Introduction

Antigen presentation on the cell surface, which is mediated by HLA molecules, is not sufficient to initiate an efficient T cell response. Accordingly, TCR co-stimulatory pathways are crucial for maintaining immune system homeostasis by regulating T cell activation. After recognition of an antigen presented in the context of an HLA molecule, cellular components rearrange to form distinctive immunological synapses upon immune cell polarization. The CD28 family of receptors, which includes CD28, CTLA-4, ICOS, and PD-1, comprises key elements of the immunological synapse. When these receptors interact with their corresponding ligands, they generate potent co-stimulatory or inhibitory signals [1]. Notably, the receptors that generate inhibitory signals prevent T cell-mediated damage to self-tissue by inhibiting the T cell response.

Importantly, tumor cells can engage these T cell pathways by expressing ligands for the inhibitory receptors on the cell surface. The first of these receptors to be successfully targeted by therapeutic monoclonal antibodies was CTLA-4 or (CD152). CTLA-4 is upregulated after T cell stimulation via the TCR. CTLA-4 binds to B7 with a higher affinity than CD28, inhibiting T cell priming. Work in animal models shows that blocking CTLA-4 can shift the immune system balance toward T cell activation and, consequently, exert anticancer effects. These effects have been confirmed in human clinical trials, and, in 2011, the anti-CTLA-4 antibody ipilimumab was the first-in-class therapeutic monoclonal antibody to be approved by the FDA for the treatment of metastatic melanoma (MM) on the basis of a phase III trial showing improved survival [2].

PD-1 is a central regulatory surface protein that delivers inhibitory signals to maintain the functional silence of T cells against their cognate antigens. The PD-1 receptor was identified in 1992 as a protein that was upregulated during apoptosis in lymphocytes [3]. PD-1 is expressed on monocytes, DCs, T cells, B cells, and NK cells. Persistent expression of PD-1 is a marker for T cell exhaustion, as recently reviewed by Wherry [4]. The PD-1 ligand PD-L1 (B7-H1) was discovered in 1999 and is a 290 amino acid transmembrane protein encoded by the CD274 gene [5]. The extracellular portion of PD-L1 comprises IgV- and IgC-like domains, while the intracellular part comprises a 30 amino acid tail. PD-L1 is expressed on non-hematopoietic cells as well as on antigen-presenting cells and on placental cells that are located in an inflammatory microenvironment [6]. PD-L1 is upregulated in a JAK-/STAT-dependent manner by type I and type II IFNs via IFN regulatory factor-1 (IRF-1) [6, 7].

In general, interactions between PD-L1 and PD-1 regulate the induction and maintenance of peripheral T cell tolerance throughout regular immune responses [5]. The interactions between PD-1 and PD-L1 negatively regulate T cell proliferation and cytokine production. Thus, PD-L1 is a critical negative regulator of self-reactive T cells during both the induction and effector phases of the immune response. PD-L1 acts as an inhibitor in multiple ways. For example, in addition to being a ligand for PD-1, PD-L1 binds B7-1 (CD80) preventing B7-1 co-stimulation [8]. Ligation of PD-L1 results in IL-10 production and may augment the apoptosis of activated T cells [9]. In addition, PD-L1 plays a critical role in the conversion of naïve T cells to regulatory T cells (Tregs) [7].

PD-L1 and cancer

It is clear that the immune system can recognize and kill malignant cells in patients with cancer. However, the immunosuppressive tumor microenvironment results in vast immune dysregulation, eventually leading to an insufficient immune response and the out-of-control growth of cancer cells. Notably, cancer cells can directly suppress anticancer immune mechanisms. In addition, cancer cells attract and/or convert immune cells to generate and maintain an immune-suppressive microenvironment. PD-1 and its ligands play central roles in the creation of an immune inhibitory tumor microenvironment that protects cancer cells from immune cell-mediated cell death [10–13]. Thus, PD-L1 helps protect malignant cells from immune destruction and, notably, is expressed by cancer cells in many different malignancies [14–22]. PD-L1 was first depicted as a marker of tumor aggressiveness in renal cell carcinoma [23]. PD-L1 expression on tumor cells correlates with increased tumor aggressiveness and with a poor prognosis in a number of solid cancers, including pancreatic cancer and ovarian cancer [24–26]. Additionally, PD-1 expression by TILs is a negative prognostic factor in several cancers [27–30].

Surface expression of PD-L1 has been described not only in solid tumors but also in several hematological cancers [17, 19, 21, 22]. PD-L1 is expressed both on malignant cells and on infiltrating immune cells in subsets of aggressive B cell lymphomas [31]. In myeloma, PD-L1 upregulation on malignant cells induces T cell apoptosis and tumor-specific T cell anergy, and it enhances the aggressive characteristics of myeloma cells [21]. In multiple myeloma, myeloma cells that overexpress PD-L1 inhibit the generation of CTLs in vitro [32, 33]. In addition, co-culture of CD4+ T cells with myeloma cells results in the generation of Tregs in a contact-dependent manner. These Tregs have a suppressive phenotype and show increased PD-1 expression compared with naturally occurring Tregs. Furthermore, the PD-1/PD-L1 pathway not only promotes the progression of myeloma indirectly by leading to immune control failure; in addition, bone marrow stromal cells induce myeloma cells to express PD-L1, which results in increased tumor cell proliferation and reduced susceptibility to anti-myeloma chemotherapy. Accordingly, clinical progression is observed in patients that have myeloma cells that express high levels of PD-L1.

Multiple studies of anti-PD-1 and anti-PD-L1 blockade report the subsequent restoration of T cell effector function and proliferation as well as increased infiltration of tumors by CTLs. This alters the CTL/Treg ratio and ultimately results in the death of tumor cells [25, 34]. The blockade of either PD-1 or PD-L1 by monoclonal antibodies has produced outstanding clinical responses [35, 36], and the Food and Drug Administration (FDA) recently approved the anti-PD-1 antibodies pembrolizumab and nivolumab in September and December of 2014, respectively. Blocking the PD-1 pathway shows great clinical promise, and there is high commercial interest and intense competition among drug companies to develop agents that target PD-1 or PD-L1. Anti-PD-1 antibodies block PD-1:PD-L1 and PD-1:PD-L2 interactions, whereas anti-PD-L1 antibodies block PD-1:PD-L1 and PD-L1:CD80 interactions. This distinction results in slightly different modes of action and in different adverse events and response patterns. Another example in which the PD-1 pathway is targeted is the use of a recombinant B7-DC-Fc fusion protein that has a unique mode of action. Specifically, this fusion protein depletes T cells that express high levels of PD-1, thus allowing a more vigorous anticancer response [37]. Interestingly, it was also shown recently that the immune system itself has an anticancer mechanism that works via PD-L1-specific effector T cells (Fig. 1) [38, 39].

Fig. 1
figure 1

PD-L1-specific T cells target immune regulatory cells as well as cancer cells. Cancer cells (purple) as well as other regulatory immune cells [e.g., tumor-associated dendritic cells (TADC) (dark red) and MDSC (light red)] express checkpoint inhibitors (e.g., PD-L1), inhibitory cytokines as well as metabolic enzymes that restrain the antitumor activity of anti-tumor-specific T cells (green) in the tumor microenvironment. Specific T cells recognizing HLA-restricted PD-L1-derived epitopes (yellow), which are generated from intracellular degraded PD-L1, are able to eliminate (red arrows) regulatory immune cells as well as cancer cells. Hence, the activation of PD-L1-specific T cells by vaccination may directly target immune inhibitory pathways in the tumor microenvironment, modulate immune regulation, and potentially alter tolerance to tumor antigens. The addition of PD-L1 epitopes to therapeutic cancer vaccines would thus be a simple and highly synergistic means to increase the outcome

Full size image

PD-L1-specific T cells

Our group was the first to describe spontaneous CD8+ and CD4+ T cell reactivity against PD-L1 in the peripheral blood of both patients with various cancers and healthy donors. These PD-L1-specific CD8+ T cells release IFN-γ and TNF-α. Notably, a few individuals in whom we were able to measure specific T cell responses directly ex vivo had a relatively high number of PD-L1-specific T cells. With very few exceptions, it is not feasible to evaluate tumor-associated antigen-specific T cells either by tetramer staining or by ELISPOT in PBMCs ex vivo without in vitro peptide stimulation [40]. We verified that the PD-L1-specific T cells in PBMCs were cytolytic effector cells using the Granzyme B ELISPOT assay. In addition, we generated PD-L1-specific T cell cultures by re-stimulating PBMCs with the PD-L1 peptide in vitro and showed that the subsequent T cell lines were PD-L1 specific. We further established that the PD-L1-specific CD8+ T cells were cytolytic effector cells that recognize and kill PD-L1-expressing melanoma cells as well as cutaneous T cell lymphoma cells. Recently, Minami et al. [41] described HLA-A24-restricted PD-L1-specific T cells that could lyse PD-L1+ HLA-A24+ renal cell carcinoma cells. In addition to recognizing tumor cells, PD-L1-specific CTLs can recognize and kill normal immune cells in a PD-L1-dependent manner. Thus, using siRNA transfection to knockdown PD-L1 protects DCs from death due to PD-L1-specific T cells [38].

Cross-presentation is defined as the processing of exogenous antigens into the HLA class I pathway [42]. We showed that long peptides (20 amino acids) derived from PD-L1 are readily cross-presented by B cells and T2 cells in the absence of antigen-presenting cells such as DCs or macrophages. This result is interesting in light of the observation that patients with renal cell carcinoma produce soluble PD-L1 that retains its immune-suppressive activity [43]. The ability of T2 cells to process the long PD-L1 peptide and, to some extent, to process the full-length recombinant PD-L1 protein demonstrates the transporter associated with antigen processing (TAP)-independent nature of the cross-presentation. Non-professional APCs were shown previously to cross-present HLA class I-restricted epitopes in a similar TAP-independent way, e.g., from exogenous NY-ESO polypeptides [44].

We additionally described that by reacting to PD-L1-expressing cells, PD-L1-specific T cells directly and indirectly augment other T cell responses [45, 46]. First, since the PD-L1/PD-1 pathway is important for the regulation of both viral and anticancer CTL responses, we considered using PD-L1-specific CTLs to influence anti-viral immunity. Indeed, in culture the addition of PD-L1-specific CTLs 1 week after virus epitope stimulation resulted in an vast increase in the number of virus-specific CD8+ T cells [45]. A similar increase in virus-specific T cells was observed in cultures after co-stimulation with the PD-L1 peptide epitope compared to cultures that were co-stimulated with an irrelevant epitope from HIV-1 [46]. Hence, PD-L1-specific CTLs may efficiently augment the effector phase of the immune response by suppressing PD-L1-expressing regulatory cells that restrain PD-1-expressing effector T cells. Second, we began investigating the possibility of influencing the immunogenicity of a DC-based vaccine using co-stimulation with two PD-L1-derived epitopes. We stimulated PBMCs from DC-vaccinated MM patients with the DC-based vaccine used in the clinical study either with or without the PD-L1-derived peptide epitopes. We observed a significant increase in the number of vaccine-reactive T cells in cultures that were co-stimulated with the PD-L1 peptide epitope compared to cultures co-stimulated with an irrelevant HIV epitope (unpublished observation). Thus, boosting PD-L1-specific T cells may directly modulate the immunogenicity of a DC-based vaccine (Fig. 2). If these findings translate to the clinic, co-vaccination with PD-L1 epitopes may be useful for boosting the immunogenicity of the vaccine. Thus, adding PD-L1 epitopes to cancer vaccines may be an easy and attractive way to increase vaccine efficiency.

Fig. 2
figure 2

Co-stimulation with PD-L1 epitopes boosts the immunogenicity of a DC-based vaccine. PBMC (numerous colors) was stimulated with an autologous DC-based vaccine (blue) in the presence of IL-2. Subsequently, DC-reactive T cells (green) expand, and this is augmented when PD-L1-specific T cells are activated by co-stimulation with PD-L1-derived epitopes (yellow) assessed in cultures co-stimulated with an HIV control epitope (red)

Full size image

It should be noted that the function and effects of PD-L1-specific CTLs may vary according to the microenvironment and the condition of the immune response. The major role of the PD-1 pathway is thought to be its involvement in regulating effector T cell responses to control tissue damage rather than its actions at the initial T cell activation stage [9]. Hence, the occurrence of PD-L1-specific CTLs during the activation phase of an immune response may not enlarge this response. In fact, adding PD-L1-specific CTLs simultaneously with virus antigen stimulation somewhat decreases the number of virus-specific T cells [45], possibly due to the expression of PD-L1 on APCs or on resting T cells.

Owing to the vital functions of PD-L1 in immune regulation, it may seem surprising that there is a natural specific T cell response against PD-L1. However, Yu and colleagues recently described that clonal deletion in the thymus prunes the T cell repertoire, but it does not eliminate self-reactive T cell clones [47]. The authors proposed that a complete deletion of self-reacting T cells would create holes in the immune repertoire that could be exploited by infectious pathogens. Hence, self-peptide-specific CD8+ T cells are present at levels similar to those specific for non-self-antigens in the blood of healthy humans. These self-reactive T cells are substantially anergic compared to non-self-specific T cells; however, they can be activated by strong activation signals. PD-L1 is highly expressed during inflammation and/or stress in professional antigen-presenting cells. Self-reactive T cells that recognize PD-L1 may therefore be activated by the strong activation signals of their cognate targets.

Conclusions and perspectives

Regulatory feedback mechanisms are essential for limiting the strength and magnitude of immune responses that might otherwise harm their host [48, 49]. However, immune evasion is detrimental in the framework of cancer immunotherapy. Thus, it may be very beneficial to target one or more immunosuppressive pathways in combination with anticancer immunotherapy. Immune regulatory cells suppress anticancer immunity in many different ways: by checkpoint inhibitors like PD-L1 and PD-L2, expressing cytokines like TGF-β and IL-10, via metabolic enzymes like tryptophan 2,3-dioxygenase (TDO) and IDO [50] and via arginase, as well as by inducible nitric oxide synthases (iNOS) and adenosine [51, 52]. In addition, regulatory cells release chemokines like CCL22 that attract additional immune regulatory cells. Several different therapeutic strategies are being utilized to target immunosuppression in cancer, including blocking inhibitory pathways such as the PD-1/PD-L1 pathway. In practice, antibodies that target the PD-L1 checkpoint have been shown to elicit impressive, dynamic, and durable tumor regression. We suggest the use of specific T cells as yet another approach to target immune suppression. This review describes naturally occurring specific T cells that recognize PD-L1 in immune-suppressive cells and in malignant cells. A major difference between targeting PD-L1 with monoclonal antibodies versus utilizing PD-L1-specific T cells is that in addition to decreasing the immunoregulatory effects of PD-L1, the PD-L1-specific T cells also inhibit other routes of immune suppression that are mediated by PD-L1+ target cells. Accordingly, a vaccine targeting PD-L1 should attract PD-L1-specific pro-inflammatory T cells to the tumor microenvironment. PD-L1-specific T cells may directly support anticancer immunity by killing target cells and indirectly support it by releasing pro-inflammatory cytokines in the microenvironment to boost additional anticancer immunity. Thus, a PD-L1-based vaccine should be viewed as complementing rather than competing with other forms of immunotherapy. Vaccine-activated PD-L1-specific T cells may, for example, be further boosted by PD-L1 blockade, since PD-L1 mAbs target the same cells as vaccine-induced T cells; this therapeutic strategy will therefore make cells more vulnerable targets (Fig. 3).

Fig. 3
figure 3

A PD-L1 vaccine and checkpoint inhibitors are complimentary. a PD-L1-expressing regulatory immune cells (red) degrade intracellular PD-L1 into peptides (yellow) that are subsequently processed into peptides and presented on the cell surface by HLA molecules, where they are recognized by PD-L1-specific T cells (green). Hence, PD-L1-specific T cells can promote local immune suppression by the secretion of effector cytokines or by killing regulatory immune cells directly (red arrow), thereby influencing general immune reactions. Similarly, they can eliminate PD-L1-expressing malignant cells, b PD-1-positive, PD-L1-specific T cells are themselves hampered by the suppressive effects of PD-L1 expression on their targets and c PD-L1-specific T cells may thus be further boosted by PD-L1 blockade, since PD-L1 mAbs target the same cells as vaccine-induced T cells; this therapeutic strategy will therefore make cells more vulnerable targets. Thus, a PD-L1-based vaccine should be viewed as complementing rather than competing with checkpoint inhibitors

Full size image

Cancer vaccines represent a promising way to eliminate minimal residual disease without inducing significant toxicity and secondary malignancies. However, so far they have been largely failed to demonstrate a significant improvement in patient outcome [53]. This probably reflects the ability of malignant cells to suppress the function of the induced immune cells. The addition of PD-L1 epitope-based therapy to current cancer vaccine strategies would be easy to implement and is likely to be highly beneficial. It should be noted that the loss of PD-L1 expression in cells during vaccination therapy might result in immune escape, i.e., it might protect target cells from immune-mediated killing by vaccine-activated T cells. However, this should reduce local immune suppression, thereby permitting circulating effector T cells to function or to become activated. PD-L1 may thus serve as a widely accessible target for immunotherapeutic strategies that has an entirely different function and expression pattern than previously described antigens.

In conclusion, these findings justify clinical testing to evaluate the efficacy of PD-L1-based vaccination. We plan to conduct the first PD-L1 vaccine study in humans at Herlev Hospital (Denmark) in which PD-L1 epitopes will be administered to patients with MM. The vaccine will consist of two PD-L1-derived peptides [54]. Long-peptide vaccines that combine MHC class I and II TAA epitopes can efficiently potentiate broad T cell effector function and long-term immunity [55]. The phase I/II trial will explore the safety and toxicity (primary objective) of vaccinating MM patients with two PD-L1 epitopes. The secondary objectives include (a) induction of PD-L1-specific immune responses and (b) obtaining clinical response. To summarize, a PD-L1-based cancer vaccine represents a completely novel immuno-oncological therapeutic approach. PD-L1-specific T cells is a fascinating example of the immune system’s ability to effect adaptive immune reactions by directly acting on the immune-suppressive mechanisms of cancerous cells.