Our recent study has shown that SRSF1 regulates the terminal maturation of thymocytes by post-transcriptionally regulating the abundances of Il27ra and Irf7 functional transcripts via alternative splicing (28). By reviewing the phenotype of thymocytes from Srsf1 ^fl/fl Lck ^Cre/+ mice, we found that the numbers of CD8 single-positive (SP) cells are more severe reduction than those of CD4⁺ SP cells, resulting in the substantially altered ratio of CD4⁺ to CD8⁺ cells ( Figures S1A–C ). In addition, we performed gene set enrichment analysis (GSEA) by using our published RNA-seq data (GSE141349). The results indicated that CD8⁺ cell-specific genes were enriched in wild-type DP cells relative to SRSF1-deficient DP cells, suggesting that the differentiation capacity of DP cell toward CD8⁺ SP was more significantly reduced in absence of SRSF1, although both CD4⁺ and CD8⁺ SP thymocyte-related genes exhibited the enrichment in wild-type DP cells ( Figure S1D ). To address the potential role of SRSF1 involved in the lineage choice of CD4-versus-CD8 thymocytes, we established the genetic mouse model with conditional inactivation of SRSF1 in DP stage by crossing Srsf1 ^fl/fl mice with Cd4-Cre mice (30), which is widely applied for the lineage determination analysis of late thymocytes ( Figure S2A ). The deletion efficiency of SRSF1 was further confirmed in district subsets along with the sequential developmental phases, indicating the effective deletion of Srsf1 was achieved in DP and CD4/CD8 SP thymocytes from Srsf1 ^fl/fl Cd4-Cre mice compared with those in their littermate control mice (henceforth called Control) ( Figure S2B ).

We next analyzed the phenotype of these conditional knock out mice. Compared with their controls, Srsf1 ^fl/fl Cd4-Cre mice exhibited comparable size and cellularity of thymus and spleen, but diminished cell number in lymph nodes ( Figures 1A, B ). The frequency of both CD4⁺ and CD8⁺ thymocytes from SRSF1-deficient mice was significantly decreased ( Figures 1C, D ), whereas the percentage of DP thymocytes was correspondingly increased, reflecting a blockade of DP thymocyte development. The cell numbers of CD8⁺ thymocytes in SRSF1-deficient mice were significantly reduced, but no statistical difference in absolute numbers of DP and CD4⁺ thymocytes was observed. The ratio of CD4⁺ cells to CD8⁺ cells was notably altered ( Figure 1E ), implying more severe impacts on CD8⁺ lineage development caused by conditional Srsf1 deletion in DP thymocytes. To determine the specific developmental stage of thymocytes that was impaired in Srsf1 ^fl/fl Cd4-Cre mice, we carved up thymocytes at five distinct developmental phases defined by the expression of TCRβ and the activation marker CD69 as previous described (8, 9, 31) ( Figure 1F ). There was no significant difference observed from populations 1 to 3 between Srsf1 ^fl/fl Cd4-Cre mice and their controls, implying the DP thymocytes at pre-selection and the initial stage of positive selection were not affected in absence of SRSF1 ( Figures 1F, G ). In contrast, Srsf1 ^fl/fl Cd4-Cre mice had significantly fewer cells in populations 4 to 5 which include post-selected DP, immature SP, and mature SP thymocytes, respectively. These results indicate that ablation of SRSF1 at DP thymocytes mainly impairs the T cell development beyond the post-selection phase.

We next focused on the post-selection TCRβ^hi thymocytes with an additional maturation marker CD24 staining combined with the activation marker CD69 of thymocytes as previously described (28). The frequency and cell numbers of TCRβ^hiCD69^-CD24^- mature subset were decreased from Srsf1 ^fl/fl Cd4-Cre mice compared with those from Controls ( Figures 2A, B ). The frequency of TCRβ^hiCD69⁺CD24⁺ immature T cell exhibited a relative increase, but the numbers were slightly diminished ( Figures 2A, B ). By further analysis of the expression of CD4 and CD8 in TCRβ^hiCD69⁺CD24⁺ immature subsets, we found that the frequency and numbers of DP, CD4⁺CD8^lo intermediate cells, and CD4⁺ SP subsets were not significantly alerted, but the frequency and numbers of CD8⁺ SP were remarkably decreased in Srsf1 ^fl/fl Cd4-Cre mice ( Figures 2C, D ). In SRSF1-deficient TCRβ^hiCD69^-CD24^- mature population, the numbers of CD4⁺ and CD8⁺ SP were dramatically diminished, though the frequency of CD4⁺ SP cells was increased whereas the frequency of CD8⁺ SP cells was reduced ( Figures 2C, D ). Moreover, the ratio of CD4⁺ to CD8⁺ SP cells was notably increased in both TCRβ^hiCD69⁺CD24⁺ immature and TCRβ^hiCD69^-CD24^- mature thymocytes from Srsf1 ^fl/fl Cd4-Cre mice ( Figure 2E ). Collectively, these data indicated that SRSF1 deficiency impaired the terminal maturation of both CD4⁺ and CD8⁺ SP cells, and led to the aberrant ratio of CD4⁺ to CD8⁺ SP cells.

We next checked whether the peripheral T cell pool was affected in Srsf1 ^fl/fl Cd4-Cre mice. The mature CD4⁺ and CD8⁺ T cell populations in spleens, LNs and PBCs were remarkably diminished in Srsf1 ^fl/fl Cd4-Cre mice ( Figures 3A, B ). By further analysis of the proportion of CD4⁺ to CD8⁺ cells in peripheral tissues, we found the frequency of CD4⁺ T cells was increased in SRSF1-deficient TCRβ⁺ cells, and the ratio of CD4/CD8 in peripheral tissues was increased, accordingly ( Figures 3C–E ). These results suggested the critical requirement of SRSF1 in maintaining the numbers of mature T cells, especially CD8⁺ cells in periphery T cell pool.

To determine whether the developmental defects in Srsf1 ^fl/fl Cd4-Cre were T cell autonomous, we generated bone marrow chimeric mice as described in Figure 4A . We found thymocytes derived from Srsf1 ^fl/fl Cd4-Cre mice had a phenotype identical to that of thymocytes in primary SRSF1-deficient mice as described above ( Figures 4B – H ). The severe defects were detected in population 4 and 5 of thymocytes derived from Srsf1 ^fl/fl Cd4-Cre mice ( Figures 4B, C ), and the frequency of TCRβ^hiCD69^-CD24^- mature population was substantially reduced ( Figures 4D, E ). In chimeric mice transplanted with Srsf1 ^fl/fl Cd4-Cre donor cells, the frequency of donor-derived CD8⁺ SP cells was remarkably reduced in both TCRβ^hiCD69⁺CD24⁺ immature and TCRβ^hiCD69^-CD24^- mature thymocytes, and the ratio of CD4⁺ to CD8⁺ SP cells was notably increased, accordingly ( Figures 4F – H ). These data thus demonstrated the impacts on maturation of late thymocytes and CD8 lineage fate were T cell intrinsic.

To further evaluate how SRSF1 contributes to CD8⁺ lineage choice, we crossed Srsf1 ^fl/fl Cd4-Cre mice with MHC class II-deficient (H2ab1^-/- ) mice, which lack mature CD4⁺ SP thymocytes ( Figure 5A ). We found the frequency of CD8⁺ SP cells in both immature and mature thymocytes from H2ab1^-/-Srsf1 ^fl/fl Cd4-Cre mice was substantially lower compared with those in their control mice ( Figure 5B ). The frequency of CD4⁺ SP cells in mature thymocytes from H2ab1^-/-Srsf1 ^fl/fl Cd4-Cre mice was significantly higher than those from their control mice ( Figure 5B ). The number of both mature and immature CD8⁺ SP cells was dramatically lower in H2ab1 ^{− / −} Srsf1 ^fl/fl Cd4-Cre mice, accordingly ( Figure 5C ). In contrast, the number of immature CD4⁺ SP cells was comparable from H2ab1 ^{− / −} Srsf1 ^fl/fl Cd4-Cre and Control mice, whereas the number of mature CD4⁺ SP cells was diminished in H2ab1 ^{− / −} Srsf1 ^fl/fl Cd4-Cre due to SRSF1 deficiency ( Figure 5C ). These data collectively indicated that SRSF1 deficiency impaired the CD8 lineage identity. We next detected the expression of genes involved in lineage selection in immature TCRβ⁺ DP, CD4⁺CD8^lo, and mature CD8⁺ SP thymocytes, including Runx3, Thpok (Zbtb7b), Tle3, Bcl11b, Tcf7, Tox, Gata3, IL7Rα and Mazr. The abundance of CD8 master regulator Runx3 was substantially reduced in all three stages, and the significant elevation of Tox and Mazr was observed in DP stage but no changes in CD4⁺CD8^lo and mature CD8⁺ SP thymocytes in SRSF1-deficient cells ( Figure 5D ). Although the expression of Tle3, Bcl11b, and IL7Rα was dramatically decreased in CD8⁺ SP thymocytes, most of detected lineage commitment-related genes were not altered in the essential transient stages (DP and CD4⁺CD8^lo), such as Thpok, Tcf7, Tle3, Bcl11b, and Gata3 ( Figure 5D ). These results imply that SRSF1 may contribute to the CD8 lineage fate by primarily controlling Runx3 expression.

We next attempted to explore whether enforced expression of Runx3 could rectify the defects in the CD8 lineage fate caused by SRSF1 deficiency. To achieve this goal, the retrogenic mouse models were established and analyzed as described in the flowchart ( Figure 6A ). We confirmed the transduced efficiency of BM LSK cells was more than 50% before transplantation ( Figure S3 ) to ensure the successful construction of chimeric mice. By analyzing donor-derived TCRβ^hi post-selection thymocytes, we found that the reduction of mature (TCRβ^hiCD69⁻CD24⁻) thymocytes was substantially restored by forced expression of SRSF1, but not by forced expression of Runx3 compared with those derived from Control-MigR1 or Srsf1 ^fl/fl Cd4-Cre-MigR1 donors ( Figures 6B, C ). Meanwhile, the ectopic expression of SRSF1 also rectified the ratio of CD4⁺ to CD8⁺ SP cells in both TCRβ^hiCD69⁺CD24⁺ immature and TCRβ^hiCD69⁻CD24⁻ mature thymocytes ( Figures 6D – F ). However, overexpression of Runx3 could largely restore the ratio of CD4⁺ to CD8⁺ SP cells in TCRβ^hiCD69⁻CD24⁻ mature stage while no rescue was observed in the TCRβ^hiCD69⁺CD24⁺ immature stage ( Figures 6D – F ). These data collectively revealed that Runx3 serves as a regulator downstream SRSF1 for CD8 lineage fate decision, but other regulators and more complicated mechanisms may involve in the SRSF1-dependent regulatory network of late thymocyte maturation and lineage fate decision.

All mice used in this study were between 7 and 10 weeks of age on a fully C57BL/6J background. Srsf1 ^fl/fl mice were kindly provided by Dr Xiang-Dong Fu (University of California, San Diego). Cd4-Cre and H2ab1 ^−/− mice from Jackson Laboratories were maintained in the animal facility of China Agricultural University. Mice were housed in specific pathogen-free conditions under controlled temperature (22 ± 1°C) and exposed to a constant 12-hour light/dark cycle. All institutional and national guidelines for the care and use of laboratory animals were followed and all animal protocols used in this study were approved by the Institutional Animal Care and Use Committee at China Agricultural University.

Single cell suspensions obtained from thymus (Thy), spleen, lymph node (LN), and peripheral blood cells (PBCs) were stained with fluorochrome-conjugated antibody as described previously (42). The fluorochrome-conjugated antibodies listed below: CD4 (RM4-5), CD8a (53-6.7), CD24 (M1/69), CD69 (H1.2F3), TCRβ (H57-597), B220 (RA3-6B2), CD11b (M1/70), CD11c (N418), CD45.1 (A20), CD45.2 (104), CD49b (DX5), Gr.1 (RB6-8C5), TER119 (TER-119), TCRγδ (GL-3), ScaI (D7), cKit (2B8) and 7AAD (00-6993-50) were purchased from eBiosciences. The fluorochrome-conjugated streptavidin (554063) was purchased from BD Biosciences. Samples were acquired on a LSRFortessa or FACSVerse (BD Biosciences) and analyzed with FlowJo software v10.4.0 (Tree Star, Inc.). For cell sorting, cells were surface-stained with indicated fluorochrome-conjugated antibodies and subjected to sorting on a FACSAria II (BD Biosciences).

The gene expression was measured by qPCR as previously described (43). Briefly, total RNA was extracted from sorted cells using RNasey Mini Kit (Cat. # 74106, Qiagen) according to manufacturer’s instructions. FastQuant RT Kit (Cat. # KR106-02, Tiangen) was used to synthesize cDNA. Quantitative RT-PCR (qPCR) was performed with SYBR Green Master Mix (Cat. # FP205-02, Tiangen) using CFX96 Connect™ Real-Time System (Bio-Rad). The primers were shown in Supplementary Table 1 . Fold differences in expression levels were calculated according to the 2^−ΔΔCT method and the relative expression of indicated genes was normalized to Gapdh.

The BM chimeric mice were generated as previously described (44). Briefly, the lethally irradiated B6.SJL (CD45.1⁺) mice were transferred intravenously with a 1:1 mixture of 1 × 10⁶ BM cells from Srsf1 ^fl/fl Cd4-Cre (CD45.2⁺) or control mice together with BM cells from congenic B6.SJL (CD45.1⁺) mice. After 10 weeks reconstitution, recipients were sacrificed and analyzed.

The retrogenic chimera mouse models were generated by a modified protocol as previously described (28, 45). Briefly, retroviral packaging was carried out by transfection of HEK293T cells with Runx3 cDNA bearing retroviral vector or empty pMigR1 vector along with pCLeco using Lipofectamine 2000 (Cat. # 11668019, Invitrogen), and the retrovirus-containing medium was collected at 24- and 48-hours post-transfection. After being filtered by 0.45 µm filters, the retrovirus-containing medium was loaded and centrifuged onto RetroNectin-coated [10 μg/mL (Cat. # T100A, TaKaRa)] non-tissue culture 24 well plates (Cat. # 351147, Falcon). BM cells from Control and Srsf1 ^fl/fl Cd4-Cre mice were depleted of lineage positive cells and cultured for 24 hours in IMDM medium in the presence of thrombopoietin (20 ng/mL), stem cell factor (50 ng/mL), 15% FBS, 2-mercaptoethanol (50 µm), streptomycin and penicillin (100 μg/mL) in retrovirus contained RetroNectin plate as described above. Then, cells were infected with fresh retrovirus-containing medium in the presence of 8 μg/mL Polybrene (Cat. # H9268, Sigma-Aldrich) by centrifuging at 1,000 rcf for 90 min at 32°C. Subsequently, the cells were cultured for 2 hours at 37°C 5% CO₂ incubator and resuspended in IMDM medium supplemented with components and cytokines as above. On the next day, the cells were spino-infected again. The infected cells were collected and analyzed by flow cytometry 24 hours later, and then these cells containing 5,000 GFP⁺ lineage⁻ScaI⁺cKit^hi (LSK) cells were transplanted into lethally irradiated (7.5 Gray) recipients (CD45.1⁺). The recipients were sacrificed to analyze at 8 weeks after transplantation.

GSEA (v4.0.2) was used to analyze RNA-Seq data (GSE141349) from the GEO database, and the gene sets used in the article were obtained from MSigDB.

Statistical analysis was carried out through using GraphPad Prism software (version 8.0). Statistical significance was determined by one-tailed Student’s t-test. *, P < 0.05; **, P < 0.01; ***, P < 0.001.