ISWI ATPase Smarca5 Regulates Differentiation of Thymocytes Undergoing β-Selection

This article is featured in In This Issue, p.3337

The production of mature T and B cells is a multistep process of differentiation from a multipotent progenitor that requires a coordinated regulation of gene expression, replication, DNA rearrangement, and repair. Progenitors of T cells migrate from the bone marrow (BM) into the thymus where they respond to a new environment by initiating a transcriptional program of T cell specification while proliferating extensively (1). During this process, CD4⁻CD8⁻ double-negative (DN) CD44⁺ positive early T lineage precursors (immature DN1) permanently silence the group of progenitor-related regulatory genes leading to the gradual upregulation of CD25 and the downregulation of c-Kit surface markers and resulting in the commitment completion at the end of the DN2 stage (CD44⁺CD25⁺c-Kit^int) (2). Thymocytes at the subsequent DN3 stage (CD44⁻CD25⁺) cease from cycling and, importantly, undergo a random rearrangement of gene segments at the TCRb locus and commence the expression of components related to the β-selection program. Upon the successful rearrangement that yields functional pre-TCR complexes, thymocytes proliferate rapidly, become rescued from the p53-regulated cell cycle arrest and apoptosis (3), and then are allowed to progress into the DN4 stage (CD44⁻CD25⁻). This transient population hence upregulates the expression of CD4 and CD8 to become double-positive (DP) cells and initiates TCRa locus rearrangement. DP cells with productive TCRαβ are positively and negatively selected so that only those with “proven” TCR can undergo differentiation into CD4 or CD8 single-positive (SP) cells (4).

Eukaryotic cells evolved numerous epigenetic regulatory mechanisms of gene expression, DNA replication, and repair to accomplish the T cell development. During early T cell differentiation, NURD and SWI/SNF chromatin-remodeling complexes were shown to play important roles in both activating as well as silencing the gene transcription (5, 6). The SWI/SNF-related matrix-associated actin-dependent regulator of chromatin subfamily A member 5 (Smarca5) represents a widely expressed and conserved chromatin-remodeling factor required for the early development in mouse and lower organisms (7). Smarca5 is an ATPase from the ISWI subfamily that functions as a molecular motor for nuclear complexes that assemble and slide basic chromatin subunits, nucleosomes. Smarca5-containing complexes have diverse nuclear functions: guiding the transcription of ribosomal (in NoRC and B-WICH complexes) and some coding genes (within the ACF or RSF complexes), participating in regularly spacing the nucleosomal array before and after DNA replication, facilitating the recruitment of DNA repair machinery (CHRAC and WICH complexes), and finally, orchestrating higher-order chromatin structure formation of centromeres and chromosomes (RSF) (8). Although several members of SWI/SNF and CHD family have had their roles established in T cell development through studies involving gene inactivation mouse models, such a role for the ISWI subfamily has not been determined yet.

Currently, there is only a limited knowledge of how Smarca5, which is highly expressed in lymphocytes (9), participates in lymphopoiesis. We previously showed that deletion of the Smarca5 gene resulted not only in the depletion of myeloerythroid precursors but also affected the earliest development of lymphoid progenitors in the mouse fetal liver (10). Additionally, Smarca5 was implicated in the V(D)J cleavage of the polynucleosomal substrate in a cell-free system (11). Another report implicated that Smarca5 in the ACF complex represses the IL2-Rα gene (CD25) via chromatin organizer Satb1 (12). Lastly, Smarca5 regulates the expression of key ILs (IL-2, IL-3, and IL-5) in murine EL4 T cell lymphoma (13). Although the role of Smarca5 in lymphopoiesis was previously suggested, the knockout models of Smarca5-interacting partners revealed no alterations in lymphoid development, including the deletion of Acf1/Baz1a (ACF and CHRAC complexes) (14, 15) or Tip5/Baz2a (NoRC) (15) genes in mice. Interestingly, it has been shown in vitro that Smarca5 can also remodel nucleosomes alone without being part of the complexes (16). As several of Smarca5-interacting partners are dispensable, studying the requirement of a catalytic subunit of the ISWI complexes by targeting experimentally Smarca5 during lymphoid development may represent a successful strategy to reveal its function in lymphopoiesis.

We, in this study, focused on deciphering the role of Smarca5 in thymocyte development and studied the molecular consequences of conditional Smarca5 deficiency in mice. Our work suggests that Smarca5 controls early T cell development by guiding early differentiation-coupled transcriptional programs at the DN3 stage and its deficiency results in thymocyte proliferation and survival defects through the activation of the DNA damage response.

Smarca5^flox conditional knockout mice contain loxP1 sites flanking the exon 5 of the Smarca5 gene (17), deletion of which produced a frame-shift mutation (Smarca5^del) that disrupts the expression of the Smarca5 protein. The murine strain expressing a codon-improved Cre recombinase (iCre) under hCD2 promoter [B6.Cg-Tg(CD2-cre)4Kio/J] was purchased from The Jackson Laboratory. The OT-II strain [B6.Cg-Tg(TcraTcrb)425Cbn/J] was kindly provided by Dr. T. Brdička (Institute of Molecular Genetics, Czech Academy of Sciences). Rag1 knockout (B6.129S7-Rag1tm1Mom/J) strain was kindly provided by Dr. T. Brabec (Institute of Molecular Genetics, Czech Academy of Sciences). Trp53 knockout strain (B6.129S2-Trp53tm1Tyj/J) was kindly provided by Dr. W. Edelmann (Albert Einstein College of Medicine).

A single cell suspension from thymi and spleens of 4–6-wk-old mice was obtained using a Dounce homogenizer. Cells were first preincubated for 10 min on ice with Fc receptor-blocking anti-CD16/32 (clone 93) Ab in PBS-1% biotin-free BSA solution and then stained for 20 min with specific primary Abs. Biotinylated primary Abs were revealed with streptavidin-conjugated fluorescent dyes (SAv-PE/Cy7, SAv-APC, and SAv-APC/Cy7). Labeled cells were analyzed on BD FACSCanto II (BD Biosciences) or CytoFLEX (Beckman Coulter) flow cytometers, and data analysis was performed using FlowJo software. The clones of mAbs were as follows: anti-CCR6 (29-2L17), anti-CD3ε (145-2C11), anti-CD4 (GK1.5), anti-CD5 (53-7.3), anti-CD8 (53-6.7), anti-CD11c (N418), anti-CD24 (M1/69), anti-CD25 (PC61), anti-CD27 (LG.3A10), anti-CD28 (E18), anti-CD44 (IM7), anti-CD45.1 (A20), anti-CD45.2 (104), anti-CD71 (RI7217), anti-CD73 (TY/11.8), anti-CD117 (c-Kit, 2B8), anti-B220 (RA3-6B2), anti-CD11b (Mac1, M1/70), anti-γδ T cell (GL3), hamster IgG-PE/Cy7 (HTK888), anti-Ly6G/Ly6C (GR1, RB6-8C5), anti-Nk1.1 (PK136), anti-TCRβ (H57-597), anti-TCRβ(Vα2) (B20.1), and Ter-119 (all from BioLegend). Splenocytes were analyzed as previously described (18).

For BM reconstitution experiments, 1 × 10⁷ BM cells from adult (8-wk-old) control C57BL/6J Ly5.1 (CD45.1) mice and S5^fl/flhCD2iCre Ly5.2 (CD45.2) donors were reciprocally transplanted into lethally irradiated (7.5Gy) adult (8 wk) S5^fl/flhCD2iCre Ly5.2 and C57BL/6J Ly5.1 control recipients, respectively. After 1 mo posttransplantation, thymi were tested for the presence of donor-derived cells using flow cytometric analysis. The Ab panel included CD45.1, CD45.2, CD44, CD25, CD4, lineage mixture (CD8, B220, Mac-1, Gr-1, Nk1.1, CD11c, and Ter119) and CD45.1, CD45.2, CD4, CD5, CD8, and lineage mixture (B220, Mac-1, Gr-1, Nk1.1, CD11c, and Ter119) for thymus.

Smarca5 conditional knockout mice and their age- and gender-matched respective controls were i.p. injected with 1 mg of BrdU in 100 μl PBS. After 3 h, the thymocytes were isolated and Ab-stained for flow cytometry analysis or cell sorting. BrdU Ag recovery and its detection by fluorescently labeled Ab were performed using the APC BrdU Flow Kit (BD Biosciences).

FACS-sorted DN3 thymocytes (small CD4⁻CD8⁻CD25⁺) and Lin⁻Sca1⁺Kit⁺ (LSK) BM cells were cocultured with OP9 stromal cells expressing the Delta-like ligand (OP9/N-DLL1) in the presence of 1 ng/ml mIL-7 (407-ML-005) and 5 ng/ml hrFLT3 ligand (308-FK-005; PeproTech) as previously described (19). In BM cell cocultures, the concentration of IL-7 was lowered to 0.1 ng/ml from day 12 to allow differentiation. For CFSE labeling, 3 × 10⁴ DN3 cells were stained with 2.5 μM CFSE following the manufacturer’s guidelines (CellTrace CFSE Cell Proliferation Kit; Invitrogen) and plated onto OP9 stromal cultures. Their proliferation and survival were analyzed by flow cytometry at 2, 4, 6, and 8 d of cocultures. OP9/N-DLL1 cells were kindly provided by Hiroshi Kawamoto (Kyoto University).

Thymi and spleens were fixed in 4% buffered formaldehyde for 48 h, transferred in 70% ethanol, and paraffin embedded. Sections were obtained at 3 μm thickness and stained in H&E and Giemsa. Cleaved Caspase-3 was detected by immunohistochemistry using a 1:1000 dilution of Ab (ab52293; Abcam) and visualized on a Zeiss Axio Scan.Z1. Casp3-positive cells were quantified using the Zen Blue Edition software (ZEISS).

CD4⁺ CD8⁺ (B220⁻ Gr-1⁻ Nk1.1⁻ CD11c⁻ CD11b⁻) thymocytes were sorted by FACS, and the total RNA was isolated using TRIzol Reagent (Invitrogen). DP cells pooled from 25 to 37 S5^fl/flhCD2iCre thymi (each pool of 3 × 10⁶ DP cells), 6 S5^fl/flhCD2iCre Trp53^−/− thymi (∼2.5 × 10⁶ DP cells/pool), and single nonpooled control thymi provided sufficient amounts of RNA for RNA sequencing analysis. Strand-specific cDNA libraries were prepared from a minimum of 1.7 μg of each DNase-treated (AM1906; Ambion DNA-Free Kit) RNA sample using the TruSeq Stranded mRNA LT Kit (Illumina). The RNA libraries were sequenced on an Illumina HiSeq 2500 instrument in Rapid Run mode with paired-end 100-bp sequencing length. Reads were mapped and aligned to mouse reference genome assembly GRCm38.p6, and transcripts were annotated and counted with Ensembl Release 94 (October 2018) using a HISAT2 aligner (20). The two RNA-seq technical replicates for each sample were combined. Differential expression analysis of RNA-seq data were performed in R Studio (21) using package DEseq2, which uses a median of ratios normalization method that accounts for sequencing depth as well as RNA composition (22). Volcano plots were drawn using the ggplot2 (23) package in R suite. Expression levels (transcripts per kb million) for Ensembl 94 genes were calculated in R using gene lengths retrieved by EDASeq package (24) and in-house scripts. The shrinkage of log₂ fold change values (log₂FC) was estimated using DESeq2 lfcShrink function using the adaptive t prior shrinkage estimator “apeglm” (25). The RNA-seq data are publicly available at the ArrayExpress database under accession number E-MTAB-7758 (https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-7758/).

Quantitative measurements of TCRb rearrangement were done on genomic DNA (DNeasy; QIAGEN) isolated from the FACS-sorted DN subpopulations. Quantitative PCR was run on 7900HT using Power SYBR Green PCR Master Mix (no. 4367659, Applied Biosystems) with the following primers: Dβ1_fwd: 5′-GTGGTTTCTTCCAGCCCTCAAG-3′, Dβ1_rev: 5′-GGCTTCCCATAGAATTGAATCACC-3′ and Dβ2_fwd: 5′-CAGGCTCTGGGGTAGGCAC-3′, Dβ2_rev: 5′-CCTCTTCCAGTTGAATCATTGTGG-3′. Primers for the control region were as follows: Apob_ex29_fwd: 5′-CTGCCGTGGCCAAAATAAT-3′ and Apob_ex29_rev: 5′-AATCCTGCAGATTGGAGTGG-3′. Cycle threshold (c_t) values of Dβ1 and Dβ2 regions were normalized to c_t from Apob control region.

Freshly isolated thymocytes were collected by centrifugation and resuspended in PBS with inhibitors of phosphatases (PhosSTOP; Roche) and proteases (cOmplete ULTRA; Roche). The cell suspension was then diluted 1:1 by adding a solution of 50 mM Tris-Cl (pH = 8) and 2% NaDodSO₄ (SDS) and incubated for 30 min at 97°C. Protein concentration was determined by bicinchoninic acid assay (no. 23228; Pierce Biotechnology). Proteins were then resolved on SDS gradient 4–15% polyacrylamide gels (Mini-PROTEAN TGX Stain-Free Gels; Bio-Rad Laboratories) and wet blotted (1 h at 100 V) onto PVDF membranes (no. 162-0177; Bio-Rad Laboratories). PVDF membranes were blocked for 1 h in 5% nonfat milk in TBS/0.1%Tween-20 (TBST) and incubated overnight at 4°C in 3%BSA/TBST 0.1% sodium azide with the following Abs: Smarca5 (1:1000, no. A301-017A-1; Bethyl Laboratories) and histone H3 (1:1000, no. ab791; Abcam). Membranes were 3 × 5 min washed in TBST buffer and incubated with peroxidase-conjugated F(ab′)2 Ab fragment, either donkey anti-rabbit or donkey anti-mouse (1:10,000; Jackson ImmunoResearch), in 5% nonfat milk in TBST for 1 h. Membranes were 3 × 5 min washed in TBST, and the protein signal was visualized by Pierce SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific) and detected with the ChemiDoc Imaging System (Bio-Rad Laboratories).

To delineate specific roles of Smarca5 in vivo, we previously created the Smarca5^fl allele containing two loxP1 sites surrounding exon5, the deletion of which results in removing the portion of evolutionarily conserved helicase domain and introducing a frame-shift mutation (10, 17) (Fig. 1A). To execute the T cell–specific deletion of the Smarca5 gene, we crossed Smarca5^fl/fl mice with hCD2iCre transgenic mice, in which the hCD2 promoter and locus control region drive the expression of the iCre recombinase (26). Use of the hCD2-iCre transgene to study thymocyte development was chosen as this model alone caused no defect in thymus development (27) as well as the cellularity and subset composition of the major lymphoid organs (including thymus, spleen, and lymph nodes) did not differ between hCD2-iCre–expressing mice and wild-type (wt) (28). We initially evaluated the onset of the Smarca5^fl gene deletion by PCR. Genomic DNA was prepared from FACS-sorted CD4/CD8 DN thymocyte subpopulations of Smarca5^+/flhCD2iCre mice and three primer sets amplifying wt, Smarca5-floxed (S5^fl), and recombined-floxed (S5^del) allele were used. Consistent with previous reports that used the hCD2iCre transgene (26), we observed Cre-mediated recombination of loxP1 sites as early as the DN1 stage in which the recombined S5^del allele was detectable (Fig. 1B). However, the deletion of the S5^fl allele at DN1 was only partial, whereas its complete deletion was observed at the DN3 stage. Western blot analysis of the whole thymic cellular extracts from S5^fl/flhCD2iCre mice confirmed a high efficiency of recombination as we observed a marked reduction of Smarca5 protein levels (Fig. 1C).

S5^fl/flhCD2iCre pups were born healthy and at normal Mendelian ratios. Following the third month of age, the Smarca5 mutants, however, displayed rectal prolapses, suggesting the disturbance of immune functions. Gross dissection of the Smarca5 mutant animals revealed severe thymic defects. Thymi from the mutant mice were dramatically reduced in size and weight (Fig. 1D), which was reflected by a 17-fold reduction in cellularity (Fig. 1E). Histological evaluation revealed an alteration of thymic corticomedullary architecture (Fig. 1F) and a higher number of cleaved Caspase-3–positive apoptotic death events in the cortex (Fig. 1G). The quantification of multiple cortical sections from different animals showed there was a 3-fold increase of apoptotic cells in the cortex of S5^fl/flhCD2iCre thymi. These observations indicated a developmental defect in thymocytes, accompanied by an increased cell death within the thymi of S5^fl/flhCD2iCre animals.

To gain insight into the developmental defects of thymocytes, we used flow cytometry of thymic cell suspensions of 4–6-wk-old S5^fl/flhCD2iCre mice. CD4/CD8 immunostaining revealed a marked reduction of DP and CD4-SP cell populations with a corresponding relative increase in DN thymocytes in the mutant thymi (Fig. 2A). In absolute counts, the DP cells were depleted by 60-fold, CD4-SP 90-fold, and CD8-SP cells 12-fold, whereas the DN thymocyte population was similar to controls (Fig. 2B). Because the Smarca5 allele deletion is completed by the DN3 stage (Fig. 1B), we analyzed each subpopulation of DN thymocytes to more precisely determine a stage in which the block of development occurs. CD44/CD25 Ag expression profiles of lineage-negative DN thymocytes revealed a relative increase in DN3 cells to the detriment of DN4 thymocytes (Fig. 2C). Translated into absolute counts, it was an almost complete absence of DN4 thymocytes in mutants, whereas the DN3 population was present in comparable numbers to controls (Fig. 2D).

We also examined the impact of Smarca5 deficiency on γδ T cell development. Using flow cytometry, we observed that Smarca5-deficient animals contain twice as many TCRδ⁺ thymocytes (CD4⁻CD8⁻TCRδ⁺) compared with controls (Fig. 2E, 2F). However, out of these TCRδ⁺ thymocytes, only 18.4% (compared with 42% in controls) were able to adopt the γδ fate (Fig. 2E) as indicated by the CD73 surface marker that discriminates TCRδ⁺ thymocytes committed to the γδ lineage (29). Additionally, the expression of surface CD24, which is normally enriched on immature γδ thymocytes and downregulated upon maturation into effector cells (30), was reduced in Smarca5 mutants (Fig. 2E). In contrast to controls, the lower expression of CD24 surface marker observed in the mutants compromised a clear separation of the CD73⁺ population into immature and mature subsets. The mature γδ thymocytes (CD24^lowCD73⁺) were further distinguished along the expression of the mutually exclusive CD27 and CCR6 surface markers into IFN-γ(CD27⁺)–producing or IL-17a(CCR6⁺)–producing γδ subsets (31, 32). The IFN-γ– and IL-17a–producing γδ subsets display a subtle imbalance in favor of an IL-17a–producing subset in the Smarca5 mutants (Fig. 2E). We conclude that the commitment to γδ lineage and CD24 expression by TCRδ⁺ thymocytes and the development of mature γδ subsets are impaired in the S5^fl/flhCD2iCre mice. Although Smarca5 plays important roles in thymocyte development during the DN3 to DN4 transition of αβ subsets, it also guides the development of the γδ compartment.

As hCD2-iCre transgene initiates deletion also in other murine hematopoietic cell subtypes (18), we tested whether the DN3 to DN4 transition defect is cell autonomous to thymocytes or a result of the impaired thymic microenvironment in which they develop. Using syngeneic transplantation, we assessed the ability of control BM (marked by CD45.1 isoform) to repopulate thymocytes in lethally irradiated S5^fl/flhCD2iCre mice (marked by CD45.2 isoform) and vice versa. At day 35 after transplantation, we observed that engrafted thymocytes from controls (CD45.1) developed normally to produce CD4/CD8 double and SP cells in the thymic microenvironment of S5^fl/flhCD2iCre (CD45.2) animals (Fig. 3A, 3B). In turn, the engrafted donor S5^fl/flhCD2iCre BM-derived thymocytes (CD45.2) recapitulated the developmental defect at DN3 to DN4 transition within the control acceptor animals (CD45.1). These results suggested that the DN3 to DN4 transition became defective independently of the thymic stromal components (of S5^fl/flhCD2iCre mice) but rather intrinsically to the thymocytes lacking Smarca5. To further settle the point of whether the developmental defect of thymocytes observed in vivo is cell autonomous, we used ex vivo cocultures with a BM-derived stromal OP9/N-DLL1 cell line (19). Sorted preselection early DN3e cells (small CD4⁻CD8⁻CD25⁺ thymocytes) were added on the OP9/N-DLL1 cells to evaluate the formation of DN4 and DP thymocytes in a time course of 8 d (Fig. 3C). Although control DN3e cells apparently proliferated and progressed into more mature developmental stages under ex vivo conditions, the majority of Smarca5-depleted cells were held at the DN3 stage till day 8, and their absolute numbers remained similar to the starting cocultures (Fig. 3C, 3D). Thus, the outcome of the ex vivo experiment was highly reminiscent of the phenotype of S5^fl/flhCD2iCre mice. We used yet another approach to test whether the defect of DN3 to DN4 transition could have emerged from a secondary effect, mainly because the DN3e thymocytes in the OP9/N-DLL1 cocultures were sorted from S5^fl/flhCD2iCre mice with the potentially impaired thymic microenvironment. We isolated LSK BM hematopoietic progenitors from S5^fl/flhCD2iCre or control mice and kept them differentiating ex vivo on the OP9/N-DLL1 cells. Again, both the control as well as S5^fl/flhCD2iCre-derived hematopoietic progenitor cells developed normally into the DN2 stage (day 9); however, from the 16th day of culture, the S5^fl/flhCD2iCre thymocytes were progressively underrepresented, and by day 22, the DN3 to DN4 transition defect was revealed (Fig. 3E, 3F). Thus, the loss of Smarca5 in developing T cells caused a defect intrinsic to the thymocytes undergoing DN3 to DN4 transition, as evidenced by the cocultures using the OP9 cells, and was not a result of a secondary effect due to impaired thymic stromal components or thymic microenvironment.

We next investigated whether reduced cell numbers in S5^fl/flhCD2iCre thymi and ex vivo OP9/N-DLL1 cocultures (Figs. 1E, 3D) may be attributed to premature cell death and/or impaired proliferation. As determined by flow cytometry, the ex vivo cultivation of purified DN3e thymocytes eventually resulted in a gradual increase of Annexin V positivity up to 71% by day 8 that did not exceed 15% in controls (Fig. 4A), indicating that the loss of Smarca5 induced apoptosis in developing T cell precursors. Similarly, we examined the effect of Smarca5 loss on the proliferation by labeling purified DN3e thymocytes with the intracellular fluorescent dye (CFSE) immediately before plating them on OP9/N-DLL1 cells. As seen in Fig. 4B, the analysis of the CFSE signal dilution indicated that the DN control thymocytes proliferated rapidly ex vivo. Conversely, those DN cells that did survive the ex vivo conditions exhibited a decreased division rate, confirming an impaired proliferation of S5^fl/flhCD2iCre thymocytes. We then used the BrdU incorporation assay to analyze the ability of Smarca5-depleted cells to progress through the various stages of the cell cycle. After a 3 h pulse of BrdU in vivo, we observed that the percentage of proliferating BrdU⁺ DN3 cells (after exclusion of postreplicative BrdU⁺ events) was almost the same in mutants as in controls (14.8% versus 15.8%), and thus, the G1-to-S progression was not altered after the Smarca5 loss at DN3 stage (Fig. 4C). However, the portion of postreplicative cells at this stage, which is defined as strictly diploid BrdU⁺ events (33), was 2.5-fold decreased in mutant DN3 cells. These data indicate that Smarca5-depleted DN3 cells normally enter S phase and begin to replicate their DNA; however, they are limited in completing the cell cycle to re-emerge as G1 cells. Mutant DP cells were also impaired because a substantial fraction of the DP cells became arrested at the G2/M phase (Fig. 4D). Thus, the developmental defects observed upon the loss of Smarca5 are likely the consequence of cell death and impaired cell cycle progression at late S through G2/M phase. This is in contrast to defects in G1/S checkpoint mechanisms as previously observed in human cancer cell lines upon SMARCA5 knockdown (34).

The accumulation of Smarca5-depleted DN3 cells resembles the phenotype of mice that have a defect in pre-TCR signaling or TCRb locus rearrangement (35). To test whether the induction of pre-TCR signaling was affected by a loss of Smarca5, we performed i.p. injections of anti-CD3ε Ab into S5^fl/flhCD2iCre and into RAG1 (Rag1^−/−)–deficient mice. Anti-CD3ε Ab can mimic pre-TCR signaling in vivo and stimulates preselection DN3 thymocytes to proliferate and differentiate into the DN4 cells even in the Rag1^−/− mice lacking the TCR β-chain expression (36). We observed that the anti-CD3ε Ab stimulated the downregulation of surface CD25 molecules on Rag1^−/− as well as on Smarca5-deficient DN3 thymocytes (Fig. 5A), which suggested that pre-TCR signaling pathway was not disrupted after Smarca5 loss. However, we noted that compared with highly proliferating Rag1^−/− DN cells, the S5^fl/flhCD2iCre DN cells were almost completely absent at day 2 of treatment (Fig. 5B), further confirming the poor survival of differentiating and proliferating Smarca5-deficient DN cells. We next examined the expression of CD2 and CD5 on the DN and DP cells to assess their upregulation upon pre-TCR signaling (37, 38). Data from flow cytometry show that β-selected DN cells that survived Smarca5 loss are still capable of upregulating the expression of CD2 and CD5 molecules (Supplemental Fig. 1A). Additionally, at the DP stage, the expression of these molecules was almost identical compared with controls (Supplemental Fig. 1B). In summary, the ability of S5^fl/flhCD2iCre DN cells to upregulate CD2 and CD5 indicated that Smarca5 deficiency in thymocytes does not perturb the pre-TCR signaling.

Next, we analyzed the intracellular expression and rearrangement of the TCR β-chain, another prerequisite for the DN3 to DN4 transition. Analysis of intracellular (i)TCRβ expression together with membrane-bound CD28 has been shown as a tool to distinguish between preselection and β-selected DN3 cells that have successfully rearranged their TCRb locus (39). Using this approach, we observed that mutant DN3 thymocytes contain a significantly reduced fraction of iTCRβ⁺/CD28⁺ β-selected cells compared with controls (Fig. 5C). Conversely, the genomic DNA analysis of the Smarca5 mutant DN thymocytes by quantitative PCR showed that the recombination rate of Dβ1-Jβ1 and Dβ2-Jβ2 gene segments was not altered (Supplemental Fig. 1C). As determined by RNA-seq analysis of DP cells (see further), the expression pattern of the constant (Trbc) and variable (Trbv) gene segments was similar to controls (Supplemental Fig. 1D), indicating that Smarca5 deficiency did not abolish the TCRb locus rearrangement. To test whether the defective formation of the β-selected DN3 cells was a result of impaired TCRβ rearrangement, we crossed S5^fl/flhCD2iCre mice with TCRb/a transgenic mice (OT-II). Expression of the fully rearranged TCRb/a construct in OT-II background is able to suppress V(D)J recombination at endogenous loci and also “rescue” thymocyte development in mice lacking essential factors for the TCRβ/α rearrangement (40, 41). Analysis of S5^fl/flhCD2iCre OT-II animals revealed that expression of the transgenic TCR β/α-chains failed to rescue the Smarca5 knockout phenotype. The absolute and relative counts of DN and DP subpopulations remained similar to the S5^fl/flhCD2iCre mice (Fig. 5D–F). Particular exceptions were mature SP thymocytes, where the CD4 SP cells whose 3-fold increase to the detriment of the CD8 SP cells likely reflected the positive selection of thymocytes toward CD4 lineage that normally occurs in the OT-II strain (Fig. 5D). Thus, rather the poor survival of β-selected DN cells than defects in pre-TCR signaling or TCRb locus rearrangement could best explain the phenotype of S5^fl/flhCD2iCre mice.

As the surface expression of TCRβ in DP thymocytes is essential for the production of SP subpopulations, we focused on stages beyond DN3 to decipher how Smarca5 deficiency influenced the TCRb expression. Normally, the surface expression of TCRβ is low or none in the DN3 stage, becomes induced at the DN4 stage, and further upregulated in SP thymocytes. We examined the level of surface TCRβ expression on individual developmental stages of thymocytes and mature peripheral T cells in the spleen. We observed that surface TCRβ expression was detectable in all developmental stages from a subfraction of DN3 (low expression) to mature SP populations (high expression) in both Smarca5-deficient as well as control thymocytes (Fig. 5G, Supplemental Fig. 1E). However, the fraction of cells with upregulated TCRβ expression was reduced within each analyzed Smarca5-deficient thymic subpopulation compared with controls. Taken together, Smarca5-deficient thymocytes are able to express and upregulate the surface TCRβ expression during their development with an exception at the DN4 stage that almost lacks fraction of TCRβ-positive cells.

It has been shown that during differentiation from DN to DP stage, the ACF complex containing Smarca5 and Acf1 represses, in cooperation with Satb1, the Il2ra (CD25) gene (12, 42). Indeed, flow cytometry indicated that DP cells of S5^fl/flhCD2iCre mice inappropriately express both CD44 and CD25, the markers of earlier developmental stages. Whereas the CD25 molecule becomes partially downregulated, the CD44 remains upregulated in DP cells, implicating the dysregulation of early expression programs (Fig. 6A). Our previous studies suggested that Smarca5 participates in the regulation of gene expression programs associated with survival and differentiation of lens, cerebellum, and hematopoietic progenitor cells (10, 17, 43). To gain a global view on the gene expression programs dysregulated by Smarca5 loss in β-selected thymocytes, we purified DP cells from S5^fl/flhCD2iCre mice and used RNA-seq to compare the gene expression profiles with those from control DP cells. Of >21.500 expressed genes, a total of 3.318 transcripts were differentially expressed with false discovery rate (FDR) <0.05 and BaseMean value >10. From these, 1.503 mRNAs were (<2-fold) downregulated and 1.815 mRNAs were (>2-fold) upregulated. Gene ontology analysis of the differential expression using gene set enrichment analysis (GSEA) (44) showed the enrichment of mRNAs involved in expected categories such as apoptosis and the p53 pathway; however, most of them were either immunologic or lymphocyte associated (Fig. 6B). By dividing the differentially expressed genes into previously published mRNA clusters with similar behaviors during thymocyte differentiation (45), we observed that the group of upregulated genes in the mutant DP cells overlapped to those mRNAs that peaked in expression at the early (DN1–DN2) or pre–β-selection DN3a stage (Fig. 6C). Such genes were, for example, receptors (Il7r, Ctla4, Ptcra, Ly6a and also Il2ra/Cd25 and Cd44), signaling molecules (Dtx1, Hes1, Notch1, Lfng, Rab44), and transcription factors (Irf7, Tcf7l2, Spib) (Fig. 6D). Accordingly, the group of downregulated genes (total 97) fall into the category of genes that are gradually expressed by β-selected cells during the transition into DP, such as transcription factors (Klf7, Ets2, Ikzf3, Bcl6), surface molecules (Plxnd1, Slamf1, Cd81), signaling (Themis), and others (Tdrd5, Cacna1e). To evaluate whether the mutant DP cells more closely resembled pre– or post–β-selected cells, we used a previously published transcriptome analysis of microarray data of wt C57BL/6J DN3a and DP stage cells (45) and created two sets of the 200 most upregulated and downregulated genes in wt DN3a compared with wt DP stage (Supplemental Fig. 2A). Hence, the GSEA analysis of mRNAs dysregulated upon Smarca5 deficiency revealed a strong enrichment of upregulated (normalized enrichment score = 2.37, FDR <0.001) and downregulated (normalized enrichment score = −2.53, FDR <0.001) mRNAs of the wt DN3a stage (Fig. 6E). This finding indicates that although Smarca5-deficient cells express post–β-selection surface markers (e.g., CD4, CD8, CD5, CD2) as normal DP cells, RNA-seq data reveals that they retain a transcriptional program of pre–β-selection DN cells. Thus, Smarca5 ablation greatly disorders the developmental programming of T cell progenitors.

Additional analysis of S5^fl/flhCD2iCre mice revealed also a dramatic reduction of the spleen cellularity (Fig. 7A). Besides the reduction of splenic T cells and NKT cells, which both developed in the thymus from the CD4⁺CD8⁺ DP precursors (46), we also observed a marked depletion of B lymphocytes, whereas the numbers of myeloid cells were not significantly altered (Fig. 7B, 7C). Indeed, immunohistochemistry of mutant spleens showed a prominent follicular hypoplasia affecting both T cell as well as B cell zones (Fig. 7D), suggesting also a defect in the B cell development of S5^fl/flhCD2iCre mice. To investigate a stage at which the developmental defect occurred, the early B cell progenitor populations from BM were analyzed. Flow cytometry analysis revealed an almost complete loss of pre-B cells (B220⁺CD43⁻) in mutants, whereas the proportion of pro-B cells (B220⁺CD43⁺) was virtually unperturbed (Fig. 7E). Closer examination of the pro-B population showed a developmental arrest between early preselected pre–B-I cells (CD117⁺) and pre–B-II (CD25⁺) cells that underwent productive IgH gene loci rearrangement (47). Taken together, Smarca5 deficiency affects development of early B220⁺CD43⁺ pro-B cells in BM, implicating the requirement of Smarca5 for both early T as well as B lymphocyte development.

It has been previously noted that deletion of Smarca5 gene in hematopoietic progenitors induces the expression of p53 transcriptional targets (10). Our current RNA-seq data suggested that increased cell death and impaired cell cycle progression of thymocytes coincide with the activation of the p53 program (Fig. 8A). To test the biological significance and requirement of the Trp53 gene for the Smarca5 mutant phenotype in developing lymphocytes, we used an additional mouse strain homozygous for the Trp53 null allele (48) to create the S5^fl/flhCD2iCre Trp53^−/− mice. Interestingly, the Smarca5 deletion slightly prolonged the survival of the Trp53 knockout mice (Fig. 8B). We observed that the introduction of the Trp53 knockout allele improved thymic cellularity of the Smarca5-deficient mice (Fig. 8C). Flow cytometry analysis revealed a proportional increase (from 26.3 to 55.1%) as well as absolute cell number expansion (4-fold) of the double knockout DP population (Fig. 8D, 8E), whereas the absolute numbers of DN cells were unchanged, indicating a DN to DP transition rescue (Fig. 8E, 8F). We performed RNA-seq of samples derived from the double knockout DP cells and compared gene expression profiles with previous data. Expression analysis confirmed that the p53 targets, especially those that are associated with the induction of apoptosis in response to DNA damage such as p21/Cdkn1a, Noxa/Pmaip1, and Bax, were upregulated specifically in the Smarca5-deficient DP cells, whereas upon the introduction of the Trp53^−/− allele, their expression became normalized (Fig. 8G). Although the thymic cellularity in double knockouts was partially recovered, we still observed markedly dysregulated expression of the mRNAs connected to normal thymocyte development (Supplemental Fig. 2B). Thus, the activation of p53 targets was rather a modifier of the severity of the phenotype and not contributory to differentiation defects observed in S5^fl/flhCD2iCre mice. Indeed, flow cytometry analysis showed that Smarca5 and Trp53 double knockout DP cells contained up to a 3.5-fold excess of H3S10^phos positive (mitotic) cells as compared with single Smarca5 knockout (Fig. 8H). However, the G2/M blockade in the DPs of the Smarca5 and Trp53 double knockout mice persisted (Supplemental Fig. 2C).

To test whether the p53 loss could also recover the development of early B cell progenitors, BM cells from S5^fl/fl hCD2iCre mice with or without Trp53^−/− loci were analyzed for the expression of B220 and CD43 molecules. However, unlike the partial rescue observed in the thymocyte compartment, the data from flow cytometry show that the p53 loss failed to rescue the survival and/or maturation at pro-B (B220⁺CD43⁺) to pre-B stage (B220⁺CD43⁻) transition (Fig. 8I). To conclude this part, the introduction of the Trp53 knockout allele into the Smarca5-deficient strain significantly improved the proliferation and/or survival of thymocytes but not of early B cells. Importantly, the rescue experiment was unable to restore the dysregulated differentiation pathways in both B and T lineages.

Although the SWI/SNF and CHD chromatin-remodeling factors have been implicated in the regulation of lymphocyte progenitor-specific transcription and differentiation (6, 49), the role of ISWI proteins in the development of early T and B cells has not been addressed. This study brings yet unknown evidence that the ISWI ATPase Smarca5 regulates early lymphocyte development by promoting stage-specific gene expression and, secondarily, cell survival and proliferation. Interestingly, the role of Smarca5-containing remodeling complexes was previously implicated in the DNA double-strand break (DSB) repair in human immortalized cell lines (50, 51). These reports showed that SMARCA5 is rapidly recruited to DSBs, and the knockdown of SMARCA5 sensitizes cells to DNA damage. SMARCA5 protein was shown recruited to the sites of DSB by histone deacetylase Sirtuin 6 (SIRT6) (51). Additional pathways and interaction partners of Smarca5 participating at the sites of DNA damage were also established (52, 53). Thus, Smarca5 seemed to be a suitable candidate for testing its in vivo role in lymphocytes, in which the developmentally programmed DSBs occur. Indeed, our data showed that S5^fl/flhCD2iCre mice initially exhibited a marked reduction of those early progenitors that productively rearranged Ag receptor loci, the TCRβ expressing DN3 thymocytes (Fig. 5C) and B220⁺CD43⁺CD25⁺ early B cells (Fig. 7E). Coincidently, the upregulation of p53 target genes (Fig. 8G) and the partial recovery of thymus cellularity in the S5^fl/flhCD2iCre Trp53^−/− mice (Fig. 8C) could also be interpreted as Smarca5 being involved in the DSB repair during the Ag receptor gene rearrangement. However, further experiments challenged this view. The developmental defect, at least in thymocytes, could not be attributed mainly to the DSBs repair aberration, as the OT-II transgene was unable to rescue the DN3 to DN4 transition defect of Smarca5-deficient cells (Fig. 5D). Notably, the data from RNA-seq showed that the expression pattern of the constant (Trbc) and variable (Trbv) gene segments was similar compared with the control DP cells (Supplemental Fig. 1D), indicating that the relative use of the different gene segments during TCRβ rearrangement was not affected in the mutants. In addition, the S5^fl/flhCD2iCre Trp53^−/− mice displayed a comparable life span as S5^fl/fl Trp53^−/− mice (Fig. 8B), which is contrasting to the mouse knockout models of genes employed directly in the NHEJ that upon codeleting with the Trp53, accelerated tumorigenesis with shortened animal survival (54). Thus, the Smarca5 deletion-mediated maturation defect is not primarily mediated via the disrupted repair of developmentally programmed DSBs.

Generally, except in the SP CD8 and CD4 T cells that expressed a lower level of surface TCRβ (Fig. 5G, Supplemental Fig. 1E), the iTCRβ expression and proximal pre-TCR signaling seem to be preserved in the Smarca5-deficient mice as S5^fl/flhCD2iCre DN3 cells gave rise (albeit with a very low rate) to some DP-like cells and could be normally stimulated by anti-CD3 Ab. The DN3 stage and all the following developmental stages were considerably altered in the S5^fl/flhCD2iCre mice. Once Smarca5 was inactivated, β-selected DN3, DN4, DP, and also pre-B cells lost their ability to accumulate and became depleted. Smarca5 loss leads to a marked increase in the number of cells undergoing apoptosis (Fig. 1G). Our data indicate that this was not caused by the induction of a generalized apoptotic response as resting DN3e cells displayed very low level (up to 5%) of cell deaths even after 6 d of ex vivo cultivation (data not shown). This result, together with the observation that Smarca5-deficient DN3 stage lacks postreplicative cells (Fig. 4C), rather suggested that the disruption of Smarca5 function triggers apoptosis of highly proliferating cells and especially those that have already entered the S phase. Others reported that depletion of Smarca5 in murine lens (using Le-Cre system) results in a reduction of BrdU and Ki67 positive presumptive lens epithelial cells, leading to the lens developmental defect (43). Also, the early deletion of Smarca5 in cerebellar progenitors (using the Nestin-Cre system) resulted in a lower number of BrdU-positive Purkinje cells and of granule neuron progenitors at E17.5, possibly because of massive cell death (17). Moreover, the defective S phase progression in the DN3 stage could also explain the formation of G2/M-arrested S5^fl/fl hCD2iCre DP cells (Fig. 4D). In the erythroid cell compartment, the Smarca5 loss caused the emergence of tetraploid cells permanently exiting the cell cycle in populations of highly proliferating proerythroblasts-to-basophilic erythroblasts (10). This, along with a substantial number of apoptotic events, was interpreted mainly as a consequence of the activation of the DNA damage-associated p53 program. Indeed, stressed replicating cells activate their replication checkpoint to delay S phase progression and G2/M transition (55). The p53 and its downstream molecules are then required to maintain a G2 or tetraploid G1 arrest, which afterward promotes cell senescence (56–58). However, although S5^fl/flhCD2iCre DP cells also upregulated some proapoptotic (Noxa, Bax, Puma) and cell cycle regulating (Cdkn1a/p21) p53 target genes, the rescue of phenotype was incomplete, and the tetraploid events were still present in the Smarca5 and Trp53 double knockouts (Supplemental Fig. 2C), indicating that the cell cycle arrest and induction of apoptosis were predominantly p53 independent.

We hypothesize that the dysregulated expression of the stage-specific mRNAs (including the surface markers) in Smarca5-depleted β-selected cells stay behind the thymocyte defects observed in the S5^fl/flhCD2iCre mice. Indeed, the ablation of chromatin remodelers Brg1 or Chd4/Mi-2β leads to differentiation defects, cell cycle arrest, or apoptosis in β-selected thymocytes partially because of dysregulated gene expression (reviewed in 59). Unlike the Brg1 or Chd4 chromatin remodelers that regulate differentiation from DN4 to DP stages (60, 61), the Smarca5 knockout phenotype appears early and is relatively unique as the pre–β-selected DN3 cells lacking Smarca5 are unable to downregulate marker molecules (CD25, CD44). Among the mechanisms behind inappropriate CD25 molecule expression might be the previously reported participation of Smarca5 in the Satb1-directed repression of Il2ra/Cd25 loci (12, 42). Genome-wide characterizations of binding into chromatin by chromatin immunoprecipitation sequencing revealed that Smarca5 is enriched mostly at the gene promoters and the regulatory regions (62, 63). Transcriptome analysis by RNA-seq confirmed a considerable number of ectopically expressed transcripts associated with β-selection and also genes that were not activated and remained downregulated during the transition into the DP stage (Fig. 6D), suggesting that the absence of Smarca5 disables a crucial component of the β-selection transcription machinery. The function of Smarca5 as a transcriptional activator and repressor was shown in studies using murine EL4 T lymphoma cell line upon small interfering RNA–mediated depletion. Although Smarca5 participates in repression of IL genes, including Il2, Il5, Il13, Il17a, and in activation of Il3 after stimulation with PMA and ionomycin (13), we have not observed this during the transition into the DP stage as those ILs are expressed mainly by mature T cells. Perturbation of Smarca5 functions may have also affected other profiles during the transition from pre– to post–β-selected DN3 thymocytes by dysregulating genes related to proliferation, metabolism, and β-selection (45). To conclude, Smarca5 represents an important transcriptional regulator that participates indispensably during early T cell development. To further address the role of Smarca5 in regulating T cell promoters, we are currently preparing a transgenic mouse line with tagged Smarca5 protein.

We thank Christian Lanctôt and Vladimír Divoký for helpful discussion, Ivan Kanchev for histopathology reports, Emanuel Nečas and Martin Molík for CD45.1 mice and assistance with the BM transplantations, Dana Mikulenková and Ilona Jirasková for cytospin preparations, Luděk Šefc for help with flow cytometry analysis, Jan Kubovciak for help with RNA-seq analysis, and Martin Drobiš for style and writing.

The authors have no financial conflicts of interest.