Contribution of the PD-1 ligands/PD-1 signaling pathway to dendritic cell-mediated CD4+ T cell activation

Authors


Abstract

Dendritic cells (DC) are extremely proficient inducers of naïve CD4+ T cell activation due to their high expression level of peptide-MHC and an array of accessory molecules involved in cell migration, adhesion and co-signaling, including PD-1 ligand 1 (PD-L1) and PD-1 ligand 2 (PD-L2). Whether PD-L1 and PD-L2 have a stimulatory or inhibitory function is a matter of debate, and could be partially dependent on the model system used. In this study we examined the role of PD-L1 and PD-L2 expressed by DC in naïve CD4+ T cell activation in a more physiologically relevant model system, using OVA-specific T cells in combination with various levels of TCR stimulation. Overexpression of PD-L1 or PD-L2 by DC did not inhibit T cell proliferation, even when B7–1 and B7–2 mediated costimulation was absent, although IL-2 production was consistently decreased. Surprisingly, blocking PD-L1 and PD-L2 with soluble programmed death-1 (sPD-1) also inhibited T cell activation, probably via reverse signaling via PD-L1 and/or PD-L2 into DC, leading to reduced DC maturation. This study suggests a relatively minor contribution of PD-1 ligands in DC-driven CD4+ T cell activation and provides evidence for reverse signaling by PD-L1 and PD-L2 into DC, resulting in a suppressive DC phenotype.

Abbreviations:
B7KO DC:

CD80- and CD86-deficient DC

hIgG1:

human IgG1

PD-1:

Programmed death-1

PD-L1:

PD-1 ligand 1

PD-L2:

PD-1 ligand 2

sPD-1:

soluble PD-1

Introduction

DC are extremely proficient inducers of T cell activation due to their high expression level of MHC-peptide complexes and an array of accessory molecules involved in cell migration, adhesion and co-signaling. In particular the B7–1/2-CD28/CTLA-4 axis of co-signaling molecules (‘signal 2′) has been shown to play an important role in this respect. A new layer of complexity has been added with the discovery of novel B7 family members, with stimulatory as well as inhibitory properties. Two of these newly described molecules are PD-1 ligand 1 (PD-L1 also known as B7-H1) 1, 2 and PD-1 ligand 1 (PD-L2 also known as B7-DC) 3, 4. PD-L1 is expressed on a broad variety of murine tissues and cells, including DC, and can be up-regulated by inflammatory stimuli. PD-L2 expression is restricted to macrophages and DC 5. Both costimulatory molecules share the same receptor programmed death-1 (PD-1), which is expressed on activated T cells. The majority of experimental data suggest that ligation of PD-1 inhibits T cell activation or induces cell death 6. PD-1 contains an immunoreceptor tyrosine-based inhibitory motif, and mice deficient in PD-1 develop autoimmune disorders, suggesting a defect in peripheral tolerance 7, 8. Whereas there is consensus that PD-1 delivers an inhibitory signal to T cells, the precise function of PD-L1 and PD-L2 is less clear, as various studies have shown that PD-L1 and PD-L2 can act as simulators 2, 4, 911 or inhibitors 1, 3, 1214 of T cell activation in vitro. A possible explanation for the observed discrepancies could be the fact that most studies varied with respect to (highly artificial) T cell stimuli, consisting of CD3- or TCR-specific agonistic antibodies or artificial APC lines expressing MHC-peptide complexes in combination with transfected PD-L1 or PD-L2. Although these model systems are very useful to initially probe for the mechanism of newly identified costimulatory molecules, they do not take into account several important aspects of professional APC such as DC. First, mature DC display a plethora of costimulatory molecules on their cell surface besides MHC-peptide complexes, which will influence the strength and type of induced T cell response. Second, APC function may also be altered by PD-L1 and PD-L2 reverse signaling. Reverse signaling through B7.1 and/or B7.2 by CTLA-4 and CD28 ligation can alter DC function 15, 16. Thirdly, DC have the potential to secrete cytokines such as IL-2, IL-7 and IL-15, which have been shown to reverse PD-1-mediated inhibition of human CD4+ T cells 17.

To investigate the role of PD-L1 and PD-L2 in CD4+ T cell activation in a more physiological setting, we decided to modulate the expression of these proteins by either constitutive overexpression of pdl1 or pdl2 in DC or blocking of PD-L1 and PD-L2 signaling with soluble PD-1 (sPD-1).

In agreement with previous reports, overexpression of PD-L1 and PD-L2 by DC inhibited CD4+ T cell activation, although less profound in comparison with artificial systems lacking costimulatory molecules. Quite surprisingly, sPD-1 inhibited T cell activation as well, and this effect may be mediated via reverse signaling into DC.

Results

Effect of constitutive PD-L1 and PD-L2 overexpression by DC

To obtain insight in the role of PD-L1 and PD-L2 in a more physiological system, we transduced DC with retroviral vectors expressing either murine PD-L1 or PD-L2, bicistronically expressed with GFP to permit single cell analysis of transgene expression. As shown in Fig. 1, PD-L1 was highly expressed on untransduced DC (GFP, dashed line), but was further increased in PD-L1-transduced DC (GFP+, bold line). The difference in PD-L2 expression between untransduced and PD-L2-IRES-GFP RV-transduced DC was higher, increasing approximately fivefold (MFI of 608 versus 3339 for GFPand GFP+ cell populations, respectively).

Figure 1.

Transduction of bone marrow-derived DC with PD-1 ligands. Bone marrow cells, depleted of lineage positive cells, were transduced at day 2, 3 and 4 with either mPD-L1-IRES-GFP RV or mPD-L2-IRES-GFP RV and cultured with GM-CSF. At day 11, cells were stained for PD-L1 or PD-L2 and expression analyzed by FACS. Numbers denote MFI.

Next, we wondered whether constitutive overexpression of PD-L1 and PD-L2 would result in an additional inhibitory effect on T cell proliferation and IL-2 production, as previously observed with artificial APC. To address this issue, we transduced bone marrow-derived DC with either mPD-L1 (PDL1-DC), mPD-L2 (PDL2-DC) or control vector (control-DC) and subsequently isolated transduced DC by FACS sorting based on GFP expression. Next, these transduced DC where co-cultured with CFSE-labeled DO11.10 CD4+ T cells at various OVA peptide concentrations. After 4 days, analysis of the CFSE-dilution pattern between control-DC stimulated and PDL1-DC- or PDL2-DC-stimulated T cells revealed no differences in T cell proliferation at any of the OVA peptide concentrations tested (Fig. 2A, left panels). IL-2 levels, however, were decreased in the cultures containing PDL1-DC or PDL2-DC, in particular at an OVA peptide concentration of 0.1 µg/mL (Fig. 2B, filled bars). Thus, although DC constitutively expressing PD-L1 or PD-L2 were not capable of decreasing CD4+ T cell proliferation, production of the autocrine growth factor IL-2 was impaired, demonstrating that even in a strong stimulatory setting as provided by mature DC, constitutive overexpression of inhibitory ligands could convey an inhibitory signal to T cells.

Figure 2.

Overexpression of PD-1 ligands by DC has a partial inhibitory effect on CD4+ T cell responses independent of CD80/86 expression. CFSE-labeled CD4+ T cells were co-cultured with marrow-derived DC from wild-type (wild-type DC) or CD80 and CD86-deficient mice (B7KO DC), transduced with either IRES-GFP-RV (control-DC), mPD-L1-IRES-GFP RV (PDL1-DC) or mPD-L2-IRES-GFP RV (PDL2-DC) in the presence of indicated concentrations OVA peptide. At day 4, T cell division was analyzed and levels of IL-2 in the supernatant determined. (A) CFSE profile of gated TOPRO-3+, KJ1–26+ cells. (B) IL-2 levels in supernatants of DC-T cell cultures. Numbers above each culture indicate levels of IL-2 in pg/mL. No CD4+ T cell division occurred in the absence of antigen (data not shown). Error bars indicate SD of duplicate wells. BD; below detection. Data shown are representative for one out of two to three independent experiments.

Effect of constitutive PD-L1 and PD-L2 overexpression by B7KO DC

Considering the small inhibitory effect seen on T cell activation when DC constitutively expressing PD-L1 or PD-L2 were used as APC, we next wondered whether this inhibitory effect would be more pronounced if strong costimulatory molecules for naïve T cells were absent. To address this question, we transduced DC obtained from CD80/CD86-deficient bone marrow cells (B7KO DC) with mPD-L1, mPD-L2 or control vector and used these DC to stimulate naïve OVA-specific CD4+ T cells as described above. T cell proliferation was decreased at any indicated concentration of antigen when B7KO DC were used to stimulate T cells compared to wild-type DC (Fig. 2A). However, in the absence of B7 costimulation, we did not observe any inhibition in cell proliferation when PD-L1- or PD-L2-transduced DC were used compared to control-DC (Fig. 2A, right panels). IL-2 concentrations in T cell cultures stimulated with B7-1- and B7-2-deficient APC were lower compared to wild-type DC cultures, in agreement with the observed decrease in T cell proliferation in general (Fig. 2B, open bars). Within the B7KO DC group, constitutive expression of PD-L1 and PD-L2 resulted in a decrease in IL-2 production at a high antigen dose (Fig. 2B, open bars).

To summarize, DC deficient in CD80 and CD86 are poor stimulators of naïve CD4+ T cells, as expected. However, even in this setting overexpression of PD-L1 or PD-L2 by DC did not lead to decreased T cell proliferation.

Soluble PD-1 inhibits CD4+ T cell activation

In an attempt to further dissect the role of PD-L1 and PD-L2 expressed by DC in CD4+ T cell stimulation, we cultured OVA-specific T cells with DC presenting OVA peptide in the presence or absence of sPD-1 and measured the proliferation of T cells. sPD-1 consists of the extracellular domain of murine PD-1 fused to the constant region of human IgG1, which will block PD-L1 as well as PD-L2 interaction with PD-1. To take into account any aspecific effects of the immunoglobulin heavy chain, T cells were exposed to human IgG1 (hIgG1) as a control protein. As it has been shown that contribution of the PD-L1 and PD-L2 pathways to T cell stimulation is dependent on the TCR stimulus 1, 3, we used increasing concentrations of antigen. When hIgG1 was present during the T cell stimulation, proliferation was maximal at 0.1 µg/mL OVA peptide and decreased at higher concentrations, presumably due to activation induced cell death (Fig. 3). However, when sPD-1 was added to the culture, CD4+ T cell proliferation was decreased at lower antigen concentrations but increased at higher concentrations (1 and 10 µg/mL; dashed line). This suggests that sPD-1 inhibits CD4+ T cell activation by increasing the threshold of T cell activation.

Figure 3.

sPD-1 blocks T cell proliferation in vitro at low antigen concentrations. OVA-specific CD4+ T cells (1 × 105) were plated with 5 × 103 bone marrow-derived DC in the presence of indicated concentrations of OVA peptide and sPD-1 (dashed line; triangles) or hIgG1 (solid line; squares) at 30 µg/mL. Proliferation was measured after 4 days by [3H]thymidine incorporation (0.5 µCi/well) for the last 8 h of culture. Results are expressed as mean ± SEM of three wells.

Because measurement of T cell proliferation through [3H]thymidine incorporation only captures the proliferation during the last 8 h of the assay, we repeated the assay with CFSE-labeled CD4+ T cells which permits visualization of the entire division history. The division profile of T cells obtained through CFSE labeling correlated with the proliferation data of radioactive thymidine incorporation, with decreased proliferation at lower antigen concentrations when sPD-1 was present during the T cell stimulation (Fig. 4A). Thus, these proliferation data support a costimulatory role for the PD-L1 and/or PD-L2 pathway rather than an inhibitory function.

Figure 4.

T cell stimulation in the presence of sPD-1 decreases IL-2 production but augments IL-10 levels. CFSE-labeled, OVA-specific, CD4+ T cells (5 × 105) were plated with 2.5 × 104 bone marrow-derived DC in the presence of indicated concentrations of OVA peptide and sPD-1 or hIgG1 (30 µg/mL). After 4 days, cells and supernatant were harvested. (A) Division profile of TO-PRO-3, KJ1–26+ cells. (B) Cytokine levels in the supernatant. All cytokine concentrations are expressed in pg/mL. Error bars indicate SD of duplicate wells. Shown results are representative for five independent experiments.

Next, we analyzed whether sPD-1 also influences the cytokine profile of stimulated CD4+ T cells. After 4 days of the T cell stimulation, supernatants were harvested and assayed for cytokines known to be involved in T cell proliferation or differentiation. Levels of the autocrine growth factor IL-2 were significantly decreased when sPD-1 was present during the T cell stimulation, suggesting a mechanism for the observed inhibition of T cell proliferation (Fig. 4B). Analysis of the levels of the prototypic Th1 cytokine IFN-γ and the key Th2 cytokine IL-4 revealed a reduction of both cytokines in the sPD-1-containing DC-T cell cultures (Fig. 4B), indicating that sPD-1 does not influence CD4+ T cell polarization. Another cytokine with known T cell-suppressive properties is IL-10. In our in vitro CD4+ T cell culture system, levels of IL-10 were increased in the presence of sPD-1 compared to hIgG1, irrespective of antigen dose (Fig. 4B). Thus, to summarize, sPD-1 blocks T cell proliferation when the TCR signal is limited, possibly due to inhibition of IL-2 production and/or an increase in IL-10 production.

sPD-1 cannot inhibit T cell proliferation in the absence of DC

It has been reported that the known ligands for PD-1, PD-L1 and PD-L2, are expressed on murine CD4+ T cells (5, 18 and data not shown), and that reverse signaling via these ligands can occur in human CD4+ T cells 19. Therefore, we examined the direct effects of sPD-1 on CD4+ T cells, without any other cell population present. Purified CD4+ T cells were activated with anti-CD3 mAb and anti-CD28 mAb in the presence of sPD-1 or hIgG1 and after 72 h the proliferation was measured.

As shown in Fig. 5, proliferation in response to increasing CD3 stimulation in the presence of sPD-1 was nearly identical to hIgG1-exposed T cells, although a slightly lower but consistent inhibition of proliferation was observed when T cells were exposed to sPD-1. This suggest that although signaling via PD-1 ligands on T cells might have a minor contribution to suppression of CD4+ T cell activity, this pathway probably does not solely account for the strong inhibition of T cell proliferation and cytokine expression pattern as observed in the antigen-specific assays.

Figure 5.

sPD-1 does not exert its effect directly on T cells. Purified BALB/c CD4+ T cells (2 × 105) were stimulated with indicated concentrations of plate-bound anti-CD3, soluble anti-CD28 (2.5 µg/mL) in the presence of sPD-1 or hIgG1 (30 µg/mL) for 72 h. Proliferation was measured after 4 days by [3H]thymidine incorporation (0.5 µCi/well) for the last 8 h of culture. Results are expressed as mean ± SEM of three wells. Results are representative for two independent experiments.

DC exposed to sPD-1 acquire a suppressive phenotype

Because direct PD-1 stimulation of CD4+ T cells did not result in a significant inhibition of proliferation, we next focused our attention on the direct effects of sPD-1 on DC. Bone marrow-derived DC were harvested at day 10 and cultured for 24 h in the presence of sPD-1 or hIgG1, followed by analysis of maturation-associated cell surface markers (CD40, CD80 and CD86) and cytokine production. The percentage of DC, as defined by CD11c and MHC class II expression, was similar in the sPD-1 culture compared to hIgG1 exposed DC (data not shown). However, when sPD-1 was added, the levels of the cell surface expression of the maturation markers were lower compared to the hIgG1-exposed DC culture (Fig. 6A, bold solid line versus bold dashed line). To determine whether this effect was sPD-1 specific, we pre-neutralized sPD-1 with anti-PD-1 antibody.

Figure 6.

DC exposed to sPD-1 acquire a suppressive phenotype. Bone marrow-derived DC were harvested at day 10 and reseeded at 1 × 106 DC/mL in 96-wells U-bottom plates in the presence of sPD-1 or hIgG1 (30 µg/mL). In some conditions sPD-1 was preincubated in vitro with equal concentrations of anti-PD-1 antibody or isotype antibody. After 24 h, cells and supernatants were harvested. (A) Cell-surface expression of CD40, CD80 and CD86 on mature (CD11c+, MHC class II+) DC. (B) IDO gene expression. Stimulated DC were harvested, snap frozen, RNA isolated and gene expression determined with quantitative RT-PCR. Expression levels are normalized to ubiquitin C gene expression and relative to RNA transcript levels in hIgG1-treated DC (indicated by dashed line). As a positive control for IDO expression, DC were exposed to recombinant murine IFN-γ (200 U/mL). (C) IL-10 production. Error bars indicate SD of duplicate wells. Data shown are representative for two to four independent experiments.

Blocking sPD-1 with anti-PD-1 antibody prevented down-regulation of CD40 and CD80, and could partially inhibit the decrease of CD86 expression (Fig. 6A, dashed bold line versus dashed line), implying that sPD-1 signals into DC, probably via PD-L1 and/or PD-L2, leading to a decrease in maturation marker expression.

This phenomenon of ligands becoming receptors, termed ‘reverse signaling’ has been shown for other B7 family members as well. For example, CTLA4-Ig up-regulated expression of the inhibitory enzyme IDO via B7–1 and/or B7–2 signaling 15. Therefore, we analyzed ido expression in sPD-1-stimulated DC. As shown in Fig. 6B, sPD-1 stimulation, in contrast to IFN-γ, did not lead to up-regulation of IDO gene expression as assessed by quantitative RT-PCR. We did observe an increase in IL-10 production when DC were stimulated with sPD-1, however (Fig. 6C), which might explain the increased IL-10 production in DC-T cell co-cultures. We conclude that DC stimulated with sPD-1 acquire a suppressive phenotype as defined by decreased expression of maturation markers and moderate increase in IL-10 expression, which may explain the observed inhibition of CD4+ T cell activation, although this inhibition is IDO independent.

Discussion

The majority of the studies that have addressed the role of PD-1 ligation on T cell activation have used either antibody-mediated delivery of TCR and CD28 costimulatory signals, or cell lines artificially expressing peptide-MHC complexes and CD80/86 to stimulate T cells. This approach carries the risk of over-interpreting the results due to the high signal strength delivered and the limited number of co-signaling pathways used. Moreover, use of artificial APC or direct ligation of purified T cells by antibodies largely ignores the role of indirect effects of APC, such as cytokine release, which may affect T cell activation 2023. Three studies that did analyze the contribution of PD-1 signaling pathways in DC-CD4+ T cell interactions, did so in either an alloreactive setting or used DC in combination with anti-CD3 antibody, thereby providing a supraphysiological antigenic stimulus 12, 24, 25.

In this study, we investigated the contribution of the ‘signal 2′ molecules PD-L1 and PD-L2 to the total signal strength provided by mature, bone marrow-derived DC through both overexpression of PD-L1 or PD-L2 and blocking of PD-L1/PD-L2 with sPD-1, at various levels of TCR stimulation.

First, we analyzed the effects of constitutive PD-L1 or PD-L2 overexpression. DC differentiated in vitro from bone marrow cells in the presence of GM-CSF already express high levels of PD-L1 and PD-L2 (Fig. 1 and 5). Constitutive overexpression of molecules already endogenously expressed by DC such as IL-12 or OX40L have shown enhancing effects in CD4+ T cell polarization and immunostimulatory capacity, respectively 26, 27 and overexpression of PD-L1 by DC could prevent the induction of EAE 28, further demonstrating the validity of this approach. Furthermore, it has been postulated that the immunological T cell response initiated by DC is dependent on a fine balance between positive and negative costimulatory signals 6, 29, 30. This was demonstrated for PD-L1 by a report from Smith and colleagues 31, showing that macrophages of S. mansoni-treated mice exhibit higher levels of PD-L1, resulting in a PD-L1-dependent induction of T cell anergy. Another recent study elegantly revealed that the PD-1 pathway is required for intrinsic T cell tolerance (deletion and/or anergy) induced by resting DC 32. Indeed, despite the high expression of PD-L1 and PD-L2 by mature bone marrow-derived DC, constitutive overexpression of these inhibitory molecules resulted in decreased IL-2 production by T cells, demonstrating that alteration of the balance of stimulatory and inhibitory signals influences T cell activation. However, at none of the antigen concentrations used in this study did we observe an inhibition of T cell proliferation. Latchman and colleagues 33, using a similar OVA-restricted TCR T cell assay, reported similar findings. Using artificial APC, they did not observe inhibition of T cell proliferation at higher antigen concentrations. Another study, using a bead-based T cell activation system with agonistic anti-CD3 antibodies, showed profound inhibition of T cell proliferation in the presence of PD-L1 protein. However, when costimulation was provided by anti-CD28 antibodies, the inhibitory effect of PD-L1 was absent or reduced. Interestingly, exogenous IL-2 could overcome inhibition of proliferation at all time points examined 14. Although the different experimental setups make direct comparison between IL-2 levels difficult, we observed much higher levels of IL-2 in our culture systems. As CD28-mediated costimulation acts to a large extent by promoting IL-2 production 34, it appears that mature bone marrow DC as used in our experimental setup provide sufficient costimulation for T cell proliferation, even at low antigen concentrations. These findings were supported by an in vitro study blocking PD-L1/2-PD-1 pathway in an allogeneic monocyte-derived human DC-T cell assay. The effects of PD-L1/2 inhibition were more pronounced when immature DC were used as APC compared to mature DC, indicating that the expression of costimulatory molecules such as CD40, CD80 and CD86, which highly correlates with the maturation level of DC, can counterbalance the negative signals provided by PD-L1 and PD-L2 12. To test this hypothesis, PD-L1 or PD-L2 was overexpressed by DC lacking the costimulatory molecules CD80 and CD86 (B7KO DC). T cells stimulated with B7KO DC exhibited decreased proliferation and IL-2 production when compared with wild-type DC, as expected 35. However, even in the absence of costimulation provided by CD80 and CD86, we could not detect any differences in proliferation when B7KO DC were transduced with PD-L1 or PD-L2. Ito et al. 36 reported a breakdown of transplantation tolerance when B7KO recipients receiving allograft transplants were systemically treated with anti-PD-L1 antibody, suggesting PD-1-dependent T cell inhibition in the absence of B7–1/2 signaling.

Our experiments blocking the PD-1 pathway using sPD-1 did not yield the expected results as predicted from the overexpression experiments. We observed inhibition of T cell proliferation and IL-2 production in DC-T cell co-cultures upon addition of sPD-1 as well. Others have dissected the role of the PD-1 ligands in DC-T cell interaction, in two studies using human CD4+ T cells cultured with allogenic DC in the presence of antibodies to PD-L1, PD-L2 or both 12, 25. Contrarily to our observations, they found increased T cell proliferation when antibodies against PD-L1 or PD-L2 were present, favoring an inhibitory role for PD-L1 and PD-L2 during T cell stimulation. A possible explanation for this discrepancy between studies might be the use of different blocking reagents, the extracellular domain of PD-1 in this study versus antibodies against PD-l and PD-L2 in the other two studies. Two previous studies reported the use of sPD-1 to disrupt PD-L1/2 signaling systemically in a murine tumor model 37, 38. The improved anti-tumor immunity suggested an inhibitory role for the PD-1 pathway as well, but type (eukaryotic expression plasmid) and route (systemic) prevent a direct comparison with our findings. Finally, there might be differences in co-signaling effects of PD-L1 or PD-L2 between species, as previously noted for the B7 family member B7-H3 39, 40.

We found that the most likely explanation for the inhibitory effect of sPD-1 was reverse signaling through DC. Several reports have shown the existence of reverse signaling of B7 family ligands into DC, as shown in studies using soluble CTLA-4 or CD28, and has been illustrated for the HVEM-BLTA pathway as well 15, 16, 41. More specifically, a recent study demonstrated that reverse signaling via PD-L2 expressed by DC led to activation of DC, as measured by enhanced T cell activation ability, migration capacity, IL-12 production and survival. Interestingly, CD80 and CD86 expression levels remained similar between the treatment and control group 42. In our experiments, DC stimulated with sPD-1 acquired a suppressive phenotype, as judged by decreased expression of CD40, CD80 and CD86, an increase in IL-10 production, while we did not detect increased IL-12 production (data not shown). The ability of sPD-1 to induce IL-10 production has been demonstrated before for T cells 19. Importantly, we could reverse this suppressive phenotype through neutralization with anti-PD-1 antibody, implying that the acquisition of the suppressive DC phenotype occurs via PD-L1, PD-L2 or perhaps a yet unknown ligand for PD-1. Whether these contrasting effects of PD-1 ligand-mediated reverse signaling are dependent on the type of stimulus used (PD-1-fusion protein versus a patient-isolated anti-PD-L2 IgM) remains to be seen. Moreover, we have not formally established whether signaling occurs via PD-L1, PD-L2 or perhaps another ligand for PD-1. DC stimulation experiments using bone marrow-derived DC from PD-L1- and PD-L2-deficient animals are necessary to verify this. Another point of interest is whether these DC induce regulatory T cells 43. Although increased IL-10 production and decreased IL-2 production of T cells activated in the presence of sPD-1 may implicate this induction, preliminary data we obtained on phenotype and function, as well as other findings in the literature 32 do not support this hypothesis.

As PD-L1 is also expressed on activated CD4+ T cells, we also analyzed the effect of sPD-1 on CD4+ T cell proliferation directly. Proliferation was similar to hIgG1 control, indicating no reverse signaling occurred via PD-L1 expressed by CD4+ T cells. Such signaling is possible, as antibodies have been isolated from serum of a rheumatoid arthritis patient that stimulated CD4+ T cell proliferation and IL-10 production via PD-L1 19. Intriguingly, in this study sPD-1 was also able to deliver these costimulatory signals. This increased proliferation was accompanied by increased apoptosis, which could be partially blocked by anti-IL-10 antibody. Although we have not directly assessed apoptosis of sPD-1-stimulated T cells and we also found increased IL-10 levels in DC-T cell cultures, we do not observe increased proliferation in response to PD-1 ligand signaling. Moreover, both anti-IL-10 mAb and anti-IL-10R could not restore the sPD-1-mediated inhibition of T cell proliferation in DC-T cell cultures (data not shown), suggesting that IL-10-mediated CD4+ T cell apoptosis is not responsible for the inhibition in cell cycling. This is in agreement with findings of Latchman et al. 3, who observed cell cycle arrest but not apoptosis upon stimulation of CD4+ T cell with PD-L2-Ig.

In conclusion, we have shown that overexpression of PD-L1 or PD-L2 by DC as well as addition of sPD-1 to DC-T cell co-cultures both lead to decreased CD4+ T cell activation. The data obtained from PD-L1 and PD-L2 overexpression studies confirmed the proposed inhibitory role of these receptors in CD4+ T cell activation. However, addition of sPD-1 to CD4+ T cells cultured with DC illustrate the importance of physiological settings to analyze such pathways as the data implicated reverse signaling exerted by sPD-1 into DC as the mechanism of suppression. It could also explain the sometimes contrasting results obtained with studies analyzing the PD-1 pathway in vivo6.

Materials and methods

Mice

Female BALB/c mice (6–10 wks old) were purchased from Harlan (Horst, The Netherlands). OVA323–339-specific, MHC class II-restricted, TCR transgenic (DO11.10) mice 44 were obtained from The Jackson Laboratory (Bar Harbor, ME) and bred in-house. Mice deficient in CD80 and CD86 on the BALB/c background 45 were a kind gift from Dr M. Oosterwegel (Utrecht University, Utrecht, The Netherlands). Mice were housed in microisolators under specified pathogen-free conditions and experiments were performed under approval of the Erasmus MC committee for animal ethics.

Vector construction

The retroviral vector expressing murine PD-L1 was constructed by PCR amplification of the murine PD-L1 cDNA using pMET7-mPD-L1 as a template with the forward primer 5′-AGATCTTCTCCTCGCCTGCAGATAGT-3′, containing a BglII restriction site, and the reverse primer 5′-CTCGAGAAGCTGCCAATCGACGATCA-3′, containing a XhoI restriction site. The PCR product was cloned into pGEMTeasy (Promega, Madison, WI) and sequenced. The BglII/XhoI PD-L1 cDNA fragment was ligated into the BglII/XhoI-digested retroviral vector pIRES-GFP RV (46; provided by K.M. Murphy), resulting in pmPD-L1-IRES-GFP RV.

The retroviral vector expressing murine PD-L2 was constructed by PCR amplification of the murine PD-L2 cDNA using pBluescriptSK+-mPD-L2 as a template with the forward primer 5′-GAAGATCTCACCATGCTGCTCCTGCT-3′, containing a BglII restriction site, and the reverse primer 5′-CCTCGAGCCCTGCTCTAGATTAGATCCT-3′, containing an AvaI restriction site. The BglII/AvaI-digested PCR product was cloned into BglII/XhoI-digested pIRES-GFP-RV, resulting in pmPD-L2-IRES-GFP RV. For verification, the cloned mPD-L2 cDNA fragment was fully sequenced.

DC transduction

Retroviral particles were produced by transient transfection and used to transduce bone marrow-derived DC as described 26. At day 11, DC were harvested by gentle pipetting, resuspended in PBS and GFP-positive DC were sorted with a FACSDiva or FACSAria flow cytometer (BD Biosciences, Erembodegem, Belgium). Purity of GFP-positive DC after sorting was on average 88% of alive cells (85–92%, 95% confidence interval). DC transduced with IRES-GFP-RV, mPD-L1-IRES-GFP RV or mPD-L2-IRES-GFP RV are hereafter designated control-DC, PDL1-DC and PDL2-DC, respectively.

T cell proliferation assays

Spleen and lymph node (LN) cells were obtained from DO11.10 mice and untouched CD4+ T cells isolated by negative selection using the CD4 T cell isolation kit and autoMACS (Miltenyi Biotec, Bergisch Gladbach, Germany) The resulting population was typically >95% CD4+. To visualize T cell divisions, cells were labeled with CFSE as described previously 47.

sPD-1 consists of the extracellular domain of murine PD-1 fused to the Fc fragment of human IgG1 (Chimerigen Laboratories, Allston, MA). As a control, human IgG1 (hIgG1, Sigma-Aldrich) was used. sPD-1 or hIgG1 were present in cell cultures at a concentration of 30 µg/mL.

To visualize CD4+ T cell divisions, 5 × 105 CFSE-labeled CD4+ T cells were cultured with 2.5 × 104 bone marrow-DC, generated as described 48 in 48-wells plates and indicated concentrations of OVA323–339 peptide (Ansynth, Roosendaal, The Netherlands). After 4 days, cells and supernatants were harvested and OVA-specific T cell proliferation and cytokine levels determined.

To measure proliferation in DC-T cell co-cultures through [3H]thymidine incorporation, 1 × 105 CD4+ T cells were cultured together with 5 × 103 bone marrow-DC and various concentrations of OVA323–339 peptide in 96-wells U-bottom plates for 4 days. [3H]Thymidine (0.5 µCi/well) was added for the last 8 h.

To determine the effect of sPD-1 on CD4+ T cell independent of APC, 96-wells U-bottom plates were coated with anti-CD3 mAb (145–2C11; BD Bioscience) at various concentrations. Next, 2 × 105 purified CD4+ T cells from BALB/c mice were stimulated in the presence of 2.5 µg/mL anti-CD28 mAb (37.51; BD Biosciences) and sPD-1 or hIgG1 (30 µg/mL). Proliferation was measured by [3H]thymidine incorporation (0.5 µCi/well) for the last 8 h of a 72-h culture.

DC stimulation experiments

Bone marrow-DC were harvested at day 10, washed, replated at 2 × 105 DC/well in 96-wells U-bottom plates and stimulated for 24 h with sPD-1 or hIgG1. To determine the specificity of the inhibitory effects of sPD-1 on DC phenotype, sPD-1 or hIgG1 were incubated in vitro with equal concentrations anti-PD-1 antibody (J43; eBioscience, San Diego, CA) or control Ig (hamster IgG; eBioscience) for 30 min before addition to the DC cultures. In some experiments, DC were cultured in the presence of recombinant murine IFN-γ (200 U/mL; Peprotech, Rocky Hill, NY) to up-regulate IDO expression.

After 24 h, cells were harvested to determine cell surface marker expression and IL-10 production by ELISA. For quantitative RT-PCR, cell pellets were snap frozen in liquid nitrogen and stored at –80 °C until RNA isolation.

Real-time quantitative RT-PCR

Frozen cell pellets were homogenized, RNA isolated with RNeasy mini-prep columns (Qiagen, Hilden, Germany) and treated on-column with DNaseI, according to the manufacturer's protocol. RNA (100 ng) was reverse transcribed using SuperscriptII (Invitrogen) and random hexamers (Amersham Biosciences, Roosendaal, The Netherlands) for 50 min at 42°C. Quantitative PCR was performed with Taqman Universal PCR Mastermix (Applied Biosystems, Foster City, CA) and preformulated primers and probe mixes (‘Assay on Demand’, Applied Biosystems). PCR conditions were 2 min at 50°C, 10 min at 95°C, followed by 40 cycles of 15 s at 95°C and 60°C for 1 min using an ABI PRISM 7900 HT (Applied Biosystems). PCR amplification of the housekeeping gene encoding ubiquitin C was performed during each run for each sample to allow normalization between samples.

Cytokine measurements

Levels of cytokines in culture supernatants were measured using OptEIA kits (BD Biosciences) according to manufacturer's instructions.

Flow cytometry

Anti-FcγRII/III antibody (2.4G2, ATCC, Manassas, VA) was included in all cell surface stainings to reduce nonspecific antibody binding. Dead cells were excluded by labeling with TOPRO-3 (Molecular Probes, Leiden, The Netherlands) prior to acquisition. Functional expression of murine PD-L1 and PD-L2 after transduction was confirmed by staining with biotinylated anti-PD-L1 (MIH5) and biotinylated anti-PD-L2 (TY25) (both purchased from eBioscience) against these molecules, followed by streptavidin-APC (BD Biosciences). To study DC-mediated CD4+ T cell division, cells were labeled with CFSE and with the anti-clonotypic DO11.10 TCR mAb KJ1–26 (Caltag Laboratories, Burlingame, USA) 44. For the DC stimulation experiments, DC were stained with anti-I-A/I-E-FITC (2G9) in combination with anti-CD11c-APC (HL3) and anti-CD80-PE (16–10A1), anti-CD86-PE (GL1) or anti-CD40-PE (3/23) (all antibodies from BD Biosciences).

Events were acquired on a FACSCalibur flow cytometer (BD Biosciences) and analyzed with FlowJo software (Treestar, Ashland, OR).

Acknowledgements

This work was supported by a grant from The Netherlands Asthma Foundation (NAF 32.00.45; to H.K. and B.L.) and a Netherlands Organization for Scientific Research VIDI fellowship (to B.L.). We would like to thank Dr. C. Malizwesky (Amgen, Seattle, WA) for providing rhFlt3-L, Dr. M. Oosterwegel (Utrecht University, The Netherlands) for the generous gift of CD80/CD80 KO mice, Prof. K. Thielemans (Vrije Universiteit Brussel, Brussels, Belgium) for rmGM-CSF and Prof. K. Murphy, Washington University, WA) for IRES-GFP RV.

Footnotes

  1. 1

    WILEY-VCH

  2. 2

    WILEY-VCH

  3. 3

    WILEY-VCH

  4. 4

    WILEY-VCH

  5. 5

    WILEY-VCH

  6. 6

    WILEY-VCH

Ancillary