The Tumor Microenvironment Impairs Th1 IFNγ Secretion through Alternative Splicing Modifications of Irf1 Pre-mRNA

The tumor microenvironment (TME) is conditioned by complex interactions between malignant and nonmalignant cells. Tumor cells mainly secrete immunosuppressive mediators, which contribute to immune surveillance evasion (1, 2). The TME consists of a wide variety of cells, immune and not. These cells dynamically communicate among themselves, as well as with tumor cells, leading to tumor progression or regression. TGFβ is secreted by malignant and nonmalignant cells. This cytokine has an antagonist role in the TME and can either inhibit tumor cell proliferation or disturb activity of antitumor immune cells (3, 4). An increasing number of studies show that alternative splicing is involved in immune system differentiation and homeostasis (5, 6). This molecular mechanism provides cell flexibility and diversity because it increases the number of protein isoforms from a unique gene. For example, it has been shown that 1,319 alternative splicing events occur upon T-cell receptor engagement (7). Alternative splicing allows the immune system to adapt to different environments, such as those within lymph nodes and tumors. Splicing is controlled by a huge protein complex called the spliceosome, which is regulated by splicing factors that control its binding to pre-mRNA (8).

Here, we showed that Irf1, encoding crucial transcription factor, IRF1, in Th1 cells, undergoes alternative splicing in tumors, giving rise to a shortened IRF1 isoform. TGFβ was involved in SFPQ (splicing factor, proline- and glutamine-rich) splicing factor activation and was responsible for the appearance of the short IRF1 isoform. This isoform acted on the full-length IRF1 transcriptional activity, decreasing Il12rb1 expression and subsequently Ifng expression. Targeting IRF1Δ7 could, therefore, potentiate Th1-cell antitumor responses.

Four tumor samples were kindly provided by the Cancer Center George François Leclerc (Dijon, Burgundy, France), and 13 healthy donors' buffy coats were obtained from Etablissement Français du Sang. Tumor samples were dissociated using the gentleMACS Dissociator (Miltenyi Biotec) in association with Tumor Dissociation Kit (human; 130-095-929, Miltenyi Biotec) or purified with the RosetteSep Human CD4⁺ T-cell Enrichment Cocktail (StemCell Technologies). The study on patient samples was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Centre Georges François Leclerc (Dijon, Burgundy, France), the Comité Consultatif de Protection des Personnes en Recherche Biomédicale de Bourgogne. Written informed consent was obtained from all patients before enrollment.

Cell lines were cultured at 37°C under 5% CO₂ in DMEM [for MC-38, B16-F10, LLC1, EL4, 293T, and platinum-E (plat-E)] or RPMI (for CT-26 and 4T1), supplemented with 10% [volume/volume (v/v)] FCS (Lonza), 1% penicillin–streptomycin/amphotericin B (Gibco), and 4 mmol/L of 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES, Gibco). The MC-38 (Kerafast, ENH204-FP) cell line was derived from C57BL/6 murine colon adenocarcinoma cells. B16-F10 (ATCC, CRL-6475) cell line was derived from a C57BL/6 murine melanoma. LLC1 (ATCC, CRL-1642) cell line was derived from a C57BL/6 murine lung carcinoma. EL4 (ATCC, TIB-39) cell line was derived from a C57BL/6 murine lymphoma. 293T (ATCC, CRL-3216) cell line was derived from a human embryonic kidney. Plat-E (Cell Biolabs, RV-101) cell line was generated from 293T cell line. Plat-E cells contain gag, pol, and env genes, allowing retroviral packaging with a single-plasmid transfection. The CT-26 (ATCC, CRL-2639) cell line was derived from a BALB/c murine colorectal carcinoma. The 4T1 (ATCC, CRL-2539) cell line was derived from a BALB/c mammary tumor. All cells were routinely tested for Mycoplasma contamination using the MycoAlert Mycoplasma Detection Kit (Lonza) and were found to be negative.

All the mice were maintained in specific pathogen-free conditions at Dijon Zootechnical Centre, and all experiments followed the guidelines of the Federation of European Animal Science Associations. All animal experiments were approved by the Ethics Committee of Université de Bourgogne (Dijon, Burgundy, France; approved protocol #2212), in accordance with the Federation of Laboratory Animal Science Associations. Only females between 6 and 10 weeks of age were used for the experiments. Female C57BL/6 and BALB/c mice were purchased from Charles River Laboratories. To induce tumor formation, 8 × 10⁵ MC-38, 8 × 10⁵ CT-26, and 2 × 10⁵ B16-F10 cells were injected subcutaneously into C57BL/6 mice. A total of 8 × 10⁵ CT-26 cells were injected subcutaneously into BALB/c mice. Tumor size was measured three times a week with an electronic caliper. After 1 week, animals were randomized according to tumor size before any treatment to ensure group homogeneity. Mice received three times a week an intraperitoneal injection of IgG1 control antibody (BioXCell, BE0083) or TGFβ-blocking antibody (BioXCell, BE0057; 100 μg/mouse). After 2 weeks, according to our institutional ethical board, animals were sacrificed after anesthetizing. Lymph nodes and spleens were harvested and dissociated using syringe plunger and a 70-μm filter (Miltenyi Biotec). Red blood cell lysis buffer (155 mmol/L NH₄Cl, 12 mmol/L NaHCO₃, and 0.1 mmol/L EDTA) was added for 2 minutes, and cells were centrifuged (400 × g, 5 minutes) and then suspended in Flow Cytometry Buffer (eBioscience, 00-4222-26) to be labeled. Tumors were harvested and dissociated using the gentleMACS Dissociator (Miltenyi Biotec) in association with the Tumor Dissociation Kit (Miltenyi Biotec, 130-096-730). The tumor shreds were filtered using syringe plunger and a 70-μm filter (Miltenyi Biotec) and centrifuged (400 × g, 5 minutes). Supernatants were harvested for IFNγ ELISA and CD45⁺ cells were purified with CD45 (tumor-infiltrating lymphocyte) MicroBeads (Miltenyi Biotec, 130-110-618). CD45⁺ cells were labeled in flow cytometry buffer and analyzed or sorted by flow cytometry.

CD4⁺ T cells were isolated from spleens and tumors of C57BL/6 (MC-38 and B16-F10) or BALB/c (CT-26) mice as described above. For RNA sequencing (RNA-seq) library preparation, total RNA from CD4⁺ T cells was extracted using TRizol. rRNA was removed using the Ribo-zero rRNA Removal Kit (Illumina). The RNA integrity numbers were determined using the TapeStation System (Agilent Technologies) and were always above 8. A total of 100 ng of rRNA-depleted RNA was used for the library preparation using the TruSeq Stranded Total RNA Library Prep Kit (Illumina) following the manufacturer's instructions. RNA-seq was performed on NextSeq Device (Illumina). The RNA-seq libraries were sequenced with paired-end 75-bp reads. FASTQ files were mapped by using BWA (mm10 version of Mus Musculus genome) for Illumina. The analysis was performed by using TopHat for Illumina. Generated files were processed with Cufflinks software to obtain annotated expressed genes in each studied subset. Differential expression between the samples was analyzed with Cuffdiff. Unsupervised hierarchical clustering of samples was performed by using Gene Cluster 3.0 software and viewed with TreeView viewer. Genes were normalized in transcripts per million (TPM). All sequencing data are available on the Gene Expression Omnibus website with the SuperSeries reference: GSE134855.

For RNA-seq library preparation, total RNA from CD4⁺ T cells cultured with IL12 and with or without TGFβ (3 or 10 ng/mL) was extracted using TRizol (Invitrogen). The library was prepared using the cDNA-PCR Sequencing Kit (Oxford Nanopore Technologies). Sequencing was performed on GridION Device (Oxford Nanopore Technologies). The analysis was performed in collaboration with the Genomic Platform, Institut de Biologie de l'ENS (Paris, France).

Naïve CD4⁺ T cells were isolated from C57BL/6 mouse spleen and differentiated into Th1 cells for 2 hours. Total RNA was extracted using TRizol and reverse transcribed into cDNA as described below. cDNA was then amplified by high-fidelity PCR with the Platinum Taq DNA Polymerase High Fidelity (Thermo Fisher Scientific, 11304011) using primers targeting the exons 6 and 7 or exon 6/exon8 junction and intron of Irf1 (Supplementary Table S1). Sanger sequencing was performed using the Applied Biosystems BigDye Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific) and the ABI PRISM 3130xl Sequencer (Thermo Fisher Scientific).

Intron 6, exon 7, and intron 8 mouse sequences were used to search for splicing factors binding motifs. In silico analysis was performed with ENCODE (https://www.encodeproject.org/), RBPmap (http://rbpmap.technion.ac.il/) software, SFmap (http://sfmap.technion.ac.il/), and AVISPA (https://avispa.biociphers.org/galaxy/) software. The Venn diagrams were generated using the online tool provided by VIB and Ghent University (Ghent, Belgium, http://bioinformatics.psb.ugent.be/webtools/Venn/).

Naïve CD4⁺ T cells (CD4⁺CD62L⁺) were obtained from spleen and lymph nodes of C57BL/6 wild-type mice. CD4⁺ T cells were purified from spleen and lymph nodes with anti-CD4 MicroBeads (Miltenyi Biotec, 130-093-227), and were then further sorted as naïve CD4⁺CD62L⁺ T cells using LS Column (Miltenyi Biotec, 130-042-401). The purity of the isolated naïve T-cell population routinely exceeded 95%. A total of 5 × 10⁵ naïve T cells were cultured with anti-CD3 (2 μg/mL) and anti-CD28 (2 μg/mL) in the presence of anti-IFNγ (10 μg/mL) and anti-IL4 (10 μg/mL) to obtain Th0; anti-IL4 (10 μg/mL) and IL12 (10 ng/mL) to obtain Th1; anti-IFNγ (10 μg/mL) and IL4 (10 ng/mL) to obtain Th2; anti-IFNγ (10 μg/mL), anti-IL4 (10 μg/mL), IL6 (20 ng/mL), and TGFβ (2 ng/mL) to obtain Th17; and anti-IFNγ (10 μg/mL), anti-IL4 (10 μg/mL), and TGFβ (4 ng/mL) to obtain regulatory T cells (Treg). Cells were cultured for 2, 24, or 72 hours at 37°C under 5% CO₂ in RPMI1640 with 10% (v/v) FCS, supplemented with sodium pyruvate, 1% penicillin and streptomycin, and 4 mmol/L of HEPES. Anti-CD3 (Armenian hamster IgG, clone 145-2C11, BE0001-1), anti-CD28 (Armenian hamster IgG, clone PV-1, BE0015-5), anti-IL4 (rat IgG1, clone 11B11, BE0045), and anti-IFNγ (rat IgG1, clone XMG1.2, BE0055) were obtained from BioXCell, and IL12, IL6, and IL4 were purchased from R&D Systems and TGFβ from Miltenyi Biotec. For in vitro treatments, naïve CD4⁺ T cells were pretreated for 1 hour with PI3K Inhibitor (0.1 μmol/L, Merck Chemical France) before beginning the Th1 differentiation with or without TGFβ (10 ng/mL).

Naïve CD4⁺ T cells (CD4⁺CD62L^hi) and differentiated CD4⁺ T-cell subsets (Th1, Th2, and Th17) were obtained from human healthy donor buffy coats. Cells were purified by flow cytometry and restimulated with plate-bound antibodies against CD3 (1 μg/mL) and CD28 (5 μg/mL). A total of 5 × 10⁵ naïve T cells were cultured with anti-CD3 (2 μg/mL) and anti-CD28 (2 μg/mL) in the presence of anti-IFNγ (10 μg/mL) and anti-IL4 (10 μg/mL) to obtain Th0; anti-IL4 (10 μg/mL) and IL12 (10 ng/mL) to obtain Th1; anti-IFNγ (10 μg/mL) and IL4 (20 ng/mL) to obtain Th2; anti-IFNγ (10 μg/mL), anti-IL4 (10 μg/mL), IL6 (20 ng/mL), and TGFβ (2 ng/mL) to obtain Th17; and anti-IFNγ (10 μg/mL), anti-IL4 (10 μg/mL), and TGFβ (4 ng/mL) to obtain Tregs. Cells were cultured for 2, 24, or 72 hours at 37°C under 5% CO₂ in RPMI1640 with 10% (v/v) FCS, supplemented with sodium pyruvate, 1% penicillin and streptomycin, and 4 mmol/L of HEPES. Anti-CD3 (clone OKT-3; BE0001-2), anti-CD28 (clone 9.3; BE0248), anti-IL4 (clone MP4-25D2; BE0240), and anti-IFNγ (clone B133.5; #BE0055) were obtained from BioXCell, and IL12, IL6, TGFβ, and IL4 were purchased from Miltenyi Biotec.

For retrovirus infection, full-length Irf1 and the alternatively spliced isoform (Irf1Δ7) were cloned into the pMXs-IRES-GFP Retroviral Vector (Cell Biolabs). RNA from Th1 cells, differentiated from naïve CD4⁺ T cells as described above, was extracted using TRizol and was reverse transcribed into cDNA as described below. Fragments were amplified by high-fidelity PCR with C57BL/6 mouse cDNA as template and specific primers (Supplementary Table S1). Ligation of DNA fragments was performed with T4 DNA Ligase (Promega, M1801). Insert orientation was determined by PCR and restriction enzyme digestion. Retroviral particles were generated by transfecting the Plat-E Cells (Cell Biolabs, RV-101) with Lipofectamine 2000 (Invitrogen, 11668019), according to the manufacturer's instructions. At the same time, naïve CD4⁺ T cells were isolated as described above and cultured in 24-well plates (1 × 10⁶ cells/well) coated with anti-CD3 and anti-CD28. After 2 days, 1 mL of fresh virus supernatant was harvested and mixed with 2 × 10⁶ proliferative CD4⁺ naïve T cells and protamine sulphate (10 μg/mL, APP Pharmaceuticals) in a 24-well coated plate and centrifuged for 60 minutes at 800 × g at 32°C. Transduced naïve CD4⁺ T cells were collected after 2 days, and cell sorted with a FACSAria Cell Sorter (BD Bioscience) according to GFP expression. GFP⁺ cells used for intracellular staining were fixed in 2% paraformaldehyde (PFA). Otherwise, GFP⁺ cells were differentiated as described above, and after 1 or 3 days, the cells were collected for PCR and ELISAs.

Total RNA from differentiated CD4⁺ T cells was extracted using TRizol. In total, 300 ng of RNA was reverse transcribed into cDNA using M-MLV Reverse Transcriptase (28025-013, Invitrogen), random primers, and RNAseOUT Inhibitor (10777-019, Invitrogen). cDNAs were quantified by real-time PCR using a Power SYBR Green Real-time PCR Kit (4367659, Life Technologies) on a StepOne Detection System (Thermo Fisher Scientific). Relative mRNA levels were determined using the ΔC_t method. Values were expressed relative to β-actin, unless otherwise specified. The sequences of the oligonucleotides used are described in Supplementary Table S1.

After polarization for 72 hours, cell culture supernatants of different tumor cell lines (1/100 dilution) were assayed by ELISA for mouse IFNγ (BD Biosciences, 555138). IFNγ concentration was also assessed in tumor shreds (pure). An ELISA for mouse TGFβ (R&D Systems, DY1679-05) was performed according to the manufacturer's protocol.

Differentiated CD4⁺ T cells were lysed in boiling buffer [1% SDS, 1 mmol/L sodium orthovanadate, and 10 mmol/L Tris (pH 7.4)] containing Protease Inhibitor Cocktail (Roche, 11697498001) for 20 minutes at 4°C. Cell lysates were subjected to sonication (10 seconds at 10% intensity), and protein concentration was assessed using the Bio-Rad DC Protein Assay Kit (5000112, Bio-Rad). Proteins were then denatured, loaded, and separated on SDS-PAGE and transferred on Nitrocellulose Membranes (Schleicher & Schuell). After blocking with 5% nonfat milk in PBS containing 0.1% Tween 20 (PBST), membranes were incubated overnight with primary antibody diluted (1 μg/mL) in PBST containing 5% BSA, washed, and incubated for 1 hour with secondary antibody (1/10,000) diluted in PBST 5% nonfat milk. After additional washes, membranes were incubated for 1 minute with Luminol Reagent (Santa Cruz Biotechnology). The following antibodies were used for immunoblotting: primary antibodies: rabbit anti-IRF1 N-terminal region, purchased from Aviva Systems Biology and mouse anti-HSC70 (B-6), purchased from Santa Cruz Biotechnology and secondary antibodies: peroxidase AffiniPure goat polyclonal anti-rabbit IgG (H+L) (Jackson ImmunoResearch) and peroxidase AffiniPure goat polyclonal anti-mouse IgG (H+L) (Jackson ImmunoResearch).

Chromatin isolation and shearing were performed with the truChIP Chromatin Shearing Kit (Covaris) using a Focused-Ultrasonicator M220 Device (Covaris). A chromatin immunoprecipitation (ChIP) assay was performed with the ChIP-IT Kit (Active Motif Europe) according to the manufacturer's instructions. A total of 7 μg of DNA was immunoprecipitated with 3 μg of anti-histidine or anti-YFP (Thermo Fisher Scientific), or 3 μg of negative control immunoglobulin at 4°C overnight. After chromatin elution, cross-links were reversed, and samples were analyzed by quantitative PCR, as described above. The sequences of the oligonucleotides used are described in Supplementary Table S1.

RNA immunoprecipitation (RIP) was performed with Magna RIP RNA-binding protein immunoprecipitation kit according to the manufacturer's instructions. RNA was immunoprecipitated with 5 μg of anti-SFPQ (Abcam). RNA was reverse transcribed into cDNA and analyzed by quantitative PCR as described previously.

siRNA-knockdown experiments were performed in mouse and human with validated controls, or Irf1-, Irf1Δ7-, Sfpq (mouse)-specific siRNAs (Life Technologies). In brief, naïve CD4⁺ T cells were transfected with Transit-TKO Transfection Reagent (MIR2154, Mirus) according to the manufacturer's instructions for 24 hours and then differentiated as described previously. The sequences of the siRNAs used are described in Supplementary Table S1.

The Il12rb1-luc luciferase construct was obtained by inserting 154 bp of Il12rb1 mouse promoters in the multicloning site of the pGl3 Basic Vector (Promega). Fragments were amplified by high-fidelity PCR with C57BL/6 mouse DNA as the template and specific primers (Supplementary Table S1). A total of 1 × 10⁵ human 293T cells were transiently transfected for 48 hours with the pSV-β-galactosidase (Promega) reporter plasmid and the pcDNA3.1/CT-GFP-TOPO-IRF1 (Thermo Fisher Scientific) and pcDNA3.1/CT-GFP-TOPO-IRF1Δ7 using Lipofectamine 2000 (Invitrogen). Luciferase was measured using the Dual Glo Luciferase Assay System (Promega, E2920) according to the manufacturer's instructions. Firefly luciferase was measured with an EnVision 2105 Multimode Plate Reader (PerkinElmer).

A total of 1 × 10⁶ CD4⁺ T cells per condition were used. Cells were washed, fixed for 10 minutes at room temperature with 4% PFA and permeabilized for 10 minutes on ice with methanol (100% glacial), and saturated for 45 minutes with a buffer of 5% BSA and 0.2% Triton X-100 (Sigma-Aldrich) in PBS. Samples were then incubated overnight at 4°C with primary antibodies (identified below and detailed in Supplementary Table S1), diluted in 5% BSA buffer. Cells were washed two times with a solution of 0.05% Tween 20 in PBS (PBS-Tween) and incubated for 1 hour at room temperature with the secondary antibody (identified below and detailed in Supplementary Table S1), diluted in the immunofluorescence (IF) blocking buffer. Cells were then washed two times with PBS-Tween solution and two times with ultrapure water. Stained cells resuspended in ultrapure water were dropped off on microscopy slides (Superfrost Ultra Plus, Thermo Fisher Scientific) and incubated at room temperature in dark until water evaporated. Dried slides were mounted with a drop of mounting medium containing DAPI (Molecular Probes) and covered with a cover slip (Knittel Glass). Slides were imaged with a Charge-coupled Device–equipped Upright Microscope (Zeiss) and 40× or 63× objective with a numerical aperture of 1.4 using Zeiss ApoTome system (for 63× magnification). Images were analyzed with ZEN Lite (Zeiss). Threshold was defined according to negative control fluorescence (IgG). The following antibodies were used for IF: primary antibody, rabbit anti-SFPQ (Abcam) and secondary antibody, goat polyclonal anti-rabbit IgG (H+L) Cross-Adsorbed Alexa Fluor 568 (Thermo Fisher Scientific).

The IRF1-YFP and IRF1Δ7-CFP constructs were obtained by inserting 819 bp of Irf1Δ7 in the multiple cloning site of the mVenus C1 Vector (Addgene) and by inserting 990 bp of Irf1 into mCerulean C1 (Addgene), respectively. Fragments were amplified by using high-fidelity PCR with C57BL/6 mouse cDNA as template and specific primers (Supplementary Table S1). Fluorescence lifetime imaging (FLIM) was performed, and images were collected using a time-correlated single-photon counting (TCSPC) module (PicoQuant) on a Nikon A1-MP Scanning Microscope. Imaging was carried out with a 60× Apo IR objective (NA: 1.27, Water Immersion, Nikon). Two-photon excitation at 820 nm was provided by an IR Laser (Chameleon, Coherent) that delivered femtosecond pulses at a repetition rate of 80 MHz. Fluorescence emission of CFP was collected through a FF01-494/20 Band-pass Emission Filter (Semrock) using a single-photon avalanche diode detector. TCSPC lifetime recording was performed over 200 temporal channels (final resolution, 0.64 ps). We performed a global lifetime analysis on regions of interest of the FLIM images using SymPhoTime Software (PicoQuant). The fluorescence lifetimes were calculated by fitting the tail of the fluorescence decay with a biexponential model.

To phenotype and sort distinct mouse CD4⁺ T-cell populations by flow cytometry, dissociated lymph nodes, spleens, and MC-38 tumors were analyzed. To phenotype and sort distinct human CD4⁺ T-cell populations by flow cytometry, colorectal tumors and healthy donor buffy coats were analyzed. For surface staining, CD4⁺ T cells were stained with different antibodies (Supplementary Table S1) in Flow Cytometry Staining Buffer (eBioscience) and Brilliant Stain Buffer (BD Biosciences) for 15 minutes at room temperature. For intracellular cytokine staining, CD4⁺ T cells were stimulated for 3 hours at 37°C in culture medium containing PMA (50 ng/mL, Sigma), Ionomycin (1 mg/mL, Sigma), and Monensin (GolgiStop, 1 mg/mL, BD Biosciences). Cells were stained in Flow Cytometry Staining Buffer (eBioscience) and Brilliant Stain Buffer (BD Bioscience) with different antibodies (Supplementary Table S1). FOXP3 intracellular staining was carried out according to the manufacturer's protocol using the Fixation/permeabilization Solution (eBioscience). All events were acquired by a BD LSRII, FACSCanto 10-Color, or a LSRFortessa cell analyzer, or sorted by a FACSAria Cell Sorter (BD Bioscience) equipped with BD FACSDiva Software (BD Biosciences). Data were analyzed using Flowlogic or FlowJo software.

Results are shown as mean with SD or SEM. Datasets were compared using unpaired Mann–Whitney test, Kruskal–Wallis test, and one/two-way ANOVA test, when appropriate. Statistical calculations were done using GraphPad Prism 7. All P values were two-tailed. Ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

First, we tested the pattern of alternative splicing in bulk of CD4⁺ T cells from spleens or tumors by RNA-seq (Supplementary Fig. S1). We observed a global increase in splicing events in tumors compared with spleens (Fig. 1A). Gene set enrichment analysis (GSEA) of genes with a higher expression in tumors showed a significant enrichment in splicing factor gene expression (Fig. 1B). When comparing transcriptomic profiles of bulk CD4⁺ T cells from spleens and tumors, we observed a clear impact on Th1- (Fig. 1C), Th2-, Th17-, and Treg-related gene expression (Supplementary Fig. S1B). Th1 cell presence in tumors is reported as a good prognosis indicator (9); therefore, we wondered whether TME-induced alternative splicing modifications had an impact on Th1 master controller genes Tbx21 and Irf1. To determine whether Tbx21 and Irf1 pre-mRNAs were alternatively spliced in Th1 cells, we generated Th1 CD4⁺ T cells in vitro and analyzed both Tbx21 and Irf1 by PCR and Sanger sequencing. We detected an Irf1-spliced isoform in which exon 7 had been excluded (Irf1Δ7), whereas no splice variant was found for Tbx21 mRNA (Supplementary Fig. S2A). In addition to exon 7 skipping, we identified the inclusion of intron 8 together with an in-phase stop codon by Sanger sequencing (Supplementary Fig. S2B) and through Nanopore sequencing (Supplementary Fig. S2C), avoiding nonsense-mediated decay degradation and enabling mRNA stabilization (Supplementary Fig. S2D). Quantification by digital PCR revealed that the strongest expression level of Irf1 was detected 2 hours after Th1 differentiation start (Fig. 1D), and 4 hours for Irf1Δ7 mRNA (Fig. 1E). After 4 hours of differentiation, Irf1Δ7 mRNA represented almost 20% of the Irf1 gene expression (Fig. 1F). In different subsets of in vitro–differentiated Th cells, we observed that Irf1 mRNA was 56%, 30%, and 78% less expressed in Th2, Th17, and Treg, respectively, than in Th1 cells, and that Irf1Δ7 mRNA was 83%, 46%, and 87% less expressed in Th2, Th17, and Treg, respectively, than in Th1 cells (Supplementary Fig. S3). We confirmed the expression of Irf1 (Fig. 1G) and Irf1Δ7 mRNA (Fig. 1H) in in vivo Th1 cells isolated from spleens (Supplementary Fig. S4). To determine whether the short form of IRF1 was detected at the protein level, we performed Western blotting using an antibody targeting the protein N-terminal domain. We detected two proteins of 37 and 35 kDa, corresponding to the predicted sequence molecular weight (Supplementary Fig. S5), indicating that both proteins were expressed (Fig. 1I). Collectively, these findings indicated that IRF1, a master controller of Th1 cell differentiation, could be affected by alternative splicing at the mRNA level, giving rise to another stable isoform that is translated in Th1 cells.

TGFβ is one of the most common cytokines found in TME (10) and is described as alternative splicing modifier in cancer cells (11). By comparing the transcriptomic profiles of CD4⁺ T cells isolated from spleens and tumors, we underlined TGFβ-related gene enrichment in tumor CD4⁺ T cells (Fig. 2A and B), whereas no enrichment was found with IL6 (Supplementary Fig. S6A), another cytokine in TME (12). By dosing TGFβ in the tumor shred of different cell lines (Fig. 2C) and directly in the cell culture supernatant (Supplementary Fig. S6B), we validated the secretion of this cytokine by different mouse tumor cell lines, CT-26 and MC-38. The amount of TGFβ was higher in tumor shreds than in supernatants (except for EL4 cells), suggesting that TGFβ could be secreted by tumor cells and tumor-infiltrating cells. On this basis, we hypothesized that TGFβ from TME could have an impact on Th1 CD4⁺ T-cell alternative splicing and consequently on Irf1Δ7 isoform expression. To test this hypothesis, naïve CD4⁺ T cells were cultured for 2 hours with IL12 and increasing doses of TGFβ, and then differentiated for 24 hours with IL12 only (Supplementary Fig. S7). Using Nanopore sequencing, we confirmed that TGFβ affected Irf1 mRNA exon 7 inclusion at 2 hours (Fig. 2D), and we also noted an increase in intron 8 inclusion (Fig. 2E). After 2 hours of TGFβ exposure, we did not observe, by PCR, differences in Irf1 mRNA expression (Fig. 2F), but we did see an increase in Irf1Δ7 mRNA expression, indicating an increase in exon 7 exclusion (Fig. 2G), confirming the Nanopore data. Thus, in these conditions, the fold change ratio was in favor of Irf1Δ7 mRNA (Fig. 2H). After 24 hours of Th1 differentiation without TGFβ, Tbx21 (Fig. 2I) and Foxp3 (Fig. 2J) mRNA expression remained unchanged, while Il12rb1 (Fig. 2K) and Ifng mRNA (Fig. 2L) expressions were decreased in a dose-dependent manner. To confirm the involvement of TGFβ in Irf1 mRNA alternative splicing, we monitored tumor growth in MC-38 (TGFβ producing)-bearing C57BL/6 mice treated with TGFβ-blocking antibody or control (Supplementary Fig. S8A). We observed inhibition of tumor growth upon TGFβ targeting compared with controls (Supplementary Fig. S8B). After sorting Th1 cells from spleens and tumors from anti-TGFβ–treated and untreated mice, we analyzed the Irf1Δ7/Irf1 ratio by qRT-PCR. We observed an increase in Irf1Δ7 mRNA expression in Th1 cells from tumors compared with those from spleens in controls (Fig. 2M). In parallel, we observed a significant decrease in Irf1Δ7 mRNA in tumors when mice were treated with TGFβ-blocking antibody (Fig. 2M). We also found a significant increase in IFNγ in tumors from mice treated with TGFβ-blocking antibody (Fig. 2N). This observation was validated by an increase in CD4⁺IFNγ⁺ cells in tumors from mice treated with TGFβ-blocking antibody (Fig. 2O; Supplementary Fig. S9A and S9B). We observed no significant differences in CD4⁺Foxp3⁺ cells in blocking antibody–treated and untreated tumors (Fig. 2P; Supplementary Fig. S9B). We obtained similar results in BALB/c mice injected with CT-26 cells in terms of tumor growth, Irf1Δ7 mRNA expression, and IFNγ secretion (Supplementary Fig. S10A–S10C). Together, these data showed that TGFβ was able to induce Irf1Δ7 isoform appearance in vitro and in vivo.

Because TGFβ impacted Irf1 pre-mRNA alternative splicing, we wondered what could be the mechanism at play. Using transcriptomic analysis of CD4⁺ T cells isolated from spleens and tumors, we observed 26 splicing factor genes overexpressed in CD4⁺ T cells infiltrating tumors (Fig. 3A). Through in silico analysis (Supplementary Figs. S11–S14), we analyzed intron 6, exon 7, and intron 7 of Irf1 pre-mRNA sequence to identify splicing factor binding sites. By crossing RNA-seq results and in silico results, we identified SFPQ as the splicing factor potentially involved in Irf1Δ7 mRNA expression (Fig. 3B). Transcriptomic analysis of Th1 cells isolated from spleens and B16-F10 tumors validated SFPQ and heterogeneous nuclear ribonucleoprotein F (hnRNPF) in Irf1Δ7 mRNA expression (Supplementary Fig. S15A and S15B). To determine whether any splicing factors were involved in TGFβ-induced exon 7 alternative splicing, naïve CD4⁺ T cells were cultured for 2 hours with TGFβ and IL12. We then analyzed the transcriptional expression of Hnrnpf, Sfpq, and Hnrnph1 (another splicing factor upregulated in tumors, but not selected in silico. Only Sfpq mRNA expression was significantly increased, whereas neither Hnrnpf nor Hnrnph1 mRNA was statistically impacted by TGFβ pretreatment (Fig. 3C). SFPQ activity is described to be regulated by the PI3K/AKT pathway, leading to its fixation on CD45 mRNA (13). After naïve CD4⁺ T-cell treatment using IL12 and TGFβ, we observed, by IF, that SFPQ expression in the nucleus was increased in response to TGFβ (Fig. 3D). GSEA showed a positive enrichment of PI3K/AKT pathway–related genes in tumors (Fig. 3E). To confirm the involvement of PI3K/AKT pathway, we pretreated IL12- and TGFβ-incubated naïve CD4⁺ T cells with a PI3K/AKT pharmacologic inhibitor. This completely abolished the increase of Irf1Δ7 mRNA expression in response to TGFβ (Fig. 3F). To further study the effect of TGFβ in SFPQ activity, we analyzed SFPQ interaction on Irf1 mRNA upon RIP. We observed that TGFβ increased the binding of SFPQ to Irf1Δ7 mRNA (Fig. 3G). When we transfected Th1 cells with a siRNA-targeting Sfpq (Supplementary Fig. S16A–S16C), we observed a decrease in Irf1Δ7 mRNA expression (Fig. 3H), whereas Irf1 was not impacted (Supplementary Fig. S16D). These data suggest the involvement of SFPQ in Irf1Δ7 mRNA appearance, through PI3K/AKT pathway activation in response to TGFβ.

After seeing TGFβ's effects on Irf1Δ7 occurrence, we focused on determining whether this isoform may have a direct effect in Th1 activity. Using siRNA, we saw that Irf1 depletion (Supplementary Fig. S16E) associated with a decrease in Il12rb1 and Ifng mRNA expression. Irf1Δ7 (Supplementary Fig. S16F) and Sfpq (Supplementary Fig. S16A–S16C) mRNA depletion associated with an increase in Il12rb1 and Ifng mRNA expression (Fig. 4A–D). Irf1Δ7 and Sfpq mRNA depletion induced higher IFNγ secretion (Fig. 4E). In contrast, overexpression of IRF1Δ7 using a retroviral strategy (Fig. 4F) inhibited Th1-cell differentiation, as demonstrated by the decreased production of both Il12rb1 and Ifng mRNAs (Fig. 4G and H). Th1 IFNγ⁺ cells (Fig. 4I; Supplementary Fig. S17) and IFNγ secretion (Fig. 4J) were also decreased by IRF1Δ7 overexpression.

To understand how IRF1Δ7 impacted Il12rb1 and Ifng expression in Th1 cells, we first performed a ChIP analysis to determine whether IRF1Δ7 could interact with DNA. We used 293T cells transfected with either Irf1 or Irf1Δ7 or both expression plasmids tagged with histidine or YFP. Cells were transfected with Il12rb1 plasmid containing 154 bp of the promoter region and 18 bp downstream of the transcription initiation site known to be the DNA binding site of IRF1 (14). We confirmed an interaction of IRF1 with Il12rb1 promoter, but no interaction of IRF1Δ7 was observed (Fig. 4K). When both IRF1 isoforms were present, we observed a disruption of IRF1 interaction with Il12rb1 promoter (Fig. 4K). To confirm the negative effect of IRF1Δ7 on IRF1 transcriptional activity, we cloned Il12rb1 promoter into a luciferase reporter plasmid. Il12rb1 reporter plasmid was transfected either alone or with an Irf1-expressing plasmid with an increasing yield of Irf1Δ7-expressing plasmid. IRF1 alone significantly activated the Il12rb1 promoter, as evidenced by an increased luciferase activity. The activity of the Il12rb1 promoter, due to the IRF1 isoform, was decreased in an IRF1Δ7 dose–dependent manner (Fig. 4L). We, therefore, hypothesized that IRF1Δ7 could interact with IRF1 and prevent its binding to DNA. To address this question, we performed a FRET assay in 293T cells transfected with both IRF1 and short-expression plasmids tagged with CFP and YFP, respectively. Our data indicated that the donor fluorescence (CFP) decreased faster when both isoforms were expressed together (Fig. 4M), indicating a high proximity (<5 nm) between the different isoforms. These data suggested a potential interaction between IRF1 and IRF1Δ7. Taken together, these data demonstrated that IRF1Δ7 reduced Th1 activity by decreasing DNA binding of IRF1 to the Il12rb1 promoter.

We then asked whether inhibiting the impact of IRF1 alternative splicing isoform was transposable to human cells. We validated the exon 7 exclusion in Th cells in vitro after 1 hour of differentiation (Supplementary Fig. S18). This isoform was similar to the Irf1Δ7 identified in mice. Expression of the short IRF1 mRNA isoform, detected by qRT-PCR, was similar among Th1, Th2, and Th17 cells (Fig. 5A). In vivo expression of IRF1Δ7, detected by qRT-PCR, was significantly higher in Th1 and lower in Th17 cells compared with naïve T cells (Fig. 5B; Supplementary Fig. S19). Because IRF1Δ7 impaired Th1 differentiation in mice, we wondered whether this IRF1 isoform could still have an effect on human Th1 differentiation. Targeting this isoform using siRNA (Supplementary Fig. S20A–S20C), led to an increase in Il12rb1 and Ifng mRNA expression (Fig. 5C and D), an increase in IFNγ⁺ cells (Fig. 5E; Supplementary Fig. S20D), and an increased IFNγ secretion as evaluated by ELISA (Fig. 5F). To determine whether TGFβ also had an effect on the alternative splicing profile of IRF1 mRNA in human Th1 cells, naïve CD4⁺ T cells or ex vivo Th1 cells obtained from healthy blood donors were cultured for 1 hour with IL12 and increasing doses of TGFβ, and then differentiated for 24 hours after TGFβ withdrawal (Supplementary Fig. S21). We did not note any differences in IRF1 mRNA expression (Fig. 5G and H), but we did observe an increase in IRF1Δ7 mRNA (Fig. 5I and J). Thus, in these conditions, the fold change ratio was in favor of IRF1Δ7 mRNA (Fig. 5K and L). After 24 hours of Th1 differentiation without TGFβ, IFNG expression (Fig. 5M and N) was decreased in a dose-dependent manner. To attest the effect of the TME on alternative splicing profile of IRF1 mRNA, we sorted Th1 cells from healthy blood donors and colon tumors, and observed a significant increase in the IRF1Δ7 isoform in Th1 cells isolated from tumors (Fig. 5O). Together, these data emphasize that the short IRF1 mRNA isoform is also expressed in different human Th cells, particularly Th1 cells, where it down modulates Th1 activation, as was observed in mouse models.

Alternative splicing of pre-mRNA provides a higher level of genetic diversity and complexity by increasing the number of protein forms of the same gene (15). This molecular phenomenon provides a way to respond and to adapt to a given environment (16). It is now well-recognized that the immune system is an integral part of the disrupted microenvironment in tumors (17). Our report showed an impact of TME in CD4⁺ Th1 cells. By comparing the same cells isolated from spleen or tumors, we observed a global alternative splicing increase in CD4⁺ T cells in tumors. Among them, we focused on Th1 cells, in which we identified a new splicing isoform of IRF1, a master controller of Th1 differentiation. From our results, we identified an isoform lacking exon 7 and included a part of intron 8 containing an in-frame stop codon.

TGFβ is the most common cytokine found in TME (18). It can be secreted both by tumor cells and by nontumor cells, like Tregs. TGFβ immunosuppressive mechanisms are not fully understood. Studies suggest that TGFβ signaling pathway in tumor cells has an important impact on alternative splicing of CD44, favoring epithelial-to-mesenchymal transition and metastasis (11). Here, our data showed that targeting TGFβ with blocking antibodies during tumor growth induced a decrease in Irf1Δ7 expression and an increase in IFNγ secretion in tumors. In vitro, we observed that increasing TGFβ concentration in presence of IL12 for 2 hours, corresponding to the peak level of Irf1 expression, increased Irf1Δ7 expression. This treatment did not affect the expression of Tbx21, the master regulator of Th1 cells, and did not induce the expression of Foxp3, the key transcription factor of Treg differentiation. However, we saw a decrease in Il12rb1 expression, a gene whose expression is under IRF1 control, and also a decrease in Ifng expression in a dose-dependent manner. Thus, Irf1 pre-mRNA alternative splicing could exacerbate TGFβ immunosuppressive effects on Th1 cells.

Splicing factors are key regulators of spliceosome activity because they modulate its binding to pre-mRNA (19). Their activity is finely regulated by extracellular signals, but underlying molecular mechanisms remain unclear (20). In human T cells, it has been reported that antigen stimulation induces skipping of three exons of CD45 to produce CD45RO (13), and the shorter isoform is less active than the full-length form and tempers T-cell activation. The splicing factor involved in this phenomenon is SFPQ. When phosphorylated by GSK3, SFPQ stays inactive because of its sequestration by TRAP150. Our results showed that SFPQ is the only splicing factor overexpressed in the TME that could be specific to Irf1 pre-mRNA splicing. Indeed, Sfpq had a significant increase in expression when cells were treated with TGFβ. We also found that SFPQ interacted more with Irf1 mRNA when naïve CD4⁺ T cells were exposed to IL12 and TGFβ. When SFPQ was targeted by siRNA, Irf1Δ7 mRNA expression was decreased, and the inhibition of PI3K (controlling GSK3 activity) activity with a pharmacologic inhibitor completely canceled the effect of TGFβ on Irf1Δ7 mRNA expression. Nevertheless, the full mechanism leading to SFPQ activation by this cytokine remains to be fully decrypted. However, it is known that SR (serine- and arginine-rich) proteins (21) and hnRNPs (22) activity are modulated by the PI3K/AKT pathway. Thus, the disruptions of alternative splicing mediated by TGFβ most likely impact some splicing factors, and SFPQ is responsible for Irf1Δ7 production.

The transcription factor, IRF1, has been identified to regulate the expression of a large number of genes in different cell types (23–26). Concerning immune cells, Th1 lymphocyte development is supported by IRF1 through the transcriptional control of Il12rb1 gene expression (14). In Tregs, IRF1 negatively regulates Foxp3 expression (27). However, alternative splicing of IRF1 is still largely underexplored, especially in immune cells. Nevertheless, experiments using normal and malignant human cervical tissue samples reveal five IRF1 variants lacking some combinations of exons 7, 8, and 9 (28). The study also showed that these variants were able to inhibit the transcriptional activity of IRF1 full-form. In this context, we observed that IRF1Δ7 had a negative impact on Th1-cell activity. Thus, IRF1Δ7 impaired IRF1 transcriptional activity by preventing its interaction with the Il12rb1 promoter, probably by sequestration. Th1 cells are known to have potent antitumor properties because of their ability to secrete a large amount of IFNγ, a source of activation of other antitumor cells, like CD8⁺ T cells (29). We found in vitro that, in the absence of the short isoform of IRF1, Th1 cells can secrete even more IFNγ, which might potentiate their antitumor activity.

RNA-seq has shown that around 95% of human genes are alternatively spliced (30). However, few studies have analyzed the impact of alternative splicing on human immune system homeostasis (31). Our study indicated that, as in mice, an IRF1 alternative isoform without the exon 7 exists in humans. When we targeted the IRF1Δ7 form, we observed an increase in IFNG expression that tended toward exacerbated antitumor activity. We noted that, when compared with Th1 cells from healthy donors' blood, Th1 cells from tumors had a higher expression of IRF1Δ7. Hence, a short form of IRF1 was also expressed in human Th1 cells and possessed similar negative effects on Th1-cell activity, as was observed in mouse models.

In summary, we have shown that a short isoform of IRF1 lacking exon 7 is expressed in Th1 cells, and that its expression is increased in the TME via a TGFβ-dependent mechanism. We also found that IRF1Δ7 had a negative impact on IRF1 isoform transcriptional activity in Th1 cells. Targeting IRF1Δ7 could, therefore, potentiate Th1-cell antitumor responses.

L. Apetoh reports personal fees from Orega Biotech and Roche outside the submitted work. No disclosures were reported by the other authors.

A. Bernard: Conceptualization, formal analysis, validation, investigation, methodology, writing–original draft. C. Hibos: Conceptualization, investigation. C. Richard: Conceptualization, formal analysis, validation, investigation, methodology. E. Viltard: Conceptualization, investigation. S. Chevrier: Investigation. S. Lemoine: Nanopore analysis. J. Melin: Investigation. E. Humblin: Conceptualization. R. Mary: Conceptualization, investigation. T. Accogli: Conceptualization. F. Chalmin: Conceptualization, investigation. M. Bruchard: Conceptualization. P. Peixoto: Investigation. E. Hervouet: Investigation. L. Apetoh: Resources. F. Ghiringhelli: Conceptualization, supervision, methodology. F. Végran: Conceptualization, supervision, methodology, writing–review and editing. R. Boidot: Conceptualization, resources, supervision, funding acquisition, methodology, writing–original draft, writing–review and editing.

This work was funded by the Fondation ARC Pour la Recherche Sur le Cancer and the Ligue Nationale Contre le Cancer. The authors would like to thank the Plateforme de Cytométrie (http://www.cytometrie-dijon.fr, Anabelle Sequeira, Dijon, Burgundy, France) for providing reagents and access to their cytometers, the Centre de Zootechnie (http://recherche.u-bourgogne.fr/potentiel-de-recherche/plateformes-technologiques/centre-de-zootechnie.html, Valérie Saint-Giorgio, Dijon, Burgundy, France) for providing animal breeding facilities and devices to perform ex vivo and in vivo experiments, and the DImaCell Imaging Facility for the use of FLIM microscopy and for technical assistance. DImaCell equipment was funded by the Regional Council of Bourgogne - Franche Comté and the “Fonds Européen de DEveloppement Régional (FEDER).” Finally, the authors thank Isabel Grégoire, CGFL for editing the article; “Ligue contre le Cancer de Côte d'Or” and “Ligue contre le Cancer de Saône-et-Loire” for the financial support of this work; and “Ligue contre le Cancer” and “Fondation ARC” for the financial support of A. Bernard.

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.