Glucocorticoids potentiate the inhibitory capacity of programmed cell death 1 by up-regulating its expression on T cells

The inhibitory co-receptor programmed cell death 1 (PD-1, Pdcd1) plays critical roles in the regulation of autoimmunity, anticancer immunity, and immunity against infections. Immunotherapies targeting PD-1 have revolutionized cancer management and instigated various trials of improved cancer immunotherapies. Moreover, extensive trials are underway to potentiate PD-1 function to suppress harmful immune responses. Here we found that both natural and synthetic glucocorticoids (GCs) up-regulate PD-1 on T cells without altering the expression levels of other co-receptors and cell surface molecules. GC-induced up-regulation of PD-1 depended on transactivation of PD-1 transcription mediated through the glucocorticoid receptor. We further found that a GC response element 2525 bp upstream of the transcription start site of Pdcd1 is responsible for GC-mediated transactivation. We also observed that in vivo administration of GCs significantly up-regulates PD-1 expression on tumor-infiltrating T cells. By analyzing T cells differing in PD-1 expression, we directly demonstrated that the amount of PD-1 on the cell surface correlates with its inhibitory effect. Accordingly, GCs potentiated the capacity of PD-1 to inhibit T cell activation, suggesting that this PD-1-mediated inhibition contributes, at least in part, to the anti-inflammatory and immunosuppressive effects of GCs. In light of the critical roles of PD-1 in the regulation of autoimmunity, we expect that the potentiation of PD-1 activity may offer a promising therapeutic strategy for managing inflammatory and autoimmune diseases. Our current findings provide a rationale for strategies seeking to enhance the inhibitory effect of PD-1 by increasing its expression level.

The inhibitory co-receptors programmed cell death 1 (PD-1) is inducibly expressed on T cells upon activation and inhibits T cell receptor (TCR) 4 signaling by recruiting SHP-2, a protein tyrosine phosphatase, in a manner dependent on engagement by either of its two ligands, PD-L1 or PD-L2 (1,2). PD-1-dependent regulation of the TCR signal is required for establishment and maintenance of immune self-tolerance and suppression of excess immune responses to pathogens. Mice deficient in PD-1 spontaneously develop tissue-specific autoimmune diseases and die of severe inflammatory tissue damage upon infection with pathogens that normally establish chronic infection in PD-1-sufficient mice (3,4). On the other hand, this immunoregulatory function of PD-1 is often hijacked by tumors to escape from cancer immune surveillance. Cancer immunotherapies targeting PD-1 successfully eradicate various types of tumors by restoring the tumoricidal activities of tumor-specific T cells (5,6). As anticipated from the autoimmune phenotypes of PD-1-deficient mice, targeted blockade of PD-1 activates not only tumor-specific T cells but also selfreactive T cells to provoke inflammatory tissue damage called immune-related adverse events (irAEs) (7,8).
Various kinds of immunosuppressants have been developed and are widely used for treatment of autoimmune diseases, allergic diseases, transplant rejection, and so on (9). However, most of these drugs have a low therapeutic index and can cause various side effects that are dependent on time and dose, requiring special caution regarding their use. Although the molecular mechanisms of action of immunosuppressants have been extensively analyzed, the actual effects of immunosuppressants on the expression and function of immune-related molecules, including immune checkpoint molecules, are not fully understood, which makes rational design of regimens with higher efficacy and safety difficult.
Among various immunosuppressants, glucocorticoids (GCs) have been mainstay drugs for the treatment of numerous inflammatory diseases, including irAEs of cancer immunother-This work was supported in part by the Core Research for Evolutional Science and Technology Program of the Japan Science and Technology Agency, Basic Science and Platform Technology Program for Innovative Biological Medicine of the Japan Agency for Medical Research and Development (JP18am0301007) and Grants-in-Aid by the Japan Society for the Promotion of Science JP18H05417, JP19H01029, and 19K16694. The authors declare that they have no conflicts of interest with the contents of this article. This article contains Figs. S1-S6. 1 Both authors contributed equally to this work. 2 Supported by Katsunuma and Yoshimitsu Otsuka scholarships. 3 To whom correspondence should be addressed. Tel.: 81-3-5841-7844; Fax: 81-3-5841-7843; E-mail: tokazaki@iam.u-tokyo.ac.jp. 4 The abbreviations used are: TCR, T cell receptor; irAE, immune-related adverse event; GC, glucocorticoid; GR, glucocorticoid receptor; TF, transcription factor; GRE, glucocorticoid response element; HC, hydrocortisone; Ab, antibody; NK, natural killer; Treg, regulatory T cell; cDNA, complementary DNA; EGFP, enhanced GFP; ANOVA, analysis of variance; geoMFI, geometric mean fluorescence intensity; PI, propidium iodide. cro ARTICLE apies. GCs, a class of steroid hormones playing critical roles in diverse physiological processes, have profound anti-inflammatory and immunosuppressive activities (10). The pharmacological effects of GCs are predominantly mediated through the glucocorticoid receptor (GR), a member of the nuclear receptor superfamily of ligand-dependent transcription factors (TFs). Upon interacting with GC in the cytoplasm, GR translocates into the nucleus, where it functions either as a transcriptional activator or repressor (11). GCs attenuate the expression of pro-inflammatory cytokines, induce apoptotic cell death, and impede recruitment of immune cells by inhibiting the expression of adhesion molecules and chemokines in suppression of inflammation (10). In T cells, GCs suppress T cell activation by abrogating TCR-induced gene expression or inhibiting dendritic cell maturation (12-15). Despite their potent anti-inflammatory effects, recent studies revealed that GCs might also augment immune responses by up-regulating genes involved in innate immunity (16,17). In addition to genomic effects, which involve the induction/suppression of genes, GCs have also been reported to function in a nongenomic manner (18). Thus, GCs have extremely diverse and complicated effects on immune and non-immune cells that are not fully understood.
In this study, we found that natural and synthetic GCs upregulate PD-1 on T cells by augmenting PD-1 transcription without changing the expression levels of other co-receptors. We also observed that in vivo administration of GCs significantly up-regulated PD-1 expression on tumor-infiltrating T cells. We identified a GC response element (GRE) responsible for GC-mediated transactivation in the promoter region of the PD-1 gene. By analyzing T cells expressing PD-1 to a variable degree, we directly demonstrated that the amount of PD-1 on the cell surface correlated with its inhibitory effect. Accordingly, GCs potentiated the inhibitory effect of PD-1 on antigendependent functional T cell activation. These results provide new insights into the mechanisms underlying the immunosuppressive effects of GCs and provide a rationale for the strategy to enhance the inhibitory effect of PD-1 by augmenting its expression level.

Dexamethasone strongly enhances PD-1 expression on T hybridoma cells
A variety of drugs have been developed as immunosuppressants with different mechanisms of action. First we assessed the effects of immunosuppressants on PD-1 expression using DO11.10 T hybridoma cells that endogenously express PD-1 on their surface and up-regulate PD-1 expression upon TCR-dependent activation. Treatment with cyclophosphamide monohydrate (an alkylating agent inhibiting DNA synthesis) and mizoribine (an imidazole nucleoside inhibiting the de novo synthesis of guanosine) did not affect PD-1 expression levels at all (Fig. 1, A and B). The expression levels of PD-1 were slightly enhanced by everolimus (a selective inhibitor of the serinethreonine kinase mTOR) (Fig. 1, A and B), consistent with a previous report that mTOR induces T-box transcription factor (T-bet), which represses transcription of PD-1 mRNA (18,19). High-dose cyclosporin A (a selective inhibitor of calcineurin) reduced PD-1 expression (Fig. 1, A and B), which suggests that endogenous expression of PD-1 on DO11.10 T hybridoma cells depends on Ca 2ϩ -calcineurin-nuclear factor of activated T cells (NFAT) signaling, as is the case with PD-1 expression upon T cell activation (20). Notably, we found that PD-1 expression was strongly augmented by the treatment with dexamethasone (a synthetic GC with potent anti-inflammatory activities) in a dose-and time-dependent manner (Fig. 1, A-C). Although dexamethasone also induced apoptotic cell death of DO11.10 T hybridoma cells in a dose-and time-dependent manner (Fig. 1, D and E, and Fig. S1), augmentation of PD-1 expression was observed in live cells and preceded cell death, indicating that PD-1 up-regulation was not due to nonspecific effects caused by cell death. Consistently, cell death, but not PD-1 up-regulation by dexamethasone, was canceled by benzyloxycarbonyl-VAD-fluoromethyl ketone, an inhibitor of apoptotic cell death (Fig. S1). In addition, no statistically significant correlation was detected between changes in PD-1 expression level and viability of cells treated with immunosuppressants at various concentrations (Fig. S2). We also confirmed that apoptosis inducers do not necessarily up-regulate PD-1 expression (Fig. S3). Collectively, PD-1 up-regulation by dexamethasone is likely mediated by its specific pharmacological action.

GCs selectively enhance PD-1 expression on T hybridoma cells
In addition to dexamethasone, various synthetic analogues of GC have been developed for treatment of inflammatory diseases. When we examined natural and synthetic GCs, all GCs tested enhanced PD-1 expression on DO11.10 T hybridoma cells to a similar extent as dexamethasone. Intriguingly, the expression levels of other cell surface molecules tested were not affected by dexamethasone or hydrocortisone (HC) (Fig. 2, A and B, and Fig. S4, A-C).
DO11.10 T hybridoma cells recognize the 323-339 segment of chicken ovalbumin (pOVA 323-339 ) in the context of I-A d . When we stimulated DO11.10 T hybridoma cells by co-culturing with pOVA 323-339 -pulsed IIA1.6 B lymphoma cells expressing I-A d , DO11.10 T hybridoma cells were activated to express or up-regulate activation markers such as CD69, CD28, 4-1BB, and receptor activator of nuclear factor kappa-B ligand (RANKL) as well as PD-1 to a variable degree (Fig. 2C). Addition of dexamethasone and HC to the co-culture attenuated activation-induced expression/up-regulation of most activation markers, which likely reflects the reduced levels of T cell activation by GC treatment. On the other hand, the expression level of PD-1 was further augmented by GCs, indicating that the magnitude of PD-1 up-regulation by GCs outweighs the magnitude of PD-1 down-regulation because of the reduced T cell activation by GCs ( Fig. 2C and Fig. S4D). Therefore, the augmentation of expression by treatment with GCs was highly specific to PD-1.

GCs selectively enhance PD-1 expression on primary T cells
We next tested whether GCs also augment PD-1 expression on primary T cells. Because naïve T cells do not express PD-1, we induced PD-1 expression by stimulating naïve T cells with anti-CD3⑀ and anti-CD28 Abs for 16 h. When we added GCs during stimulation, the expression levels of PD-1 were aug-

Glucocorticoids strengthen PD-1 effects
mented both on CD4 ϩ and CD8 ϩ T cells with a clear dependence on the dose of GCs (Fig. 3, A and B, and Fig. S5, A and B). In contrast, induction of activation markers such as CD25, CD44, and 4-1BB was attenuated by GCs (Fig. 3C). The levels of their down-regulation by GCs were smaller compared with those in stimulated DO11.10 T hybridoma cells, probably because anti-CD3⑀ and anti-CD28 Abs activated primary T cells so strongly that GCs could only partially inhibit T cell activation.
Addition of GCs for 24 h after stimulation also resulted in augmentation of PD-1 expression on CD4 ϩ and CD8 ϩ T cells (Fig. 3, D and E, and Fig. S5, C and D). As is the case with unstimulated DO11.10 T hybridoma cells, dexamethasone and HC did not substantially affect the expression levels of most cell surface molecules on preactivated CD8 ϩ T cells ( Fig. 3F and Fig. S5E). These results indicate that GCs selectively up-regulate PD-1 on primary T cells as well.

In vivo treatment with GC up-regulates PD-1 expression on tumor-infiltrating CD8 ؉ T cells
GCs are commonly used to mitigate irAEs in cancer patients receiving immunotherapy targeting PD-1 and CTLA-4 (8). To investigate the effect of GCs on PD-1 expression in the tumor setting, we analyzed PD-1 expression on T cells infiltrating into tumors of CT26 mouse colon carcinoma cells. As is well known, tumor-infiltrating CD4 ϩ and CD8 ϩ T cells expressed PD-1 at substantial levels (Fig. 4A). Notably, administration of dexamethasone significantly increased the expression levels of PD-1

Glucocorticoids strengthen PD-1 effects
as well as the proportion of PD-1-expressing cells among tumor-infiltrating CD8 ϩ T cells compared with control mice treated with the solvent DMSO (Fig. 4, A-C). Although not statistically significant, tumor-infiltrating CD4 ϩ T cells also exhibited a tendency toward increased levels of PD-1 expression upon dexamethasone treatment (Fig. 4, A-C). These results indicate that GCs have the potential to up-regulate PD-1 expression in vivo.

The GC-GR complex up-regulates PD-1 mRNA by binding to a GRE in the promoter region
We tested the involvement of the GR in up-regulation of PD-1 and cell death by GCs using RU486, a GR antagonist. Up-regulation of PD-1 as well as cell death by dexamethasone were canceled by addition of RU486, indicating that up-regulation of PD-1 and cell death by GCs are mediated through the GR (Fig. 5A). Then we examined the effects of GCs on PD-1 expression at the mRNA level. The amount of PD-1 mRNA was increased about 2-fold by 1 h after addition of GCs and further augmented over time, which was canceled by RU486 (Fig. 5, B and C). These results indicate that GCs augment PD-1 expression at the mRNA level through binding to the GR.
Then we investigated whether the GR-GC complex directly increases transcription of the PD-1 gene. By using the TF binding site prediction database, we identified three putative GREs 2525, 2740, and 2790 bp upstream of the transcription start site of PD-1 mRNA and termed them as GRE1, GRE2, and GRE3, respectively. To examine whether augmentation of PD-1 expression by GCs was mediated through these GREs, we generated a series of reporter constructs containing intact or mutated GREs. Dexamethasone treatment strongly augmented promoter activity in the presence of GRE1, suggesting that GRE1 is responsible for augmentation of PD-1 expression by GCs (Fig. 5D).

GCs potentiate the inhibitory effect of PD-1 by increasing the cell surface amount of PD-1
Upon antigen stimulation, DO11.10 T hybridoma cells secrete IL-2 in a manner dependent on the amount of antigen.

Glucocorticoids strengthen PD-1 effects
Thus, we can evaluate the magnitude of functional T cell activation based on the amount of secreted IL-2. As reported previously, IL-2 production from DO11.10 T hybridoma cells upon antigen stimulation was strongly inhibited when IIA1.6 cells overexpressing PD-L1 (IIA1.6 -PD-L1 cells) were used as antigen-presenting cells, and the inhibitory effect of PD-1 was completely canceled by anti-PD-L1 blocking Abs (21,22). PD-L1mediated inhibition was abolished by targeted deletion of the

Glucocorticoids strengthen PD-1 effects
PD-1 gene in DO11.10 T hybridoma cells and restored by retroviral reconstitution of PD-1 (Fig. 6, A and B). By using this system, we examined the correlation between the cell surface amount of PD-1 and its inhibitory effect. We overexpressed PD-1 at various levels using five different promoters (EF1␣, cytomegalovirus, cytomegalovirus enhancer/chicken ␤-actin, MC1, and SV40) with or without a polyadenylation signal, which plays essential roles in the stabilization of mRNA (Fig. 6,  C and D). These cells were stimulated with pOVA 323-339pulsed IIA1.6 -PD-L1 cells, and PD-1-mediated inhibitory effects were calculated by comparing the amount of secreted IL-2 in the presence or absence of anti-PD-L1-blocking Ab. We observed a strong positive correlation of the cell surface expression level of PD-1 with its inhibitory effect (Fig. 6E).
Then we evaluated the functional consequence of PD-1 upregulation by GCs. DO11.10 T hybridoma cells were treated with dexamethasone, and live cells were stimulated with pOVA 323-339 . As anticipated from the immunosuppressive activity of GCs, pretreatment with dexamethasone substantially reduced the amount of IL-2 secreted from DO11.10 T hybridoma cells upon antigen stimulation. PD-1 engagement further reduced IL-2 production from activated DO11.10 T hybridoma cells. Intriguingly, when we compared the levels of PD-1-dependent inhibition with or without dexamethasone pretreatment, PD-1 inhibited IL-2 production more efficiently when DO11.10 T hybridoma cells were pretreated with dexamethasone (Fig. 6, F and G). Thus, GCs can potentiate the inhibitory effect of PD-1 by increasing the cell surface amount of PD-1.

No substantial effect of GCs on ectopic PD-1
To confirm the specificity of PD-1 up-regulation by GCs, we evaluated the effect of GCs on ectopic PD-1. We treated DO11.10 T hybridoma cells that express exogenous PD-1 under the retroviral long terminal repeat promoter but lack endogenous PD-1 expression with GCs. As expected, GCs failed to augment the cell surface amount of ectopic PD-1 (Fig.   S6A). Accordingly, GCs failed to potentiate the inhibitory effect of ectopic PD-1 (Fig. S6, B and C). These results strongly suggest that the GC-GR complex transactivates PD-1 transcription by directly binding to GRE1 located in the promoter region of the PD-1 gene.

Discussion
GCs have been widely used as anti-inflammatory and immunosuppressive agents for treatment of a variety of inflammatory and autoimmune diseases. However, despite extensive clinical and experimental studies, the extremely diverse and complicated effects of GCs on immune and non-immune cells have not been fully understood. In this study, we found that GCs augment the expression level of PD-1 on T cells but not other cell surface molecules, including CD25, CD44, CD69, 4-1BB, RANKL, and LAG-3. Dexamethasone has been reported to enhance expression of PD-1 and CTLA-4 on T cells, but the underlying mechanism and the functional consequence of upregulation were largely unknown (23)(24)(25). We revealed that treatment with GCs leads to transactivation of PD-1 expression in a manner dependent on GR and GRE in the promoter region of the PD-1 gene. We directly demonstrated that the amount of PD-1 on the cell surface strongly correlated with its inhibitory effect. Accordingly, augmented expression of PD-1 by GCs led to proportional enhancement of the inhibitory effects of PD-1 against antigen-dependent functional T cell activation. These findings suggest that PD-1-mediated inhibition contributes, at least in part, to the anti-inflammatory and immunosuppressive effects of GCs.
Naïve T cells do not express PD-1 but rapidly express PD-1 upon antigen stimulation. This initial induction of PD-1 is driven by multiple TFs, including nuclear factor of activated T cells 1 (NFATc1) and c-Fos/activator protein-1 (AP-1), which are activated by signaling through the TCR or Notch (20,26,27). In the setting of chronic viral infection and tumors, chronic antigen exposure induces high and sustained expression of PD-1 on CD8 ϩ T cells, resulting in their functional exhaustion  (4,28). Forkhead box protein O1 (FoxO1), NFATc1, and nuclear receptor subfamily 4 group A (NR4A) have been reported to mediate the expression of PD-1 on exhausted CD8 ϩ T cells (29 -32). Although these TFs function as transcriptional activators in PD-1 expression, T-bet and Blimp-1 have been reported to function as transcriptional repressors in PD-1 expression (19,33). These TFs cooperatively or competitively regulate PD-1 expression by directly binding to the promoter region of the PD-1 gene. GRs are known to exert diverse functions by binding to promoter sequences together with other TFs or by physically interacting with other TFs (11). In this study, we identified a GRE in the promoter region of the PD-1 gene that is responsible for transactivation by GC treatment. Therefore, it is likely that GR directly binds to the promoter region of the PD-1 gene in response to GC treatment and acts as a transcriptional activator to augment PD-1 transcription. Further studies are expected to reveal a possible positive or negative cooperation of the GC-GR complex with other TFs in regulation of PD-1 expression.

Glucocorticoids strengthen PD-1 effects
Endogenous GCs (cortisol in humans and corticosterone in mice) play critical roles in the regulation of various physiological and developmental processes. It was recently reported that endogenous GCs produced upon infection with mouse cytomegalovirus induced PD-1 expression on natural killer (NK) cells to restrain IFN-␥ production from NK cells, leading to prevention of immunopathology (34). Another recent study showed that deletion of the GR in regulatory T cells (Tregs) resulted in a reduction of PD-1-expressing Tregs in the spleen (35). These findings suggest that PD-1 expression is positively modulated by endogenous GCs. It is possible that the GRE found in the current study also mediates the expression of PD-1 on NK cells and Tregs by endogenous GCs under physiological and pathological conditions. Because of the recent success of cancer immunotherapy targeting PD-1 and CTLA-4, many other inhibitory co-receptors are extensively investigated with the aim of developing new cancer immunotherapies with higher efficacy. Revisiting the significance of inhibitory co-receptors in maintaining immunotolerance to self and preventing excess immune responses, potentiation of their activities is expected to be a promising therapeutic strategy for inflammatory and autoimmune diseases. Our current findings clearly demonstrate that the expression level of PD-1 strongly correlates with its inhibitory function, just like LAG-3, another inhibitory co-receptor (36). These results provide a rationale for the strategy to treat inflammatory and autoimmune diseases by augmenting the expression levels of inhibitory co-receptors.

Mice
C57BL/6J and BALB/c mice were obtained from Charles River Laboratories Japan and Japan SLC, respectively, and housed under specific pathogen-free conditions in environmentally controlled clean rooms. All experimental procedures were planned and conducted according to institutional regulations complying with the Act on Welfare and Management of Animals and the related guidelines in Japan. All mouse protocols were approved by the Animal Experimentation Committee of Tokushima University.

Real-time quantitative PCR
Total RNA was extracted from cells using TRIzol reagent (Ambion) and then subjected to reverse transcription using a high-capacity cDNA reverse transcription kit (Applied Biosystems). Gene expression was analyzed by quantitative PCR using Power SYBR Green PCR Master Mix (Applied Biosystems) on a 7900HT fast real-time PCR system (Applied Biosystems). Values were normalized to the expression of Gapdh. The following specific primer sets were used: Gapdh forward (5Ј-TTCACC-ACCATGGAGAAGGC-3Ј) and reverse (5Ј-GGCATGGACT-GTGGTCATGA-3Ј) and Pdcd1 forward (5Ј-ACCCTGGTCA-TTCACTTGGG-3Ј) and reverse (5Ј-CATTTGCTCCCTCT-GACACTG-3Ј).

Reporter assay
Putative GREs in the promoter region of the PD-1 gene were predicted using the JASPAR database (39) and Genomatix Genome Analyzer (Genomatix). A DNA fragment containing GRE1, GRE2, and GRE3 (2790 -2339 bp upstream of the transcription start of PD-1 mRNA) was amplified from a C57BL/6N mouse BRC clone (B6Ng01-240G08, RIKEN Bio-Resource Research Center) by PCR and cloned into a retroviral expression plasmid vector modified from pSUPER.retro.puro. together with a synthetic minimal promoter sequence and EGFP cDNA. Where indicated, GRE sequences (AGAA-CAnnnTGTTCT) were mutated by overhang PCR to AGGT-CAnnnTGACCT. Reporter constructs were retrovirally introduced into DO11.10 T hybridoma cells overexpressing the mouse GR. After selection with puromycin, cells were cultured in the presence of dexamethasone (100 nM), phorbol 12-myristate 13-acetate (50 ng/ml, Sigma-Aldrich), and ionomycin (500 ng/ml, Sigma-Aldrich) for 24 h and analyzed by flow cytometry.

Preparation of tumor-infiltrating T cells
Male and female BALB/c mice (8 -10 weeks old) were inoculated subcutaneously with 1 ϫ 10 6 CT26 murine colon carcinoma cells on a shaved back. On days 10 and 12, dexamethasone (400 g) in 50% DMSO/PBS was subcutaneously injected near the site of tumor inoculation. On day 13, tumor tissues were dissected and mechanically dissociated using a gen-tleMACS Octo Dissociator with Heaters (Miltenyi Biotec) in RPMI 1640 medium containing collagenase (1 mg/ml, Wako), Glucocorticoids strengthen PD-1 effects hyaluronidase (50 g/ml, Sigma-Aldrich), and DNase I (10 g/ml, Sigma-Aldrich). Single-cell suspensions prepared from tumor tissues were used for flow cytometric analysis.

Statistical analysis
Two-way ANOVA with Dunnett's and Bonferroni's multiple comparisons test, one-way ANOVA with Dunnett's multiple comparisons test, Pearson's correlation test, and two-tailed Student's t test were used to evaluate statistical significance. p Ͻ 0.05 was considered statistically significant. These statistical analyses were performed using GraphPad Prism 8 (GraphPad Software).