PD-1 signalling in CD4⁺ T cells restrains their clonal expansion to an immunogenic stimulus, but is not critically required for peptide-induced tolerance

Abstract

The ultimate outcome of T-cell recognition of peptide–major histocompatibility complex (MHC) complexes is determined by the molecular context in which antigen presentation is provided. The paradigm is that, after exposure to peptides presented by steady-state dendritic cells (DCs), inhibitory signals dominate, leading to the deletion and/or functional inactivation of antigen-reactive T cells. This has been utilized in a variety of models providing peptide antigen in soluble form in the absence of adjuvant. A co-inhibitory molecule of considerable current interest is PD-1. Here we show that there is the opportunity for the PD-1/PD-L1 interaction to function in inhibiting the T-cell response during tolerance induction. Using traceable CD4⁺ T-cell receptor (TCR) transgenic cells, together with a blocking antibody to disrupt PD-1 signalling, we explored the roles of PD-1 in the induction of tolerance versus a productive immune response. Intact PD-1 signalling played a role in limiting the extent of CD4⁺ T-cell accumulation in response to an immunogenic stimulus. However, PD-1 signalling was not required for either the induction, or the maintenance, of peptide-induced tolerance; a conclusion underlined by successful tolerance induction in TCR transgenic cells genetically deficient for PD-1. These observations contrast with the reported requirement for PD-1 signals in CD8⁺ T-cell tolerance.

Introduction

The ultimate outcome of T-cell recognition of an antigenic peptide is determined by the molecular context in which antigen presentation is provided. Thus, an appropriate response to pathogen-derived antigen can be achieved, whereas immune tolerance to self-antigen is induced or maintained.¹ Tolerance is now acknowledged to be an active process and, rather than being completely absent, certain costimulatory molecules probably play an important role in its induction.² Using the systemic instillation of antigenic peptides without adjuvant – a tolerogenic regime with therapeutic applications – we have reported on the relative expression of various costimulatory pairs on T cells and dendritic cells (DCs), under conditions leading to tolerance versus immunity.³ Amongst these, we noted that PD-1 was rapidly expressed by T cells en route to tolerance. Given the considerable interest in targeting PD-1 signalling as a means of adjusting the balance between tolerance and immunity⁴ we examined, in this study, its role in the induction of both tolerance and immunity in CD4⁺ T cells.

PD-1 is expressed upon activation on CD4⁺ and CD8⁺ T cells, B cells, natural killer T cells and monocytes.⁴ PD-1 is a member of the immunoglobulin family and its ligation appears to provide inhibitory signals that dampen T-cell receptor (TCR) signalling;^5,6 thus, PD-1 is termed a co-inhibitory molecule.⁷ Two ligands (PD-L1 and PD-L2) have been identified, which differ in their expression patterns. In mice, expression of PD-L2 is only induced upon the activation of DCs, macrophages and bone-marrow-derived mast cells.⁸ By contrast, PD-L1 is constitutively expressed by most haematopoietic cell types and can be further up-regulated following cell activation.⁹ Interestingly, PD-L1 is also expressed on an array of non-haematopoietic cell types and at sites of immune privilege, including the brain, placenta and eye,^10–12 suggesting that it might play an important barrier role in preventing T cells entering these sites.

Experiments using experimental autoimmune encephalomyelitis (EAE),¹³ diabetes in the non-obese diabetic (NOD) mouse,¹⁴ allograft rejection¹⁵ and other immune-driven conditions⁴ have all shown enhanced severity and/or onset upon disruption of the PD-1/PD-L interaction. Despite this, few studies have examined the role of PD-1 during the early activation of a naïve cell, when the decision between T-cell tolerance and immunity is being made. Those studies that posed this question concluded that PD-1 signalling makes a vital contribution to the induction of T-cell tolerance. Probst et al.¹⁶ demonstrated that, unlike their wild-type counterparts, PD-1^−/− CD8⁺ T cells were resistant to unresponsiveness leading from in vivo interactions with resting DCs. Two groups have highlighted the importance of PD-1-mediated signals in the induction of tolerance in ovalbumin (OVA)-reactive CD8⁺ (OT-I) cells. Transfer of OT-I cells to mice expressing OVA under the rat insulin promoter (RIP-OVA mice) results in their deletion. PD-1^−/− OT-I cells avoid this fate and initiate diabetes, a result that could also be achieved by in vivo antibody blockade of PD-1 and/or PD-L1.^17,18 Using a similar model in which OVA was expressed exclusively within the small intestine, a role for PD-1 in maintaining intestinal tolerance amongst CD8⁺ T cells has recently been demonstrated.¹⁹ Furthermore, PD-1-mediated signals have been reported to be vital for the induction of peptide-induced T-cell tolerance in TCR transgenic CD8⁺ T cells.^20,21

Adoptive transfer models using TCR transgenic CD4⁺ T cells have established that administration of the relevant peptide via a variety of routes is also highly effective at inducing tolerance.^22,23 The consensus appears to be that this in vivo exposure to peptide and major histocompatibility complex (pMHC) triggers a transient activation state, with a proliferative burst that cannot be sustained. The majority of the activated T cells then enter apoptosis and are deleted from the immune system; those that persist show an unresponsive phenotype.^23,24

Using this simple, linear model of CD4⁺ T-cell responsiveness, we sought to determine the role(s) of PD-1 signals in peptide-induced tolerance. OT-IIxCD45.1 TCR transgenic mice provided a source of naïve CD4⁺ T cells recognizing the OVA (323–339) peptide (hereafter referred to as pOVA) that could be tracked following their transfer into C57BL/6 hosts. Using an antibody to block PD-1 signalling in vivo, we showed that these signals are important in restricting the size of the productive immune response. However, we found no evidence to support the notion that PD-1 signals are required for the induction of tolerance. We also utilized a second CD4⁺ transfer model to show that PD-1^−/− T cells are fully susceptible to this form of tolerance. These data contrast with current thinking and indicate that further exploration of the role of PD-1 in immunity and tolerance will instruct the most effective development of new therapeutics targeting PD-1 signalling.

Materials and methods

Mice, antigen and antibodies

C57BL/6 mice and OT-IIxCD45.1 mice^3,25 expressing an A^b-restricted, pOVA-reactive TCR were bred and maintained under pathogen-free conditions in the Institute of Immunology and Infection Research, University of Edinburgh. 2D2xCD90.1 and 2D2xCD90.1xPD-1^−/− mice, expressing the A^b-restricted, 35–55 peptide of myelin oligodendrocyte glycoprotein (pMOG)-reactive TCR,^26,27 were bred under specific pathogen-free conditions at the University of Mainz. Sex-matched, 6 to 8-week-old mice were used for all experiments. All animal experiments were approved by the relevant institutional ethical review committee.

pOVA and pMOG were prepared at the Advanced Biotechnology Centre, Imperial College, UK. The RMP1-14 anti-PD-1 Ig (rat IgG2a) has been described previously.²⁸ The MAC-1 hybridoma (obtained from the ECACC, Wiltshire, UK) provided a rat IgG2a isotype-control antibody.

T-cell transfer model

Peripheral lymph nodes and spleens from TCR transgenic mice were disaggregated and resuspended in magnetic antibody cell sorting (MACS) buffer [Hanks’ balanced salt solution (HBSS) containing 5 × 10⁻⁵ m 2-mercaptoethanol (2-ME), 100 U/ml of penicillin and 100 μl/ml of streptomycin; all from Gibco, Life Technologies, Paisley, UK]. Erythrocytes were depleted using red blood cell (RBC) lysis buffer (Sigma, Poole, UK). CD4⁺ T cells were purified using CD4-conjugated MACS beads with MS or LS columns and a VarioMACS magnet (all Miltenyi Biotec, Bergisch Gladbach, Germany), according to the manufacturer’s instructions. A consistent purity of ∼95% was confirmed by flow cytometry. Purified cells (1–2 × 10⁶) were injected intravenously (i.v.) into naïve C57BL/6 recipients.

Induction of tolerance or immunity

For the induction of tolerance, mice received a single dose of 500 μg of pOVA, or 200 μg of pMOG, i.v., 1 day post OT-II or 2D2 cell transfer respectively. Mice were then immunized with 20 μg of pOVA in complete Freund’s adjuvant (CFA) (Sigma) or with 100 μg of pMOG subcutaneously (s.c.) in CFA 7 days after tolerance induction. In some experiments immunity was induced (without prior induction of tolerance) 1 day after T-cell transfer, by the administration of 500 μg of pOVA plus 30 μg of lipopolysaccharide (LPS) i.v., or by immunization with 20 μg of pOVA in CFA s.c.

Phenotypic analysis of T cells and DC populations

Spleens were injected with 100 μl of 8 mg/ml of collagenase IV (Worthington, Lakewood, NJ) and were incubated at 37° for 20 min before manual disaggregation. DC and T-cell populations were characterized by flow cytometry. Cells were stained for fluorescence-activated cell sorter (FACS) analysis using the following antibodies (all from BD Pharmingen, except where stated); anti-CD4-allophycocyanin, anti-CD4-peridinin chlorophyll protein (PerCP), anti-CD45.1-phycoerythrin (PE)/fluorescein isothiocyanate (FITC) (clone A20), anti-CD8-allophycocyanin (clone 53–6.7; eBiosciences, Hatfield, UK), anti-CD11c-FITC (clone HL3), anti-PDCA-1-biotin (clone JF05.1C2.4.1; Miltenyi Biotec), anti-B220-PerCP (clone RA3-6B2), anti-PD-1-PE (clone J43; eBiosciences), anti-PD-L1-PE (clone MIH5; eBiosciences), anti-PD-L2-PE (clone TY25; eBiosciences), and rat IgG2a-PE. All samples were collected on a Becton Dickinson LSRII with diva software (Becton Dickinson, San Jose, CA). Data were analyzed using FlowJo software (Treestar, Ashland, OR).

Ex vivo assessment of T-cell function

Single-cell suspensions were prepared from experimental mice by manual disaggregation through gauze. Following RBC depletion, cells were plated out at a concentration of 5 × 10⁵ lymph node cells or 8 × 10⁵ splenocytes per well of a 96-well microtitre plate with X-VIVO-15™ tissue culture medium (BioWhittaker, Wokingham, UK) and stimulated with a dose range of the appropriate peptide. After 48 hr cultures were pulsed with [³H]thymidine (Amersham, Amersham, UK) at 0·5 μCi/well for the last 18 hr of culture. [³H]Thymidine incorporation was measured using a scintillation β-counter (Wallac, Milton Keynes, UK). The results are expressed as mean counts per minute (c.p.m.) ± standard error of the mean (SEM). Alternatively, after 48 hr of culture, supernatants were removed for determination of cytokine levels using enzyme-linked immunosorbent assay (ELISA) analyses. Intracellular cytokine production was measured ex vivo following overnight stimulation of lymph node or splenocyte cultures with 100 μm pOVA. Brefeldin A was added, as per the manufacturer’s instructions (1 in 1000 dilution), to each well for the last 4 hr of culture.

Statistics

Statistical analysis was performed either using an unpaired Student’s t-test, when comparing two experimental groups, or by an analysis of variance (anova) using Tukey’s multiple comparison test, when comparing three or more experimental groups. Differences were regarded as significant when P-values were < 0·05.

Results

PD-L1 expression following tolerogenic or immunogenic treatment differs to that of PD-L2

Here we examined the expression of PD-1, PD-L1 and PD-L2 following tolerogenic and immunogenic peptide administration. OT-IIxCD45.1 cells were transferred to C57BL/6 (CD45.2) hosts, which then received an i.v. injection of antigen, either in a tolerogenic form [pOVA in phosphate-buffered saline (PBS)], or in an immunogenic form (pOVA + LPS), or PBS alone. Splenocytes were then sampled after 1, 2 and 3 days to allow flow cytometric examination of PD-1 on OT-II T cells and of PD-L1 and PD-L2 levels on the OT-II T cells, on conventional DCs (CD11c⁺; cDCs) and on plasmacytoid DCs (pDCs; defined by expression of PDCA-1).

As found previously,³ PD-1 expression on OT-II cells remained negligible in the PBS-treated controls, but was up-regulated from days 1–3 during the induction of both tolerance and immunity, with peak expression found on day 2 (Fig. 1). Consistent with other studies,⁹ OT-II cells expressed PD-L1 in the absence of TCR stimulation (PBS group, Fig. 2a). This was strongly upregulated at the 24-hr time-point, and was up-regulated to a similar extent following either the treatment with pOVA alone, or the treatment with pOVA + LPS. Expression then declined almost to background levels by day 3. PD-L1 was expressed on DCs from naïve hosts, again consistent with previous reports.⁹ Analysis of DC subpopulations revealed that PD-L1 was expressed to a slightly lower level on pDCs compared with cDCs (Fig. 2b,c). Up-regulation of PD-L1 expression by cDCs and pDCs was observed only after administration of pOVA + LPS (and not pOVA alone). Expression was still enhanced on cDCs on day 2 (but at a lower level than on day 1), and by day 3 had almost returned to baseline levels (Fig. 1b). Given their lower resting expression levels, the up-regulation of PD-L1 on pDCs seen 24 hr after LPS administration was most striking (Fig. 1c). Moreover, a fraction of pDCs maintained high-level expression as far as day 3. LPS-induced expression of PD-L1 is therefore more pronounced on pDCs than on cDCs.

Figure 1.

PD-1 expression on OT-II T cells following tolerogenic or immunogenic administration of the 323–339 peptide of OVA (pOVA). C57BL/6 mice (three per group) received OT-IIxCD45.1 cells 1 day before the administration of pOVA (tolerance), pOVA + lipopolysaccharide (pOVA + LPS) (immunity) or phosphate-buffered saline (PBS) intravenously (i.v.). Spleens were sampled on days 1–3 and PD-1 expression was determined by fluorescence-activated cell sorting (FACS) analyses. (a) PD-1 expression on day 2. Filled histograms show PD-L1/2 staining and open histograms show isotype-control staining. (b) Summary of PD-1 expression (data are expressed as mean percentage of OT-II-expressing PD-1 ± standard deviation).

Figure 2.

PD-L1 and PD-L2 expression on OT-II T cells, CD11c^hi dendritic cells (DCs) and PDCA-1⁺ DCs following administration of tolerogenic peptide or immunogenic peptide. C57BL/6 mice (three per group) received OT-IIxCD45.1 cells 1 day before administration of the 323–339 peptide of OVA (pOVA) (tolerance), pOVA + lipopolysaccharide (pOVA + LPS) (immunity) or phosphate-buffered saline (PBS), intravenously (i.v.). Spleens were sampled at the indicated time-points, and the expression of PD-L1 (a–c) and PD-L2 (d–f) was determined by fluorescence-activated cell sorting (FACS). Filled histograms show PD-L1/2 staining, and open histograms show isotype-control staining. Numbers on histograms refer to the mean percentage of cells expressing each molecule per group on that day. The results are from one of two separate experiments giving consistent results.

PD-L2 was not expressed on OT-II cells at any time-point or following any treatment (Fig. 2a). PD-L2 was only expressed on a fraction of cDCs, and on a similar fraction of pDCs, following treatment with pOVA + LPS, and this was only evident on day 1 (Fig. 2b,c).

PD-1 signals are required to limit a productive immune response

Because PD-1 and PD-L1 were up-regulated on the OT-II cells, and both PD-L1 and PD-L2 were up-regulated by DCs under conditions leading to a productive immune response (pOVA + LPS), we sought to determine the role of PD-1 signals during this process (Fig. 3). At the same time as priming with pOVA + LPS i.v., mice received either 250 μg of anti-PD-1 (clone RMP1-14), or 250 μg of the isotype control (MAC-1 or rat IgG) intraperitoneally (i.p.). PD-1 blockade led to enhanced ex vivo recall responses to pOVA by splenocytes sampled 7 days after injection of pOVA + LPS (Fig. 3b,c). The production of interleukin (IL)-2 and interferon-γ (IFN-γ) (data not shown), and proliferative responses, were all enhanced following treatment with anti-PD-1. These enhanced recall responses reflected a markedly increased expansion of OT-II cell numbers in the group receiving anti-PD-1 (Fig. 3a).

Figure 3.

Administration of anti-PD-1 enhances T-cell immunity by allowing greater expansion of antigen-reactive T cells. (a–c) OT-IIxCD45.1 cells were transferred to C57BL/6 mice 1 day before administration of the 323–339 peptide of OVA (pOVA) + lipopolysaccharide (LPS) intravenously (i.v.), plus either 250 μg of anti-PD-1 or isotype-control antibody intraperitoneally (i.p.). Ex vivo recall responses were examined 7 days after treatment. (a) OT-II cells in the spleen, expressed as a percentage of all CD4⁺ cells. Ex vivo (b) interleukin-2 (IL-2) production and (c) proliferation of splenocytes to increasing concentrations of pOVA. (d–f) OT-IIxCD45.1 cells were transferred to C57BL/6 mice 1 day before immunization with pOVA in complete Freund’s adjuvant (CFA) plus either 250 μg of anti-PD-1 or isotype-control antibody i.p. (d) OT-II cells in the spleen, expressed as a percentage of all CD4⁺ cells. Ex vivo (e) IL-2 production and (f) proliferation of splenocytes to increasing concentrations of pOVA. *P < 0·04 and **P < 0·0082 as determined by the unpaired t-test. The percentage of OT-II cells in the CD4⁺ population was significantly different following administration of anti-PD-1, as determined using the unpaired t-test. Data are from one of two or more experiments giving consistent results. c.p.m., counts per minute.

Consistent data were obtained following administration of anti-PD-1 at the time of immunization with pOVA in CFA, with splenocytes sampled 10 days later, showing that in vivo blockade of PD-1 led to elevated frequencies of OT-II cells (Fig. 3d) and increased IL-2 production and proliferation in response to pOVA (Fig. 3e,f). A similar pattern of responsiveness was seen in the draining LN (data not shown). These experiments confirmed that the dose of anti-PD-1 selected was sufficient to have a profound effect on T-cell responsiveness to immunogenic peptide administered either systemically via the i.v. route, or locally, when emulsified in CFA.

Blockade of PD-1 does not convert a tolerogenic exposure to peptide to an immunogenic stimulus

Based on previous reports of the role of PD-1 in CD8⁺ T-cell tolerance,^16,20,21 we predicted that blockade of PD-1 signalling might prevent tolerance induction in our system; moreover, this might convert a tolerogenic signal into an immunogenic one. Following adoptive transfer of OTIIxCD45.1⁺ cells, host mice received an i.v. injection of pOVA in PBS together with anti-PD-1 or the isotype-control antibody (given i.p.). A cohort of mice that received pOVA + LPS i.v. served as an immunogenic control (Fig. 4a).

Figure 4.

Administration of anti-PD-1 does not convert a tolerogenic dose of the 323–339 peptide of OVA (pOVA) to an immunogenic dose. (a) OT-IIxCD45.1 cells were transferred to C57BL/6 mice 1 day before the administration of pOVA + lipopolysaccharide (pOVA + LPS) intravenously (i.v.), or pOVA intravenously (i.v.) plus either 250 μg of anti-PD-1 or isotype-control antibody intraperitoneally (i.p.). Spleens were sampled 7 days after treatments. (b) OT-II cells in the spleen, expressed as a percentage of all CD4⁺ cells. Ex vivo (c) interleukin-2 (IL-2) production and (d) proliferation of splenocytes to increasing concentrations of pOVA. OT-II cell percentages and responses of both pOVA-treated groups were significantly different to the pOVA + LPS-treated group, but not from each other, as determined by an analysis of variance (anova) using Tukey’s multiple comparison test. Data are from one of two experiments giving consistent results.

Supplementing the tolerogenic stimulus of pOVA alone with PD-1 blockade did not produce productive immunity. By contrast, anti-PD-1 had no effect on the tolerant state induced in the OT-II cells. Similar levels of proliferation (Fig. 4d), and of IL-2 (Fig. 4c) and IFN-γ production (data not shown) were seen from pOVA-treated mice given anti-PD-1 or the isotype control. Treatment of mice with anti-PD-1 did not increase the expansion of OT-II cells following treatment with pOVA (Fig. 4b); most OT-II cells were still deleted following pOVA treatment, irrespective of anti-PD-1 administration. In all aspects examined, the mice that received anti-PD-1 together with pOVA/PBS therefore responded in precisely the same way as tolerant control mice and both were significantly different from mice that received pOVA + LPS.

Blockade of PD-1 does not prevent the establishment of tolerance

Although the above data showed that PD-1 blockade could not convert tolerance to immunity, those experiments did not preclude the possibility that administration of anti-PD-1 could have had more subtle effects, preventing the establishment of the tolerance programme in those OT-II cells that remained. The precedent for this comes from previous experiments showing such a disruption of tolerance, without a switch to immunity, when agonistic anti-CD40 or anti-OX40 were employed at the time of peptide treatment.²⁹ To test this possibility, experiments were performed, as described above, with the mice subsequently immunized with pOVA in CFA 7 days after i.v. peptide injection (Fig. 5). A cohort of mice that were injected with PBS alone following the OT-II transfer served as the immunogenic control, and these mice mounted robust ex vivo recall responses 10 days after the immunization with pOVA/CFA. Administration of anti-PD-1 at the time of tolerance induction did not increase the frequency of OT-II cells present, compared with the tolerized group receiving the isotype-control antibody, in either the spleen (Fig. 5a) or the LN (data not shown). This low frequency of OT-II cells was reflected in an absence of ex vivo recall responses to pOVA, as assessed by the production of IL-2, IFN-γ and IL-17, and proliferation (Fig. 5b–e). Collectively, these data lead to the conclusion that the systemic application of soluble peptide triggers a tolerance programme in CD4⁺ T cells that works independently of PD-1 signals.

Figure 5.

Treatment with anti-PD-1 does not prevent the establishment of peptide-induced tolerance in OT-II cells. OT-IIxCD45.1 cells were transferred to C57BL/6 mice 1 day before administration of the 323–339 peptide of OVA (pOVA) (tolerance) or phosphate-buffered saline (PBS) (immunity) intravenously (i.v.) plus either 250 μg of anti-PD-1 or isotype-control antibody intraperitoneally (i.p.). Mice were immunized with pOVA in complete Freund’s adjuvant (CFA) 7 days after peptide treatment and were killed 10 days after immunization. (a) OT-II cells in the spleen, expressed as a percentage of all CD4⁺ cells. Ex vivo (b) interleukin-2 (IL-2), (c) interferon-γ (IFN-γ), (d) interleukin-17 (IL-17) production and (e) proliferation of lymph node (LN) cells to increasing concentrations of pOVA. The percentages of OT-II cells from both tolerant groups were not significantly different. The responses of both pOVA-treated groups were significantly different to those of the PBS-treated group, but not to each other, as determined by an analysis of variance (anova) using Tukey’s multiple comparison test. Data are from one of two experiments giving consistent results.

Blockade of PD-1 does not overcome tolerance once it has been established

The abortive activation triggered by soluble peptide is followed by extensive apoptosis 3–4 days after initial T-cell activation.³ However, as can be seen in Fig. 4b, a population of persisting OT-II cells could clearly be distinguished at 7 days, but these appeared to be unresponsive either to pOVA stimulation in vitro (Fig. 4c,d), or to pOVA/CFA immunization in vivo (Fig. 5). Previous reports have indicated that the disruption of PD-1 signalling in vivo can release islet cell-reactive T cells from a tolerant state, thereby precipitating autoimmune diabetes.³⁰ It was therefore possible, in our model, that PD-1 signals at the time of pOVA/CFA immunization might be important in maintaining the remaining OT-II T cells in their tolerant state. To explore this possibility we performed experiments similar to those outlined in Fig. 5, but instead the mice received the anti-PD-1, or control antibody, at the time of immunization with pOVA + CFA (Fig. 6)

Figure 6.

Anti-PD-1 treatment does not overcome established tolerance in OT-II cells. OT-IIxCD45.1 cells were transferred to C57BL/6 mice 1 day before administration of the 323–339 peptide of OVA (pOVA) (tolerance) or phosphate-buffered saline (PBS) (immunity) intravenously (i.v.). Mice were immunized with pOVA in complete Freund’s adjuvant (CFA) 7 days after peptide treatment. At the same time as immunization, mice received either 250 μg of anti-PD-1 or isotype-control antibody intraperitoneally (i.p.). Mice were killed 10 days after immunization. (a) OT-II cells in the spleen, expressed as a percentage of all CD4⁺ cells. Ex vivo (b) interleukin-2 (IL-2) production and (c) proliferation of lymph node (LN) cells to increasing concentrations of pOVA. Intracellular staining (ICS) on LN cells stimulated overnight with 100 μm pOVA; percentage of OT-II cells that were (d) interferon-γ (IFN-γ) positive and (e) IL-17 positive. OT-II cell percentages and responses of both pOVA-treated groups were significantly different from that of the PBS-treated group, but not from each other, as determined by an analysis of variance (anova) using Tukey’s multiple comparison test. Data are from one of two experiments giving consistent results.

Again we found that PD-1 blockade did not alter the percentage of OT-II cells in either the spleen (Fig. 6a) or the LN (data not shown). For all other parameters measured, anti-PD-1-treated mice responded to pOVA in the same way as the tolerant control mice. Neither pOVA-treated group produced IL-2 (Fig. 6b) and their proliferative responses were also minimal (Fig. 6c). Intracellular cytokine staining after overnight culture with pOVA confirmed that those remaining OT-II cells were hyporesponsive, producing no IFN-γ (Fig. 6d) or IL-17 (Fig. 6e).

We conclude that PD-1 signals at the time of in vivo immunogenic challenge are not required for the maintenance of the tolerant state in OT-II cells once it is established.

PD-1^−/− CD4⁺ T cells are susceptible to peptide-induced tolerance

We also assessed whether PD-1^−/− T cells could be tolerized by the administration of soluble peptide. We made use of the 2D2 mouse,²⁷ which is transgenic for a TCR that recognizes pMOG. These mice had previously been crossed to PD-1^−/− mice.²⁶ As with the OT-II response to pOVA, deletion of adoptively transferred 2D2 cells occurs following i.v. or i.p. injection of soluble pMOG (S.M. Anderton, unpublished data). The susceptibility of wild-type and PD-1^−/− 2D2 T cells (both expressing CD90.1) to pMOG-induced tolerance could therefore be compared after their transfer to C57BL/6 (CD90.2) hosts. Host mice received either PBS or pMOG/PBS i.p. 7 days before immunization with pMOG/CFA. Those that received PBS showed robust expansion of 2D2 T-cell numbers after subsequent pMOG/CFA immunization (Fig. 7) Importantly, 2D2 PD-1^−/− cells expanded to a greater degree than wild-type 2D2 cells, thereby constituting a ∼fourfold higher frequency of the splenic CD4⁺ T-cell pool of the host mice. These data are consistent with the results of the anti-PD-1 experiments, as shown in Fig. 3a. Furthermore, analysis of mice that received soluble pMOG before immunization found that the transferred 2D2 cell numbers were below the level of detection, regardless of whether they expressed PD-1 (Fig. 7). Thus, we conclude that CD4⁺ T cells that lack PD-1 are fully susceptible to peptide-induced tolerance.

Figure 7.

Peptide tolerance can be induced in PD-1^−/− T cells. 2D2xCD90.1 or 2D2xCD90.1xPD-1^−/− cells were transferred to C57BL/6 mice 1 day before the administration of pMOG (tolerance) or phosphate-buffered saline (PBS) (immunity) intravenously (i.v.). Mice were immunized with pMOG in complete Freund’s adjuvant (CFA) 7 days after peptide treatment and killed 10 days after immunization. (a) Flow cytometric analysis was performed on spleen samples stained for both CD4 and CD90.1. (b) 2D2 or 2D2xPD-1^−/− cells in the spleen, expressed as a percentage of all CD4⁺ cells. Data are from one of two experiments giving consistent results (each with two mice per group).

Discussion

In this study we have performed the first direct analysis of PD-1 involvement in peptide-induced CD4⁺ T-cell tolerance. Previous studies using disease models such as EAE, diabetes and allograft rejection^13,31,32 have concluded that PD-1 signalling is critical in limiting the size of the effector T-cell response. The enhanced proliferation of T cells upon disruption of PD-1 signals is a T-cell-intrinsic effect.¹⁷ Those disease models included the possible effects of PD-1-mediated signals delivered by the tissue. In contrast, our experiments employed a model in which the fate (tolerance versus productive immunity) of the transferred T cells is essentially determined by their interactions with antigen-presenting cells (APCs) in the lymphoid organs. Our data underline the role played by PD-1 signalling in restraining the numbers of antigen-reactive T cells that accumulate in response to an immunogenic stimulus. However, PD-1 signalling does not play any role in either the development, or the maintenance, of tolerance in response to soluble peptide administration.

PD-L2 expression was not detected on OT-II cells under any experimental condition. In keeping with other studies,⁹ we found PD-L1 to be expressed on naïve T cells, and (as with PD-1) this was up-regulated in response to TCR ligation, both under tolerogenic and under immunogenic conditions. Although both PD-1 and PD-L1 were up-regulated on the OT-II cells after exposure to i.v. pOVA alone, the fact that anti-PD-1 did not prevent tolerance in response to this treatment would argue against a T cell–T cell-mediated triggering of PD-1 signalling as an important component of this model of peptide-induced tolerance. In line with other studies suggesting that the decision between immunity and tolerance reflects the activation status of the DC,³ expression of both PD-L1 and PD-L2 by the DC was up-regulated only in response to LPS, and not when peptide alone was given. This up-regulation on the DC following immunogenic stimulation could well provide the basis for why the effects of PD-1 signalling were only manifest under priming conditions, serving to restrain the initial T-cell accumulation. It will be interesting to explore the greater modulation of PD-L1 levels on pDCs than on cDCs in response to LPS, in terms of their roles in promoting or restraining CD4⁺ T-cell proliferation.

The expression data generated here also raise an important question of the role of PD-L1 in T-cell immunity and tolerance. Because PD-L1 is constitutively expressed on T cells it may well play an important role in both these processes. Moreover, ligation of PD-L1 by CD80 has also been shown to mediate a negative signal.³³ As such, PD-L1 can mediate a negative signal to the T cell by two receptors. However, our data conclusively demonstrate that PD-1 signalling to CD4⁺ T cells is not required for peptide-induced tolerance.

Previous studies have described how the absence of PD-1 from a responding CD8⁺ T cell can turn a tolerogenic stimulus into a priming one;¹⁶ however, this was not the case here for two models of peptide-induced CD4⁺ T-cell tolerance. This was most striking in the 2D2 PD-1^−/− studies where peptide-induced T-cell tolerance was still induced despite PD-1^−/− cells being capable of greater expansion, as shown in the non-tolerant mice.

Our data show that PD-1-mediated signals play no role in peptide-induced T-cell tolerance in CD4⁺ T cells, contrasting directly with other studies identifying PD-1 as a critical component of tolerance in CD8⁺ T cells.^{17,18,20,21,30} In fact, the PD-1/PD-L interaction is critical for the induction of peptide-induced T cells in tolerance in a very similar experimental protocol studying transferred CD8⁺ OT-I cells.²⁰ Recently, Haspot et al.³⁴ reported that the PD-1/PD-L1 interaction is required for deletional tolerance of CD8⁺ T cells, but not CD4⁺ T cells, in a model of transplantation tolerance. We suggest that PD-1-mediated signalling could be more important in limiting the responses of CD8⁺ T cells than CD4⁺ T cells. Supporting this, CD8⁺ T cells have been shown to be more sensitive to PD-1-mediated inhibition than CD4⁺ T cells,³⁵ as a result of the fact that CD4⁺ T cells produce higher levels of IL-2 upon stimulation. Indeed, robust IL-2 signalling has recently been reported to down-regulate PD-1 expression on CD8⁺ memory T cells, allowing efficient recall responses in a system that is independent of CD4⁺ T-cell help.³⁶ This interplay between PD-1 and IL-2 was highlighted by data from another model of peptide-induced T-cell tolerance – administration of a peptide recognized by CD8⁺ transgenic T cells (2C).²¹ Tolerance induction was PD-1 dependent, as PD-1-mediated signals were shown to inhibit IL-2 production by transgenic TCR CD8⁺ T cells, causing unresponsiveness. Moreover, the addition of IL-2 reversed tolerance induction in that system. It appears that IL-2 can overcome PD-1-mediated inhibition in a number of settings. If CD4⁺ T cells are able to produce IL-2 en route to tolerance (our unpublished data suggests this to be the case) this might overcome PD-1 signalling in this setting. Nevertheless, tolerance is still the outcome, with the majority of the OT-II cells being deleted. A counter-argument comes from the clear ability of PD-1 signalling to provide some limit to OT-II cell accumulation under immunizing conditions (Fig. 3), a situation in which IL-2 production by the CD4⁺ T cells should be optimal. Clearly, the different requirements for PD-1 signalling in CD8⁺ T cells versus CD4⁺ T cells during tolerance induction in response to soluble peptide application are worthy of further investigation.

Despite these differences between CD4⁺ and CD8⁺ cells, others have used a different tolerance protocol (using the transfer of peptide-loaded fixed APCs) to show the PD-1/PD-L1 interaction to be important in the maintenance of CD4⁺ T-cell tolerance.³⁰ Importantly, however, that study concluded that the PD-1/PD-L1 interaction was not required during the initial APC–T cell encounter, because fixed PD-L1^−/− APCs were equally capable of tolerance induction. Those data, together with the demonstration that anti-PD-1 or anti-PD-L1 treatment at later time-points reversed tolerance, suggested an important role for PD-1 signals in maintaining tolerance within the tissues. From those data it would be predicted that, in the studies presented here, the administration of anti-PD-1 at the same time as immunizing mice with pOVA in CFA, would overcome established tolerance in the small population of remaining OT-II cells. This did not occur, and instead tolerance to pOVA was maintained. Moreover, our studies using PD-1^−/− 2D2 cells should have shown a loss of tolerance once the mice were primed with pMOG/CFA if PD-1/PD-L1 interactions were required to maintain tolerance. Again, this did not happen. Different mechanisms of tolerance induction might explain why treatment with anti-PD-1 overcomes tolerance induced by peptide-loaded fixed APCs but not by soluble peptide, with greater numbers of peptide-reactive cells persisting following administration of peptide-coupled fixed APCs versus peptide. An important role for PD-1 later in immune responses was proposed by Bluestone et al.,³⁷ who suggested that cytotoxic T-lymphocyte antigen (CTLA-4) is important in the induction of T-cell tolerance, whereas PD-1 plays a more important role in maintaining T-cell tolerance. Leading on from this using a diabetes model, they recently described a mechanism of how PD-1/PD-L1 interactions maintain the unresponsiveness of previously tolerized T cells. PD-1 signalling counteracts the TCR-driven stop signals that prevent tolerized CD4⁺ T cells from slowing and interacting with antigen-bearing APCs, both in the pancreatic lymph node and in the islets. Interestingly, they also showed that CTLA-4 signalling had no effect on the motility of previously tolerized CD4⁺ T cells.³⁸ An important experimental nuance is that Bluestone et al. transferred pre-activated TCR transgenic cells, which were subsequently tolerized by administration of peptide-loaded fixed APCs. Although this approach leaves a measurable population of ‘anergized’ cells, it is not clear how many of the transferred cells undergo apoptosis. Such a system will select for those cells that avoid this alternative fate, and high expression of PD-1 might allow this by reducing sensitivity to TCR signalling. In contrast, in our system using naïve T cells, the overwhelming outcome in response to soluble peptide administration is T-cell death, which cannot be prevented by PD-1 blockade. Furthermore, the unresponsiveness of those T cells that do persist cannot be retrieved by PD-1 blockade. Extrapolation of these differences would suggest that the efficacy of inhibiting PD-1 signalling in inactive CD4⁺ T cells therapeutically might depend on the provenance of the target T-cell population (whether they had gone through a phase of productive activation before tolerization). Further experimental investigation will be needed to determine if this is the case. Collectively, it is clear that different mechanisms are employed to induce and maintain tolerance at different sites and within different T-cell populations.

In conclusion, these studies provide new information on the role of PD-1 on naive CD4⁺ T cells at the decision point between tolerance and productive immunity. We show that PD-1 signals play no role in peptide-induced T-cell tolerance for CD4⁺ T cells. This insight adds to previous observations on the role of PD-1 in the induction of tolerance and raises questions over a differential role of PD-1 in CD4⁺ versus CD8⁺ T cells that needs to be addressed further.

Glossary

Abbreviations:

APC

antigen-presenting cell

cDCs

conventional DCs

CFA

complete Freund’s adjuvant

c.p.m.

counts per minute

DC

dendritic cell

FITC

fluorescein isothiocyanate

IFN-γ

interferon-γ

IL

interleukin

i.p.

intraperitoneally

i.v.

intravenously

LPS

lipopolysaccharide

OT-I

ovalbumin-reactive CD8⁺ cells

PBS

phosphate-buffered saline

pDCs

plasmacytoid DCs

PE

phycoerythrin

PerCP

peridinin chlorophyll protein

pOVA

the 323–339 peptide of OVA

pMHC

a complex of peptide and major histocompatibility complex

pMOG

the 35–55 peptide of myelin oligodendrocyte glycoprotein

RBC

red blood cell

s.c.

subcutaneously

TCR

T-cell receptor

Disclosures

The authors have no financial conflict of interest related to this work.