The Osa-containing SWI/SNF chromatin-remodeling complex regulates stem cell commitment in the adult Drosophila intestine

Adult tissue homeostasis is maintained by adult stem cells, which are multipotent cells that can self-renew and differentiate into functional cell types throughout the lifetime of the organism. The differentiation into multiple mature cell types and the self-renewal of adult stem cells are well balanced, and alterations in this equilibrium may cause diseases such as premature aging and carcinogenic transformation.

Like its mammalian counterpart, the adult midgut of Drosophila is maintained by multipotent intestinal stem cells (ISCs). After an asymmetrical division (de Navascués et al., 2012; Goulas et al., 2012; O’Brien et al., 2011), ISCs give rise to one new ISC (self-renewal) and one immature daughter cell, an enteroblast (EB). The EB can further differentiate into either an absorptive enterocyte (EC) or a secretory enteroendocrine (EE) cell without mitotic division (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006). Notch (N) signaling plays a major role in regulating ISC self-renewal and differentiation, and its loss leads to ISC expansion at the expense of ECs and to increased numbers of EE cells, probably because of the elevated expression of EE cell fate determinants scute (sc) and asense (ase), whereas N overactivation results in ISC differentiation into ECs (Bardin et al., 2010; Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006; Ohlstein and Spradling, 2007). The ligand of the N pathway, Delta (Dl), is specifically expressed in ISCs and unidirectionally switches the N signaling pathway on in neighboring EBs to promote differentiation toward ECs.

Stem cell fate is orchestrated by both intrinsic programs within the stem cells and extrinsic cues, namely the stem cell niche (Decotto and Spradling, 2005). Epigenetic programming, such as DNA methylation, histone modification and chromatin remodeling, which can generate variable patterns of gene expression from an invariant regulatory DNA sequence, has been identified as a major intrinsic mechanism for stem cell fate regulation (Hochedlinger and Plath, 2009; Juliandi et al., 2010). However, the epigenetic regulation of stem cell self-renewal and differentiation in vivo is not well understood.

SWI/SNF is an evolutionarily conserved and well characterized ATP-dependent chromatin-remodeling complex (Bouazoune and Brehm, 2006). A growing body of evidence indicates that many counterparts of SWI/SNF in mammals have a widespread role in tumor suppression; a high frequency of mutations in several SWI/SNF subunits have been identified in various cancers (Clapier and Cairns, 2009; Wilson and Roberts, 2011). There are at least two subtypes of the SWI/SNF (Brahma) complex in Drosophila: BAP and PBAP (Bouazoune and Brehm, 2006; Mohrmann et al., 2004). BAP and PBAP share common subunits including Brahma (Brm), Snr1 and Moira (Mor) but contain different signature proteins. Osa defines the BAP complex, which is required for normal embryonic segmentation and antagonizes Wingless signaling (Collins and Treisman, 2000; Treisman et al., 1997). The BAP complex also plays a role in the regulation of gene expression in response to Egfr signaling in the Drosophila wing (Terriente-Félix and de Celis, 2009). The mammalian homologs of Osa, BAF250a (ARID1A) and BAF250b (ARID1B), are required for maintaining the pluripotency of embryonic stem cells (Gao et al., 2008; Yan et al., 2008).

To further understand the molecular mechanisms of ISC self-renewal and differentiation, we carried out a transgenic RNAi screen and identified that the Osa-containing SWI/SNF complex regulates Drosophila ISC commitment to differentiation into discrete lineages. Loss-of-function mutations of subunits of the Osa-containing SWI/SNF complex resulted in ISC-like cell expansion at the expense of differentiated EC and EE cells. We demonstrated that Osa binds to the promoters of Dl and ase to regulate their expression, thus controlling ISC self-renewal and commitment to differentiation into EC and EE cells.

The following fly strains were used: esg-Gal4 (Shigeo Hayashi, Riken); esg-lacZ (Stephen DiNardo, University of Pennsylvania); mira-GFP (Francois Schweisguth, CNRS); Dl-lacZ (Bruce Edgar, University of Heidelberg); UAS-N^Δ34a (Ken Irvine, Rutgers); UAS-N^DN (Mark Fortini, Thomas Jefferson University); Su(H)GBE-lacZ (Sarah Bray, University of Cambridge); UAS-ase (Yuh Nung Jan, UCSF); ase-Gal4 (Tzumin Lee, Janelia Farm); FRT^82B-osa² (James Kennison, NIH); and FRT^82B-Snr1^R3 (Andrew Dingwall, Loyola University). UAS-Dl, act-Gal4, UAS-sc, UAS-2XEYFP, tub-Gal80^ts and fly strains used for MARCM clones (FRT^82B-piM; act>y⁺>Gal4, UAS-GFP; SM6, hs-flp; FRT^82B tub-Gal80) were obtained from the Bloomington Drosophila Stock Center (BDSC) at Indiana University. The following transgenic RNAi lines were obtained from BDSC or the Vienna Drosophila RNAi Center (VDRC): UAS-osa^RNAi (V7810 and BL31266), UAS-Snr1^RNAi (V12645, V108599 and BL32372), UAS-brm^RNAi (V37721 and BL31712), UAS-mor^RNAi (V6969 and V110712) and UAS-ase^RNAi (V108511).

The UAS-Snr1 transgene was generated by cloning full-length cDNA of Snr1 into pUAST (Brand and Perrimon, 1993), and injecting purified DNA into the embryo using standard protocols.

To induce MARCM clones of FRT^82B-piM (as a wild-type control), FRT^82B-osa², FRT^82B-Snr1^R3 and FRT^82B-Dl^RevF10, we generated the following flies: act>y⁺>Gal4, UAS-GFP/SM6, hs-flp; FRT^82B tub-Gal80/FRT^82B mutant. The genotype of flies used to generate N and osa² double-mutant clones is hs-flp, FRT^19A tub-Gal80/FRT^19A N^55e11; act>y⁺>Gal4, UAS-GFP; FRT^82B-osa²/FRT^82B tub-Gal80. Three- or four-day-old adult female flies were heat shocked twice at an interval of 8-12 hours, at 37°C for 60 minutes. The flies were transferred to fresh food daily after the final heat shock, and their posterior midguts were processed for staining at the indicated times.

Four male UAS-RNAi transgenic flies were crossed with eight virgin act-Gal4, esg-lacZ/Cyo; tub-Gal80^ts/tub-Gal80^ts or esg-Gal4, UAS-2XEYFP/Cyo; tub-Gal80^ts/tub-Gal80^ts virgin female flies at 18°C. One-week-old adult progenies of the correct genotype were transferred to new vials at 29°C for 7 days before dissection.

Total RNA from adult guts was isolated using the RNeasy Mini Kit (Qiagen) with on-column DNase digestion to remove genomic DNA. cDNA was synthesized using the ThermoScript RT-PCR system (Invitrogen). Real-time PCR analysis was performed on the Mastercycler Realplex real-time PCR system (Eppendorf) using SYBR Green PCR Master Mix (Clontech). qPCR results are represented as mean ± s.e.m. of three biological replicates. Primer pairs for qPCR are listed in supplementary material Table S1.

Two hundred fly intestines were dissected for ChIP assays using the Magna ChIP G Tissue Kit (Millipore) with some modification. The primer pairs used to detect immunoprecipitated DNA are listed in supplementary material Table S1.

Polyclonal rabbit antisera were raised against 6×His fusion proteins containing amino acids 2-180 of the Snr1 protein. To produce the 6×His fusion protein, a 537 bp fragment of Snr1 was amplified by PCR using primers 5′-AGTAGAATTCGCACTGCAGACATACGGGGA-3′ and 5′-AGTAGCGGCCGCTCACTCTAGCTCCATGTCCAGTC-3′ (restriction sites underlined). The amplified fragments were cloned into the EcoRI and NotI sites of PET-28a (+) (Novagen). 6×His-Snr1 fusion protein was expressed in E. coli BL21(DE3) pLysS, purified on Ni-NTA agarose columns (Qiagen) and used to immunize rabbits as described previously (Zeng et al., 2007).

The fly intestines were dissected in PBS and fixed in PBS containing 4% formaldehyde for 20 minutes. After three 5-minute rinses with PBT (PBS + 0.1% Triton X-100), the samples were blocked with PBT containing 5% normal goat serum overnight at 4°C. Then, the samples were incubated with primary antibody at room temperature for 2 hours and then with fluorescent secondary antibody for 1 hour at room temperature. Samples were mounted in Vectashield mounting medium with DAPI (Vector Laboratories). We used the following antibodies: mouse anti-β-Gal (1:200; Clontech); mouse anti-Dl [1:50; Developmental Studies Hybridoma Bank (DSHB)]; mouse anti-Pros (1:50; DSHB); nc82 (1:20; DSHB); rabbit polyclonal anti-Pdm1 (1:1000; a gift from X. Yang, Zhejiang University); rabbit anti-Spdo (1:1000; a gift from J. Skeath, Washington University in St Louis); mouse anti-Osa (1:20; DSHB); rabbit anti-Snr1 (1:1000; this study); rabbit anti-Ase (1:2000; a gift from Yuh Nung Jan); guinea pig anti-Sc (1:1000; a gift from S. Crews, UNC-Chapel Hill) and chicken anti-GFP (1:3000; Abcam). Secondary antibodies were goat anti-mouse, anti-chicken and anti-rabbit IgG conjugated to Alexa 488, Alexa 568 or Alexa 649 (1:400; Molecular Probes). Images were captured with a Zeiss LSM 510 confocal system and processed with LSM5 Image Browser (Zeiss) and Adobe Photoshop.

To quantify the number of escargot (esg)⁺ or Prospero (Pros)⁺ cells in Fig. 1 and Fig. 5, the esg⁺ or Pros⁺ cells were counted in a 5×10³ μm² area of the field. In Fig. 4, all of the images were taken with the same confocal settings and the fluorescence intensity was measured using LSM5 Image Browser. All data were analyzed using Student’s t-test and sample size (n) is shown in the text.

To identify new regulators of ISCs, we carried out a screen in which a collection of transgenic RNAi lines from the Vienna Drosophila RNAi Center and the Bloomington Stock Center (Dietzl et al., 2007; Ni et al., 2009) were crossed with act-Gal4, esg-lacZ; Tub-Gal80^ts (referred to as act^ts, esg-lacZ) flies. One-week-old adult flies were shifted to the restrictive temperature (29°C) for 1 week, dissected and stained, and then examined for esg-lacZ-labeled progenitors.

One of the first genes identified in this screen was osa. Knockdown of osa by transgenic RNAi (osa^RNAi, V7810) resulted in a dramatic expansion of esg⁺ cells (average of 49.5 esg⁺ cells/5×10³ μm², n=31; Fig. 1B,B′,F) compared with the wild-type (7.8 esg⁺ cells/5×10³ μm², n=24; Fig. 1A,A′,F) posterior midgut. To test whether Osa functions specifically in progenitors, we knocked down osa specifically in ISCs and EBs using esg-Gal4, UAS-GFP; tub-Gal80^ts/+ (referred to as esg^ts). Compared with the wild-type control (8.6 esg⁺ cells/5×10³ μm², n=30; Fig. 1C,C′,G), knockdown of osa using esg^ts also caused dramatic expansion of the esg⁺ cells (V7810; 46.1 esg⁺ cells/5×10³ μm², n=33; Fig. 1D,D′,G). Knockdown of genes by dsRNAs often produces false-positive phenotypes because of off-target effects (Kulkarni et al., 2006). We ruled out the possibility of false-positive effects and confirmed the osa phenotype with a second transgenic RNAi line (BL31266; 44.7 esg⁺ cells/5×10³ μm², n=42; Fig. 1E,E′,G) generated from independent sequences (Ni et al., 2009). We further stained wild-type and osa^RNAi midguts for phospho-Histone H3 (pH3), a specific marker for mitotic cells. More pH3⁺ cells were found in osa^RNAi posterior midguts than in wild type (supplementary material Fig. S1), indicating that these esg⁺ cells kept dividing to achieve the cell expansion.

Osa is a component of the BAP SWI/SNF complex (supplementary material Fig. S2E) (Bouazoune and Brehm, 2006; Clapier and Cairns, 2009; Collins et al., 1999; Vázquez et al., 1999). Knockdowns of three other components (Snr1, brm and mor) by RNAi also resulted in significant increases of esg-lacZ⁺ cells in the posterior midguts (supplementary material Fig. S2A-D,F). However, the phenotypes of brm or mor knockdowns are generally much weaker than those of osa or Snr1 knockdowns, suggesting that the other components in the complex might partially compensate the loss of function of brm or mor in ISCs, or that ISC fate regulation is a mechanistically novel function of a subset of SWI/SNF complexes. We focused on analyzing the loss-of-function phenotypes of osa and Snr1.

Based on the important function of Osa and Snr1 in ISCs, we examined their expression patterns in the midgut using antibodies. In the wild-type flies, Osa and Snr1 were expressed in all cells, including ISCs and EBs, in the posterior midgut and displayed nuclear localization (supplementary material Fig. S3A,A′,C,C′), which is consistent with their role in chromatin remodeling. As expected, in GFP-labeled osa^RNAi and Snr1^RNAi cells, the protein levels were reduced to an undetectable level, suggesting that the RNAi almost completely depletes Osa and Snr1 expression (supplementary material Fig. S3B,B′,E,E′). However, higher expression of Snr1 was detected in ISCs and EBs when UAS-Snr1 was expressed in midguts by esg^ts (supplementary material Fig. S3D,D′), which confirms the specificity of our new anti-Snr1 serum.

Esg is a marker of both ISCs and EBs (Micchelli and Perrimon, 2006). To characterize expanded esg⁺ cells in osa^RNAi midguts, we analyzed the expression of Su(H)GBE-lacZ, a marker of EBs (Ohlstein and Spradling, 2007), in wild-type and osa^RNAi posterior midguts. In wild-type midguts, only some esg⁺ cells expressed Su(H)GBE-lacZ (Fig. 2A,A′), which is consistent with its reported expression in EBs (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2007). In osa^RNAi posterior midguts (Fig. 2B,B′), Su(H)GBE-lacZ-expressing cells were missing, suggesting that the expanded esg⁺ cells in osa^RNAi posterior midguts are not EBs. miranda promoter-GFP (Mira-GFP) is a target of Daughterless-dependent bHLH transcriptional activity and is specifically expressed in ISCs (Bardin et al., 2010). Interestingly, Mira-GFP is indeed expressed in all expanded esg⁺ cells in osa^RNAi posterior midguts (supplementary material Fig. S4A,A′), which suggests that these esg⁺ cells might be ISC-like cells.

Dl and Sanpodo (Spdo) are markers of ISCs (Ohlstein and Spradling, 2007; Perdigoto et al., 2011). We first examined Dl expression in the wild-type (esg^ts>WT; Fig. 2C,C′) and osa^RNAi (esg^ts>osa^RNAi; Fig. 2D,D′) posterior midguts. In wild-type midguts, some of the esg⁺ cells that are ISCs express cytoplasmic Dl (Fig. 2C,C′). Surprisingly, none of the esg⁺ cells in the osa^RNAi midguts expressed Dl (Fig. 2D,D′). We also found that Dl expression was undetectable in Snr1^RNAi midguts (esg^ts>Sn1^RNAi; supplementary material Fig. S4B,B′). In wild-type posterior midguts, some of the esg⁺ cells that are ISCs expressed Spdo (Fig. 2E,E′). In the osa^RNAi midguts (Fig. 2F,F′), all esg⁺ cells expressed Spdo, further suggesting that the esg⁺ cells are ISC-like cells.

To further determine the function of Osa in regulating ISC self-renewal or differentiation, we generated wild-type and osa² (Vázquez et al., 1999) clones using the mosaic analysis with a repressible cell marker (MARCM) technique (Lee and Luo, 1999). Clones marked homozygous for wild type (Fig. 3A,A′,C) and osa² (Fig. 3B,B′,D) were generated in the posterior midgut and identified by GFP expression. Eight days after clone induction (ACI) we observed that, in GFP-labeled wild-type clones (Fig. 3A,A′,C), there were differentiated EC cells with large nuclei (asterisk in Fig. 3A,A′), Pros⁺ EE cells (arrowheads in Fig. 3A,A′), Dl⁺ ISCs (arrows in Fig. 3A,A′) and Spdo⁺ ISCs (arrow in Fig. 3C). However, osa² mutant clones were almost devoid of EC and EE cells, and all cells were Dl^- and Spdo⁺ ISC-like cells (Fig. 3B,B′,D).

To more precisely assess the phenotypes, we counted the number of Spdo⁺ ISC-like cells and Pros⁺ EE cells in GFP-labeled MARCM clones of wild type, osa² and Dl^RevF10 (a null allele of Dl; supplementary material Fig. S5) (Heitzler and Simpson, 1991) midguts at 4 and 8 days ACI (Fig. 3E,F). In the wild-type control clones, 23% and 18% of cells were Spdo⁺ ISCs at 4 and 8 days ACI (n=20, 21), respectively. However, 91% and 98% of cells were Spdo⁺ ISC-like cells in osa² clones at 4 and 8 days ACI (n=20, 21), respectively; and 70% and 67% of cells were Spdo⁺ ISC-like cells in Dl^RevF10 clones at 4 and 8 days ACI (n=20, 21), respectively. The Pros⁺ EE cells were dramatically reduced in osa² mutant clones. At 4 and 8 days ACI, respectively, 9.7% and 11% of cells were EE cells in wild-type clones and only 1.8% and 0.9% of cells were EE cells in osa² clones, whereas 24.9% and 31.8% of cells were EE cells in Dl^RevF10 clones.

We similarly generated Snr1^R3 (Zraly et al., 2003) MARCM clones and found that Snr1 mutation also resulted in ISC expansion without differentiation (supplementary material Fig. S6). Together, these results suggest that Osa and Snr1 might regulate ISC self-renewal and differentiation into both EC and EE cells.

The undetectable level of Dl expression, as judged by antibody staining, encouraged us to further examine the expression of Dl using Dl-lacZ, an enhancer trap at the Dl locus (Beebe et al., 2010; Jiang et al., 2009; Zeng et al., 2010). In wild-type posterior midguts, Dl-lacZ was highly enriched in ISCs (fluorescence intensity of 3991, n=56; Fig. 4A,A′,C). However, the expression of Dl-lacZ was dramatically reduced in osa^RNAi posterior midguts (esg^ts>osa^RNAi; fluorescence intensity of 200, n=58; Fig. 4B,B′,C). We also detected Dl mRNA expression by qPCR and found that the level of Dl mRNA was significantly reduced in the osa^RNAi midguts (Fig. 4D). These data suggest that Osa regulates Dl expression at the transcriptional level in the posterior midgut.

Expression of the Dl receptor N was unaffected, as N expression can be detected in the osa^RNAi midguts as well as in the wild type (supplementary material Fig. S4C-D′). These data suggest that Osa might specifically regulate Dl expression at the transcriptional level to control ISC self-renewal and differentiation of ISCs into ECs.

When an activated form of N is expressed in the posterior midgut (esg^ts>N^Δ34a), all ISCs differentiate into ECs (compare supplementary material Fig. S7B with S7A). In the osa^RNAi midgut (esg^ts>osa^RNAi; supplementary material Fig. S7C), excess ISC-like cells were found at the expense of differentiated EC and EE cells. To determine the epistatic relationship between Osa and the N signaling pathway, we expressed the constitutively activate form of N (N^Δ34a) in the osa^RNAi midgut (esg^ts>osa^RNAi+N^Δ34a; supplementary material Fig. S7D). After shifting the adult flies to the restrictive temperature (29°C) for 7 days, all ISCs had differentiated into ECs. These results support the idea that Osa functions upstream of N in regulating EC fate.

Since Osa functions upstream of N signaling, and Dl expression is blocked in the osa^RNAi midgut as well, Osa might regulate ISC differentiation into ECs by controlling Dl expression. To test this, we expressed UAS-Dl in osa^RNAi posterior midguts (esg^ts>osa^RNAi+UAS-Dl; Fig. 5C) or in GFP-marked wild-type and osa² mutant MARCM clones (Fig. 5E; supplementary material Fig. S7E-F′). UAS-Dl expression using the esg^ts driver rescues the ISC tumor phenotype in osa^RNAi midguts to a phenotype resembling wild type (8.5, 43.5 and 10.5 esg⁺ cells/5×10³ μm²; n=23, 36 and 42, respectively; Fig. 5A-C,F). In osa² mutant MARCM clones, all GFP-marked cells were ISC-like diploid cells (Fig. 3B,B′,D, Fig. 5D). UAS-Dl expression in osa² mutant MARCM clones rescued their phenotypes to that resembling wild-type clones including ISCs (or EBs) and polyploid ECs (Fig. 5E; supplementary material Fig. S7E-F′).

We also quantified the number of Pros⁺ EE cells in the midguts of wild type (Fig. 5A), osa^RNAi (Fig. 5B) and osa^RNAi with UAS-Dl expression (Fig. 5C). Dl expression does not rescue the EE cell-loss phenotype in osa^RNAi midguts (3.8, 1.8 and 1.7 EE cells/5×10³ μm²; n=23, 36 and 42, respectively; Fig. 5G).

These data suggest that Osa regulates ISC self-renewal and differentiation into ECs by controlling Dl expression and regulates EE cell formation by controlling other gene(s).

In addition to regulating Dl expression to control EC fates, the Osa-containing SWI/SNF complex might control EE cell formation by regulating other signal(s). We examined the epistatic relationship of N signaling and the SWI/SNF complex in regulating EE cell fate determination. Expressing a dominant-negative form of N in the posterior midgut (esg^ts>N^DN; 50.9 EE cells/5×10³ μm², n=33; Fig. 5H,K) resulted in the formation of excess EE cells. The expression of osa^RNAi (esg^ts>N^DN+osa^RNAi; 3.2 EE cells/5×10³ μm², n=45; Fig. 5I,K) and Snr1^RNAi (esg^ts>N^DN+osa^RNAi; 3.2 EE cells/5×10³ μm², n=25; Fig. 5J,K) in N^DN midguts completely suppressed the phenotype of excess EE cells to wild-type midgut levels (3.7 EE cells/5×10³ μm²; n=30; Fig. 5A,K).

In N null (N^55e11) mutant MARCM clones, excess Pros⁺ EE cells were found (38% of the cells are Pros⁺ EE cells, n=30; supplementary material Fig. S7G,I). As in osa² mutant MARCM clones (Fig. 3B,B′,F), very few Pros⁺ EE cells were detected in N^55e11; osa² double-mutant MARCM clones (only 1.2% of the cells are Pros⁺ EE cells, n=40; supplementary material Fig. S7H,I). Consistently, we also noted that osa² repressed Dl expression in N^55e11 clones (supplementary material Fig. S7H). To exclude the possibility that Osa directly regulates expression of the EE cell marker Pros in contexts other than in EE cell formation, we identified another EE cell-specific marker, nc82 (Bruchpilot - FlyBase), which labels the synaptic active zone in the Drosophila neuromuscular junction (Wagh et al., 2006). nc82 exhibits punctate staining at the membrane of Pros⁺ EE cells (supplementary material Fig. S8A-B′); however, unlike in the N^DN midgut where there are many nc82⁺ EE cell clusters (supplementary material Fig. S8D,D′), we could not detect the expansion of any nc82⁺ EE cells in osa^RNAi midguts (esg^ts>osa^RNAi; supplementary material Fig. S8C,C′). Together, these data suggest that the Osa-containing SWI/SNF complex functions downstream of N in regulating EE cell formation.

Two transcription factors, Sc and Ase, have been shown to play a major role in EE cell fate determination and to be upregulated in the N^DN midgut by mRNA profiling (Bardin et al., 2010). Expression of Sc and Ase in the posterior midgut was below the detection level using specific antibodies (Brand et al., 1993; Stagg et al., 2011) (supplementary material Fig. S9A,A′,C,C′). Nevertheless, Sc and Ase were readily detected in the midgut with ectopic sc and ase expression driven by esg^ts (esg^ts>UAS-sc and esg^ts>UAS-ase) for 24 hours at 29°C (supplementary material Fig. S9B,B′,D,D′). In addition, consistent with previous observations by mRNA profiling (Bardin et al., 2010), both Sc and Ase were upregulated in the N^DN midgut as assessed by qPCR (Fig. 6A,B) and antibody immunofluorescence (compare supplementary material Fig. S9G with S9A,A′ and S9I with S9C,C′). Using the more sensitive ase-Gal4 (ase-Gal4>UAS-mCD8-GFP) transgene (Zhu et al., 2006), we detected that ase is weakly expressed in ISCs and EBs but not in Pros⁺ EE cells (supplementary material Fig. S9E-F′).

To examine the relationship of Osa to Sc and Ase, we first compared mRNA levels of sc and ase in wild-type and osa^RNAi midguts by qPCR. Although sc mRNA levels were lower in osa^RNAi than in N^DN midgut, sc mRNA levels were upregulated in osa^RNAi compared with wild-type midgut (Fig. 6A). However, unlike in the N^DN midgut (Bardin et al., 2010), the ase mRNA levels were significantly lower in osa^RNAi than in wild-type control midgut (Fig. 6B). We further confirmed the expression levels of Sc and Ase in osa^RNAi midgut by antibody immunofluorescence. osa^RNAi midguts had much higher expression of sc than wild-type midguts (compare supplementary material Fig. S9H with S9A,A′), whereas ase expression was undetectable in osa^RNAi as in wild-type midguts (compare supplementary material Fig. S9J with S9C,C′).

These data suggest that the Osa-containing SWI/SNF complex might control EE cell formation through regulating the expression of ase. Indeed, expression of ase^RNAi in the N^DN midgut (Fig. 6D; esg^ts>N^DN+ase^RNAi) suppressed the excess EE cell phenotype of N^DN (compare Fig. 6D with 6C). When we tried to overexpress ase in the osa^RNAi midgut (esg^ts>osa^RNAi+ase) it suppressed the EE cell-loss phenotype of osa^RNAi (only 1.8 EE cells/5×10³ μm² in osa^RNAi midgut but 39.7 EE cells/5×10³ μm² in the osa^RNAi+ase midgut; n=36 and 33, respectively; Fig. 6E-G). These data together suggest that Osa regulates EE cell formation through Ase.

We further examined the interaction of Osa and Snr1 proteins with the promoters of Dl, ase and spdo in a ChIP assay using Osa- and Snr1-specific antibodies or an unrelated IgG control (Fig. 7A-B′). We found that Osa and Snr1 associated with the Dl and ase promoters at the chromatin level, whereas neither bound to the spdo promoter, which suggests that Osa and Snr1 might regulate Dl and ase expression directly. However, we could not exclude the possibility that Dl and ase are indirectly regulated by another transcriptional target of Osa.

In summary, these data suggest that Osa regulates ISC differentiation into ECs by controlling Dl expression in ISCs, and regulates EE cell formation by controlling ase expression (Fig. 7C).

Previous studies have demonstrated that N signaling regulates ISC fate specification in the Drosophila posterior midgut (Ohlstein and Spradling, 2007). The Dl ligand is specifically expressed in ISCs to activate N signaling in neighboring EBs to promote EB to EC or inhibit EB to EE cell differentiation. However, it is still not fully understood how ISCs are coordinately controlled to differentiate into EC or EE cells, thus maintaining tissue homeostasis. Here, we found that the commitment of stem cells to discrete differentiated cells is coordinately regulated by the Osa-containing SWI/SNF chromatin-remodeling complex. Osa regulates ISC differentiation into ECs by controlling the expression of Dl to elaborate N signaling, and regulates EE cell formation by controlling the expression of ase transcription.

Our results suggest that Osa, a component of the SWI/SNF complex, is essential for ISC differentiation into EC and EE cells in the Drosophila posterior midgut. The SWI/SNF chromatin-remodeling complex performs a crucial function in gene regulation by using energy derived from ATP hydrolysis to alter the contacts between histones and DNA in nucleosomes, resulting in increased DNA accessibility to transcription factors and other regulatory proteins. Activating the N signal in EBs with the ISC-originated N ligand Dl is essential for EB differentiation into ECs. Osa promotes the expression of Dl in ISCs and maintains activation of the N signal in EBs to promote EB to EC differentiation. Repressing Dl expression by loss of function of Osa enhances ISC self-renewal and blocks ISC commitment to ECs, resulting in ISC expansion/tumors. Dl expression alone is sufficient to rescue the ISC expansion/tumor phenotype in osa mutants. SWI/SNF components have been shown to frequently occupy transcription start sites, enhancers and CTCF-binding regions (Euskirchen et al., 2011). Indeed, Osa and Snr1 bind to the promoter region of Dl at the chromatin level. Consistently, genetic interaction between the SWI/SNF complex and Dl were found during a genetic screen (Armstrong et al., 2005), and the expression of a dominant-negative form of Brm, a core component of the SWI/SNF complex, in the adult sense organ precursor lineage causes phenotypes similar to those resulting from impaired Dl-N signaling.

Why does the Osa-containing SWI/SNF complex regulate Dl expression only in ISCs and not in EBs? One possible explanation is that the complex in ISCs and EBs has different associated factors, and only that in ISCs is able to regulate Dl expression. Another possibility is that some transcription factors recruit the complex to regulate Dl expression, and these transcription factors exist only in ISCs.

Disrupting the Dl-N signal in ISCs results in both ISC and EE cell expansion. Given that Osa regulates Dl expression in ISCs, osa mutant ISCs should have phenotypes similar to those that result from impaired Dl-N signaling; however, osa mutation only causes ISC expansion, and EE cell formation is repressed. Proneural bHLH factors of the achaete-scute complex (AS-C), Sc and Ase, have been reported to play a crucial role in EE cell fate specification (Bardin et al., 2010). sc or ase overexpression in midgut progenitors leads to dramatically increased EE cell numbers, whereas mutant ISC clones of Df(1)scB57, which lack all four AS-C genes, i.e. achaete (ac), sc, lethal of scute [l(1)sc] and ase, are devoid of EE cells (Bardin et al., 2010). sc and ase expression is upregulated when the N signal is blocked in the intestinal midgut. Similarly, we also found that the expression of sc is upregulated in the osa^RNAi midgut compared with the wild-type control, whereas ase expression is significantly decreased in the osa^RNAi midgut. Furthermore, knockdown of ase alone blocks EE cell formation in the N^DN midgut. Thus, our data suggest that Osa controls EE cell formation by regulating the expression of ase. Terminal differentiation of EBs into EC or EE cells has been proposed to be regulated by differential N signaling, whereby a high level of N activity induces EBs to differentiate into ECs, whereas a low level of N activity promotes the EE cell fate (Ohlstein and Spradling, 2007). However, EE cells can still be generated even when N activity is suppressed by either the expression of N^DN or the clonal deletion of Dl or N, suggesting that the N signal is not required for EE cells at all.

Both Sc and Ase may play important roles in regulating EE cell fate in the midgut. A high level of N activity might suppress the expression of sc and ase and thereby promote EC fate. By contrast, a low level of N activity might activate the expression of sc and ase, inducing EB cells to differentiate into EE cells. Excess EE cells in the N^DN midgut might be the result of upregulation of ase and sc caused by disruption of the N signal. Indeed, previous studies have shown that N signaling activates the expression of the Enhancer of split complex [E(spl)-C], which can, in turn, repress the expression of the proneural genes ac, sc, l(1)sc and ase (Bailey and Posakony, 1995; Lecourtois and Schweisguth, 1995). In addition to sc and ase, the overexpression of ac and l(1)sc in the midgut also results in dramatically increased EE cell numbers, but only Sc and Ase are required for EE cell formation (data not shown). Why does Osa regulate ase and not other AS-C members? Even though Ase belongs to the AS-C proneural family, it differs from the other AS-C members in its expression pattern, regulation, mutant phenotype and in some DNA-binding properties (Jarman et al., 1993). These characteristics suggest that Ase might be associated with some specific transcription factors that bind to ase DNA at particular chromatin states controlled by the Osa-containing SWI/SNF chromatin-remodeling complex.

We thank Jiangsha Zhao, Chhavi Chauhan and Ashley DeVine for critical reading of the manuscript; Shigeo Hayashi, Stephen DiNardo, Francois Schweisguth, Bruce Edgar, Ken Irvine, Mark Fortini, Sarah Bray, James Kennison, Andrew Dingwall, Yuh Nung Jan, Tzumin Lee, VDRC, BDSC and TRiP at Harvard Medical School for fly stocks; Xiaohang Yang, Jim Skeath, Stephen Crews, Yuh Nung Jan and DSHB for antibodies; and Stephen Lockett for help with confocal microscopy.