Dysfunction in T-cell antitumor activity contributes to the tumorigenesis, progression, and poor outcome of clear cell renal cell carcinoma (ccRCC), with this dysfunction resulting from high expression of programmed cell death-1 (PD-1) in T cells. However, the molecular mechanisms maintaining high PD-1 expression in T cells have not been fully investigated in ccRCC. Here, we describe a mechanism underlying the regulation of PD-1 at the mRNA level and demonstrated its impact on T-cell dysfunction. Transcriptomic analysis identified a correlation between TGFβ1 and PD-1 mRNA levels in ccRCC samples. The mechanism underlying the regulation of PD-1 mRNA was then investigated in vitro and in vivo using syngeneic tumor models. We also observed that TGFβ1 had prognostic significance in patients with ccRCC, and its expression was associated with PD-1 mRNA expression. CcRCC-derived TGFβ1 activated P38 and induced the phosphorylation of Ser10 on H3, which recruited p65 to increase SRSF3 and SRSF5 expression in T cells. As a result, the half-life of PD-1 mRNA in T cells was prolonged. SRSF3 coordinated with NXF1 to induce PD-1 mRNA extranuclear transport in T cells. We then demonstrated that TGFβ1 could induce SRSF3 expression to restrict the antitumor activity of T cells, which influenced immunotherapy outcomes in ccRCC mouse models. Our findings highlight that tumor-derived TGFβ1 mediates immune evasion and has potential as a prognostic biomarker and therapeutic target in ccRCC.
See related Spotlight on p. 1464
Clear cell renal cell carcinoma (ccRCC) is the most common subtype of renal cell carcinoma, accounting for approximately 70% of cases. Although the 5-year relative survival rate has shown some improvement with treatment, the overall prognosis is still not optimistic due to tumor growth, abnormal energy metabolism, immune evasion, etc. (1–3). The tumor microenvironment (TME) is an integral part of tumors and is necessary to establish dynamic interactions with tumor cells. These interactions, in turn, influence the behavior of cancer (2, 4). An immunosuppressive TME facilitates the growth and malignant properties of the lesion while evading the body's immune system (5). Therefore, there is significant value in discovering potential immunosuppressive features in ccRCC.
Increased expression of TGFβ1 is a particular feature of the immunosuppressive tumor microenvironment (4) and has been studied extensively as part of a prospective clinical trial (6). During tumor progression, it becomes a stimulatory molecule that promotes immune evasion, growth, invasion, and metastasis by tumor cells (7). Tumors can reduce their immunogenicity and suppress T-cell responses by stimulating T cells to express immune checkpoint molecules (e.g., PD-1, Tim-3, and CTLA-4), which results in their dysfunction (8). It has been widely reported that TGFβ1 restrains cytotoxic T-cell responses against tumor cells (9). One of the mechanisms underlying this regulation is that TGFβ1 modulates Smad3 to enhance programmed cell death-1 (PD-1) expression on antigen-specific T cells in cancer (10). The expression of PD-1 in T cells in many cancers is higher than that in activated cells or memory T cells found in the corresponding normal tissue (11). PD-1 binding to its ligand, PD-L1, on tumor cells can impair T-cell effector function (12).
Over the past few decades, the clinical therapeutic regimen for ccRCC has transitioned from a nonspecific immune approach to immunotherapeutic strategies. With the clinical application of immunotherapy agents, such as nivolumab (anti–PD-1), the survival of patients with ccRCC has been effectively prolonged (1, 13). However, the molecular mechanisms that permit high expression of PD-1 in T cells have not been fully elucidated in ccRCC. Here, we report a molecular mechanism of PD-1 regulation in T cells, in which ccRCC cell–derived TGFβ1 stabilized PD-1 mRNA required for SRSF3/5 expression, and SRSF3 promoted PD-1 mRNA nuclear export in T cells to restrict their antitumor function.
Materials and Methods
CcRCC transcriptome data were downloaded from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases. The independent datasets from GSE53000, GSE53757, GSE87508, and GSE73731 were analyzed in this study. CcRCC sequence–based gene expression (EXP-S, 534 tumor and 72 normal specimens) and array-based gene expression (EXP-A, 72 specimens) datasets were downloaded from TCGA (http://cancergenome.nih.gov/). Four independent datasets were downloaded from the NCBI Gene-Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/). GSE53000 contains microarray profiles separately profiled from 56 tumor specimens and 6 normal specimens from 9 patients with ccRCC. GSE53757 contains microarray profiles separately profiled from 72 ccRCC tumor tissues, and compared their gene expression levels to matched normal kidney. GSE87508 contains deep sequencing profiles separately profiled from 4 TGFβ-treated or 3 untreated CD8+ T cells. GSE73731 contains microarray profiles separately profiled from 265 ccRCC tumor specimens. The methods of analysis for these datasets include gene set enrichment analysis (GSEA), single-sample GSEA (ssGSEA), Gene Ontology (GO) terms, The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways analysis, and Estimation of STromal and Immune cells in MAlignant Tumour tissues using Expression data (ESTIMATE), which are described in the following sections.
3′UTR of PD-1 mRNA was uploaded to RBPmap (http://rbpmap.technion.ac.il/) in FASTA format. Human database assembly is GRCH38/hg38, and mouse database assembly is GRCm38/mm10. Stringency level was set to medium stringency. We submitted the data, and obtained the RNA-binding proteins potentially binding to PD-1 mRNA 3′UTR. The binding of these proteins on PD-1 mRNA 3′UTR have also been verified in SpliceAid (http://www.introni.it/splicing.html).
JASPAR database (http://jaspar.genereg.net/) was used to predict the binding sites of p65 in the promoter region of SRSF3 or SRSF5. Briefly, we chose JASPAR CORE Vertebrata, and input the SRSF3 or SRSF5 promoter sequences in FASTA format. We selected human transcription factor p65 and set other options as default. Click scan to obtain the potential binding sequence of p65 and SRSF3 or SRSF5 promoter region.
GSEA, ssGSEA, and ESTIMATE
The GSEA analysis was done using GSEA software version 3.0 (www.broadinstitute.org/gsea/), which used predefined gene sets from the Molecular Signatures Database (MSigDB v6.1). GSEA was used to analyze whether a priori-defined gene set exhibited statistically significant, concordant differences between two defined biological states (high expression group of SRSF3 vs. low expression group of SRSF3, or high expression group of SRSF5 vs. low expression group of SRSF5) in GSE87508 dataset, and the predefined sets (ID: M12012) can be found in the Molecular Signatures Database. SsGSEA (14), an extension of GSEA, calculated separate enrichment scores for each pairing of a sample and gene set. Each ssGSEA enrichment score represented the degree to which genes in a particular gene set were coordinately up- or downregulated within a sample. SsGSEA was used to calculate the enrichment score of every gene set for every sample and can be accessed on the genepattern website (https://www.genepattern.org/). ESTIMATE algorithm (https://sourceforge.net/projects/) was used to calculated levels of infiltrating stromal, immune cells, and tumor purity in tumor samples, and this method allowed the consideration of tumor-associated normal cells in transcriptomic researches. Briefly, we defined ssGSEA based on the signatures related to stromal tissue and immune cell infiltration as stromal and immune scores, and combined the stromal and immune scores as the ESTIMATE score (15).
We built six ccRCC microenvironment signatures and performed ssGSEA or ESTIMATE algorithm for the signatures to generate corresponding scores that reflected the presence of each gene signature in the TCGA EXP-S (n = 534), GSE53000 (n = 62), and GSE53757 (n = 144) ccRCC datasets. We used the Pearson correlation analysis to excavate the genes correlated with the mature vascular score, microvascular score, hypoxia activation score, stromal score, immune categories score (r > 0.4), and tumor purity score (r < −0.4), and then overlapped the positively correlated genelists.
GO terms or KEGG pathways analysis
GO is a structured, controlled vocabulary for the classification of gene function at the molecular and cellular levels. KEGG is a knowledgebase and was used for systematic analysis of gene functions, linking genomic information with higher order functional information of the differentially expressed genes. The GO enrichment and the KEGG pathways analysis were carried out when gene sets of the differentially expressed genes correlated with TGFB1 (Pearson r > 0.4) from GSE87508 dataset were uploaded to the DAVID website (http://david.abcc.ncifcrf.gov/home.jsp). Date were analyzed using Fisher exact test. A P < 0.05 was considered statistically significant.
SB203580 hydrochloride (a P38 inhibitor, MedChemExpress) and QNZ (a p65 inhibitor, MedChemExpress) were dissolved in DMSO at final concentrations of 10 μmol/L and 11 nmol/L, respectively, and used to treat cells as described below. A recombinant human TGFβ1 protein (10 ng/mL) was purchased from R&D Systems (7754-BH) to stimulate cells as indicated. Anti–PD-1 (nivolumab) and anti–PD-L1 (atezolizumab) were purchased from Merck and Roche, respectively, and used as indicated. The transcriptional inhibitor actinomycin D (ActD, Sigma-Aldrich, A9415) and doxycycline (Dox, Sigma-Aldrich, 24390–14–5) were dissolved in DMSO at final concentrations of 5 μg/mL and 1 μg/mL, respectively.
Patients and mice
The neoplastic and paracarcinoma tissues used in this study were from 36 patients diagnosed with ccRCC. Peripheral blood samples were collected into heparin-treated tubes from 43 healthy donors. All the patients and healthy donors included in the study were from the Second Affiliated Hospital of Harbin Medical University (Harbin, China) and enrolled between 2016 and 2019. All research performed was approved by the Institutional Review Board (IRB) of the Second Affiliated Hospital of Harbin Medical University (Harbin, China) and conducted in accordance with the principles expressed in the Declaration at Helsinki. Written informed consent was received from all participants.
Four- to 6-week-old female BALB/c mice were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. OT-1 TCR transgenic mice were obtained from the Jackson Laboratory. All mice were maintained under pathogen-free conditions in the animal facility at Harbin Medical University (Harbin, China). All animal experiments were approved and supervised by the Harbin Medical University Institutional Animal Use and Care Committee.
Cell lines and culture
786-O and Renca cells were purchased from the Chinese Academy of Sciences Cell Bank in 2015. HEK293 Tet-off cells were purchased from Takara Bio USA, lnc. (631152) in 2016. Renca-ovalbumin (OVA) cell line was generated by lentivirus transduction of a pLVX-EF1a-OVA-IRES-mCherry (Takara Bio). Cells were maintained in culture fewer than 10 passages. A MSCV expression vector containing a GFP gene (GeneChem) was used as a sorting marker for OT-1 CD8+ T cells. HEK293 Tet-off cells were cultured with complete DMEM (Gibco) containing 10% FBS (Gibco). 786-O and Renca cells were cultured in RPMI1640 medium (Gibco) containing 10% FBS (Gibco). These cells were authenticated using STR assay (Genetic Testing Biotechnology, China, last authentication September 2017). Tumor cell lines were tested for Mycoplasma using the Lonza Mycoalert Mycoplasma Detection Kit (Lonza, last authentication September 2017).
Human peripheral blood mononuclear cells (PBMC) were isolated from healthy donors through Ficoll-Hypaque (Histopaque, Sigma-Aldrich) density gradient centrifugation. Tumor tissues of OT-1 TCR transgenic mice were processed into a single-cell suspensions through mechanical (Miltenyi Biotec gentle ACS) and enzymatic dissociation followed by filtration through a 70-μm cell strainer (BD Biosciences). CD8+ T cells were then isolated from the PBMCs or the single-cell suspension from OT-1 TCR transgenic mice using the negative selection kit (EasySep Human CD8+ T-Cell Isolation Kit/EasySep Mouse CD8+ T Cell Isolation Kit, StemCell Technologies) according to the manufacturer's instructions, and the purity of the isolated population was higher than 90%. The isolated human or mouse CD8+ T cells were activated for 72 hours with anti-CD3/anti-CD28 Dynabeads (5 μg/mL, Invitrogen). All cells were cultured at 37°C in an atmosphere of 5% CO2.
The sequences of the siRNAs are listed in Supplementary Table S1 (Scramble_siRNA, P38_siRNA, SRSF3_siRNA, SRSF5_siRNA, NXF1_siRNA, Srsf3_siRNA, Srsf5_siRNA) and were purchased from RiboBio. Plasmids encoding CUG-BP, SRSF1, HNRNPA1, SRSF5, SRSF3, SRSF9, MBNL1, YBX1, PD-1, or ΔSRSF3-PD1, a constitutively active MKK6 mutant (MKK6EE), were purchased from GeneChem (Shanghai, China). A Tet-off system was purchased from GeneChem. Briefly, as described in our previous study (16), HEK293 Tet-off cells (1.5 × 106) or CD8+ T cells (1 × 106) were transfected with siRNAs or plasmids with Lipofectamine 2000 (Invitrogen). Transfected cells were further incubated in humidified atmosphere with 5% CO2. Forty-eight hours later, expression of the indicated protein was confirmed by qRT-PCR and Western blots. HEK293 Tet-off cells were transfected with 25 μg KCΔ4. The reporter construct (KCΔ4) containing the PDCD1 coding region was constructed by GeneChem.
An sgRNA for TGFB1 knockdown was inserted into a lenti-CRISPR v.2 vector (GeneChem) with a neomycin gene as the sorting marker. Transductions were performed with 1 × 105 786-O/Renca cells per well in 24-well plates. For knocking down TGFB1, LV-Cas9 lentiviruses were transfected into cells for 48 hours at an MOI of 20 and selected for 7 days with neomycin at a final concentration of 3 μg/mL. Cells were subsequently infected with lentiviruses carrying sgRNAs designed for TGFB1. Forty-eight hours later, expression of TGFB1 was confirmed by Western blots. The sgRNA sequences are listed in Supplementary Table S2.
CD8+ T cells isolated from PBMCs of healthy donors. Briefly, 1 × 104 CD8+ T cells were seeded into 96-well plates and transfected with 60 ng pcDNA3 luciferase vector (GeneChem) for 48 hours. SRSF3/5 with potential p65 binding sites or mutated p65 binding sites (Supplementary Table S3) were constructed (GeneChem) and fused to the luciferase reporter vector GV272 (GeneChem). The firefly luciferase fluorescence intensity was measured by using a dual-luciferase reporter assay system (Promega, E1960) and normalized to the Renilla luciferase fluorescence intensity, according to the manufacturer's instructions.
In vivo syngeneic tumors and treatment experiments
All procedures were approved by the Committee on the Ethics of Animal Experiments of Harbin Medical University. For the orthotopic ccRCC model construction and treatment, 4- to 6-week-old BALB/c mice were anesthetized with sodium pentobarbital (60 mg/kg, Sigma-Aldrich, 57–33–0). A skin incision (approximately 2.0 cm) was made dorsally in the region of the right kidney, and the right kidney was partially exteriorized. 3 × 105 luciferase-labeled Renca TGFβ1-knockdown or scramble-treated cells from the 8th generation were injected into the subcapsular space of the right kidney. On the fifth day after injection, the mice were treated with anti–PD-L1 (200 μg per mouse) or IgG antibody (BioXCell). On the 24th day, tumor growth was measured with bioluminescence imaging technology (Xenogen IVIS 100 Imaging System). the survival time was recorded until the experimental endpoints, and tumor tissue specimens were extracted and processed to obtain single-cell suspensions. Briefly, tumor tissue was minced into 1 to 2 mm pieces and digested with collagenase IV (1.25 mg/mL, Worthington, LS004188), 0.1% trypsin inhibitor from soybean (Sigma-Aldrich, T9128), hyaluronidase (1 mg/mL, Worthington, LS002592), and DNase I (100 μg/mL, Worthington, LS002007) in complete DMEM for 30 minutes at 37°C. Cell suspensions were passed through a 70-μm strainer and resuspended in RPMI 1640 media (Gibco, USA). Lymphocytes were isolated from processed tumor tissues by OptiPrep (Sigma-Aldrich) density gradient centrifugation. After washing twice with PBS and resuspended in 1% paraformaldehyde (PFA), 2 × 106 cells were used for antibody staining by using anti–PD-1 PE-conjugated (clone: 29F.1A12, BioLegend), anti-CD8 PerCP-conjugated (clone: 53-6.7, BioLegend), and anti-SRSF3 FITC-conjugated (clone: 7B4, Santa Cruz Biotechnology) for evaluation by flow cytometry. All samples were acquired on LSR II and LSRII Fortessa (BD Biosciences) at UNC Flow Cytometry Core Facility and analyzed by FlowJo version 10.2 (TreeStar, Inc.). The following gating strategy was used: cells were first gated by FSC/SSC to exclude debris, followed by gating FSC-A and FSC-H to eliminate nonsinglets. The target cell population for further analysis were gated by cell surface marker (mouse CD8).
For adoptive cell transfer therapy to treat ccRCC in mice, 4- to 6-week-old BALB/c mice were subcutaneously injected with 1 × 106 Renca cells in the dorsal region. On the sixth day after injection, the tumor-bearing mice were randomly divided into five groups and intravenously injected with either PBS or 1.2 × 106 OT-1 GFP+ CD8+ T cells transfected with scramble or SRSF3 siRNA. Beginning on the twelfth day after injection, tumor volume (bioluminescent images) and survival were monitored every 3 days, and the mice were treated with an anti–PD-1 (200 μg per mouse) or IgG antibody (BioXCell). To analyze the phenotype of the transferred OT-1 CD8+ T cells in vivo, ccRCC-bearing mice were sacrificed on the 25th day after injection, and the draining (inguinal) lymph nodes (dLN) and nondraining (mesenteric) lymph nodes (non-dLN) were used for further experiments. Lymph nodes were smashed and passed through a 70-μm filter to produce a single-cell suspension. The cells were stained with fluorescently labeled anti-CD8 (clone: 53-6.7, BioLegend), anti–PD-1 (clone: 29F.1A12, BioLegend), and anti–Ki-67 (clone: 16A8, BioLegend). Ki-67 expression of transferred CD8+ T cells was analyzed by using the Transcription Factor Staining Buffer Set (eBioscience, 00-5223-56) according to the manufacturer's instructions. All samples were acquired on LSR II and LSRII Fortessa (BD Biosciences) at UNC Flow Cytometry Core Facility and analyzed by FlowJo version 10.2 (TreeStar, Inc.).
Briefly, CD8+ T cells isolated from PBMCs of healthy donors were lysed with IP lysis buffer (Thermo Scientific, 87787). The lysates were incubated with 5 μg of anti-SRSF3 (Sigma-Aldrich, HPA056981) or anti-NFX1 (Abcam, ab176733) overnight at 4°C. Ten microliters of protein A agarose beads (Merck Millipore) were added to each sample and incubated at 4°C with gentle shaking for 3 hours. The mixtures were then centrifuged and washed three times with lysis buffer. The coprecipitation products were further verified by Western blot analysis.
CD8+ T cells were isolated from PBMCs of healthy donors. Briefly, protein from cultured CD8+ T cells was extracted using the lysis buffer, including radio immunoprecipitation assay (RIPA) lysis buffer (Pierce, 89901), protease inhibitor (Roche, 05056489001), and phosphatase inhibitor (Roche, 04906837001). After incubation for 20 minutes at 4°C with moderate shaking, samples were centrifuged at 14,000 rpm for 10 minutes at 4°C, and supernatants were collected for further analysis. Protein concentration was measured using Bradford Assay (Bio-Rad Laboratories), and 40 μg/lane protein was loaded per well. The primary antibodies (1:1,000 for Western blot) used were anti-p65 (Cell Signaling Technology, 8242), anti–p-p65 (Cell Signaling Technology, 3033), anti-P38 (Cell Signaling Technology, 8690), anti–p-P38 (Cell Signaling Technology, 4511), anti-SRSF3 (Abcam, ab198291), anti-SRSF5 (Abcam, ab67175), anti–p-Ser10 H3 (Cell Signaling Technology, 53348), anti-H3 (Cell Signaling Technology, 4499), anti-GAPDH (GAP, Cell Signaling Technology, 5174), anti–Ser10-Lys14 H3 (Merck, 07–081), anti–CUG-BP (Abcam, ab9549), anti-SRSF1 (Abcam, ab38017), anti-HNRNPA1 (Abcam, ab5832), anti-SRSF9 (Abcam, ab74782), anti-MBNL1 (Abcam, ab45899), and anti-YBX1 (Abcam, ab12148). The polyvinylidene difluoride membranes were incubated overnight at 4°C with primary antibodies, and were then incubated with HRP-labeled goat anti-rabbit (1:4,000, Zsbio Commerce Store, ZDR-5306) or HRP-labeled goat anti-mouse (1:4,000, Zsbio Commerce Store, ZDR-5307) at room temperature for 1 hour. The protein bands were visualized using a chemiluminescence reagent (ECL) kit and Bio-Rad Gel Doc XR Imaging System (Bio-Rad).
RNA extraction and qRT-PCR
Total RNA was extracted from cultured CD8+ T cells isolated from PBMCs of healthy donors and clinical ccRCC tissues using TRIzol (Invitrogen) according to our previously described method (17). The nuclear and cytoplasmic fractions were separated using 0.5% NP-40 (Solarbio) with an RNAase inhibitor (Promega), followed by extraction using TRIzol reagent (Sigma-Aldrich). One microgram of total RNA was used as a template for cDNA synthesis using a PrimeScript RT Reagent Kit (Takara). Real-time quantitative PCR (qRT-PCR) was performed on triplicate samples in a reaction mixture containing SYBR Green (Takara) with the CFX96 Touch Real-Time PCR Detection System (Bio-Rad). The expression of the indicated genes were normalized to GAPDH or U6 using the 2−ΔΔCt method. Thermal cycling parameters: 95°C for 10 minutes, denature at 95°C for 15 seconds, and anneal/extend at 60°C for 1 minute, and then repeated for 40 cycles. The sequences of the primers used for PCR in this study are listed in Supplementary Table S4.
Cultured CD8+ T cells isolated from PBMCs of healthy donors were performed using the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, 17–700) according to the manufacturer's instructions. Briefly, 2 × 107 cultured CD8+ T cells were suspended in 100 μL of complete RIP Lysis Buffer (100 μL of RIP Lysis Buffer, add 0.5 μL of protease inhibitor cocktail and 0.25 μL of RNase inhibitor) and then incubated on ice for 5 minutes. Two hundred microliter each of the lysate was used in nuclease-free microcentrifuge tubes. Five micrograms anti-SRSF3 (Abcam, ab198291) or 5 μg anti-NXF1 (Abcam, ab50609) and 50 μL of magnetic beads suspension was then added to each tube. RNAs bound to RBP were purified after immunoprecipitation using the kit. PCR was used to detect the precipitated fraction. The sequences of the primers used for PCR in this study are listed in Supplementary Table S4.
Chromatin immunoprecipitation (ChIP) experiments were performed using the Chromatin Immunoprecipitation Assay Kit (Millipore, 17-295) and conducted following a protocol described previously (18). Briefly, 4 × 107 cultured CD8+ T cells isolated from PBMCs of healthy donors (anti-CD3/anti-CD28 activated) were fixed with 4% formaldehyde (Sigma F1635) for 15 minutes. Crosslinking was stopped by adding 0.125 mol/L glycine for 5 minutes. The cells in each tube were resuspended in 1 mL of cell lysis buffer (with PI), vortexed every 3 minutes for a total of 15 minutes, centrifuged at 300 × g for 5 minutes, and the supernatant aspirated. Nuclei were resuspended in 2,000 mL Nuclei Lysis buffer (with PI) and sonicated (JY96-II, Ningbo Scientz Biotechnology) for 12 cycles, each with a 15-second pulse at 70% of maximal power, followed by a 45-second cooling period, on ice, to achieve an average DNA size of 200 bp, as detected through agarose gel electrophoresis. Sonicated DNA was immunoprecipitated with 3 μg anti-p65 (Cell Signaling Technology, 8242) or 3 μg anti–p-Ser10 H3 (Cell Signaling Technology, 53348), and 25 μL magnetic protein A/G beads (Millipore, 17-295) was added to each sample with rotation at 4°C overnight. DNA was purified and rehydrated according to manufacturer's instructions. The immunoprecipitated chromatin was detected with 5% agarose gel electrophoresis (AGE). The sequences of the primers for SRSF3 and SRSF5 are listed in Supplementary Table S5.
Neoplastic and paracarcinoma tissues were collected from patients with ccRCC. Formalin-fixed, paraffin-embedded ccRCC tissue sections were incubated at 80°C for 15 minutes, dewaxed in xylene, rinsed in graded ethanol, and rehydrated in double distilled water. For antigen retrieval, the slides were pretreated by steaming them in sodium citrate buffer for 15 minutes at 95°C. Washing with PBS for 3 minutes. For cells, 4 × 104 cultured CD8+ T cells isolated from PBMCs of healthy donors were plated in confocal dishes and stained using standard procedures. Primary antibodies, including anti-SRSF3, anti-SRSF5, and anti–PD-1 antibodies were diluted (1:100) in PBS with 1% BSA. After overnight incubation at 4°C, the cells or tissue sections were washed three times with PBS and incubated with FITC-labeled anti-IgG (Alexa Fluor 488 or 594, Thermo Fisher Scientific) for 1 hour at room temperature. DNA was stained with DAPI (Sigma-Aldrich), and the cells or tissue sections were visualized with a fluorescence microscope (Nikon C2). The proportion of stained cell counts per field was used for the statistical analysis. All images were processed and analyzed using ImageJ software (NIH, Bethesda, MD).
Nascent RNA labeling
Nascent RNA labeling was conducted using the Click-iT Nascent RNA Capture Kit (Thermo Fisher Scientific, C10365) according to the manufacturer's protocol. Briefly, CD8+ T cells isolated from PBMCs of healthy donors were treated with scramble siRNA (KD_Scra) or SRSF3_siRNA (KD_SRSF3) and were fed 5-ethynyl-urudine (EU) at a concentration of 200 mmol/L for 4 hours. After 4 hours, RNA was isolated from whole cells and nuclear and cytoplasmic portions. Click chemistry was performed on RNA, precipitated in ethanol, and enriched using provided streptavidin beads. cDNA synthesis was performed on the beads using SuperScript VILO (Invitrogen). RT-PCR was performed using FastStart Essential DNA Green Master in a LightCycler 96 (Roche).
Electrophoretic mobility shift assay
CD8+ T cells isolated from PBMCs of healthy donors were stimulated with TGFβ1 (10 ng/mL) for 24 hours. Nuclear extracts of T cells were prepared using the Nuclear Protein Extraction Kit (Solarbio), and an emobility shift assay (EMSA) Gel Shift Kit was purchased from Affymetrix. Ten micrograms of nuclear extract was then incubated with biotin-labeled p65 annealed probes (Affymetrix; Supplementary Table S6) at room temperature for 15 minutes. The samples were separated on a nondenaturing polyacrylamide gel, and the gel was then dried at 80°C for 2 hours. The gel was transferred to a nylon membrane and detected using streptavidin-HRP and chemiluminescent substrate, followed by analysis with an imaging system (Bio-Rad ChemiDoc MP Imaging System).
IHC staining was performed according to a previously described protocol (19). Sections of paraffin-embedded ccRCC tissues were incubated with anti-TGFβ1 (1:50, ProteinTech, 21898–1-AP) and visualized with a microscope (Leica DM2500P). We obtained ccRCC tissues samples from patients who provided informed consent for an Institutional Ethics Committee–approved study at the Second Affiliated Hospital of Harbin Medical University (Harbin, China). Manual scoring of intensity, location, and cell types of staining was completed by a pathologist. The intensity (strength, 0–3) of TGFβ1 staining was scored as negative (0–0.5), moderate (1), or strong (2–3).
Differences in variables were assessed by Student t test for two groups or one-way ANOVA for at least three groups. Kaplan–Meier survival curves and the log-rank tests were used to describe survival distributions and assess statistical significance between two groups. All statistical analyses were performed using GraphPad software version 7.0 (GraphPad Software) or IBM SPSS Statistics 23.0 (SPSS). R version 3.3.2 with the extension packages, such as ggplot2 and circlize, were used to produce the figures. All results are expressed as the mean ± SD. A P value less than 0.05 was considered statistically significant.
CcRCC-derived TGFβ1 prolongs PD-1 mRNA half-life
To investigate the underlying molecules involved in the ccRCC microenvironment, we built six ccRCC microenvironment signatures: a mature vascular signature (mature vascular); a microvascular signature (microvascular); a hypoxia activation signature (hypoxia activation; ref. 20); and tumor purity, stromal, and immune signatures (21, 22). We performed ssGSEA and applied the ESTIMATE algorithm to generate corresponding scores that reflected the presence of each gene signature in the ccRCC samples analyzed (Supplementary Fig. S1A). We overlapped the above three correlated gene lists and identified six ccRCC microenvironment-associated genes: IL4R, TGFB1, CXCR4, ITGA5, MKNK1, and ACTN1 (Supplementary Fig. S1B). TGFβ1 has a broad spectrum of biological actions in the development of cancers (10, 23). Kaplan–Meier analysis showed that patients with ccRCC expressing high TGFβ1 mRNA had worse survival than those expressing low TGFβ1 mRNA (Supplementary Fig. S2A). Through an analysis of TGFβ1 mRNA expression in the TCGA, GSE47032, and GSE53757 ccRCC datasets, we observed that TGFβ1 mRNA had higher expression in tumor tissues than in normal tissues (Supplementary Fig. S2B) and higher expression in neoplastic tissues than in paracarcinoma tissues (Supplementary Fig. S2C). We enrolled 36 paired ccRCC and paracarcinoma tissue samples and confirmed that neoplastic tissues exhibited higher TGFβ1 mRNA and protein expression than the paracarcinoma tissues (Supplementary Fig. S2D–S2F). Cancer staging is the process of determining the malignancy of a primary tumor and the extent of its spread (24). The expression of TGFβ1 mRNA was also increased in high pathologic stages of ccRCC (Supplementary Fig. S2G). These results indicated that TGFβ1 might be involved in the malignant progression of ccRCC.
To define the mechanistic role of TGFβ1 in ccRCC, we performed KEGG pathway analysis of genes correlated with TGFβ1 expression in TCGA ccRCC datasets and found that TGFβ1 was associated with negative regulation of the immune system (Supplementary Fig. S3A). We then evaluated immunosuppressive factors and found that TGFβ1 mRNA was significantly positively correlated with PD-1 mRNA (Supplementary Fig. S3B and S3C). Dysfunctional T cells in the tumor microenvironment have abnormally high expression of PD-1 (25). Kaplan–Meier analysis showed that patients with ccRCC expressing high PD-1 mRNA had worse survival than those expressing low PD-1 mRNA (Supplementary Fig. S3D). We found that PD-1 mRNA expression was also increased in high pathologic stages and associated with malignant progression in ccRCC (Supplementary Fig. S3E). Compared with normal or paracarcinoma tissues, ccRCC tissues exhibited higher mRNA and protein expression of PD-1 (Supplementary Fig. S3F and S3G). These results suggested that TGFβ1 cooperates with PD-1 and that both are involved in T-cell dysfunction in ccRCC.
To further investigate the process by which TGFβ1 regulates PD-1 in T cells, we added exogenous TGFβ1 to human-derived CD8+ T cells stimulated with anti-CD3/anti-CD28 dynabeads. We found a relative increase in PD-1 mRNA expression and a prolonged PD-1 mRNA half-life in CD8+ T cells after TGFβ1 treatment, whereas the half-lives of TNF, IL10, β-actin, and α-tubulin mRNAs did not change (Fig. 1A and B). CD8+ T cells cultured with ccRCC-conditioned medium (CCM) from 786-O cells had slowed PD-1 mRNA degradation rates following the addition of the transcriptional inhibitor actinomycin D (ActD; Supplementary Fig. S4A). Knockdown of TGFβ1 expression in 786-O cells resulted in rapid endogenous PD-1 mRNA degradation in CD8+ T cells (Supplementary Fig. S4B). To test the role of TGFβ1 in promoting PD-1 mRNA stability, we employed reporter plasmids controlled by a tetracycline-responsive element through the action of the Tet transactivator (tTA) protein (Fig. 1C). HEK293 Tet-off cells stably expressing the tTA protein were transfected with a reporter construct (KCΔ4) containing the PDCD1 coding region. After this construct was transfected into the HEK293 Tet-off cells, TGFβ1 was added to the cells, and PD-1 mRNA was readily detected after inhibiting transcription by adding doxycycline, indicating that TGFβ1 prolonged the PD-1 mRNA half-life (Fig. 1D).
A series of proteins with the capacity to either accelerate or diminish the translation or decay rate of mRNAs have been identified (26). We predicted potential protein-binding sites in the 3′-UTR of the PD-1 mRNA transcript (Fig. 1E; Supplementary Fig. S4C). To further test the predicted proteins that might be important components in the stabilization of PD-1 mRNA, we cotransfected HEK293 Tet-off cells with KCΔ4 and a plasmid encoding CUG-BP, SRSF1, HNRNPA1, SRSF5, SRSF9, MBNL1, YBX1, or SRSF3, or the vector alone (pcDNA3) to maintain the protein of interest at a stable level (Supplementary Fig. S4D) and observed that SRSF3 and SRSF5 (SRSF3/5) each had the ability to prolong the PD-1 mRNA half-life (Fig. 1F). SRSF3/5 protein was increased after CD8+ T cells were stimulated with exogenous TGFβ1 (Fig. 1G). Compared with control and TGFβ1-knockdown CCM, CCM from 786-O cells overexpressing TGFβ1 increased SRSF3/5 protein expression in CD8+ T cells (Supplementary Fig. S4E). However, the degradation rate of PD-1 mRNA was accelerated in T cells after knockdown of SRSF3 and/or SRSF5 expression, but there was no effect on the degradation of TNF, IL10, β-actin, or α-tubulin mRNA (Fig. 1H). We also found that overexpression of SRSF3/5 prolonged the half-life of PD-1 mRNA in CD8+ T cells (Supplementary Fig. S4F). These results provide evidence that SRSF3/5 are required for TGFβ1 to extend the PD-1 mRNA half-life in CD8+ T cells.
TGFβ1 increased SRSF3/5 expression by activating the P38/p-Ser10 H3 axis in T cells
To further test the hypothesis that TGFβ1 is an important regulator of SRSF3/5 in T cells, we used an RNA-seq dataset (GSE87508), in which human CD8+ T cells were exposed to TGFβ1, and gene expression in TGFβ1-stimulated or unstimulated T cells was performed by deep sequencing. SsGSEA and GO analysis results revealed that CD8+ T cells stimulated with TGFβ1 exhibited P38MAPK pathway activity (Supplementary Fig. S5A and S5B). T cells stimulated with exogenous or CCM-derived TGFβ1 displayed enhanced phosphorylation of P38MAPK (p-P38; Fig. 2A; Supplementary Fig. S5C). We showed that conditional induction of a constitutively active MKK6 mutant (MKK6EE; refs. 27, 28), a direct upstream kinase of P38, prolonged the half-life of PD-1 mRNA in T cells, and on the contrary, inhibition of P38 by SB203580 reduced the stability of PD-1 mRNA (Fig. 2B and C; Supplementary Fig. S5D). Consistently, P38 knockdown or inhibition suppressed the extension of the half-life of PD-1 mRNA induced by exogenous or CCM-derived TGFβ1, but had no effect on the half-lives of TNF, IL10, β-actin, and α-tubulin mRNAs in T cells (Fig. 2D and E; Supplementary Fig. S5E and S5F). GSEA also revealed that SRSF3/5 were relevant to the P38MAPK pathway in CD8+ T cells (Supplementary Fig. S5G). Our results showed that activation or inhibition of P38 increased or decreased SRSF3/5 in T cells, respectively (Fig. 2F).
Inflammatory stimulation induces P38-dependent phosphorylation and phosphoacetylation of histone H3, regulating gene expression (29). We found that TGFβ1 stimulation resulted in strong histone H3 phosphorylation at Ser10 and phosphoacetylation at Ser10-Lys14 (p-Ser10/Ac-Lys14 H3) in CD8+ T cells (Fig. 3A). Consistent with the increased p-Ser10/Ac-Lys14 H3 expression, p-P38 was activated in response to TGFβ1 stimulation and peaked at 30–60 minutes in CD8+ T cells (Fig. 3B). Treatment with SB203580 resulted in maximal inhibition of TGFβ1-induced p-Ser10/Ac-Lys14 H3 in CD8+ T cells (Fig. 3C). ChIP analysis demonstrated that the enrichment of p-Ser10 in the promoter regions of SRSF3/5 was increased after TGFβ1 stimulation in CD8+ T cells (Fig. 3D); however, P38 inhibition impaired the binding of p-Ser10 to the SRSF3/5 promoter regions (Fig. 3E). Immunofluorescence revealed that TGFβ1 stimulation or P38 inhibition promoted or restrained SRSF3/5 levels in CD8+ T cells, respectively (Fig. 3F and G).
P38 activation is required for NFκB (p65) recruitment to promoters to regulate gene expression undergoing P-Ser10 H3 modification (29). P65 overexpression could prolong the half-life of PD-1 mRNA, and MKK6EE-induced PD-1 mRNA stability could be suppressed by QNZ treatment. These treatments did not affect the half-lives of TNF, IL10, β-actin, and α-tubulin mRNAs (Fig. 3H and I). We then performed a bioinformatic analysis to identify the DNA-binding elements (DBE) recognized by p65 in the SRSF3/5 promoter regions, and we predicted four and two DBEs, respectively, by using the JASPAR database (Fig. 3J). ChIP-PCR results showed enrichment of p65 in the SRSF3/5 promoter regions (Fig. 3K and L). Dual luciferase reporter assays showed that p65 directly increased the transcriptional activity of SRSF3/5 (Fig. 3M). Overexpression or inhibition of p65 in CD8+ T cells promoted or inhibited, respectively, the expression of SRSF3 and SRSF5 (Fig. 3N). TGFβ1 stimulation induced the appearance of DNA binding-competent p65 dimers in the nucleus. Consistently, ChIP analysis showed recruitment of p65 to the SRSF3/5 promoter regions (Fig. 3O). P38 inhibition impaired the enrichment of p65 in the SRSF3/5 promoter regions (Fig. 3P). In general, our findings demonstrated the mechanism by which TGFβ1 regulates SRSF3/5 expression via the activation of the P38/p-Ser10 H3/p65 axis to prolong the half-life of PD-1 mRNA in T cells.
SRSF3 binding with NXF1 promotes the extranuclear transport of PD-1 mRNA
RNA transportation from the nucleus to the cytoplasm is a crucial step in gene expression. SRSF3 coordinates a pluripotency gene expression program through mRNA export (30). Knockdown or overexpression of SRSF3 decreased or increased, respectively, the nucleocytoplasmic fractionation levels of PD-1 mRNA in CD8+ T cells (Fig. 4A), in agreement with the proposal that SR proteins promote nuclear export (31). To further explore the effect of SRSF3 on the exportation of PD-1 mRNA, we labeled nascent RNA with 5-ethynyl uridine. Upon knockdown of SRSF3 expression, CD8+ T cells were observed to have nuclear accumulation of PD-1 mRNA; however, cytoplasmic PD-1 mRNA was reduced in abundance (Fig. 4B). In parallel, a RIP assay showed that knockdown of SRSF3 expression or exogenous TGFβ1 stimulation decreased or increased, respectively, the enrichment of PD-1 mRNA associated with SRSF3 in CD8+ T cells (Fig. 4C and D). To validate whether mechanistically directed SRSF3 binding to PD-1 mRNA was required for PD-1 mRNA extranuclear transport, we synthesized PD-1 expression constructs carrying either the wild-type PD-1 cDNA sequence that included the 3′-UTR (WT-PD-1) or a PD-1 cDNA sequence with silent mutations abolishing the SRSF3-binding sites (ΔSRSF3-PD-1). As expected, RIP analysis demonstrated significantly reduced binding of SRSF3 to ΔSRSF3-PD-1 mRNA (Fig. 4E). Nucleocytoplasmic fractionation revealed that in scramble and SRSF3-deficient CD8+ T cells, ΔSRSF3-PD-1 mRNAs mainly accumulated in the nucleus. the nuclear export of WT-PD-1 mRNAs was restrained in SRSF3-deficient CD8+ T cells compared with scramble cells (Fig. 4F). Nuclear export factor-1 (NXF1) is recruited to mRNA by SRSF3 (32). In our study, we found that NXF1 interacted with SRSF3 in CD8+ T cells (Fig. 4G). Knockdown of SRSF3 expression decreased the interaction between SRSF3 and NXF1, whereas SRSF3 showed an enhanced interaction with NXF1 in CD8+ T cells stimulated with TGFβ1 (Fig. 4H and I). RIP with an anti-NXF1 showed NXF1 binding to WT-PD-1 mRNA in CD8+ T cells (Fig. 4J). RIP analysis demonstrated significantly increased binding of SRSF3 to WT-PD-1 mRNA compared with ΔSRSF3-PD-1 mRNA in the context of NXF1 overexpression (Fig. 4K). Nucleocytoplasmic fractionation revealed that in scramble and NXF1-deficient CD8+ T cells, ΔSRSF3-PD-1 mRNAs mainly accumulated in the nucleus. The nuclear export of WT-PD-1 mRNAs was reduced in NXF1-deficient CD8+ T cells compared with scramble cells (Fig. 4L). These results directly demonstrated the functional importance of SRSF3 coordination with NXF1 in the regulation of PD-1 mRNA extranuclear transport in T cells.
TGFβ1 induced SRSF3 to restrict the antitumor capability of T cells in ccRCC
TGFβ1 impairs the antitumor response to PD-1/PD-L1 blockade via exclusion of T cells (33). We next used BALB/c mice to test the importance of TGFβ1 in antitumor immunity. In an orthotopic model, on the fifth day after injection, mice were treated with anti–PD-L1 or IgG antibody. We found that knockdown of TGFβ1 expression enhanced the effect of PD-L1 inhibition on ccRCC and prolonged the survival of mice (Fig. 5A–D). On the 40th day, TGFβ1 knockdown had decreased PD-1 and SRSF3 expression in tumor-infiltrating CD8+ T cells, further supporting the hypothesis that TGFβ1 is a critical mediator of PD-1 expression in the ccRCC microenvironment (Fig. 5E). We predicted Srsf3/5 binding sites for the Srsf3/5 proteins in the 3′-UTR of the mouse Pd-1 mRNA sequence in the RBPmap and SpliceAid databases (Supplementary Fig. S6A). We performed a RIP assay and observed that knockdown of Srsf3 or Srsf5 expression decreased the enrichment of Pd-1 mRNA on Srsf5 or Srsf3 in mouse CD8+ T cells (Supplementary Fig. S6B and S6C). We also conducted RNA stability assays with mouse CD8+ T cells by knocking down Srsf3 or Srsf5 expression and observed that the half-life of Pd-1 mRNA was decreased in the knockdown groups compared with the scramble groups (Supplementary Fig. S6D and S6E). These results suggested that the mechanism by which Srsf3/5 regulates the Pd-1 mRNA half-life exhibited cross-species conservation in mice.
To verify the intrinsic role of SRSF3 in the antitumor immune activity of CD8+ T cells, we conducted adoptive T-cell transfer in the ccRCC model. On the sixth day after tumor cell injection, unstimulated OT-1 cells overexpressing Srsf3 or control OT-1 cells were subcutaneously transferred into BALB/c mice bearing Renca- OVA ccRCC tumors (Fig. 5F). The number of OT-1 cells in the Srsf3-overexpressing group was lower than the number in the scramble group in the tumor-draining lymph nodes. However, the numbers of Srsf3 OE and scramble cells were comparable in non-dLNs (Fig. 5G). OT-1 cells overexpressing Srsf3 revealed higher Pd-1 expression and lower Ki-67 expression than scramble cells in the dLNs (Fig. 5H). The ability to control tumor progression was poor in the group overexpressing Srsf3, which also had a worse survival. However, anti–PD-1 therapy rescued the defect in the antitumor function of T cells overexpressing Srsf3 (Fig. 5I and J). Taken together, these findings indicated that TGFβ1 increased SRSF3 expression to restrict the antitumor capability of T cells by increasing PD-1 expression. Repression of the TGFβ1-mediated pathway can rescue T-cell dysfunction to enhance antitumor activity.
There are various types of immune processes involved in different antitumor and tumor-promoting lymphocyte actions, and tumor cells use many strategies to evade immune responses. The TME composition modulates which immunosuppressive pathways become activated to repress antitumor immunity (34), and ccRCC immune evasion is an essential step in tumor progression, leading to unfavorable outcomes in patients (4). Here, we described a mechanism of PD-1 regulation at the mRNA level and demonstrated the importance of this mechanism in T-cell dysfunction (Fig. 6). We identified six microenvironment-associated genes: IL4R, TGFB1, CXCR4, ITGA5, MKNK1, and ACTN1. Our findings showed that TGFβ1 expression was increased in neoplastic tissues and advanced malignant stages and that high TGFB1 expression predicted a poor outcome in patients with ccRCC. We performed KEGG pathway analysis of genes correlated with TGFβ1 expression in the TCGA ccRCC datasets and found that TGFβ1 associated with negative regulation of the immune system process.
TGFB1 expression was observed to be significantly positively correlated with the expression of PD-1, which was highly expressed on the surface of T cells in neoplastic areas. PD-1 immune checkpoint blockade shows activity in patients with advanced ccRCC. A previous study reveals that FBXO38 mediates PD-1 ubiquitination and regulates the antitumor immune activity of T cells (35). TGFβ1 transcriptionally regulates the expression of PD-1 selectively through Smad3 (10). However, the mechanism by which PD-1 expression is promoted in T cells in ccRCC seems unclear but of great interest. Here, we found that ccRCC-derived TGFβ1 prolonged the half-life of PD-1 mRNA in T cells. The regulation of the mRNA half-life is a critical determinant of the magnitude of expression. High mRNA instability can prevent the inappropriate expression of the encoded protein in resting cells, and unstable mRNAs often require stimulus-induced prolongation of the half-life for the protein to be expressed (36). We then analyzed proteins that potentially bound to the 3′-UTR of PD-1 mRNA, and found that CUG-BP, SRSF1, HNRNPA1, SRSF5, SRSF9, MBNL1, YBX1, and SRSF3 were predicted RNA-binding proteins. CUG-BP, also known as CUGBP1 (CUG triplet repeat, RNA binding protein 1), regulates pre-mRNA alternative splicing and may also be involved in mRNA editing (37). SRSF1, SRSF3, SRSF5 and SRSF9, members of the SR protein family, act as common splicing factors (38, 39). HNRNPA1 and MBNL1, which belong to the subfamily of heterogeneous nuclear ribonucleoproteins (hnRNP), regulate pre-mRNA processing and other aspects of mRNA metabolism and transport (40). YBX1 has been implicated in the regulation of transcription and translation, pre-mRNA splicing, DNA repair, and mRNA packaging (41).
We further identified that SRSF3/5 could bind to and stabilize PD-1 mRNA in CD8+ T cells. SR proteins share a modular structure of one or two RNA recognition motifs (RRM) at their amino terminus and an arginine-serine-rich RS domain of variable length at the carboxyl terminus (38), and both domains can directly contact RNA (42). SRSF3 and SRSF5 display unique RNA-binding properties underlying diverse cellular regulatory mechanisms, and are involved in the occurrence of various diseases. Many target mRNAs of SRSF3 have been identified in cycling and neurally induced P19 cells, and the functional homogeneity indicates SRSF3 is related to transcriptional regulation, developmental processes, cell proliferation, and differentiation (43). SRSF3 displays clusters of binding sites at the 3′-UTR junction of the majority of histone mRNAs, implicating SRSF3 in histone mRNA metabolism (44). SRSF5 has been attributed to functions like splicing, maintaining mRNA stability and translation, facilitates the production of p19 H-RAS, and increases sensitivity to doxorubicin in human U-2 OS cells (45). The RNA recognition motifs of SRSF5 are sufficient to activate pre-mRNA splicing, whereas proteasome-mediated proteolysis of SRSF5 requires the presence of the C-terminal RS domain of the protein (46). The 3′-UTR variations at position 29742 could affect the interaction of SRSF5 and modulate genome stability in SARS-CoV2 (47). SRSF3/5 could affect biological functions via directly binding the target RNAs.
The export of mRNA from the nucleus to the cytoplasm is a regulated step in gene expression (48). SRSF3 causes aberrant splicing of nascent RNAs, which is crucial for cell differentiation and metabolic function (49). A previous study reveals that SRSF3 interacts with NXF1 and regulates NXF1 mRNA splicing (30, 32). Our results demonstrated that mutation of the SRSF3 binding sites abolished SRSF3 binding to PD-1 mRNA and resulted in nuclear accumulation of PD-1 mRNA, suggesting a function for SRSF3 in PD-1 mRNA extranuclear transport in T cells. SRSF3/5 expression was upregulated by TGFβ1 stimulation. To investigate the precise mechanism by which TGFβ1 increased SRSF3/5 expression to prolong the half-life of PD-1 mRNA, we have performed ssGSEA analysis and GO/KEGG pathway analysis, which revealed that TGFβ1 stimulation activated the P38 pathway activity in CD8+ T cells
It has been reported that P38 activation-dependent p-Ser10 H3 may mark promoters for increased p65 recruitment (29). GSEA analysis from the GSE87508 dataset showed SRSF3 and SRSF5 were involved in P38MAPK pathway in CD8+ T cells and that activated P38 could recruit p65 via p-Ser10 H3, induce SRSF3/5 expression, and prolong the half-life of PD-1 mRNA. These results suggested that ccRCC-derived TGFβ1 mediates immune escape and could be used as a therapeutic target to enhance antitumor immunity to improve the prognosis of patients with ccRCC. However, a randomized phase II trial evaluating the addition of dalantercept (an inhibitor of TGFβ1 superfamily receptor type I) to axitinib did not find improved treatment-related outcomes in previously treated patients with advanced RCC (50), which indicates that TGFβ1 pathway inhibition in combination with immunotherapy is an inexorable trend in the future.
Blocking TGFβ1 signaling and PD-L1 unleashes a potent and enduring cytotoxic T-cell response against colon and urothelial cancers and prevents metastasis (9, 33). We found that knockdown of TGFβ1 expression enhanced the efficiency of anti–PD-L1 in an orthotopic ccRCC model. We predicted Srsf3/5 binding motifs in the 3′-UTR of the mouse Pd-1 mRNA transcript, and found binding sites for the Srsf3/5 proteins. We observed that knockdown of Srsf3 or Srsf5 decreased the enrichment of Pd-1 mRNA on Srsf5 or Srsf3 in mouse CD8+ T cells, consistent with the evolutionarily conserved roles of SR proteins in mRNA processing (51, 52). Our results showed that knocking down Srsf3 or Srsf5 expression could decrease the half-life of Pd-1 mRNA in mouse CD8+ T cells. TGFβ1 signaling, P38 signaling, and NF-κB also affect the occurrence, development, or immune escape of tumors in mice (53–55). These results suggest that these Srsf3/5 pathways exerted biological functions with cross-species conservation in mice. To confirm the underlying role of SRSF3 induced by TGFβ1 in mediating CD8+ T-cell dysfunction in ccRCC, adoptive T-cell transfers were performed, and CD8+ T cells overexpressing SRSF3 were less effective at eradicating ccRCC cells than control T cells, but could be rescued by anti–PD-1 therapy. Our findings reveal a mechanism of increased PD-1 expression induced by TGFβ1 via upregulation of SRSF3/5 expression and provide evidence that patients with ccRCC may obtain clinical benefit from immune checkpoint blockade (56, 57).
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
P. Wu: Conceptualization, resources, data curation, funding acquisition, validation. B. Geng: Investigation, methodology, writing–original draft. Q. Chen: Funding acquisition, investigation, writing–original draft, writing–review and editing. E. Zhao: Data curation, formal analysis, supervision, methodology. J. Liu: Validation, investigation, writing–original draft, writing–review and editing. C. Sun: Resources, formal analysis, visualization, methodology. C. Zha: Software, formal analysis, supervision. Y. Shao: Data curation, formal analysis, visualization. B. You: Visualization, methodology, writing–original draft. W. Zhang: Data curation, supervision, writing–original draft. L. Li: Writing–original draft, project administration, writing–review and editing. X. Meng: Software, funding acquisition, investigation, visualization. J. Cai: Software, formal analysis, funding acquisition, validation, visualization, project administration. X. Li: Conceptualization, supervision, funding acquisition, validation, visualization.
This study was supported by the National Natural Science Foundation of China (nos. 81702972, 81874204, and 81902903); the China Postdoctoral Science Foundation (2018M640305, 2019M652109, and 2019M660074); the Research Project of the Chinese Society of Neuro-oncology, CACA (CSNO-2016-MSD12); the Heilongjiang Postdoctoral Science Foundation (LBH-Z18103 and LBH-Z19029); the Research Project of the Health and Family Planning Commission of Heilongjiang Province (2017-201 and 2019-102); and the Harbin Medical University Scientific Research Innovation Fund (YJSKYCX2018-95HYD).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.