UBR5 promotes tumor immune evasion through enhancing IFN-γ-induced PDL1 transcription in triple negative breast cancer.

Background: The up-regulation of PD-L1 is recognized as an adaption of cancer cells to evade immune surveillance and attack. However, the intrinsic mechanisms of the induction of PD-L1 by interferon-γ (IFN-γ) in tumor microenvironment remain incompletely characterized. Ubiquitin ligase E3 component N-recognition protein 5 (UBR5) has a critical role in tumorigenesis of triple negative breast cancer (TNBC) by triggering specific immune responses to the tumor. Dual targeting of UBR5 and PD-L1 exhibited superior therapeutic benefits in a preclinical TNBC model in short term.
Methods: The regulation of UBR5 to PD-L1 upon IFN-γ stimulation was evaluated through in UBR5 deficiency, reconstitution or overexpression cell line models by quantitative PCR, immunohistochemistry and RNA-seq. The effects of PD-L1 regulation by UBR5 and double blockade of both genes were evaluated in mouse TNBC model. Luciferase reporter assay, chromatin immunoprecipitation-qPCR and bioinformatics analysis were performed to explore the transcription factors involved in the regulation of UBR5 to PD-L1.
Results: E3 ubiquitin ligase UBR5 plays a key role in IFN-γ-induced PDL1 transcription in TNBC in an E3 ubiquitination activity-independent manner. RNA-seq-based transcriptomic analyses reveal that UBR5 globally affects the genes in the IFN-γ-induced signaling pathway. Through its poly adenylate binding (PABC) domain, UBR5 enhances the transactivation of PDL1 by upregulating protein kinase RNA-activated (PKR), and PKR's downstream factors including signal transducers and activators of transcription 1 (STAT1) and interferon regulatory factor 1 (IRF1). Restoration of PD-L1 expression in UBR5-deficient tumor cells recoups their malignancy in vivo, whereas CRISPR/Cas9-mediated simultaneous abrogation of UBR5 and PD-L1 expression yields synergistic therapeutic benefits than either blockade alone, with a strong impact on the tumor microenvironment. Conclusions: This study identifies a novel regulator of PDL1 transcription, elucidates the underlying molecular mechanisms and provides a strong rationale for combination cancer immunotherapies targeting UBR5 and PD-L1. © The author(s).

Introduction

Tumor cells can adapt immune regulatory signaling pathways to evade immune recognition and elimination. One of the mechanisms utilized by tumor cells is the upregulation of PD-L1, which has been identified as an indicator of poor prognosis in various tumor types, including breast cancer 1, 2. It has been reported that the mRNA and protein levels of PD-L1 are elevated in triple negative breast cancer (TNBC) cells 3. In breast cancers, higher expression level of PD-L1 is associated with larger tumor size, higher tumor grade, and increased positive lymph node number 4, 5. The PD-1/PD-L1 interaction can cause T cells to enter a state of anergy/exhaustion, which manifests impaired active proliferation, cytokine production and cytotoxicity 6, 7. Thus, targeting the PD-1/PD-L1 axis is a beneficial approach in the treatment of different cancers. Blockade of PD-1 or PD-L1 with monoclonal antibodies can reverse many of these phenomena and restore T cell function 8. Recently, atezolizumab and pembrolizumab showed durable antitumor activity as first-line therapies for patients with PD-L1-positive TNBC by blocking the interaction between PD-1 and its ligand PD-L1 9, 10. However, the patient response rate was still lower than expected. Thus, a better understanding of PD-L1 regulation may help predict patient responses and improve treatment.

The expression of PD-L1 can be exogenously induced by various cytokines including interferon-γ (IFN-γ), tumor necrosis factor-α (TNF-α), interleukins (ILs) and epidermal growth factor (EGF) through the Janus kinase (JAK)/signal transducer and activator of transcription 1 (STAT1)/interferon regulatory factor1 (IRF1), nuclear factor‑κB (NF-κB), phosphatidylinositol 3-kinase (PI3K)/AKT/mammalian target of rapamycin (mTOR) or JAK/STAT3 signaling pathways 11, 12. Among these factors, IFN-γ in the tumor microenvironment (TME) affects both tumor and immune cells in both immunoactivating and immunosuppressive ways 13 which explains why that early approaches targeting IFN-γ in the TME largely failed to provide any clinical benefit 14-16. IFN-γ-induced adaptive immune resistance highlights the importance of utilizing IFN-γ-mediated immunotherapies by simultaneously blocking the expression or activity of PD-L1 and other factors 13. However, the intrinsic mechanism controlling IFN-γ-induced PDL1 transcription remains incompletely characterized.

Human ubiquitin protein ligase E3 component N-recognin 5 (UBR5) was originally identified in a screen for progestin-regulated genes in breast cancer cells 17. UBR5, a member of a rare “homologous to E6-AP C-terminus” (HECT)-domain E3 ubiquitin ligase family 17, is highly conserved in metazoans and is essential for early embryonic development in mice 18, 19. UBR5 is frequently amplified and overexpressed in many cancer types, especially in human breast cancer and ovarian cancer 20, 21. Our previous work revealed a critical role of UBR5 in the aggression of a murine TNBC model 21. Overexpression of UBR5 was shown to correlate with poor overall survival in breast cancer 22. Two key functional domains, the HECT and poly adenylate binding C terminal (PABC) domains of UBR5, are well characterized. The HECT domain associates mainly with E3 ligase activity, and the PABC domain is thought to be a protein-protein interaction motif 23, 24 and may regulate ubiquitin transfer catalyzed by the HECT domain 25. Frameshift mutations tend to occur in the PABC/HECT domain in tumors 22. UBR5 has been reported to directly interact with various proteins implicated in a wide variety of cellular processes, including cell cycle, transcriptional and translational machinery, and DNA damage response. Known targets of UBR5 ligase activity include β-catenin 26, pregnane X receptor 27, and E6-AP 28.

We previously reported that UBR5-deficiency can facilitate the processing and presentation of tumor antigens by antigen-presenting cells to host T cells, triggering specific immune responses to the tumor 21. Dual targeting of UBR5 and PD-L1 exhibited superior therapeutic benefits in a preclinical TNBC model in short term 29. Here, we report for the first time that UBR5 globally regulates IFN-γ-mediated pathways and stimulated genes, particularly PD-L1 expression and uncover the underlying molecular mechanism. We also showed here that simultaneous abrogation of Ubr5 and Pdl1 expression has synergistic therapeutic benefits in long term.

Materials and Methods

Cell lines

Murine TNBC cell lines 4T1, and its derivative 4T1/GFP, 4T1/Ubr5, human TNBC cell lines BT549, MDA-MB-231, ER+ breast carcinoma cell lines MCF7 and human embryonic kidney cell lines HEK293T were stored in Ma lab at SJTU. 4T1 cells were cultured with RPMI 1640 (Invitrogen) containing 10% fetal bovine serum (BI) and 100 μg/mL penicillin/streptomycin (Invitrogen). BT549, MDA-MB-231, MCF7 and HEK293T cells were maintained in DMEM (Invitrogen) supplemented with 10% FBS (BI) and 100 μg/mL penicillin/streptomycin (Invitrogen). All cell lines were incubated under an atmosphere of 5% CO2 at 37℃.

Mice and mouse tumor model

Wild-type (BALB/c) female mice (6-8 weeks old) were purchased from the Charles River Laboratories (Pinghu, China) and maintained in a pathogen-free facility, supplied with sterile food and water.

For 4T1 tumor model, 1×106 cells were injected into the 4th mammary fat pads of BALB/c mice. Tumor growth was measured every 3 days with a caliper and tumor volume was calculated as volume=1/2×length×width2. For lung metastasis experiment, 5×105 4T1 cells were suspended in 100 μL PBS and then intravenously injected into BALB/c mice through the tail vein. Twelve days later, mice were sacrificed and the lungs were collected and made single cell suspension to perform tumor cell metastasis assay in vitro with 6-thioguanine as described previously 21.

Plasmids and vectors

For the constructs used in luciferase reporter assay, mouse Pdl1 and human PDL1 promoters were amplified by PCR from genomic DNA isolated from 4T1 or BT549 cells using the primers listed in , then cloned into the multicloning site (MCS) of the pGL3 Basic Vector (Promega Corporation). Specific deletions of the putative binding sites were carried out by the protocol described elsewhere 11, with primers listed in . For UBR5 mutant UBR5-ΔPABC, 78 amino acids of the PABC domain were deleted using overlapping PCR. The mEif2ak2 cDNA were amplified by PCR from 4T1 cells cDNA pool using primers listed in Table S1. The RNA interference (RNAi) from a lentiviral vector were generated with specific short hairpin RNA (shRNA) expression for each gene. All shRNA sequences were listed in .

CRISPR/Cas9-mediated knockout

4T1 and its derivative cells were subjected to CRISPR/Cas9-mediated knockout of Pdl1 by transient transfection of lentiCRISPR v2 based vector carrying the guide sequences specific for PD-L1. Three guide sequences used per gene were listed in . Positive single-cell clones were screened using 4 μg/mL puromycin. Disruption was confirmed finally by western blot and FACS analysis.

Cell transfections and infections

For the reconstitution of human UBR5 or mouse Pdl1, 4T1/Ubr5 cells were transfected with hUBR5 and mPdl1 plasmids by Lipofectamine 2000 reagent (Invitrogen). Stable hUBR5 or mPdl1 reconstited-4T1/Ubr5 cells were obtained by transfecting plasmids pCDH-hUBR5 or pCDH-mPdl1 and selecting by puromycin. Western blot or FACS were used to confirm the efficiency of reconstitution. 4T1/GFP cells were transiently transfected with siRNA targeting JAK3, STAT1, STAT2, IRF1, and IRF7 individually by Lipofectamine 2000 reagent. siNC (non-target control) was used as the negative control. All siRNA were designed and purchased from GenePharm. Lentiviruses were produced by cotransfection of 293T cells with PSPA, pMD2G, and pGIPZ-dtTomato-shUBR5 or shEIF2AK2 with polyethyleneimine (PEI). Virus supernatants were collected at 24 h and 48 h post-transfection. MDA-MB-231 and BT549 cells were infected with shUBR5 containing lentivirus, then were selected with puromycin. 4T1 cells were infected with shEIF2AK2 containing lentivirus and treated with puromycin. A scrambled shRNA was used as the negative control (shNC).

Quantitative RT-PCR (RT-qPCR)

Total RNA was extracted with Trizol (Invitrogen) and cDNA was synthesized using the cDNA Synthesis Kit (Vazyme). RT-qPCR was performed with Hieff qPCR SYBR Green Master Mix (YEASEN). The cDNA was quantified using SYBR mRNA expression assays by CFX96 Touch Real-Time PCR Detection System (Bio-Rad). The primers sequence of target genes were listed in .

Western blot

Cells were lysed in RIPA buffer (Beyotime Biotechnology) on ice for 10 min. Cell lysates were centrifuged at 12,000 rpm for 15 min at 4℃, and supernatant was collected. Protein concentration was quantified by Beyotime protein assay (Beyotime Biotechnology, 5000006). Proteins were resolved on a 10% SDS PAGE gel and transferred to the NC membrane, blocked with 5% milk and probed for monoclonal antibodies against UBR5 (Santa Cruz, sc-515494), PD-L1 (Proteintech) #66248-1-lg, EIF2AK2 (Beyotime) #AF2125, STAT1 p84/91 (C-136) (Santa Cruz, sc-464), STAT1 (D1K9Y) Rabbit (Cell Signal Technology) mAb #14994, Phospho-STAT1 (Tyr701) Rabbit (Beyotime) #AF5935, IRF1 (E-4) (Santa Cruz, sc-514544) or anti-GAPDH (Proteintech).

Flow cytometry analysis

For IFN-γ stimulation and PD-L1 staining in cell lines, TNBC cell lines were seeded into 6-well plates on Day 1, targeting 70-80% of confluence on the day of surface staining. On Day 2, cells were treated with 10 ng/mL IFN-γ (mouse IFN-γ: Sino Biological #50709-MNAH; human IFN-γ: PEPROTECH #300-02) for 24 h. On Day 3, cells were trypsinized and stained with allophycocyanin (APC) labelled anti-PD-L1 antibodies (Biolegend) on ice for 30 min, then washed with staining buffer for three times. For single cell staining, cells were dissociated from tumors or lymph nodes, then stained with antibodies accordingly. Antibodies against CD4 (RM4-5), CD3 (17A2), CD8 (53-6.7), Foxp3 (150D), CD25 (PC61), Granzyme B (GB11), and IFN-γ (XMG1.2) were purchased from Biolegend. Antibodies against CD45 (30-F11), CD11c (HL3), and MHCⅡ (2G9) were purchased from eBioscience. Flow cytometry data were analyzed by FlowJo software.

Quantification of the micrometastases in lungs

To quantify micrometastases, mice were sacrificed twelve days after 5×105 4T1 cells injected into mice through i.v. Lungs were excised, minced, and digested with tissue dissociation buffer [0.25% collagenase IV (384 unit/mg, Worthington), 0.2% Dipase II (Roche), and 0.01% DNase I (Sigma) in PBS] with periodic votexing for 1 h in 37℃ water bath. Single cell suspension was washed and strained with 70 mm strainer, then plated in 60 μmol/L 6-thioguanine selection (serve as duplicates). After 1 to 2 weeks of selection, tumor colonies were stained with crystal violet for 10 min, rinsed with ultrapure water and dried overnight prior to counting.

RNA-seq analysis

Total RNA was isolated for RNA-seq analysis. Second generation of RNA sequencing was performed by Genomic Core Facilities at Weill Cornell Medicine. High-quality reads were aligned to the mouse reference genome (vM25) using Histat2. We next used the featureCounts function of the subread software package to count the number of reads that mapped to a reference gene and performed differential expression with DEseq2. The enrichment analyses were based on differential expressed genes (padj <0.05, |foldchange|>2) using clusterProfiler R package.

Bioinformatic analysis of TCGA database

The KEGG pathway analysis was used by R clusterProfiler. Based on the data of all cancer expression profiles (FPKM) of TCGA, Pearson Correlation was used to calculate pairwise expression correlations between UBR5 and ISGs.

Transient luciferase reporter assays

Cells were seeded in 6-well plates for 18 h and then transfected using Lipofectamine 2000 reagent with 0.5 µg plasmid each (0.05 µg Renilla DNA was used for normalization). Cells were then treated with or without 10 ng/mL IFN-γ. 24 h later, relative luciferase units (RLUs) were measured using the Dual-Luciferase Report Assay System and GloMax 96 Microplate Luminometer (Promega) according to the manufacturer's instructions. RLUs from firefly luciferase signal were normalized by RLUs from Renilla signal.

ChIP-qPCR

Formaldehyde-cross-linked chromatin was prepared from 1×107 WT, Ubr5 and hUBR5-reconstituted Ubr5 4T1 cells, and chromatin immunoprecipitation (ChIP) was performed using the EZ ChIP Kit (#17-371) from Millipore according to the manufacturer's instructions. Normal Mouse immunoglobulin G (IgG, sc-2025), anti-STAT1 p84/91 (C-136) (sc-464) and anti-IRF1 (E-4) (sc-514544) antibodies were purchased from Santa Cruz. Normal rabbit IgG (#2729) was purchased from Cell Signal Technology. Anti-H3K27ac (ab4729) antibody was purchased from Abcam. Anti-H3K4me1 antibody (#39498) was purchased from Active Motif. Real-time qPCR was performed in a CFX96 real-time PCR system (Bio-Rad) using Hieff qPCR SYBR Green Master Mix (YEASEN) and the primers for the GAPDH and PD-L1 promoters were listed in .

Statistical analysis

All values were presented as mean ± SEM, and the Student t test was used to determine statistical differences between groups. Values of P < 0.05 were considered statistically significant. These analyses were carried out using the GraphPad Prism 6 for statistical software.

Results

UBR5 is required for IFN-γ-induced PDL1 gene expression

Tumor cells can respond to elevated IFN-γ levels in the tumor microenvironment by upregulating the expression of PD-L1 to evade immune surveillance 11. Interestingly, we observed that the IFN-γ-mediated induction of PD-L1 in previous generated 29 4T1/Ubr5 cells (Figure ) was weaker (~50%) than that in control 4T1/GFP cells at both the mRNA and protein levels (Figure -B). To determine whether the same phenomenon occurs in vivo, 4T1/GFP and 4T1/Ubr5 cells were subcutaneously injected into the 4th mammary fat pads of BALB/c mice individually, and tumors were dissected at Day 28 after injection. Immunohistochemistry staining data (Figure ) showed that the PD-L1 levels were lower in 4T1/Ubr5 than in 4T1/GFP tumors.

To confirm that the effect of UBR5 on IFN-γ-induced PD-L1 expression was not due to cell line specificity, we evaluated the expression level of PD-L1 in short hairpin RNA (shRNA)-mediated UBR5 knockdown MDA-MB-231 and BT549 cells. Upon IFN-γ stimulation, only 32-54% and 49-66% of PDL1 mRNA and protein were expressed in UBR5 knockdown MDA-MB-231 and BT549 cells compared with scramble control cells, which was consistent with the finding in 4T1 cells (Figure ). The PD-L1 surface protein levels in these cells were correlated with the mRNA levels (Figure ). Reciprocally, overexpression of UBR5 in BT549, MDA-MB-231 and MCF7 cell lines increased IFN-γ-induced PD-L1 expression at both the mRNA and surface protein levels (Figure ). Taken together, the findings indicate that there is a strong correlation between UBR5 and IFN-γ-stimulated PD-L1 expression.

Next, to explore whether the positive correlation between UBR5 and PD-L1 exists in other cancers beyond breast cancer, we evaluated the correlation between UBR5 and PD-L1 in TCGA database. TCGA-based analyses also highlight that the mRNA expression levels of UBR5 and PDL1 are highly correlated in many other cancer types, such as pancreatic adenocarcinoma (PAAD), thymoma (THYM), uveal melanoma (UVM), and prostate adenocarcinoma (PRAD) ().

Restoration of PD-L1 expression in UBR5-deficient tumor regains malignancy

Given that UBR5 regulates IFN-γ-induced PD-L1 expression and that Ubr5 tumor growth is arrested from Day 10 onward 21, it is important to determine the role of PD-L1 in the impaired growth of Ubr5 tumors. We thus rescued PD-L1 in 4T1/Ubr5 cells with murine Pdl1 to levels similar to those in WT 4T1 cells (Figure ) without affecting the expression of UBR5 (Figure ). Upon inoculation in mice, the mPdl1-reconstituted 4T1/Ubr5 cells exhibited substantially enhanced growth compared with Ubr5 cells (Figure ). Analyses of tumor-infiltrating immune cells revealed that the number of CD8+ T cells and their cytolytic activity, manifested as increased granzyme B expression and decreased PD-1 expression, was strongly increased in Ubr5 tumors, which were subsequently reversed following the rescue of Pdl1 or Ubr5 expression (, Figure ). The restoration also effected IFN-γ production of these CD8+ T cells (), as well as the presence of CD25+ and FoxP3+ regulatory T cells (, Figure ). Interestingly, the increased CD11c+ MHC Ⅱ+ mature dendritic cells (DCs) observed previously in mice bearing 4T1/Ubr5-/- tumor 21, was also lost when either mPdl1 or hUBR5 was reconstituted (). Accordingly, a significantly poorer prognosis was observed in the mPdl1-reconstituted 4T1/Ubr5 group than that of the Ubr5 group (Figure ). Given that the UBR5 deficiency decreased PD-L1 levels in MDA-MB-231 cells (Figure -F), we tested the idea if this could enhance the function of effector T cells, using the c-Met specific human chimeric antigen receptor (CAR) T cells on the human TNBC cells. We found that, indeed, UBR5 deficiency in the target cells rendered them more susceptible to CAR T-mediated killing (Figure , ).

Combined genetic targeting of Ubr5 and Pdl1 yields synergistic long term therapeutic effects

Next, we investigated whether combined abrogation of intrinsic UBR5 and PD-L1 expression has synergistic therapeutic benefits. GFP/Pdl1 and Ubr5 cell lines derived from the GFP and Ubr5 4T1 cell lines were generated by CRISPR/Cas9 editing (Figure -B). The cells were inoculated into the mammary pad of mice and the tumor growth was monitored over time (Figure ). The tumor growth of the Ubr5group was dramatically reduced within 30 days, as we have reported previously 21, but gradually increased beyond 30 days. Tumor growth was considerably more arrested in the Ubr5 group compared with the other groups. Furthermore, the tumor did not recur in 83% of mice (4/6) in the Ubr5 group until Day 122. Markedly, no tumor recurrence was observed in these mice for more than 1 year. In contrast, no therapeutic benefit was observed in the GFP/Pdl1 group, which is consistent with the TNBC “cold tumor” theory 30.

To evaluate the effect of dual targeting UBR5 and PD-L1 on the spontaneous lung metastasis of 4T1 tumors, 5×105 tumor cells were administrated i.v. to mice and lung metastasis was measured at Day 12 post injection using the 6-thioguanine clonogenicity assay (Figure ). We observed that the number of lung-colonizing Ubr5 tumor cells was significantly less than the number of colonizing Ubr5 tumor cells (Figure ). Notably, ~2 folds higher CD8+ T cell infiltration was observed in Ubr5 tumors than in Ubr5 tumors (Figure ). These infiltrating T cells were more active shown by an increased GzmB+/PD-1- effector T population and decreased regulatory T cells (Figure ) and more GzmB and IFN-γ production (). Increased production of IFN-γ by CD4+ T cells were also observed in the tumors of mice bearing 4T1/Ubr5 tumor (). Furthermore, CD11c+ MHC Ⅱ+ mature DCs in 4T1/Ubr5-/-Pdl1tumor significantly increased (Figure ). Consequently, the Ubr5 bearing mice exhibited prolonged survival for up to 420 days (Figure ). These results demonstrate that simultaneous blockade of UBR5 and PD-L1 expression has synergistic therapeutic benefits with a strong impact on the infiltrating immune cells.

UBR5 globally regulates IFN-γ-mediated pathways and stimulated genes

Since UBR5 could enhance IFN-γ-induced PD-L1 expression to promote tumor growth, it was of interest to explore whether there are other IFN-γ responsive factors affected by UBR5, we performed transcriptome profiling with RNA-sequencing (RNA-seq) in 4T1/GFP and 4T1/Ubr5 cells treated with or without IFN-γ. The RNA-seq data showed that there were more genes downregulated than upregulated in Ubr5 cells compared with GFP 4T1 cells no matter with or without IFN-γ stimulation (Figure ). A total of 555 genes were induced with IFN-γ in 4T1/GFP cells, while significantly fewer (289 genes) were induced in 4T1/Ubr5 cells (Figure ). In addition, more genes were downregulated in 4T1/GFP cells (123 genes in total) than in 4T1/Ubr5 cells (68 genes in total) (Figure ). RNA-seq data of IFN-γ-treated WT and Ubr5 4T1 cells were further evaluated both by Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment and Gene Ontology (GO) functional annotation analysis. KEGG pathway enrichment analysis revealed that UBR5 was highly correlated with multiple signaling pathways including the JAK-STAT pathway and cytokine-receptor interactions, which suggests that UBR5 is involved in relatively broad regulation of the IFN-γ stimulation pathway (Figure ). The pathways of positive response to external stimulus and regulation of T cell activation/leukocyte proliferation ranked in the top 20 GO biological processes in the enrichment analysis (). Interestingly, in selected interferon-stimulated gene (ISG) subsets, a set of genes respond to IFN-γ in GFP cells but completely not or only mildly responsive in Ubr5cells (Figure ). Genes responding to IFN-γ differently in 4T1/Ubr5 cells, including the immune costimulatory and checkpoint genes Cd40 31 and Siglec15 32 significantly decreased in IFN-γ-induced 4T1/Ubr5 cells compared with IFN-γ-induced 4T1/GFP cells (Figure ), while expression of ISG resistance signature (ISG.RS) genes, such as ISG Isg15 33, JAK2, OAS3, OAS1, IRF7, OAS2, IFIT3 and IFIH1 decreased (Figure ). These genes are predominantly expressed in cancer cells, albeit with variable expression 34. Further analysis utilizing WT, Ubr5 and hUBR5-reconstituted Ubr5 4T1 cells confirmed that the transcription of the Cd40, Siglec15 and Isg15 genes was affected by the expression of UBR5 (Figure ).

UBR5 is crucial for IFN-γ-mediated activation of STAT1 and IRF1 transcription

IFN-γ is generally considered the most prominent soluble inducer of PD-L1 and the JAK1/2-STAT1/3-IRF1 signaling axis has been shown to play a central role in the IFN-γ-mediated induction of PD-L1 11, 35. RNA-seq-based transcriptomic analyses suggested that the JAK-STAT signaling pathway (Figure ) played a role in the regulation of IFN-γ-stimulated PD-L1 by UBR5. The expression levels of JAK3, STAT2 and IRF7 were also significantly different between IFN-γ-treated GFP and Ubr54T1 cells (Figure ). We then evaluated PD-L1-targeting transcription factors including JAK1, JAK2, JAK3, STAT1, STAT2, STAT3, IRF1, IRF7, and TYK2 (), and found that the IFN-γ-stimulated mRNA and protein levels of IRF1 and STAT1 were decreased in UBR5-silenced 4T1 (Figure ), BT549 (Figure ) and MDA-MB-231 (Figure ) cells. Reciprocally, overexpression of UBR5 increased the mRNA (Figure ) and protein levels (Figure ) of STAT1 and IRF1 in different human breast cancer cell lines. The protein levels of pSTAT1 increased in UBR5 overexpression BT549 cells compared to the control cells (). The phosphorylation level of STAT1 also markedly decreased in 4T1/Ubr5 cells (Figure ). Consistently, both the mRNA (Figure ) and protein (Figure ) levels of STAT1 and IRF1, as well as the levels of pSTAT1 were restored after hUBR5 was reconstituted in 4T1/Ubr5 cells. The mRNA levels of Jak1, Jak2, Stat3, Tyk2 altered in a similar pattern as those of Stat1 and Irf1 in GFP, Ubr5 and hUBR5-reconstituted Ubr54T1 cells; however, Jak1 gene did not respond to IFN-γ treatment, and the mRNA levels of Jak2, Stat3 and Tyk2 increased less than 2 fold in IFN-γ-induced 4T1/GFP cells, which is inconsistent with Pdl1 mRNA level changes in 4T1 cells with or without IFN-γ stimulation (). To further evaluate the roles of these genes in IFN-γ-induced PDL1 transcription in 4T1 cells, JAK3, STAT1, STAT2, IRF1, and IRF7 were silenced individually through siRNA (). The expression of STAT1 and IRF1 transcripts, but not JAK3, STAT2 or IRF7 transcripts were required for IFN-γ-induced PDL1 transcription (Figure ). These results suggest that upon IFN-γ-activation, UBR5 enhanced PDL1 transcription is mediated through STAT1 and IRF1.

In eukaryotic cells, mRNA homeostasis is achieved through a balance between mRNA synthesis and degradation. Therefore, we sought to investigate whether UBR5 affects STAT1 and IRF1 mRNA by affecting its stability or transcription. 4T1 cells were treated with the transcriptional inhibitor actinomycin D and the mRNA levels of STAT1 and IRF1 were evaluated over time. The half-lives of both STAT1 and IRF1 mRNA showed no differences in WT, Ubr5 and hUBR5-reconstituted Ubr5 4T1 cells (). Next, we explored whether UBR5 affects the transcription of STAT1 and IRF1 via dual-luciferase reporter assay. Human STAT1 (from -972 to +884) and IRF1 (from -820 to +138) promoters and mouse Stat1 (from -2000 to +100) and Irf1 (from -2000 to +100) promoters were cloned into pGL3 plasmids separately. These reporter plasmids were transiently cotransfected with pCDH-UBR5/pCDH-eGFP plasmids into MDA-MB-231, BT549 or 4T1 cell lines separately to detect STAT1 and IRF1 transcription activity. The results showed that UBR5 indeed enhanced the transcriptional activity of the STAT1 and IRF1 promoters in BT549, MDA-MB-231 and 4T1 cells (Figure ). These data confirm that UBR5 plays an essential role in IFN-γ-induced activation of STAT1 and IRF1 transcription by enhancing synthesis rather than slowing degradation.

To further explore whether the effect of UBR5 on the transcription of PDL1 is mediated by the binding of STAT1-IRF1 in the PDL1 promoter region, we carried out chromatin immunoprecipitation (ChIP)-qPCR assays to analyse the enrichment of Stat1 and Irf1 in the mPdl1 promoter region. Six putative Stat1 binding sites and six putative Irf1 binding sites were predicted by the ALGGEN website (Figure ). Enhanced enrichment of Stat1 and Irf1 (in predicted binding sites 1 and 4 and predicted binding sites 5 and 6) in the mPdl1 promoter was observed in WT and hUBR5-reconstituted Ubr54T1 cells compared to Ubr54T1 cells, which further supported the idea that the UBR5-mediated enhancement of IFN-γ-induced PDL1 transcription is dependent on STAT1 and IRF1 (Figure , ).

Increased post-translational histone modification, such as H3K4me1 and H3K27ac, have been reported to be responsible for the overexpression of PD-L1 as well as immune evasion in cancer 36, 37. Although the status of H3K4me1 and H3K27ac differed when cells were stimulated with IFN-γ at sites 1, 2, 3, 4, 5, and 6, there were no differences among the 4T1 WT, Ubr5 and hUBR5-reconstituted Ubr5cell lines (). In addition, treatment with trichostatin A (TSA), a histone deacetylase (HDAC) inhibitor, and 5-aza-2'-deoxycytidine (5'-AZA-dC), a DNA methylation inhibitor did not change the expression of PD-L1 in 4T1 cell surface regardless of whether UBR5 was expressed (). These results suggest that histone modifications such as H3K4me1 and H3K27ac are not involved in the regulation of PD-L1 expression by UBR5.

UBR5-mediated transactivation of PDL1 is independent of the E3 ligase activity

Given that UBR5 was found to be responsible for IFN-γ-induced STAT1 and IRF1 transcription, we further explored that which domain of UBR5 (Figure ) is critical to its regulation. The human PDL1 promoter was cloned and inserted it into a pGL3-based firefly luciferase reporter vector (Figure ), and transiently cotransfected the vector with plasmids encoding wild type UBR5, the HECT domain mutant C2768A or the PABC domain deletion (∆PABC) of UBR5 into MDA-MB-231 cells. The results showed that UBR5 increased the transcriptional activity of the PDL1 promoter (Figure ). The UBR5-ΔPABC mutant completely lost the IFN-γ-inducibility on the PDL1 promoter, suggesting that this domain is pivotal for regulation. PDL1 promoter constructs with site-directed mutations in the binding sites of IRF1 or STAT1/3 11 were generated (Figure ), and transiently cotransfected into MDA-MB-231 cells with expression vector of UBR5 and the variants. Luciferase activity data indicated that the IRF1 and STAT1/3 binding sites were essential for the transcriptional activity of the PDL1 promoter by UBR5 upon IFN-γ stimulation (Figure ). Consistently, a highly analogous response was observed in IFN-γ-stimulated 4T1 cells on the mouse Pdl1 promoter (Figure ).

To further confirm that PABC is required for IFN-γ-induced PDL1 transcription, we transiently transfected plasmids encoding hUBR5, hUBR5-ΔPABC or hUBR5-C2768A respectively into 4T1/Ubr5cells (Figure ) and performed flow cytometry assays to detect PD-L1 surface protein levels. Consistent with the dual-luciferase reporter assay results, the PABC domain, but not the E3 ligase catalytic site, was essential for the regulation of IFN-γ-induced PD-L1 expression (Figure ). Additionally, we corroborated the dispensability of UBR5 ubiquitin ligase activity by treating BC cells with MG132, an inhibitor of the 26S proteasome. The reduced surface PD-L1 level observed in UBR5-depleted cells did not result from increased protein degradation through the proteasome (). Taken together, UBR5 promotes PD-L1 in a manner dependent on its PABC domain and independent of the E3 ligase activity.

UBR5 transactivation of STAT1 and PDL1 is mediated by PKR

Through RNA-seq (Figure and previous mass spectrometry (MS) analysis 29, we observed a significant downregulation of EIF2AK2 in RNA and protein levels, the gene that encodes protein kinase RNA-activated (PKR), in Ubr5 cells compared with WT 4T1 cells 29. PKR can be activated by IFN-γ mRNA 38 and plays an important role in interferon and dsRNA‐signaling pathways by modulating the transcriptional function of STAT1 39. In addition, PKR can indirectly activate the modification of IRF1 and activation of its DNA-binding activity in a protein kinase-dependent manner 40. Therefore, we speculated that UBR5 might regulate STAT1 and IRF1 through PKR. To test this possibility, we first evaluated Eif2ak2 mRNA levels in 4T1 WT, Ubr5, hUBR5-reconstituted, hUBR5-ΔPABC and hUBR5-C2768A-reconstituted Ubr5 cells with or without IFN-γ treatment. Interestingly, the alterations in Eif2ak2 mRNA (Figure ) and protein (Figure ) levels were consistent with the patterns of Stat1, Irf1 and Pdl1 alterations with IFN-γ induction when the expression of UBR5 changed. Consistently, the Eif2ak2 mRNA (Figure ) and protein (Figure ) levels were decreased in BT549 cells after UBR5 was knocked down. The EIF2AK2 was expressed in response to IFN-γ in hUBR5-C2768A-reconstituted Ubr5 cells but not in hUBR5-ΔPABC-reconstituted Ubr5 cells (Figure -B), which was consistent with our previous observation in Figure that PD-L1 regulation by UBR5 is dependent on the PABC domain.

To further confirm that EIF2AK2 affects the mRNA levels of STAT1 and IRF1, we generated Eif2ak2-knockdown 4T1 cells via shRNA (Figure ). The mRNA levels of Pdl1, Stat1 and Irf1 (Figure ) and the protein levels of STAT1, IRF1 (Figure ) and PD-L1 (Figure ) decreased in 4T1 cells after Eif2ak2 knockdown. Conversely, the mRNA and protein levels of STAT1, IRF1 and PD-L1 were restored after we reconstituted mEif2ak2 expression in Ubr5cells compared with GFP and Ubr5 4T1 cells (Figure ). However, compared with WT cells, the cell surface protein level of PD-L1 restoration was not completely rescued in mEif2ak2-reconstituted cells, suggesting that additional mechanisms might be involved in the regulation of UBR5-mediated PD-L1 surface upregulation other than PKR. These data suggest that the transactivation of STAT1 and PDL1 by UBR5 is mediated by PKR.

To investigate whether our observations match the scenario in clinical human breast cancer samples, we performed bioinformatic analysis with data from TCGA breast cancer invasive carcinoma (BRCA) RNA-seq database by GEPIA and starBase database respectively. Moderate and above correlations between EIF2AK2 and UBR5, STAT1, IRF1 or PDL1 expression (Pearson correlation: R=0.447, R=0.704, R=0.43 and R=0.471, respectively) were observed (Figure -L), which supports the idea that EIF2AK2 expression is highly positively correlated with those of UBR5, STAT1, IRF1 and PDL1.

PKR can mediate the activation of IFN-γ-stimulated STAT1 by regulating p38 MAPK and is required for efficient activation of JNK by IFN-γ 41. To further explore whether UBR5 activates PKR-mediated STAT1 and PD-L1 through p38 or JNK, we used the PKR inhibitor oxindole/imidazole compound (C16) to treat GFP and Ubr5 4T1 cells with or without IFN-γ stimulation. The results showed that there were no differences of the protein levels of STAT1, p38 and JNK (and their phosphorylated forms) with or without PKR inhibition in GFP and Ubr5 4T1 cells in the presence of IFN-γ (). Also, the surface levels of PD-L1 protein were not different between GFP cells and Ubr5 4T1 cells treated with C16 (), suggesting that PKR is involved in UBR5-mediated activation of STAT1 and PDL1 transcription in a kinase-independent manner. The kinase activity independent PKR was also observed in PKR-activated NF-κB signaling pathway 42, 43, as well as in broad mechanisms of inflammasome-mediated caspase-1 activation 44.

Since the PABC domain is required for the regulation of EIF2AK2 by UBR5 (Figure -B) and plays a role in enhancing mRNA stability 45, we further explored whether UBR5 controlled Eif2ak2 by maintaining the latter's mRNA stability. IFN-γ-pretreated cells were treated with the transcriptional inhibitor actinomycin D for different times, and then the mRNA levels of Eif2ak2 were evaluated. The half-lives of Eif2ak2 mRNA were not different among 4T1 WT, Ubr5 and hUBR5-reconstituted cells (). This result suggests that UBR5 regulates Eif2ak2 not by enhancing its mRNA stability, but by increasing the transcription of Eif2ak2.

Discussion

As PD-L1 is a ligand for the critical immune checkpoint molecule PD-1, the regulation of PD-L1 expression is complex and can occur at genetic, transcriptional and posttranscriptional levels. Here, we demonstrate that UBR5 is required for IFN-γ-induced PD-L1 expression in breast cancer, and that transactivation of PDL1 gene expression by UBR5 is mediated through the PKR/STAT1/IRF1 pathway. Furthermore, we show that PD-L1 regulation by UBR5 is dependent on the PABC domain but not the E3 ligase activity of UBR5. More significantly, the augmented expression of PD-L1 by UBR5 can enhance the latter's function in promoting tumor growth and metastasis. Lastly, combined blockade of UBR5 and PD-L1 leads to a synergistic therapeutic benefit in eradicating tumor growth and prolong the survival or even to a potential cure (Figure ), although we need to more rigorously corroborate the synergism by other ways of targeting of UBR5 and PD-L1 than the genetic means.

However, since the UBR5 protein does not contain a DNA binding motif, it is still unclear how UBR5 regulates EIF2AK2, although there is evidence that UBR5 can enhance its transcription through the UBR5 PABC domain. Previous studies have implied that UBR5 can regulate gene transcription through several mechanisms involving three different stages: initiation of RNA transcription by transcription factors and their associated co-activators 46, RNA elongation, and RNA processing and nuclear export 47. It seems from our observations that UBR5 does not affect EIF2AK2 mRNA stability, thus UBR5 may promote EIF2AK2 expression through other mechanisms at one or more of these three stages. UBR5 is overexpressed in many cancer types 20, 21, 48-54 and associated with poor prognosis in a variety of cancers 48-51, 55, 56. EIF2AK2-encoded PKR also plays a critical role in tumorigenesis. High PKR expression is linked with prognosis 57-62 and tumor malignancy in multiple cancer types 63-65. In addition, high PKR expression is related to genomic instability 66, reduced survival and shortened remission period 67. We showed that UBR5 could influence the transcription of EIF2AK2. Interestingly, we observed significant co-expression between UBR5 and EIF2AK2 in the TCGA mRNA database (Figure ). It's worth noting that PKR may also be upregulated by IFN-γ indirectly by UBR5-induced changes in the tumor microenvironment and it is of interest to further delineate the role UBR5 in the regulation of PKR expression in vivo.

We demonstrate that UBR5 is broadly and critically involved in the IFN-γ signaling pathway and in the regulation of immune checkpoints (Figure ). Furthermore, our previous study 21 showed that IFN-γ mRNA levels were elevated in 4T1/Ubr5 tumors compared to WT. As a cytokine, IFN-γ plays a dual role in tumor immune response. On the one hand, IFN-γ inhibits the occurrence of tumor and promotes the apoptosis of tumor by regulating immune responses and cell cycle, promoting cell apoptosis, and inhibiting angiogenesis 68, 69. On the other hand, IFN-γ can promote immune escape of tumor cells by regulating the TME 70-79. However, there is little research on the conversional mechanisms of IFN-γ between the two contrasting activities, which also illustrates that IFN-γ has great prospects in the field of tumor immunotherapy.

It is noted that UBR5 can upregulate PDL1 transcription to promote tumor immune evasion and that restoration of PD-L1 expression in UBR5-deficient tumor decreases T cell infiltration and restores malignancy, as was indirectly proven by the combination therapy targeting both UBR5 and PD-L1 in the TNBC mouse model. Therefore, dual targeting UBR5 and PD-L1 has better efficacy than single targeting for breast cancer, and potentially for other cancer types as well. These areas represent interesting directions for future work.

Together, our in vivo and in vitro results reveal for the first time, that UBR5 is a key player in the transcriptional regulation of IFN-γ-induced PDL1 and ISGs, and in directing cancer immune evasion. These findings provide strong evidence and rationale for targeting both UBR5 and PD-L1 as a novel approach to enhance the efficacy of immune checkpoint blockade-based therapy against breast cancer, particularly TNBC.

Supplementary figures and tables.

Click here for additional data file.