The Role of Brg1, a Catalytic Subunit of Mammalian Chromatin-remodeling Complexes, in T Cell Development

Coordinated programs of gene expression underlie the development, functional differentiation, and homeostatic regulation of the mammalian immune system. Perhaps one of the most intensely studied systems of transcriptional regulation is that of T lymphocyte ontogeny (1). Activation or repression of lineage-specific and developmental stage–restricted cell surface molecules has been correlated with cell fate choices in the αβ T cell differentiation pathway. Up-regulation of both CD4 and CD8 and subsequent down-regulation of one of these two coreceptors in developing thymocytes leads to mature T helper (CD4⁺ CD8⁻) and cytotoxic (CD4⁻ CD8⁺) T cells in the periphery.

In addition to lineage-restricted transcription factors, it is becoming increasingly clear that chromatin-modifying factors are necessary to achieve proper temporal and spatial patterns of gene expression in the immune system. Modification of local chromatin structure is achieved, in part, through the activity of multisubunit protein complexes that use the energy of ATP hydrolysis to disrupt the conformation and position of nucleosomes (2). The yeast SWI–SNF complex was the first such complex to be identified. It has been characterized in the most detail and is conserved throughout evolution. Mammalian SWI–SNF-related complexes consist of 9–11 subunits that utilize either brahma (Brm; also called SNF-α or Smarca2) or Brm-related gene 1 (Brg1; also called SNF2-β or Smarca4) as the catalytic subunit (2, 3).

Targeted mutations of the Brg1 catalytic subunit or two other complex members (INI1–Snf5 and Srg3–Baf155) confer early embryonic lethality in the mouse, indicating that mammalian SWI–SNF-related complexes are essential (4–7). Although few direct target genes have been confirmed in mammalian cells, biochemical interactions with gene regulatory elements and transcription factors important for hematopoietic lineage specification have implicated SWI–SNF-related function in these processes (8). For example, interactions of chromatin-modifying complexes, including SWI–SNF-related, with Ikaros, a zinc finger–containing, sequence-specific DNA-binding transcription factor important in lymphocyte development, suggest a link between chromatin regulation and hematopoietic cell fate decisions (9). Consistent with this notion, chromatin immunoprecipitation assays have correlated down-regulation of IL-2Rα (CD25) at the double negative (DN) stages of T cell development with the dissociation of Brg1 and histone acetylase p300 from promoter elements (10). Moreover, dominant negative expression of BAF57, an HMG-containing, DNA-binding subunit of the SWI–SNF-related complexes, in the T cell lineage revealed a role of chromatin remodeling in CD4 silencing and CD8 activation (11). These data provide evidence that SWI–SNF-related complexes are important regulators of stage-restricted gene expression in the T cell lineage. In addition to this developmental gene regulation, chromatin-modifying activity has been associated with proliferation of peripheral T lymphocytes. For example, antigenic activation of T cells induces rapid association of SWI–SNF-related complexes with chromatin in a Brg1-dependent manner (12). Furthermore, inactivation of Lsh, a member of the SNF2 family of proteins including Brg1, in lymphoid precursors results in a severe proliferation defect in response to Con A activation of peripheral T lymphocytes (13). Thus, chromatin-remodeling activity has been associated with a number of processes in T cell development and function, although the early lethality or minimal phenotypes associated with targeted mutations in SWI–SNF-related complex members do not permit loss of function studies in this lineage.

To investigate the in vivo role of Brg1 and the SWI–SNF-related complex function in T cell biology, we specifically inactivated Brg1 in the T cell lineage using the Cre-loxP conditional mutagenesis system. Here, we report that loss of Brg1 results in a block in T cell development at the immature stages and investigate the effect of Brg1 mutation on mature T cells in the periphery. These data provide in vivo evidence implicating ATP-dependent chromatin-remodeling factors as important regulators of T lymphocyte development and immune response.

The Brg1 germline-null allele (4), the lck-Cre (14), and the Zp3-Cre (15) mice have been described. The Brg1 floxed allele mice were provided by P. Chambon (Centre National de la Recherche Scientifique, Strasbourg, France) and harbor previously described loxP sites integrated into the Brg1 genomic locus (16). All mice were maintained on a mixed genetic background at the University of North Carolina, Chapel Hill Animal Facility.

Genotyping for the Brg1-targeted mutation has been described (4). For the detection of the lck-Cre and Zp3-Cre transgenes, similar conditions were used except for 1.5 mM MgCl₂, 62°C annealing temperature, and the following primers specific for Cre: forward, 5′caatttactgaccgtacac3′; reverse, 5′catcgccatcttccagcag3′. For the amplification of the unrecombined and recombined floxed allele depicted in Fig. 1 B the sequences are: primer 1, 5′-gatcagctcatgccctaagg-3′; primer 2, 5′-gtcatacttatgtcatagcc-3′; and primer 3, 5′-gccttgtctcaaactgataag-3′ (1 mM MgCl₂ and 51°C annealing). RT-PCR conditions for Brm, TCRβ, and CD3ε have been reported (4, 17, 18).

Quantification of PCR products was performed on a GelDoc 2000 (Bio-Rad Laboratories) using Multianalyst software (Bio-Rad Laboratories; see Supplemental Materials and Methods, available at http://www.jem.org/cgi/content/full/jem.20030714/DC1).

mAbs used for immunofluorescence are listed in Supplemental Materials and Methods, available at http://www.jem.org/cgi/content/full/jem.20030714/DC1. Annexin-V/propidium iodide (PI) staining was performed according to the manufacturer's protocol (Roche). LN cells were stained and analyzed on a FACScan™ flow cytometer (BD Biosciences). Cell sorting was performed on a MoFlow (DakoCytomation) instrument. The FACS^® data were analyzed with the Summit V3.1 data analysis software (DakoCytomation).

Thymi from 4-wk-old animals and livers from 3–6-mo-old animals were dissected, fixed in 4% paraformaldehyde, and processed for the production of paraffin sections. 5–8-μm sections were stained with hematoxylin and eosin using standard conditions.

BrdU (Sigma-Aldrich) was administrated by intraperitoneal administration of 1 mg/mouse injected two times 4 h apart as previously described (19). Thymocytes were harvested in 1 h after the second injection and stained with anti-CD4 and anti-CD8 mAbs. The BrdU staining of those cells was performed by a conventionally modified method as previously described (19).

The complete medium (RPMI 1640 supplemented with 10% FCS, 2 mM l-glutamine, 50 U/ml pen/strep) was used for cell culture. CD3 and CD3/CD28 stimulation was performed as previously described (20). For in vitro mitogen stimulation, 5 × 10⁵ LN cells were cultured in 200 μl medium, with or without 5 or 10 μg/ml Con A (Sigma-Aldrich), or with 0.5 ng/ml PMA and 0.5 μM ionomycin in 96-well flat-bottom plates. Incubation was performed in complete RPMI for 48 h in 37°C with pulse [H³]thymidine at 1 uCi/well for the last 12 h.

See Supplemental Materials and Methods, available at http://www.jem.org/cgi/content/full/jem.20030714/DC1, for details regarding genotyping, RT-PCR, and PCR quantification.

To investigate the potential function of Brg1 during thymocyte development, we used a previously described Brg1 “floxed” allele in which loxP sites flank exons that encode a portion of the conserved ATPase domain (Fig. 1 A; references 4 and 16). Cre-mediated recombination of the loxP sites creates a deletion corresponding to the putative ATP hydrolysis site and also introduces a frameshift mutation. This “Δ floxed” configuration encodes a COOH terminally truncated protein product and is predicted to be a null allele, which was confirmed by crossing the floxed allele to a Zp3-Cre transgenic mouse that recombines loxP sites in the female germline (15). Constitutive Δ floxed/+ intercrosses failed to yield Δ floxed/Δ floxed homozygotes and no homozygotes were recovered upon dissection of litters at embryonic days 7.5 and 8.5 (0/15; 0.05 > P > 0.025). The fact that no resorption moles or empty decidua were observed indicates that lethality occurs before decidualization is initiated as reported for the targeted null allele (4). Thus, Δ floxed is a null allele.

Mice carrying the Brg1 floxed allele (+/flox) were mated with mice heterozygous for a Brg1-null mutation generated by gene targeting (+/null) and mice expressing Cre recombinase under the control of the lck-proximal promoter (+/+; lck-Cre) to obtain compound heterozygous mice harboring the lck-Cre transgene (null/floxed; lck-Cre, henceforth referred to as tBrg1^−/− for T cell lineage–specific Brg1 mutation; references 4 and 14). Mice harboring one floxed allele, one wild-type allele, and the lck-Cre transgene (+/floxed; lck-Cre) showed no differences from wild-type mice in total thymocyte number or cell type composition (unpublished data), consistent with previously published reports of no gross abnormalities in T cell development in mice heterozygous for the Brg1-null mutation (11). Therefore, both wild-type and +/floxed; lck-Cre mice were used as controls.

The lck-Cre transgene has previously been shown to express Cre specifically in the T cell lineage, with Cre-mediated recombination detected as early as the DN1 (CD44⁺ CD25⁻) stage and complete by the double positive (DP; CD4⁺ CD8⁺) stage (14). PCR analysis of whole thymic DNA preparations from Brg1 +/floxed; lck-Cre mice showed efficient recombination of loxP sites, as judged by the loss of the floxed amplicon (Fig. 1 B, top) using primers that amplify both the wild-type and floxed allele (Fig. 1 A, primers 2 and 3). Accordingly, the Δ floxed allele showed significant amplification in thymus and not in tail DNA preparations using primers specific for the recombined floxed allele (Fig. 1 B, bottom, and A, primers 1 and 3). To investigate timing of recombination of the Brg1 floxed allele, we analyzed the triple negative (TN; CD4⁻ CD8⁻ CD3⁻) DN1 through DN4 cell populations from Brg1 +/flox; lck-Cre mice. Cre-mediated deletion began at DN1 and proceeded through DN4 (Fig. 1 C, bottom) stage, consistent with previous reports using this lck-Cre transgene (14). As shown in Fig. 1 C, top, the floxed allele was only partially deleted at the DN1, DN2, and DN3 stages, but was virtually complete by the DN4 stage. Loss of the floxed allele amplicon at DN4 indicates high efficiency of recombination by this stage.

Gross dissection of tBrg1^−/− mutant animals revealed severe thymic defects. The overall size of the mutant thymus was markedly reduced compared with the large, bi-lobed thymus observed in control littermates at 4 wk of age (compare A and B in Fig. 2). Thymocyte cell counts demonstrated a dramatic 20-fold reduction in total thymocytes present in the mutant thymus, indicating a T cell developmental defect (1.3 ± 0.7 × 10⁷ in mutant vs. 27.3 ± 8 × 10⁷ in control 4-wk-old thymus; Fig. 2 D). In support of this finding, histological analysis revealed an almost complete loss of corticomedullary architecture in tBrg^−/− mutant animals (compare E and F in Fig. 2) and enlarged cortical thymocytes containing irregular oval nuclei with more open chromatin, which is indicative of more immature thymocytes (Fig. 2 H). Additionally, higher magnification histological analysis revealed an increase in the number of mitotic figures as well as apoptotic bodies in the mutant thymus (compare G and H in Fig. 2).

tBrg1^−/− mice displayed a significant reduction in body weights at 3–4 wk of age compared with sex-matched control littermates (Fig. 2 C). Although the severity of runting was variable between litters, differences in body weights between mutants and control animals remained relatively constant throughout the aging process.

To provide insight into the nature of the T cell developmental defect, FACS^® analysis of a number of stage-specific cell surface markers was performed on thymocytes from tBrg1^−/− animals versus controls. Significantly, CD4/CD8 staining revealed an almost complete loss of the DP (CD4⁺ CD8⁺) and CD8 single positive (SP; CD4⁻ CD8⁺) cell populations, with a corresponding relative increase in DN (CD4⁻ CD8⁻) T cells in the mutant thymus (Fig. 3, A and B). Interestingly, mutant thymocytes showed an increased percentage of CD4 SP cells, although the absolute number of CD4⁺ CD8⁻ cells was reduced (Fig. 3, B and C). In absolute numbers per thymus, DP thymocytes were reduced 250-fold, CD4 SP fourfold, and CD8 100-fold, whereas the total DN cell population was largely unaffected (Fig. 3 C). Recently, SWI–SNF-related complexes have been implicated in the coordinated transcriptional regulation of CD4 (repression) and CD8 (activation) in developing thymocytes (11). Therefore, we hypothesized that the CD4⁺ SP cells present in the tBrg1^−/− mutant thymus constitute an immature T cell population that aberrantly express CD4 (derepression) at the DN stages in the absence of Brg1 function. Examination of TCR surface levels in mutant and control CD4⁺ SP cells confirmed this hypothesis, with <5% of CD4⁺ mutant thymocytes expressing high levels of the TCR compared with 90% TCR⁺ observed in control mature CD4⁺ SP cells (Fig. 3 D). Moreover, direct examination of CD4 expression at the DN stages demonstrated significant increases of CD4 levels in mutant thymocytes at DN2, DN3, and DN4 (Fig. 4, B and C). Finally, PCR analysis on mutant CD4⁻ and CD4⁺ DN3 thymocytes revealed a much higher percentage of mutant cells in the CD4⁺ derepressed class than the CD4⁻ cell population (see Fig. 6, bottom). Taken together, these data indicate that loss of Brg1 function in developing thymocytes results in CD4 derepression and a block at the DN stages of T cell development.

To determine more precisely at which stage the developmental block occurred in the tBrg^−/− mutant mice, DN thymocytes were subdivided into four distinct developmental stages (DN1, DN2, DN3, and DN4) based on cell surface expression of CD25 and CD44 expression on TN thymocytes (Figs. 4 A and 1 C). On a percentage basis, tBrg1^−/− mice showed a relative increase in TN cells, as predicted by the loss of the DP population (see above), and this increased percentage translated into an increase in absolute terms as the total number of TN cells was significantly higher than in controls (Fig. 4 A). Examination of the DN1→DN4 populations from the mutant thymus revealed the total TN increase was attributed to increased DN2 and DN3 populations (Fig. 4 A), suggesting a developmental block and/or increased proliferation at these stages that should be reflected by decreased or increased DN4 population, respectively (Fig. 4 A). However, DN4 cell counts were similar in mutant compared with controls. This is most likely explained by CD4 derepression occurring in mutant thymocytes transitioning from DN2/DN3→DN4 stages (Fig. 4 B), which would exclude this population from DN4 cell counts. Given the almost complete deletion of the floxed allele at the DN4 stage, it is not clear why this population was not uniformly affected (CD4 derepression). In DN4 cells that have undergone Cre-mediated recombination but do not express CD4, it is likely that residual Brg1 protein–loaded prior loss of the floxed allele is sufficient to repress CD4 expression at this stage. Regardless, the remaining DN4 cells were blocked in differentiation based on the absence of DP cells. In support of a DN4 to DP block, immature CD8 SP cells, the immediate precursors of the DP stage, are severely reduced in the mutant thymus (Fig. 4 A). Complicating this interpretation, however, is data implicating the SWI–SNF complex in direct regulation of CD8 expression (11). Therefore, it is unclear whether the reductions in cell numbers at the immature SP stage are due to block before this stage or fail to activate CD8 expression. Taken together, these data are consistent with developmental abnormalities beginning at the DN2/DN3 stage and subsequently with a block at the DN4→DP transition in tBrg1^−/− thymocytes.

To further investigate the increases in DN2/DN3 populations and determine the fate of the CD4⁺ CD3⁻ thymocytes, the proliferative status of mutant thymocytes was measured through in vivo BrdU incorporation. In tBrg1^−/− mice, BrdU incorporation was significantly higher in both the DN and CD4 derepressed immature cell populations compared with controls (Fig. 5 A). Additionally, a slightly higher percentage of DP cells stained positive for BrdU⁺, however the low numbers of DP cells present in the mutant thymus make this data unreliable. To characterize the proliferative status at the immature stages, DN cell populations were divided into two subclasses based on CD44 expression: CD44⁺ (DN1+2) and CD44⁻ (DN3+4). Interestingly, both subclasses showed an increased percentage of BrdU⁺ cells in the mutant thymus (Fig. 5 B), indicating increased cell proliferation throughout the DN stages. This result is consistent with the histological observation of increased mitotic figures in the mutant thymus and increased total cell numbers at DN2 and DN3 in mutant mice (Figs. 2 H and 4 A).

Although the increased proliferation observed in the tBrg1^−/− mice is consistent with an increase in total TN cells (Fig. 4 A), the severe reduction in total thymocytes present in mutant thymus cannot be fully explained by the inability of cells to proliferate. Therefore, we used the annexin V staining method to assay the percentage of cells undergoing apoptosis (21). Analysis of annexin V staining revealed a slight increase in the percentage of apoptotic DN thymocytes in mutant compared with controls (Fig. 5 C). Although these differences are unlikely to account for the massive cell loss in the mutant thymus, examination of annexin V staining in the mutant CD4 derepressed class (CD4⁺ CD8⁻ CD3⁻) did demonstrate high percentages of cells undergoing apoptosis (Fig. 5 C; P < 0.01). These data are consistent with increased numbers of apoptotic bodies observed through histological analysis (Fig. 2 H). Taken together, these data suggest that loss of Brg1 triggers an increase in cell proliferation, derepression of CD4, and subsequently, programmed cell death at the DN stages of T cell development.

Based on the timing and severity of defects in Brg1-deficient thymocyte development, we investigated potential molecular abnormalities associated with the observed thymic phenotype. Specifically, we examined whether the loss of Brg1 affected the expression of Brm, another SWI–SNF-related catalytic subunit often coordinately down-regulated in human tumor cell lines (22). As shown in Fig. 6, CD4⁺ derepressed DN3 thymocytes exhibit efficient recombination of the floxed allele, indicating high proportion of mutant cells, but similar levels of Brm expression through semiquantitative RT-PCR analysis, indicating that Brm expression is independent of Brg1 and does not contribute to the severity of the thymic phenotype. Additionally, we examined the expression of two components of the pre-TCR complex (TCRβ and CD3ε), which, when mutated, block thymocyte development at the DN3→DN4 transition (23). Both TCRβ and CD3ε were expressed at similar levels in mutant and control DN3 thymocytes (Fig. 6), suggesting that Brg1 deletion impairs thymocyte developmental processes downstream of pre-TCR assembly.

To assess the degree of developmental block in the thymus and to determine whether Brg1 function is required for the survival of mature T cells, we used FACS^® on LN and spleen cells to examine peripheral T cell populations in tBrg1^−/− mice. Surprisingly, CD4/CD8 staining of peripheral lymphocytes indicated mature CD4⁺ and CD8⁺ cells present in the periphery of the mutant animals, although at significantly reduced levels, suggesting an incomplete developmental block in the thymus and/or failure of Cre-mediated recombination in a low percentage of “escaping” cells. To address this issue, we performed PCR amplification for the presence of the deleted floxed allele in whole LN preparations from mutant animals. Results from this analysis demonstrated that Brg1 mutant cells were present in the LN, suggesting that the peripheral T cells were not exclusively cells that had escaped Cre-mediated recombination (Fig. 7 G).

Although total LN cell counts were similar in mutant compared with control (Fig. 7 A), TCR⁺ cells were reduced fourfold in mutant mice with both the CD4⁺ and CD8⁺ populations significantly underrepresented (sevenfold and 20-fold, respectively; Fig. 7 B). Increased numbers of B cells and myeloid cells compensate for the reduced numbers of T cells in the mutant LN and account for the equivalent total LN cellularity observed (Fig. 7, C and D). Interestingly, analysis of CD4 and CD8 expression on CD3⁺ cells revealed a significant increase in DN TCR⁺ cells (CD4⁻ CD8⁻ CD3⁺; Fig. 7 E, middle, and F). This result could potentially be explained by increased percentages of γδ-TCR (CD4⁻ CD8⁻) T cells due to loss of αβ-TCR T cells in the periphery. However, calculation of total DN CD3⁺ T cells indicated an increase in absolute number of this cell type in tBrg1^−/− compared with controls (1.5 ± 0.2 × 10⁵ vs. 0.6 ± 0.1 × 10⁵; P < 0.01). FACS^® analysis for γδ-TCR cells in LN populations revealed that 80 ± 9.4% of mutant DN CD3⁺ cells are γδ-TCR⁺ compared with 28.7 ± 3% in controls. Similarly, absolute numbers of γδ-TCR T cells were increased in mutants (1.2 ± 0.12 × 10⁵ in mutant vs. 0.18 ± 0.05 × 10⁵ in controls; P < 0.01). This result was surprising given that the size of the γδ-TCR T cell population has been shown to be regulated independently of the αβ-TCR T cell population (24). Additionally, significantly higher numbers of CD4⁺ T cells stained positive for γδ-TCR in mutant compared with controls (20.1% or 0.15 ± 0.03 × 10⁵ vs. 0.05% or 0.07 ± 0.010⁵ in controls; P < 0.05) and likely represent γδ-TCR T cells that have undergone CD4 derepression (see above). The lck proximal promoter has been shown to be active at the DN1 stage before αβ/γδ lineage commitment (Fig. 1 C; reference 25) and raised the possibility that Brg1 inactivation in these early stages of T cell development influenced αβ/γδ cell fate decisions in the thymus. Arguing against this interpretation, absolute numbers of γδ-TCR⁺ cells were not significantly increased in the mutant thymus (1.45 ± 0.6 × 10⁶ in mutant vs. 1.04 ± 0.3 × 10⁶ in control) and suggested the increase in γδ-TCR⁺ cells in mutants is due to peripheral expansion of this population. If this expansion in γδ-TCR cells were a direct result of loss of Brg1 function, a prediction would be high percentages of Brg1 mutant cells in this cell type. Therefore, we performed PCR analysis on CD4⁺ CD8⁻ CD3⁺ and DN CD3⁺ flow-sorted populations to test whether mutant cells contributed exclusively to the DN CD3⁺ population. This prediction was not borne out experimentally, however, because both the CD4⁺ SP and the DN populations amplified with primers specific for the recombined floxed allele (Fig. 7 H), indicating mutant cell contribution to both cell types. To examine relative contribution of mutant cells to these cell populations, we amplified the unrecombined floxed allele and null alleles (the targeted null mutation and the wild-type allele amplify the same 230-bp fragment, which is different from the 387-bp fragment for the Δ floxed allele). Comparison of unrecombined floxed versus wild-type band intensity allowed us to estimate the percentage of undeleted floxed allele lost due to Cre-mediated conversion to Δ flox. We determined ∼25% of the CD4⁺ CD8⁻ CD3⁺ cells and 40% of the DN CD3⁺ population were mutant for Brg1. Given that 20% of CD4⁺ CD8⁻ CD3⁺ and 80% of DN CD3⁺ cells are γδ-TCR⁺, it raises the possibility that most of the Brg1 mutant cells are γδ-TCR cells. In this scenario, loss of Brg1 might be less detrimental to γδ-TCR cells and therefore allows for survival of this cell type. However, the inability to separate mutant (undergone Cre-mediated recombination) and heterozygous (escaped Cre-mediated recombination) T cells precludes direct examination of TCR or CD4/CD8 expression on mutant T cells. Although, significantly, these results suggest that, at least in the mature T cell lineage, Brg1 is not a general cell survival factor. In addition, loss of Brg1 function in this lineage does not lead to increased incidence of cancer due to the absence of T cell neoplasia (leukemia and/or lymphomas) in a cohort of mutants aged over 18 mo.

Taking advantage of this “mosaic” (with respect to Brg1 functional status) population of T cells, we sought to investigate potential T cell functional abnormalities, either cell autonomous or noncell autonomous, attributed to Brg1 mutation. FACS^® analysis of CD4⁺ T lymphocytes from tBrg1^−/− mutant mice revealed reduced surface levels of the TCR (Fig. 7 E), which has been implicated as a hallmark of TCR-mediated activation of T cells (26). Moreover, expression levels of CD44 and CD69, markers that are up-regulated upon naive T cell activation, were found to be increased on these T cells (Fig. 8 A). In addition to marking activated T cells, CD44 up-regulation has been shown to persist into the memory pool (27). Therefore, the dramatic up-regulation observed in tBrg1^−/− peripheral T cells (>70% of cells CD44⁺ in mutant vs. <10% CD4⁺ CD44⁺ in controls) indicates that these cells are effector cells or that they have undergone prior activation and are memory-like T cells. To distinguish between these possibilities, we used IL-2Rα (CD25), another marker shown to be expressed on activated T cells, in concert with CD44 and CD69, an early marker of activation. Triple staining experiments indicated a moderate increase in CD4⁺ CD25⁺ CD44⁺ (78.7 ± 2.1% in mutant vs. 32.8 ± 2.5% in controls) and CD4⁺ CD25⁺ CD69⁺ (45.0 ± 9.0% in mutant vs. 31.5 ± 6.0% in controls) cells in mutant mice (Fig. 8 A). Collectively, these data demonstrate that mature T cells in the periphery of tBrg1^−/− mice display features indicative of conversion of naive T cells to activated and/or memory-like state. Similar memory-like T cell phenotypes are characteristic of naive T cells that proliferate (called homeostatic proliferation) in response to conditions of lymphopenia after adoptive transfer into sublethally irradiated or genetically modified T cell–deficient hosts (28). The severe block in thymocyte development and the low numbers of DP cells in the tBrg1^−/− mice suggests that the lymphopenic environment created by Brg1 mutation in the mouse induces homeostatic proliferation of peripheral T cells.

To examine the functional consequences of loss of Brg1 function in peripheral mature T cells, we determined in vitro proliferative responses to a number of stimuli including TCR cross-linking using anti-CD3 or anti-CD3 plus anti-CD28, Con A, and PMA/Ion. Activation of total LN cells from tBrg1^−/− showed a significant reduction in T cell proliferative response for CD3 alone or CD3/CD28, two different concentrations of Con A, and PMA/Ion stimulation when compared with controls (Fig. 8, B–D). Significantly, these results indicate that tBrg1^−/− mutant LNs cannot mount a significant T cell response and therefore can be classified as immunodeficient mice. Taken together, these results indicate that in tBrg1^−/− mice, a small number of mutant and unrecombined heterozygous T cells escape the developmental block in the thymus and undergo limited homeostatic proliferation in the periphery that converts them to activated/memory-like T cells. These populations of T cells do not, however, completely fill the periphery, which results in immunodeficiency.

The biological consequence, in terms of overall animal health, attributed to Brg1 mutation in the T cell lineage was manifest at weaning (runting; Fig. 2 C) and throughout subsequent aging over a 12-mo period. Aging of mutant animals revealed low penetrance lethality. Roughly 15% of mutant animals (7 out of 49) died prematurely over a 12-mo period, whereas no lethality was observed in littermate control (+/floxed; lck-Cre and wild-type) animals (0 out of 72) over the same period. Furthermore, >20% (11 out of 49 in mutant vs. 0 out of 72 in controls) of mutant animals exhibited rectal prolapse over this aging period, although there was no correlation between the prolapsed rectums and premature lethality phenotypes. Based on the immunodeficiency observed in tBrg1^−/− mice, these whole animal phenotypes were unexpected, as other immunologically compromised mice, such as RAG-deficient and nude mice, thrive in specific pathogen-free facilities similar to the one housing tBrg1^−/− mice. Recently, it has been reported that infection of SCID/Ncr and nude mice with Helicobacter (Helicobacter hepaticus and Helicobacter bilis) results in large bowel disease clinically detected by rectal prolapse (29, 30) and, in the case of H. hepaticus, mild periportal lymphocyte infiltration of the liver (31). Therefore, we sought to investigate whether endogenous Helicobacter infection was the causative agent of illness in tBrg1^−/− mice. Indeed, fecal samples from our colony tested positive for both H. hepaticus and H. bilis using PCR analysis specific for these pathogenic bacteria (unpublished data). Moreover, hematoxylin and eosin–stained sections of mutant liver revealed mild to moderate periportal lymphocyte (mixed lineage) infiltration in tBrg1^−/− mice and not in controls (compare G and H in Fig. 8). These results suggest that endogenous Helicobacter infections contribute to the health problems in immunodeficiency associated with Brg1 mutation in the T cell lineage.

Considerable interest in ATP-dependent chromatin-remodeling complexes has focused on delineating biochemical interactions and molecular mechanisms functioning at promoters to influence transcriptional regulation. These studies do not, however, address the function of these complexes in vivo and significantly less is known about the function of these complexes in the context of global gene regulation. The peri-implantation lethality observed in Brg1 homozygous mutant mice demonstrated the importance of these complexes in early development but precluded further functional studies in cell lineages later in embryogenesis or adults. Here, we provide further insight into the function of Brg1 in an adult cell lineage by showing that the Brg1 complex is required for T cell development at the DN→DP transition but not for cell survival of mature T lymphocytes.

In tBrg1^−/− mice, abnormalities in thymocyte differentiation are first manifest by an increased number of cells at the DN2/DN3 stage and subsequently by a block in development at the DN4→DP transition. This transition from DN3→DN4→DP is characterized by β selection checkpoints, the importance of which is underscored by mutations in genes involved in these processes. For example, mutations in the genes involved in TCRβ rearrangement (Rag1 and Rag2), assembly of the pre-TCR (TCRβ, pTα, or CD3ε), or mediation of the signal transduction cascade from the pre-TCR (p56^lck/p59^fyn, Syk/ZAP70, or SLP-76) results in a block in development at the DN3 stage (23). Therefore, based on the phenotypic similarities between these mutants and tBrg1^−/− mice, it is possible that the Brg1 complex(s) plays a role in the β selection process. In support of this notion, the human SWI–SNF complex was shown to stimulate the cleavage and processing of DNA by RAG1 and RAG2 in V(D)J recombination (32). However, this is an unlikely explanation for the developmental block observed in tBrg1^−/− thymocytes because TCR-β recombination and expression are unaffected by the loss of Brg1 function (Fig. 6; reference 33). Interestingly, a similar block beyond the DN3 stage has been found in mice mutant for N-CoR, a chromatin-modifying complex that has been shown to contain Brg1, suggesting that the coordinated effort of multiple chromatin-modifying complexes are required for transition to the DP stage (34). The loss of the Brg1–N-CoR complex function is unlikely to fully explain the tBrg1^−/− phenotype, however, because injection of anti-CD3ε, an antibody that mimics pre-TCR signaling, promotes DN→DP transition in N-CoR^−/− thymocytes whereas in SWI–SNF-related mutant thymocytes it does not (33, 34). This result implicates the Brg1-containing complex(s) in mediating events downstream of the pre-TCR complex.

Signals emanating from the pre-TCR induce the cell proliferation and differentiation required for transition to the DP stage, raising the possibility that Brg1 is required for regulation of genes essential for these processes. Indeed, BrdU incorporation in tBrg1^−/− thymocytes revealed increased cell proliferation at the DN stages, indicating a cell cycle defect. Interestingly, Brg1 has been shown to interact with the retinoblastoma (Rb) and cyclin E gene products, both of which are required for the G1/S phase transition (35). Based on these data, it has been suggested that Brg1 is a critical regulator of cell cycle progression and tumor suppression. This notion is consistent with an increase in total TN thymocytes in mutants, however, the severe reduction in total thymocytes, owing to an almost complete loss of DP cells, suggests a mechanism of active cell elimination before the DP stage. Indeed, using CD4 derepression to mark mutant cells, we observed increased apoptotic rates in DN populations, suggesting a coupling of cell proliferation to programmed cell death. In addition to the well-established role of Rb loss in tumor progression, a similar coupling of proliferation to apoptosis (p53-mediated) has been associated with Rb deficiency in a variety of cell types in animal models (35). In support of a similar Rb/E2F/p53 pathway operating in the thymic development, disruption of E2F1 leads to suppression of apoptosis and results in excess mature T cells (36). Moreover, p53 deficiency rescues the block in pre–T cell differentiation associated with CD3γ mutation (37). Therefore, given the interaction between Brg1 and Rb pathway, loss of Brg1 leading to aberrant Rb function, increased E2F-mediated transcription, and induction of p53-mediated apoptotic pathways provide a plausible explanation for the defects in thymocyte development in tBrg1^−/− mice. However, we cannot rule out a role for Brg1 in the cell survival–independent differentiation program underlying transition to the DP stage.

Due to the high efficiency of Cre-mediated recombination at the DN4 stage in +/floxed; lck-Cre mice and the severity of developmental block (low numbers of DP cells) in the thymus of tBrg1^−/− mice, an unexpected result of our study was the presence of mature T cells in the periphery of these mutants. These T cell populations were shown to be a mixture of mutant and heterozygous (unrecombined floxed) cells, ruling out defective Cre-mediated recombination as the sole explanation for survival of these T lymphocytes. A parsimonious explanation for the survival of mutant cells is that in a small percentage of thymocytes, Cre-mediated recombination occurs relatively late in the DN4 stage and protein stability allows for clearing of key checkpoints required for further maturation. The mere existence of mature mutant T cells indicates that in contrast to developing thymocytes, Brg1 is not required for survival of this more differentiated cell type. One could argue that because the Brg1 mutation induced cell death at a highly proliferative stage of thymocyte development (β selection), perhaps Brg1 would not be expected for survival in the more quiescent circulating mature T cell. Arguing against this view, however, is evidence for homeostatic proliferation occurring in the lymphopenic environment of tBrg1^−/− mice. Interestingly, homeostatic proliferation of the limited number of surviving cells did not fully reconstitute peripheral T cell populations, contrary to the notion that a single T cell clone stimulated to generate memory T cells can fill the activated peripheral T cell pool (24, 38). Our results are consistent with more recent reports showing that limited numbers of naive T cells cannot fully reconstitute the periphery (39), indicating that loss of Brg1 in these cells might not affect homeostasis, per se. The loss of Brg1 function did not, however, lead to deregulated cell proliferation and T cell–derived tumor formation in mutant animals aged over 18 mo, which stands in contrast to the high incidence of CD8⁺ lymphomas observed through conditional inactivation of SNF5, another SWI–SNF-related complex member (40). Importantly, this finding supports the notion that the putative tumor suppressor function associated with Brg1 loss of function mutations in neoplasms of humans and mice is tissue and cell type dependent and further, does not predispose cells of the mature T cell lineage to oncogenic transformation. Thus, taken together these results suggest that Brg1 has inherently different functions with respect to cell survival and proliferation at different stages of T cell lineage specification.

We thank Daniel Metzger and Pierre Chambon for the Brg1 floxed allele, Yuan Zhuang for the lck-Cre mice, and Gail Martin for the Zp3-Cre mice. We are also grateful to members of the Magnuson and Su labs for useful discussions and advice.

This work was supported by grants from the National Institutes of Health to T. Magnuson and L. Su.