Egr2-independent, Klf1-mediated induction of PD-L1 in CD4+ T cells

Programmed death ligand 1 (PD-L1)-mediated induction of immune tolerance has been vigorously investigated in autoimmunity and anti-tumor immunity. However, details of the mechanism by which PD-L1 is induced in CD4+ T cells are unknown. Here, we revealed the potential function of Klf1 and Egr2-mediated induction of PD-L1 in CD4+ T cells. We focused on the molecules specifically expressed in CD4+CD25−LAG3+ regulatory T cells (LAG3+ Tregs) highly express of PD-L1 and transcription factor Egr2. Although ectopic expression of Egr2 induced PD-L1, a deficiency of Egr2 did not affect its expression, indicating the involvement of another PD-L1 induction mechanism. Comprehensive gene expression analysis of LAG3+ Tregs and in silico binding predictions revealed that Krüppel-like factor 1 (Klf1) is a candidate inducer of the PD-L1 gene (Cd274). Klf1 is a transcription factor that promotes β-globin synthesis in erythroid progenitors, and its role in immunological homeostasis is unknown. Ectopic expression of Klf1 induced PD-L1 in CD4+ T cells through activation of the PI3K-mTOR signaling pathway, independent of STATs signaling and Egr2 expression. Our findings indicate that Klf1 and Egr2 are modulators of PD-L1-mediated immune suppression in CD4+ T cells and might provide new insights into therapeutic targets for autoimmune diseases and malignancies.

CD25 + Tregs in concert with peripherally induced Foxp3-independent Tregs. We reported CD4 + CD25 − Tregs that characteristically express LAG3 and Early growth response gene 2 (Egr2) and produce immune suppressive cytokine IL-10 14,15 . Egr2 is a zinc-finger transcription factor that plays an important role in the maintenance of T cell anergy by negatively regulating T cell activation 16 . We previously reported that forced expression of Egr2 in naïve T cells induces LAG3, IL-10, and transcription factor Blimp-1, indicating that Egr2 confers the phenotype of CD4 + CD25 − LAG3 + Tregs (LAG3 + Tregs) on CD4 + T cells 14,17 . LAG3 + Tregs develop in the periphery and exert their suppressive activities in a Foxp3-independent manner 14 . LAG3 + Tregs also produce large amounts of the inhibitory cytokine transforming growth factor-β3 (TGF-β3), which improves the pathology of lupus in a murine model (MRL-Fas lpr/lpr ) 10 . LAG3 + Tregs lacking Egr2 attenuate TGF-β3 production and lose suppressive ability for B cells 10 . Interestingly, LAG3 + Tregs express high levels of PD-L1. Moreover, TGF-β3-mediated suppression of B cells is dependent on PD-1 expression on B cells 10 . Those observations indicate that PD-L1 expression on LAG3 + Tregs play a critical role in the maintenance of immune tolerance. Taken together, these findings suggest that elucidating the molecular control of the PD-1/PD-L1 axis in CD4 + T cells is important for clinical applications targeting costimulatory molecules.
In the current study, we show new functions of transcription factor Krüppel like factor 1 (Klf1) and Egr2 in the induction of PD-L1 in CD4 + T cells. Klf1 promotes β-globin synthesis in erythroid progenitor cells and its role in immunological homeostasis has not been investigated. Mechanisms that induce PD-L1 expression have been studied mainly in immune cells and tumor cells, and the induction mechanisms of PD-L1 appear to be dependent on cell type. However, the detailed mechanism by which PD-L1 is induced in CD4 + T cells remains to be determined. Moreover, it is important to elucidate the cell type-specific mechanism of PD-L1 induction in CD4 + T cells. Here, we reveal that Klf1 and Egr2, both of which are characteristically expressed in PD-L1-expressing LAG3 + Tregs, induce PD-L1 in CD4 + T cells independently of each other.

Results
Egr2 induces PD-L1 expression in CD4 + T cells. Constitutive high level expression of PD-L1 in the steady state occurs only in Tregs, such as CD25 + Tregs and LAG3 + Tregs 9,10 . We first confirmed our previous findings 10, 14 that among CD4 + T cell subsets, both Egr2 and PD-L1 were most highly expressed in LAG3 + Tregs ( Fig. 1a-c). Then, to elucidate whether overexpression of Egr2 induced PD-L1 protein, we utilized the pMIG retroviral vector containing IRES-regulated green fluorescent protein (GFP) as a reporter to generate a pMIG-Egr2 vector (Fig. 1d). As expected, forced expression of Egr2 in naïve CD4 + T cells induced PD-L1 protein expression (Fig. 1e). These findings suggest the important role of Egr2 in the induction of PD-L1 in CD4 + T cells.
Egr2 is not essential for the induction of PD-L1 expression in CD4 + T cells. To examine whether expression of PD-L1 in LAG3 + Tregs was dependent on Egr2, we employed T cell-specific Egr2 conditional knockout mice (Egr2 fl/fl CD4Cre + : Egr2 CKO). Unexpectedly, under physiological conditions, LAG3 + Tregs from Egr2 CKO mice expressed PD-L1 protein to the same extent as the littermate control mice (Fig. 2a). It has been reported that TCR stimulation induces the upregulation of PD-L1 in CD4 + T cells. 8 Here, we found that the expression levels of PD-L1 in Egr2-deficient CD4 + T cells stimulated with anti-CD3ε and anti-CD28 antibodies in vitro was not affected (Fig. 2b). These results indicate that Egr2 is not necessary for the induction of PD-L1 in either LAG3 + Tregs or activated CD4 + T cells, suggesting the involvement of another PD-L1 induction mechanism in CD4 + T cells.

Klf1 induces PD-L1 expression in CD4 + T cells independent of Egr2.
To analyze other transcription factors that induce PD-L1 expression, we selected 1562 LAG3 + Treg-specific genes from microarray data of T cell subsets, including freshly isolated naïve T cells, CD25 + Tregs, LAG3 + Tregs and CD4 + CD25 − CD45RB low LAG3 − memory T cells registered in the ArrayExpress database (access number E-MEXP-1343) 14 . Furthermore, we performed a transcription factor binding site enrichment analysis using the JASPAR Database 18 , an open-access database for eukaryotic transcription factor binding profiles. Then, we combined a transcriptomics and in silico binding prediction approach to identify PD-L1-inducible genes in LAG3 + Tregs. As shown in Figs 3a, 5 genes overlapped in these two data sets: signal transducer and activator of transcription 1 (Stat1), Stat3, Klf1, T-cell acute lymphocytic leukemia protein 1 (Tal1), and Egr2. Stat1 and Stat3 directly bind to the promoter region of the Cd274 gene 19,20 . Intriguingly, transcription factor genes Klf1 and Tal1 were picked up as novel candidate PD-L1-inducible genes. To validate the expression profiles obtained by microarray data, qRT-PCR was performed on Klf1 and Tal1 genes, and both genes were characteristically upregulated in LAG3 + Tregs (Figs 3b and S1). PD-L1 protein was induced in pMIG-Klf1 transfected cells, but not pMIG-Tal1 transduced cells (Figs 3c,d, and S2). We next focused on the mechanism by which Klf1 induced PD-L1. We examined the dependency of Egr2 expression on Klf1-mediated induction of PD-L1 in CD4 + T cells. Expression of Egr2 was not induced in Klf1-transduced CD4 + T cells (Fig. 3e). On the other hand, Klf1 was not induced in Egr2-transduced CD4 + T cells (Fig. 3f). Moreover, CD4 + T cells from Egr2 CKO mice expressed PD-L1 when Klf1 was overexpressed (Fig. 3g). Although we and others have shown that Egr3 compensates for the function of Egr2 during Egr2 deficiency 17,21 , PD-L1 expression was induced by Klf1 in CD4 + T cells from Egr3 fl/fl Egr2 fl/fl CD4Cre + (Egr2/3 DKO) mice (Fig. S3). These results clearly indicate that Klf1 induces PD-L1 in CD4 + T cells independently of Egr2/Egr3 expression.

Marked induction of the Klf1 gene in CD4 + T cells modulates the gene expression profile.
To clarify the mechanism by which PD-L1 is induced by Klf1 in CD4 + T cells, changes in the gene expression profile in pMIG-Klf1-or pMIG-Mock-transduced cells were divided into three groups according to GFP expression levels and analyzed by RNA sequencing (Fig. 4a). Principal component analysis (PCA) of these cells showed that Klf1-transduced cells (Klf1 GFP high ) had unique characteristics compared with the other cells (Fig. 4b). Pathway analysis of the differentially expressed genes (DEGs) was conducted using Ingenuity Pathways Analysis software (IPA). IPA analysis showed candidate intracellular signal transduction pathways induced by Klf1, which include mammalian target of rapamycin (mTOR) signaling pathway as the top canonical pathway, indicating that mTOR signaling played a crucial role in Klf1-mediated PD-L1 expression (Fig. 4c).  (Fig. 3a), previous reports showed that Stat1 and Stat3 directly bind to the promoter region of Cd274 19,20,22,23 , suggesting the important role of STATs signaling in the induction of PD-L1. The induction mechanism of PD-L1 is cell-type dependent (see Discussion). To clarify the nature of the Klf1-mediated PD-L1 induction mechanism, we employed Stat1 KO mice, T cell-specific Stat3 KO mice (Stat3 fl/fl CD4Cre + : Stat3 CKO), Stat4 KO mice, Stat5a KO mice, Stat6 KO mice and wild-type (WT) mice. Unexpectedly, the expression levels of PD-L1 on Klf1-transduced CD4 + T cells were not affected by Stat1, Stat3, Stat4, Stat5a nor Stat6 deficiency, indicating that these STATs do not regulate Klf1-mediated upregulation of PD-L1 in CD4 + T cells (Fig. 5a).

Discussion
Previous work has established that PD-L1 has critical functions as an immunosuppressive cell surface molecule. However, the precise mechanisms by which PD-L1 is induced on CD4 + T cell have long remained elusive. Here, we showed that the expression of PD-L1 on CD4 + T cells is induced by Klf1 and Egr2. Klf1 was predicted by in silico analysis from the gene expression profile of Egr2-expressing LAG3 + Tregs that characteristically express PD-L1. Klf1-mediated PD-L1 induction by CD4 + T cells, and it was partially dependent on the PI3K-AKT-mTOR signaling pathway, but independent of Egr2 and STATs signaling. Our findings have important implications for understanding the molecular mechanisms underlying cell-mediated immune tolerance modulated by CD4 + T cells that express PD-L1.
Interaction of PD-1 and its ligand PD-L1 or PD-L2 negatively regulates intracellular signaling by recruiting protein tyrosine phosphatases, SHP-1 and SHP-2 28 . PD-L1 KO mice show enhanced Th1 responses, exacerbation of experimental autoimmune encephalomyelitis and autoimmune hepatitis 29,30 . In human, the clinical relevance of genetic polymorphisms in the human CD274 gene, encoding PD-L1, has been reported in several autoimmune diseases, such as Addison's disease and Graves' disease 31,32 . On the other hand, deficiency of PD-L2 results in orally administered antigen intolerance 33 , indicating different roles for PD-L1 and PD-L2 in immunity and pathogenesis. Although the expression of PD-L2 is restricted mainly to activated dendritic cells and macrophages 6 , PD-L1 is expressed in various immune cells, including activated CD4 + T cells and Treg subsets. PD-1/PD-L1 interaction is also known as an inhibitory mechanism in CD25 + Tregs 13 and LAG3 + Tregs. These findings clearly suggest that PD-L1 expression on CD4 + T cells is important for maintenance of immune tolerance.
We previously reported that forced expression of Egr2 in CD4 + T cells provides the functional traits of LAG3 + Treg, such as induced expression of Prdm1, IL-10 and LAG3 14 . However, the role of Egr2 in induction of PD-L1 in CD4 + T cells has not been clarified. We demonstrated that Egr2 is not required for PD-L1 expression on CD4 + T cells. It has been reported that among the 4 Egr family members, Egr3 compensates for the function of Egr2 when the latter is deficient 21 . We reported that expression of latent TGF-β binding protein (Ltbp)-3 was maintained by both Egr2 and Egr3 and was required for TGF-β3 production from LAG3 + Tregs 17 . However, T cells that were doubly-deficient for Egr2 and Egr3 were not affected in their Klf-1-mediated induction of PD-L1, indicating that Klf-1 induces PD-L1 in CD4 + T cells via an Egr2/Egr3-independent manner.
These findings indicate a possibility of the involvement of another Egr2-independent PD-L1 induction mechanism in CD4 + T cells, and we focused on Klf1 based on the expression on LAG3 + Tregs. The Klf family is a transcription factor group composed of 17 genes, many of which are important for immune homeostasis. Klf1 is also called erythroid KLF (EKLF) and is predominantly expressed in erythroid progenitor cells. Klf1 has important functions in a variety of mechanisms, including gene activation and repression, regulation of chromatin arrangement, transcription initiation and elongation 34 . Klf1 binds to the CACCC site in the promoter region of the β-globin gene 35,36 . In humans, KLF1 also plays a role in transcribing the β-globin gene. A mutation in the CACCC region causes β-thalassemia 37,38 . Although Klfl is characterized primarily as a transcriptional activator, it can also act as a repressor that interacts with the components of the corepressor Sin3A and the NuRD complex 39,40 . KLF1 participates in the repression of the fetal γ-globin genes in human adult erythroid progenitors by regulating the expression of BCL11A. However, the role of Klf1 in the immune system has not been clarified since deficiency of Klf1 results in embryonic lethality due to congenital thalassemia 41 . In the present study, we revealed a new immunological function of Klf1 in PD-L1 induction in CD4 + T cells that was achieved in an Egr2/Egr3-independent manner.
The induction mechanism of PD-L1 appears to depend on the cell type. Extracellular stimuli and intracellular signaling mechanisms that contribute to the expression of PD-L1 have been studied mainly in tumor cells. A cell line derived from lung cancer cells (A549) expresses PD-L1 following interferon (IFN)-γ stimulation, but PD-L1 expression is attenuated by a Jak/Stat inhibitor 42 . Mouse and human T cell lymphoma cells express PD-L1 in the steady state, and Stat3 binds to the promoter region of the Cd274 gene encoding PD-L1 19,22,23 . In Hodgkin's lymphoma cells, expression of PD-L1 was reduced through inhibition of STAT3 expression by a MAPK inhibitor 20 . PD-L1 expression is promoted by Toll-like receptor 4-(TLR4) stimulation in human multiple myeloma-derived plasma cells, and this PD-L1 expression is dependent on the mitogen-activated protein kinase (MAPK) signal pathway 43 . In human dermal fibroblasts, phosphorylation of ERK1/2 and PI3K induces expression of PD-L1 44 . On the other hand, in dendritic cells, Stat4-or Stat6-deficiency does not affect PD-L1 expression 45 . Although accumulating evidence has revealed the induction mechanisms of PD-L1 in various cell types, that of PD-L1 in CD4 + T cells has yet to be determined. In the present study, we revealed the unique signaling pathway that cooperates to induce PD-L1 expression in CD4 + T cells. Although our in silico analysis suggests the involvement of Stat1 and Stat3, deficiencies of various STATs, including Stat1 and Stat3, did not affect Klf1-induced PD-L1 expression levels. Thus, it appears that Stat1 and Stat3 are not necessary for the Klf1-mediated PD-L1 induction in CD4 + T cells.
Previous studies revealed that the PI3K-Akt-mTOR pathway plays a critical role in PD-L1 expression. In murine squamous cell carcinoma cells, breast cancer, prostate cancer 46 and human glioma cells, inactivation of phosphatase and tensin homolog deleted from chromosome 10 (PTEN) promotes PD-L1 expression through the PI3K-Akt pathway and mTOR activation 47,48 . In human non-small cell lung cancer, PD-L1 is also induced by mTOR activation 49 . In T cells, Akt-mTOR signaling is a key driver of activation, differentiation and function. T cells possess a unique metabolic profile and a corresponding set of mTOR signal requirements. In the present study, we revealed that Klf1-mediated PD-L1 induction in CD4 + T cells is dependent on the PI3K-Akt-mTOR signaling pathway.
In recent years, treatments focused on PD-1/PD-L1 signaling have been clinically applied to the field of tumor immunity, but this approach has not been expanded to autoimmune diseases. Targeting PD-L1 and PD-1 is a potentially novel treatment strategy, and it has attracted attention. In the field of tumor immunity, several monoclonal antibodies against PD-1 or PD-L1 have been developed and clinical studies initiated. These include treatments with the anti-PD-1 antibody Nivolumab and the anti-PD-L1 antibody Avelumab for treatment of skin cancer and non-small cell lung cancer [50][51][52] . Tumor immunity and autoimmunity are inseparable, and autoimmune side effects are observed in about 5% of subjects treated with anti-PD-1 antibody and anti-PD-L1 antibody 53,54 . Pancreatic islet cells highly express PD-L1 in nonobese diabetic (NOD) mice, and exacerbation of the disease is observed by administration of anti-PD-1 antibody or anti-PD-L1 antibody, whereas no change was observed with anti-PD-L2 antibody 55 . Similar trends are also seen in a contact hypersensitivity model 56 and a coronary artery lesion model following heart transplantation 57 . These reports indicate that PD-1/PD-L1 interaction is a novel therapeutic target for autoimmune diseases.
Here, it was shown that Klf1 and Egr2, transcription factors specifically expressed in LAG3 + Tregs, induce PD-L1 expression in CD4 + T cells independent of each other. These novel findings shed light on the understanding of PD-L1-mediated immune tolerance and suggest the potential therapeutic targeting of Klf1 and Egr2 in the treatment of autoimmune diseases and malignancies.

Materials and Methods
Mice. C57BL  Patrick Charnay (INSERM, France)) and CD4Cre + mice. Egr2/3 DKO mice (Egr2 fl/fl Egr3 fl/fl CD4Cre + ) were generated by crossing Egr2 CKO mice with Egr3 fl/fl mice 17 . Control mice were littermates of Egr2 fl/fl , CD4Cre + or wild-type (wild type: WT) mice. All mice were housed in a specific pathogen-free environment. For the experiments, mice were at least 6-8 weeks of age. Mice of 24-28 weeks of age were used for analysis of LAG3 + Tregs. All animal experiments were approved by the ethics committee of the University of Tokyo Institutional Animal Care and Use Committee, and all experiments were conducted based on the approved experimental plan according to the guidelines of the University of Tokyo.
Flowcytometry. Mouse spleens were treated with type IV collagenase (Sigma, St. Louis, USA), and CD4 + T cells were collected using a magnet-assisted cell sorting (MACS) system (Miltenyi). Cells were magnetically labeled using biotinylated anti-CD8a, CD11c, and CD19 antibodies and streptavidin microbeads, and the negative fraction that passed through the MACS LS column (Miltenyi) was collected as CD4 + T cells. LAG3 + Tregs were collected by negative selection using biotinylated anti-CD45 RB antibody. Collected cells were labeled using FITC-anti-CD45RB antibody, PE-anti-LAG3 antibody, APC-anti-CD25 antibody, APC/Cy7-anti-CD4 antibody after blocking Fc receptors. Cell collection was performed with CellQuest (BD Biosciences) using a FACS Vantage flow cytometer (BD Biosciences), and analysis was performed by FlowJo (Tree Star, USA). The purity of cells concentrated by MACS was 90% or more, and the purity of cells collected by flow cytometry (FCM) was 99%.
Quantitative real-time PCR. RNA was extracted using an RNeasy Micro kit (QIAGEN, Germantown, USA) and reverse transcribed into cDNA using SuperScript III and Random Primers (both Invitrogen). Quantitative RT-PCR (qRT-PCR) was performed with CFX Connect (Bio-Rad) using QuantiTect SYBR Green PCR Kit (QIAGEN). The RNA expression level of the target gene was evaluated relative to β-actin (Actb). Sequences of primers are shown in Supplementary Table S1.
Microarray analysis. Microarray analysis was performed as described elsewhere 14 . Briefly, the gene expression profile of murine CD4 + T cell subsets was obtained from an ArrayExpress database, access number E-MEXP-1343 data. Background adjustment and standardization were performed using Bioconductor's Mas5 package in statistical software R. Hierarchical clustering was performed in GeneSpring GX (Agilent Technologies, Santa Clara, CA, USA) for genes whose expression was enhanced 1.5-fold compared to CD4 + CD25 − CD45RB high T cells (naïve CD4 + T cells), or expression decreased at least 0.6-fold. Gene expression level was evaluated as the expression level relative to naïve CD4 + T cells.
T cell isolation. T cells were isolated from mouse splenocytes and then collected in the MACS system using a naïve CD4 + T cell isolation kit (Miltenyi Biotec) as follows. Mouse splenocytes were harvested, passed through a 70 μm cell strainer (BD Bioscience), and hemolyzed using an ACK lysis buffer. A solution containing antibodies against a biotin-antibody cocktail (anti-CD8a, anti-CD11b, anti-CD11c, anti-CD19, anti-CD25, anti-CD45R (B220), anti-CD49b (DX5), antiCD105, anti-MHC Class II, anti-Ter119 and anti-CD16/CD32 (Fcg III/II Receptor) was added and reacted at 4 °C for 10 min. Then, 200 μL of anti-biotin microbeads were added and reacted at 4 °C for 15 min, and T cells were separated by negative selection using an LS column. After separation, staining was carried out using anti-CD44 and anti-CD62L antibodies and analyzed by FCM; the purity was >90%.
Gene transduction using retroviral vectors. DNA encoding Egr2 protein (NM-010118) was isolated from mouse T cell cDNA and subcloned into the retroviral vector pMIG. The full-length mRNA of mouse Klf1 (NM-010635) was synthesized by Takara Bio and subcloned into the retroviral vector pMIG. PMIG-Mock, pMIG-Egr2 and pMIG-Klf1 were transfected into packaging cells (Plat-E) using the Extreme Gene 9 transfection reagent (Roche) and the virus-containing culture supernatant was collected. Mouse splenocytes were cultured for 48 h in the presence of Concanavalin A (ConA) (10 μg/mL) (Sigma) and IL-2 (50 μg/mL) to prepare ConA blasts, and gene transduction was conducted using a retroviral supernatant. Specifically, a viral supernatant was added to a Falcon 24-well plate coated with a recombinant human fibronectin fragment CH296 (RetroNectin, Takara), centrifuged at 1200 × g at room temperature for 3 h, and the same operation was repeated three times. Cell culture. Naïve CD4 + T cells were isolated from splenocytes from Egr2 CKO mice and Egr2 fl/fl , Egr2 fl/+ , CD4Cre + (control) mice using the mouse naïve CD4 + T cell isolation kit (Miltenyi Biotech) with a MACS system. The cells were cultured for 72 h on anti-CD3ε antibody (2-10 μg/mL) and anti-CD28 (5 μg/mL) antibody-coated 96-well flat bottom plates, followed by FCM analysis.

RNA-sequencing.
CD4 + T cells that were transfected with retroviral vectors were divided into 3 fractions according to GFP fluorescence (strongly positive, weakly positive and negative) using a FACSVantage flow cytometer. Total RNA was then extracted. Sequence libraries were prepared according to standard methods using Smart-seq2 (Illumina), and they were analyzed with a next generation sequencer, a MiSeq system (Illumina). Resultant data were mapped to UCSC mm10 using the STAR program, and the read count was calculated using HTSeq. Principal component analysis (PCA) was performed using Prcomp (R package), and visualization in three dimensions was performed by Plot 3D (R package). Differentially expressed genes were analyzed using Ingenuity Pathway Analysis (QIAGEN).
Statistical analysis. Statistically significant differences were analyzed using GraphPad Prism 7 (GraphPad Software, Inc.). Comparative study of qRT-PCR and mean fluorescence intensity was conducted by Student's t-test in the case of two groups. Comparative studies between multiple groups were conducted by one-way ANOVA and then a multiple comparison test was performed using the Bonferroni method.
Data availability. Please contact the corresponding author for data requests.
Accession number. The microarray and RNA-sequencing data are available online at in the ArrayExpress database (access number E-MEXP-1343, E-MTAB-6778).