of Foxp3-GFP⁺ cells into Th17 cells in the setting of PDL1 blockade (Figure 9E). After transfer of Foxp3-GFP⁻ cells, we observed about a 5% conversion to Foxp3-GFP⁺ cells in both groups (Figure 9D). Interestingly, we noticed an increased conversion of Foxp3-GFP⁻ cells into IL-17-producing T cells (Figure 9F), in agreement with our in vitro data above. Overall these data show that PDL1 blockade increases the conversion of Foxp3-GFP ⁻ cells into IL-17-producing cells and decreases the number of Tregs, but does not seem to convert Tregs into IL17-producing T cells.

Figure 9. Fate of Thy1.2⁺Foxp3-GFP cells after adoptive transfer and their effect on litter size.

Flow-sorted ABM Thy1.2⁺Foxp3-GFP⁺ (Tregs) and Thy1.2⁺Foxp3-GFP⁻ cells (0.3 × 10⁶) were injected intraperitoneally into Thy 1.1 B6 female mice mated with Bm12 males and treated with either anti PDL1 or Control IgG. We then harvested these cells 72 hours after injection to evaluate their fate. Some pregnant females transferred with Tregs were also followed long-term to evaluate effect on litter size. A, Representative flow cytometry plots of Thy1.2⁺ cells recovered from secondary lymphoid organs after transfer of Thy1.2⁺Foxp3GFP⁺ cells (antigen-specific Tregs), with mean total cells obtained by multiplying total number of cells by the percentage of Thy 1.2⁺ cells (n = 3–5/group). B, Representative histograms of Foxp3 expression on Thy 1.2⁺ cells with mean percentage between groups after transfer of Tregs. C, Representative flow cytometry plots of Thy 1.2⁺ cells recovered from secondary lymphoid organs after transfer of Thy1.2⁺Foxp3GFP⁻ cells, with mean total cells of 3 animals per group. D, Representative histograms of Foxp3 expression on Thy 1.2⁺ cells with mean percentage between groups after transfer of Thy1.2⁺Foxp3-GFP⁻ cells. E, F, IL17 production by Luminex after restimulation with irradiated Bm12 cells three days after cell transfer of Thy1.2⁺Foxp3GFP⁺ or GFP⁻, respectively. G. Average litter size in Thy1.1 B6 females mated with Bm12 males and adoptively transferred with Thy1.2⁺Foxp3-GFP⁺ cells or Thy1.2⁺Foxp3-GFP⁻ or no cells following anti PDL1 antibody or Control IgG treatment (n=4 mice per group).

In order to confirm the role of Tregs in PDL1-induced fetomaternal tolerance, we performed similar adoptive transfer of Thy1.2⁺Foxp3-GFP⁺ (Tregs) and GFP⁻ from ABM Tg.FoxP3-GFP mice into Thy 1.1 B6 female mice mated with Bm12 males and treated with either anti PDL1 or Control IgG. However, this time we evaluated the pregnant B6 mice for litter size. A group of control mice without any adoptive transferred cells was also set up and litter size of all groups was assessed. The litter size in Thy1.1 B6 females treated with anti PDL1 mAb was close to 6 pups per litter. Following transfer of Tregs, there was a significant increase in the litter size in this group to 8 pups per litter (p=0.0033) (Figure 9G). There was no increase in control IgG treated group receiving Tregs versus not receiving any Tregs (Figure 9G). This could presumably be because of no defect in the litter size in the Control IgG group and hence an additional dose of Tregs did not change the litter size outcome. This beneficial effect was specific to GFP⁺ cells as transfer of GFP⁻ cells into anti PDL1 or Control IgG treated Thy1.1 B6 (mated with Bm12) mice did not result in rescue of phenotype in anti PDL1 group (litter size, 5.0+/− 0.71 in GFP⁻ + anti PDL1 vs 5.0+/−0.91 in no cells +/− anti PDL1 group; n=4, p=ns). Litter size in GFP⁻ cells + Control IgG and no cells+ Control IgG control was normal and close to 8 pups per litter.

DISCUSSION

In this report, we demonstrate that PDL1 blockade reduces fetal survival rate by affecting various immune cells and regulatory molecules both in the periphery and at the uteroplacental interface. For this purpose we used an alloantigen-specific CD4⁺ transgenic model with a Bm12 × B6 mating. First, PDL1 blockade inhibits the negative regulation of the immune response against alloantigens, and contributes to fetomaternal tolerance by controlling alloantigen-specific Tregs in vivo in the periphery. The immunoregulatory role of PD1-PDL1 costimulatory pathway in peripheral tolerance has been previously described in a fully allogeneic heart transplantation model (25), and in a class II mismatch model of skin transplantation (26). In both cases PDL1 blockade resulted in an acceleration of allograft rejection. This was attributed to either a failure in immune regulation as a result of a decrease in Treg numbers, and/or ineffective suppressive effect. Wang et al. demonstrated that the deficiency of PDL1 molecules alters the balance of peripheral lymphocytes by decreasing the number of Tregs. In addition, there is an increase in the number of antigen presenting cells (APCs) that express a higher level of costimulatory molecules, thereby enhancing allogeneic immune responses in PDL1^−/− recipients, and accelerating heart allograft rejection (27). In this study, we did not look specifically at the APCs. Kitazawa et al. have recently shown that the PD1-PDL1 pathway is required for Tregs in order to effectively suppress the alloreactive responses of Teff (28). In addition, PDL1 expression by dendritic cells (DCs) is necessary for the induction of Foxp3⁺ adaptive Tregs both in vitro and in vivo (29). Finally, PD1-PDL1 costimulatory pathway and Tregs have been demonstrated to strictly participate in the development of fetomaternal tolerance (14) and in the prevention of fetal rejection in an abortion-prone experimental model of murine pregnancy (30).

On the other hand, our results document that PDL1 blockade also modulates the balance of alloimmune response locally at the fetomaternal interface with the involvement of different molecules and pathways. In line with this, Yang et al. demonstrated that PDL1 blockade in Bm12 into B6 heart transplant significantly accelerates chronic rejection depending on the expression of PDL1 on donor tissue that regulates recipient alloimmune responses (31). Similarly, Tanaka et al. described a shift in the balance between Tregs and Teff in favour of Teff during PDL1 blockade in a fully allogeneic heart transplant due to an inhibition of the delivery of the negative costimulatory signal and also showed that PDL1 expression in donor tissue was essential to prevent graft infiltration and rejection (10).

Our results in the alloantigen-specific model also support the role for PD1-PDL1 costimulation in promoting antigen-specific Tregs and in limiting antigen-specific Teff cell expansion in the context of fetomaternal tolerance.

Furthermore, our data suggests that PD1-PDL1 costimulation also contributes to the maintenance of fetomaternal tolerance by promoting apoptosis of maternal antigen-specific activated cells. It has been reported earlier that PD1-PDL1 costimulatory pathway plays an important role in fetomaternal tolerance by inducing apoptosis of paternal antigen-specific T cells and that PD1 action is the main mechanism that prevents T cell accumulation in lymph nodes (32). Even though the authors did not confirm directly a role for PDL1, they hypothesized that within the network of immunomodulatory pathways present at the fetomaternal interface, PD1-PDL1 is able to tip the maternal immune response towards a favourable milieu by limiting the accumulation of alloantigen-specific cells through apoptosis and modulation of cytokine secretion (32). In addition to that, a study from Bai et al. on TRAIL apoptotic pathway in humans showed that DR5 expression on cytotrophoblast cell surface promotes susceptibility to apoptosis, and that lack of DR5 cell surface expression may direct TRAIL-induced cytotoxicity away from the placenta, and towards the maternal system (33). This is in accordance with our observation that DR5 expression at the fetomaternal interface is reduced in anti PDL1 treated animals, and accounts for a diminished responsiveness to TRAIL-mediated apoptosis.

As part of PD1-PDL1 immunomodulatory function, our study also shows that PDL1 blockade is associated with a switch in Th balance towards Th17 which leads to a reduction in fetal survival and increase in the fetal resorption rate in our single-class II mismatch B6 × Bm12 mating model. Habicht et al. described a similar increase in IL-17 production in the placentae of PDL1^−/− females (on B6 background) mated with CBA males compared to B6 (wild type WT) control female mated with CBA male (14). Here we show, in an antigen-specific model that the Th1 response is abolished for the most part in anti PDL1 treated animals, and despite detecting a Th2 cytokine production, the development of a Th17 population overcomes it, and creates a break in tolerance. Despite the recently discovered plasticity between Th17/Tregs (34), the Th17 population in anti PDL1 treated animals seems to arise from CD4⁺Foxp3⁻ cells and not Tregs, as evident in our in vitro and adoptive transfer models (Fig. 8B, 9F).

It has been previously reported that successful pregnancy is a phenomenon mediated by Th2 cells and that Th1 cytokines such as IFN-γ and TNF-α may favor fetal loss (35). Chaouat et al, on the contrary, shed light on the fact that some Th1 cytokines play a beneficial role in promoting implantation (IFN-γ) and that other new cytokines have been found to contribute to tipping the balance of immune response during pregnancy such as IL-15, IL-18, IL-12, IL-16 and IL-27 (36). However, whether these cytokines promote fetomaternal tolerance or not is not entirely clear.

Therefore, we focused mainly on confirming the role of IL-17 in contributing to fetal rejection and its complex interplay with both PD1-PDL1 costimulation and Tregs modulation in interrupting the state of tolerance. Our study demonstrates that neutralization of IL-17 abrogates the anti PDL1 effect on fetal survival rate. It also confirms a detrimental role for IL-17-producing cells in promoting and sustaining fetal rejection as evidenced by the IL-17 KO data. Our data with these mice indicates that IL-17-producing cells play a major role in disrupting the state of tolerance early on but further studies are necessary to investigate the exact mechanism in details.

Moreover, it has been shown that PDL1 blockade in transplantation is correlated with a net increase in alloreactive T cells due to a reduction in apoptosis mechanisms. It has also been correlated with a reduction in Tregs and a decreased Treg suppressive effect (9, 26). Here we show that PDL1 blockade is associated with a reduction in the absolute number of Tregs. This results in a failing suppressive effect on alloreactive T cells and in particular in those producing IL-17. Different mechanisms have been reported in an attempt to explain the link between PDL1 and Tregs. Francisco et al. described a direct role for PDL1 in regulating induced Treg development and function. The authors showed that in the absence of PDL1 naive CD4⁺ T cells minimally convert to Tregs and PDL1 itself enhances and sustains Foxp3 expression and the suppressive function of induced Tregs (8). Despite our initial in vitro findings suggesting similar effect, our adoptive transfer experiments in vivo did not show a significant decrease in conversion in the setting of PDL1 blockade. Possible explanations for the divergent results might be related to the number of Tregs transferred, timing of analysis or model used. Nonetheless, other authors have suggested that the expression of PDL1 on tumor cells exerts an antiapoptotic effect and renders them resistant to apoptosis (37–38), which findings are in accordance with the observed increased apoptosis of Tregs in our results.

In summary, it has been shown that Tregs are able to prolong graft survival not only because of their suppression of Teff but also because of their anti-inflammatory properties (39). It has also been reported that Tregs represent a key mechanism in inducing fetal acceptance during pregnancy both in mice and humans (6, 40). PD1-PDL1 negative costimulation mediates the induction of peripheral tolerance both in a transplantation model as well as in a fetomaternal experimental model by favouring the regulatory function exerted by Tregs (9–10, 13–14, 30). In this study we show that PD1-PDL1 negative costimulation is essential to maintain fetomaternal tolerance because of its role in limiting Teff expansion and promoting Tregs in an antigen-specific manner. We have demonstrated that tolerance at the fetomaternal interface results from the balance of different factors that facilitate the regulation of the immune response. In this context, we propose that skewing the response towards Th17 away from Tregs by inhibiting PDL1 costimulation can be crucial in breaking fetomaternal tolerance and this effect seems to be mediated by a conversion of CD4⁺Foxp3⁻ cells in Th17 cells and increased apoptosis of Tregs. Furthemore, the neutralization of IL17 or the adoptive transfer of antigen-specific Tregs is able to abrogate the deleterious effect of PDL1 blockade. Therefore, PDL1 expression may represent the key factor in balancing Tregs and Th17 cells and maintaining tolerance at the fetomaternal interface. Further studies are urged to assess a possible role for PD1-PDL1 costimulation in affecting the interplay between Tregs and Th17 in allograft rejection during transplantation.