Adult stem cells maintain tissue homeostasis by controlling the proper balance of stem cell self-renewal and differentiation. The adult midgut of Drosophila contains multipotent intestinal stem cells (ISCs) that self-renew and produce differentiated progeny. Control of ISC identity and maintenance is poorly understood. Here we find that transcriptional repression of Notch target genes by a Hairless-Suppressor of Hairless complex is required for ISC maintenance, and identify genes of the Enhancer of split complex [E(spl)-C] as the major targets of this repression. In addition, we find that the bHLH transcription factor Daughterless is essential to maintain ISC identity and that bHLH binding sites promote ISC-specific enhancer activity. We propose that Daughterless-dependent bHLH activity is important for the ISC fate and that E(spl)-C factors inhibit this activity to promote differentiation.
The homeostasis of adult tissues is controlled through the renewal of differentiated cells by adult stem cells or committed progenitors. A detailed understanding of how stem cells are controlled in their in vivo environment has proved challenging owing to the requirement for precise genetic manipulation and in vivo lineage-labeling techniques. The posterior midgut of the adult Drosophila intestine is a simple model system in which to understand how adult stem cells are maintained. The posterior midgut epithelium contains intestinal stem cells (ISCs) distributed throughout the tissue that self-renew and produce differentiated cells during the adult lifetime (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006). Upon cell division, each ISC produces one daughter cell that retains the ISC fate and one enteroblast (EB) that differentiates into either an absorptive enterocyte (EC) or a secretory enteroendocrine cell (ee) (Ohlstein and Spradling, 2006) (see Fig. 1N). This simple cell lineage greatly facilitates the analysis of fate decisions using precise lineage labeling and genetic manipulation of the ISC and its progeny.
Recent studies have begun to address the control mechanisms of stem cell self-renewal and proliferation. Self-renewal of ISCs is influenced by Wingless (Wg) (Lee et al., 2009; Lin et al., 2008). Wg, however, is not strictly required to maintain ISC identity, as ISCs are still detected in the absence of Wg signaling (Lin et al., 2008). Additionally, ISC proliferation is modulated by Insulin and Jak/Stat signaling, at least in response to intestinal damage (Amcheslavsky et al., 2009; Buchon et al., 2009; Jiang et al., 2009; Lin et al., 2010). Despite these significant recent advances, the signals and transcriptional programs that control ISC identity and maintenance are unknown. One important clue, however, has come from studies on the role of Notch in this lineage that showed that differentiation of ISC progeny requires Notch signaling and that forced expression of activated Notch results in the differentiation of ISCs (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006; Ohlstein and Spradling, 2007).
In this study, we find that ISC maintenance requires transcriptional repression of Notch target genes by Hairless. We identify the Enhancer of split complex [E(spl)-C] genes as the key Notch targets that need to be repressed by Hairless to ensure ISC maintenance, while being upregulated in EBs to promote differentiation. Additionally, Daughterless (Da), a basic helix-loop-helix (bHLH) transcriptional activator, is also essential for ISC maintenance and bHLH E-box binding sites are required for expression of an ISC-specific enhancer. We propose a model in which Da and E(spl)-C factors act antagonistically to regulate maintenance of the ISC fate versus EB differentiation.
MATERIALS AND METHODS
Drosophila stocks and clone analysis
Drosophila stocks and crosses were kept at 18-25°C. Adults were aged at 25°C unless stated otherwise. Clones were induced in 3- to 6-day-old flies and analyzed in well-fed females. MARCM clones (Lee and Luo, 2001) were generated using the X chromosome y w P[hs-FLP] P[pTub-GAL4] P[UAS-nlsGFP] combined with FRT P[pTub-GAL80] chromosomes on the second and third chromosomes. MARCM clones on the X chromosome were induced with hsflp122 P[pTub-GAL80] FRT19A; P[pAct-GAL4] P[UASGFP] (Lin et al., 2008). The following mutant alleles were used to generate recombinant lines and for experiments: HE31, H1, H2, neurIF65, P[gro+] Df(3R) gro32.2, P[gro+] on II, P[l(2)35Bg+] Su(H)Δ47 [a small deletion of Su(H) with a genomic rescue construct of the neighboring gene (Morel and Schweisguth, 2000)], numb2, numb15, da10, da3, Df(1)scB57, amos1, and ato1. Control wild-type MARCM clones were generated using FRT82B P[w+]90B, FRT40A and y w FRT19A chromosomes. We generated the E(spl)-CΔmδ-m6 deletion by FLP-mediated recombination using lines XPd08311 and RBe00084 (Exelixis). This deficiency removes a genomic region containing the coding sequences (CDSs) of the E(spl)-C genes mδ, mγ, mβ, mα, m1, m2, m3, m4, m5 and m6. It also removes promoter sequence from m7, but leaves the coding sequence intact. The E(spl)-CΔmδ-m6 deficiency does not remove m8 or groucho. The entire E(spl)-C and groucho sequences are removed by Df(3R)gro32.2 and groucho is rescued by P[gro+].
We have adapted the MARCM technique to induce positively marked Su(H) H double-mutant clones in y w P[hs-FLP] P[pTub-GAL4] P[UAS-nlsGFP]; FRT40A P[l(2)35Bg+] Su(H)Δ47 / FRT40A P[pTub-GAL80]; FRT82B P[w+]90B HE31 / FRT82B P[pTub-GAL80] flies, in which GFP is expressed only upon recombination at both FRT40A and FRT82B sites. A two heat-shock protocol was used to alleviate residual Su(H) protein. The first heat shock creates three types of clones: unmarked Su(H)Δ47 mutant clones that grow into large clones, unmarked HE31 mutant clones that do not proliferate, and marked double-mutant clones that do not proliferate; these double-mutant clones behave as H single-mutant clones (data not shown), probably owing to residual levels of Su(H) protein. A second heat shock, applied 5 days later, creates the three types of clones described above as well as a fourth type: marked double-mutant clones produced by a single recombination event at the FRT82B occurring in Su(H)Δ47 mutant cells.
The esg-GAL4 UAS-GFP; pTub-GAL80ts driver (Jiang et al., 2009) was used to ectopically express UAS-scute and UAS-asense in adults raised at 18°C and shifted to 29° C at 3 days of adulthood. The MARCM system was used to overexpress UAS-Hairless and UAS-m7-VP16 (Jimenez and Ish-Horowicz, 1997). E(spl)-mβ1.5-lacZ (Cooper et al., 2000), EE4-lacZ (Culi and Modolell, 1998; Giagtzoglou et al., 2005) and Su(H)-GBE-lacZ (Furriols and Bray, 2001) flies were used.
The wild-type mira-promoter-GFP transgene consists of PCR-amplified genomic sequence flanking the mira CDS up to adjacent genes (for details, see Fig. S6 in the supplementary material), with the mira CDS replaced by nuclear-targeted mGFP6 sequence (Haseloff et al., 1999). The resulting product was cloned with NotI/SpeI into the NotI/XbaI sites of the pattB vector (Bischof et al., 2007). The E-box-mutated version was generated by PCR amplifying seven individual regions (the E-box sequence CAGCTG was mutated to CAAATG within the primer sequences). All PCR was performed using Phusion polymerase (Finnzymes). The mira-promoter-GFP and miraΔEbox-promoter-GFP constructs were sequenced and injected into ϕX-22A flies (Bischof et al., 2007) to allow site-specific integration of the promoter constructs.
Immunostaining and in situ hybridization
Adult female intestines were dissected in PBS and fixed for 2 hours at room temperature (RT) in 4% paraformaldehyde. Intestines were rinsed in PBT (PBS containing 0.1% Triton X-100), trimmed and incubated for at least 30 minutes in PBS containing 50% glycerol, then equilibrated in PBT to osmotically clean the lumen. For Fig. S1 in the supplementary material and Fig. 6A, intestines were fixed for 15 minutes in 4% formaldehyde/heptane, dehydrated in methanol and rehydrated in PBT as described (Lin et al., 2008). Primary antibody incubations were either overnight at 4°C or 3-5 hours at RT. Secondary antibodies were incubated 3-5 hours at RT. DAPI (1 μg/ml) was added to the final wash. Intestines were mounted in 4% N-propyl-galate, 80% glycerol and imaged on Leica SP2 and SPE confocal microscopes. Images are selected layers of confocal stacks. For Fig. 4B,C, intestines were processed as previously described for embryos (Bardin and Schweisguth, 2006). The following antibodies were used: mouse anti-Delta [C594-9B, developed by the laboratory of S. Artavanis-Tsakonas (Harvard Medical School, Boston, MA, USA); 1:2000, DSHB], guinea pig anti-Hairless A (A. Preiss, Institute of Genetik, University of Hohenheim, Stuttgart, Germany; 1:500), mouse anti-Pros [MR1A, developed by the lab of C. Doe (University of Oregon, Eugene, OR, USA); 1:10, DSHB], rabbit anti-Pros (Y.-N. Jan, University of California, San Francisco, CA, USA; 1:2000), rabbit anti-Daughterless (Y.-N. Jan; 1:1000), rabbit anti-Asense (Y.-N. Jan; 1:5000) and goat anti-β-gal (1:1000, Biogenesis).
For the scute in situ hybridization, the protocol developed by the Bier laboratory was used (O'Neill and Bier, 1994), followed by anti-DIG-POD (1:2000, Roche, 11207733910). In situs were developed using the TS Plus Cyanine 3 System following the supplier's instructions (Perkin-Elmer, NEL741001KT).
Quantification and statistical analysis
Only clones located in the posterior midgut were considered for analysis and numerous clones of representative midguts were analyzed. In Fig. 1M and Fig. 5E, the number of cells per clone was counted. In Fig. 3E and Fig. 4H, Delta– GFP+, Delta+ GFP+, as well as Pros– GFP+ and Pros+ GFP+ cells were counted on images taken using a CoolSNAP Camera (Ropers Scientific) on a Leica DMRXE epifluorescence microscope. Two planes of focus, apical and basal, were analyzed for each field. Differences in total cell number (Fig. 1M, Fig. 5E) or in the proportion of Delta+ and Pros+ cells relative to GFP+ cells (Fig. 3E, Fig. 4H) between genotypes were tested for statistical significance using a standard normal distribution where ϵ≥1.96 indicates significance with 95% confidence. In Fig. 1M: wild type, n=215 clones; HE31, n=334. Differences in the proportions of wild-type versus HE31 clones are significant for: 1-cell clones, ϵ=3.1; 2- to 5-cell clones, ϵ=3.2; ≥6-cell clones, ϵ=8.7. In Fig. 5E: wild type, n=139 clones; da10, n=62. Differences in the proportions of wild-type versus da10 clones are significant for: 1-cell clones, ϵ=5.78; 2- to 5-cell clones, ϵ=5.62. Since there were no da10 clones with greater than 5 cells, the difference in proportions could not be tested.
For counting Delta+ cells in Hairless mutants and in Fig. 3E at the 10 day time point: wild type, n=422 GFP+ cells; HE31, n=202 GFP+ cells; E(spl)-CΔmδ-m6, n=983 GFP+ cells; E(spl)-CΔmδ-m6 HE31, n=331 GFP+ cells. The following differences in proportions were statistically significant: wild type versus HE31, ϵ=3.7; HE31 versus E(spl)-CΔmδ-m6 HE31, ϵ=17.1. The difference in proportions between E(spl)-CΔmδ-m6 and E(spl)-CΔmδ-m6 HE31 is not significant (ϵ=1.7). For counting Pros+ cells in Fig. 4H, we used a 6 day time point: wild type, n=294 GFP+ cells; HE31, n=197 GFP+ cells; E(spl)-CΔmδ-m6, n=622 GFP+ cells; neurIF65, n=327 GFP+ cells. Wild type is significantly different from neurIF65 (ϵ=17.5). This statistical test was not used for cases in which the number of incidents was less than five.
EC density in Fig. 5D was determined on confocal stacks in which nuclear Pdm1 expression and nuclear size were measured and divided by the total area of the mutant or wild-type tissue. We found that Delta+ cells and Pros+ cells have nuclei ranging between 3 and 5 μm in diameter; we therefore use a cut-off of 7 μm to define the EC. Several representative stacks were analyzed. Error bars represent s.d. from the mean. In Fig. 6J, the number of clones at 10 days AHS that contained at least one Pros+ cell was assessed in wild-type versus Df(1)scB57 clones from Leica SPE confocal stacks. Twelve out of 32 wild-type clones contained at least one Pros+ cell, compared with none out of 37 Df(1)scB57 clones. This difference is statistically significant (P=0.0000256, Fisher's exact test).
For Fig. S2 in the supplementary material, confocal sections were used to count the total number of cells and the Delta+, Pros+ and phosphorylated Histone H3 (PH3)+ cells per clone. Error bars represent s.d. from the mean. In Fig. S2A,B in the supplementary material: for the overexpression of Hairless, n=20 clones; wild type, n=25 clones. In Fig. S2D in the supplementary material: for the overexpression of Hairless, n=25 clones; wild-type, n=28 clones. Thirteen out of 25 clones overexpressing Hairless have at least one PH3+ cell, compared with 3/28 in wild type (P=0.0008, two-tailed Fisher's exact test). Additionally, 4/25 clones overexpressing Hairless contained more than one dividing cell, compared with 0/28 for wild type (P=0.043, two-tailed Fisher's exact test).
Hairless is required for self-renewal of ISCs
Upon division of the ISC, Delta is inherited by both daughter cells and Notch is similarly present in both daughter cells (Ohlstein and Spradling, 2006; Ohlstein and Spradling, 2007). This raises the possibility that Notch receptors are activated in both daughter cells. Since Notch activity in ISCs can cause terminal differentiation (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2007), we hypothesized that expression of Notch target genes might be repressed in ISCs. The Hairless-Su(H) co-repressor complex acts in a limited number of developmental contexts to repress Notch target gene expression (Furriols and Bray, 2000; Koelzer and Klein, 2003; Koelzer and Klein, 2006; Morel and Schweisguth, 2000). The adaptor protein Hairless binds to Su(H) and is required for the repressor activity of Su(H) by recruiting the co-repressors CtBP and Groucho, as well as the Histone H3/H4 chaperone Asf1, thereby promoting efficient silencing through repressive chromatin (Bang and Posakony, 1992; Barolo et al., 2002; Brou et al., 1994; Castro et al., 2005; Goodfellow et al., 2007; Morel et al., 2001; Nagel et al., 2005). Upon Notch receptor activation, Notch is cleaved and the active form of Notch, NICD (Notch intracellular domain), translocates to the nucleus where it interacts with Su(H) and replaces Hairless (reviewed by Kopan and Ilagan, 2009; Bray, 2006; Schweisguth, 2004). First, we examined the expression of Hairless in the intestine of adult flies. Hairless was detected in the nuclei of ISCs and EBs, as well as in a small number of differentiated progeny (see Fig. S1 in the supplementary material). By contrast, lower or undetectable levels of Hairless were present in more mature ECs (see Fig. S1 in the supplementary material). This expression is consistent with a role of Hairless in ISCs and/or ISC daughter cells.
Next, we examined the role of Hairless in ISC maintenance. Growth of clones that were homozygous for the null allele HairlessE31 (HE31) was examined using the mosaic analysis with repressible cell marker (MARCM) technique (Lee and Luo, 2001), in which GFP was heritably expressed in either the ISC and its progeny cells, forming a growing stem cell clone, or in the EB, forming a transient single-cell clone with a ∼1-week turnover rate (Ohlstein and Spradling, 2006). Whereas wild-type clones proliferated over time and formed large clones by 14 days after heat shock (AHS) (Fig. 1A-D), HE31 mutant clones failed to grow, forming either very small groups of cells or remaining as single cells at 14 days AHS (Fig. 1E-H). To quantify the self-renewal potential of Hairless mutant clones, we counted the number of large stem cell clones (6 cells or more), small stem cell clones (2-5 cells), and single-cell clones (transient clones and stem cell clones that are non-proliferating) at 6 days AHS (Fig. 1M). This analysis revealed that only 4% of the HE31 mutant clones contained 6 cells or more, compared with 31% for wild-type clones. Furthermore, 55% of the HE31 mutant clones were single-cell clones (n=215), compared with only 42% of wild-type clones (n=334). These differences are statistically significant (see Materials and methods). H1 and H2 clones similarly lost self-renewing capacity over time and failed to produce large clones at 14 days AHS (data not shown), indicating that this phenotype is unlikely to result from unrelated mutations on this chromosomal arm. These results clearly show that Hairless mutant clones have a considerably diminished capacity to grow.
We next examined the fate of ISCs in Hairless mutants using Delta as a marker. At 10 days AHS, only 7% of cells in HE31 mutant clones were Delta+ (n=202; Fig. 1I-J″′ and Fig. 3E), whereas 18% of cells in wild-type clones were Delta+ (n=422), indicating that the ISC fate is not properly maintained upon loss of Hairless activity. We further examined differentiation in Hairless mutant clones. At 6 days AHS, 40% (n=158) of the HE31 mutant cells expressed the EC marker Pdm1 (Nubbin – FlyBase) (Lee et al., 2009) and 59% (n=675) had a large polyploid nucleus, a characteristic of ECs (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006), whereas only 0.5% (n=202) of the HE31 mutant cells expressed the ee marker Prospero (Pros) (Fig. 3E). This indicates that Hairless mutant cells can differentiate and, when they do so, they become ECs and not ees. This fate choice is consistent with the known role of Hairless in antagonizing Notch, as strong Notch signaling favors adoption of the EC fate (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2007). Nevertheless, 27% (n=675) of the HE31 mutant cells had a small diploid nucleus and 14% had an intermediate size nucleus. Thus, many Hairless mutant cells appear to have lost ISC characteristics (i.e. the expression of Delta and the ability to form large clones over time) without properly differentiating into ECs.
The effect of Hairless overexpression on clone growth and cell fate was tested. Hairless overexpression for 6 days caused a large increase in average clone size (85±57 cells, n=20 clones; see Fig. S2A in the supplementary material), as compared with wild-type clones (14±10 cells, n=25 clones). Additionally, Hairless overexpression resulted in many clones containing at least one mitotic cell (52%, n=25 clones; see Fig. S2C,D in the supplementary material), with one clone containing five cells undergoing division at the same time, including two that were directly adjacent, strongly suggesting that these two cells had acquired ISC fate. In addition, Hairless overexpression was sufficient to promote the formation of many ectopic small cells that expressed Delta and/or Pros (Fig. 1K-K″ and see Fig. S2 in the supplementary material). The increased number of both Delta+ and mitotic cells indicates that the overexpression of Hairless induces ectopic ISCs. We conclude that Hairless is both necessary and sufficient to produce ISC-like cells, indicating that Hairless regulates the maintenance of ISCs.
By contrast, Numb, another negative regulator of Notch, plays no essential role in ISC self-renewal: both numb2 (Fig. 1L) and numb15 (data not shown) mutant cells produced large clones over time. We therefore conclude that Hairless, but not numb, is required for ISC self-renewal and maintenance.
Hairless acts in an Su(H)-dependent manner to maintain the ISC fate
Repression of Notch target genes by Hairless requires the sequence-specific DNA-binding protein Su(H), which recruits the repression complex onto DNA (Barolo et al., 2002; Furriols and Bray, 2001; Morel et al., 2001; Morel and Schweisguth, 2000). To test whether Hairless acts via Su(H) to regulate ISC maintenance, we generated clones lacking both Hairless and Su(H) activities (see Materials and methods). The loss of Hairless activity resulted in defective self-renewal (Fig. 2C-D′; see also Fig. 1), whereas loss of Su(H) activity led to an overspecification of small Delta+ ISC-like cells (Fig. 2A-B′) (see Ohlstein and Spradling, 2007). Positively marked Su(H)Δ47 HE31 double-mutant clones generated within Su(H)Δ47 single-mutant unmarked clones proliferated and expressed the ISC marker Delta (Fig. 2E-H′). Therefore, the activity of Su(H) is required for the ISC loss seen in Hairless mutant clones and we conclude that Su(H) is epistatic to Hairless with respect to Delta expression and ISC self-renewal. This suggests that Hairless acts via Su(H) to repress Notch target genes in ISCs in order to promote stem cell maintenance. Interestingly, the loss of Su(H) led to an increased number of ISC-like cells, similar to Notch loss-of-function, whereas loss of Hairless activity led to a loss of ISC self-renewal, similar to the overexpression of an activated Notch receptor. This suggests that loss of Hairless/Su(H)-mediated repression, on its own, does not lead to ISC loss and that Su(H)-mediated target gene activation is also required for ISC loss in this context.
Deletion of the E(spl)-C suppresses the Hairless loss-of-ISC phenotype
We next sought to identify the Notch target genes that need to be repressed by Hairless in ISCs. The E(spl)-C genes are well characterized targets of the Notch pathway (Bailey and Posakony, 1995; Jennings et al., 1994; Lecourtois and Schweisguth, 1995). We found that an E(spl)mβ-lacZ construct (Cooper et al., 2000) was highly upregulated in EB cells, whereas it was expressed at only low levels in ISCs (Fig. 3F-F″). Moreover, loss of E(spl)-C genes, using either E(spl)-CΔmδ-m6 or the Df(3R) gro32.2 combined with the groucho rescue construct, P[gro+], resulted in an increased number of small Delta+ ISC-like cells (Fig. 3A-A″,E; see Fig. S3A-C″′ in the supplementary material). This indicates that E(spl)-C genes are required in ISC daughter cells to promote EB differentiation and/or block self-renewal in presumptive EB cells. To test whether this phenotype was due to the loss of the E(spl)-C bHLH transcriptional repressors, we took advantage of an `activator' version of m7 (HLHm7 – FlyBase), m7-VP16, in which fusion of the transcriptional activator VP16 to the bHLH m7 converts m7 from a transcriptional repressor into an activator (Jimenez and Ish-Horowicz, 1997). Expression of m7-VP16 similarly resulted in an increase in small Delta+ ISC-like cells and a loss of Pros+ cells (see Fig. S3F-F″ in the supplementary material). We therefore conclude that EB differentiation and/or inhibition of ISC self-renewal are likely to result from the loss of one or more E(spl)-C bHLH genes.
To test whether the loss of ISCs seen in the absence of Hairless activity was due to derepression of E(spl)-C genes, we examined the phenotype of E(spl)-C Hairless double-mutant clones. We found that deletion of the E(spl)-C, using either E(spl)-CΔmδ-m6 or the Df(3R) gro32.2 P[gro+], rescued the growth defect of Hairless mutant clones at 6 and 10 days AHS (Fig. 3B-B″′,D; see Fig. S3D-E″′ in the supplementary material; compare with Hairless single-mutant clones in Fig. 1E-H,J-J″′). The observation of large clones at 10 days AHS clearly indicated that Hairless E(spl)-C double-mutant ISCs retained their ability to self-renew over time (Fig. 3D and see Fig. S3D-D″ in the supplementary material). Consistent with this, ISC-like Delta+ cells were seen in Hairless E(spl)-C double-mutant clones (Fig. 3B-B″,E and see Fig. S3E-E″′ in the supplementary material). Quantification of Delta+ cells showed that HE31 E(spl)-CΔmδ-m6 mutant clones contain a similar proportion of Delta+ cells (83%, n=331 cells) as E(spl)-CΔmδ-m6 mutant clones (87%, n=983 cells) at 10 days AHS, a significant rescue of the 7% of Delta+ cells in Hairless mutants clones (n=202 cells; Fig. 3E). We note, however, that E(spl)-C single-mutant clones were larger than Hairless E(spl)-C double-mutant clones (Fig. 3C,D). We interpret this difference in clone growth to suggest the existence of additional Notch targets that act to limit self-renewal and are subject to derepression in the Hairless E(spl)-C mutant cells. Expression of these targets would therefore limit the growth of Hairless E(spl)-C double-mutant clones. Taken together, our data indicate that E(spl)-C genes promote EB differentiation and/or block ISC self-renewal in presumptive EB cells and that their inhibition in the ISC by Hairless-mediated repression prevents ISC loss.
E(spl)-C genes play a non-essential role in enteroendocrine differentiation
E(spl)-C mutant cells do not express the ee markers Pros [Fig. 4A-B″′,H and see Fig. S3 in the supplementary material; only 0.5% of mutant cells are Pros+ (n=622 cells) compared with 6.1% of cells in wild-type clones (n=294 cells)] and Allatostatin (data not shown). Expression of m7-VP16 produced similar phenotypes to those when E(spl)-C is lost (see Fig. S3F-F″ in the supplementary material). By contrast, EC cells were properly specified, as shown by both large nuclear size (>7 μm) and expression of the EC marker Pdm1 (Fig. 4B-D and see Fig. S3C in the supplementary material; note that Pdm1 appeared to also mark early ECs with a nucleus smaller than 7 μm). The density of large ECs in E(spl)-C mutant tissue was similar to that of wild-type tissue, whereas the density of Pdm1+ cells was slightly increased in the E(spl)-C mutant tissue (Fig. 4B-D). Thus, loss of E(spl)-C genes led to a large excess of Delta+ ISC-like cells, a normal density of differentiated EC-like cells and a loss of Pros+ ee-like cells. Despite the large increase in small Delta+ cells, no general hyperplasia of the tissue was observed (Fig. 4B-D). We conclude that genes of the E(spl)-C are important to limit ISC fate specification and to promote ee differentiation, but are dispensable for the production of ECs. Since Notch signaling is required for EC fate (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006), this implies that Notch targets other than those encoded by the E(spl)-C act to promote EC differentiation. Furthermore, whereas the E(spl)-C genes are important for ee differentiation, this role could be bypassed by reducing Delta signaling activity using a neuralized (neur) mutation. Indeed, expression of Pros was restored in neur E(spl)-C double-mutant clones (Fig. 4E-G″′), showing that E(spl)-C genes are not strictly required for ee specification.
Daughterless is required for ISC maintenance
A major function of the E(spl) bHLH repressors in both Drosophila and vertebrate development is to inhibit the activity of the bHLH Daughterless (Da)/E47-based dimeric transcriptional activators (reviewed by Kageyama et al., 2007; Alifragis et al., 1997; Gigliani et al., 1996; Heitzler et al., 1996; Oellers et al., 1994). Consistent with this, we found that Da is expressed in cells of the ISC lineage (see Fig. S4 in the supplementary material). To test the possible role of da in this tissue, we studied the growth of da mutant clones. We found that ISCs are lost in da10 (Fig. 5C-E) and da3 (data not shown) clones, as seen by loss of clone growth and loss of Delta+ ISC cells. da mutant clones consisted of single or pairs of polyploid EC-like cells, showing that da is important for ISC maintenance.
Da-dependent bHLH transcriptional activity is mediated through E-box motifs. To test whether Da regulates ISC-specific gene expression, we examined the expression of a known Da/proneural target (Reeves and Posakony, 2005), miranda (mira), using a reporter gene in which the mira coding sequence is replaced with nuclear GFP (mira-promoter-GFP; Materials and methods and see Fig. S6 in the supplementary material). We found that mira-promoter-GFP was specifically expressed in ISCs (Fig. 5F-G″). Moreover, mutation of the seven E-box motifs present in this construct largely eliminated ISC-specific gene expression (Fig. 5H-I″), indicating that E-box sites are necessary to drive such expression. In addition, multimerized E-boxes in front of a minimal promoter were sufficient to direct expression in ISCs, albeit weakly (see Fig. S5 in the supplementary material). We conclude that Da-binding motifs can mediate ISC-specific gene expression.
achaete-acute complex genes are required for ee, but not ISC, specification
Class I bHLH family members (Da/E47) can act as homodimers (Ellenberger et al., 1994; Murre et al., 1991; Oellers et al., 1994), as heterodimers with class II bHLH proteins (Powell and Jarman, 2008), and, at least in the Drosophila wing, evidence suggests that Da acts in concert with a Zn-finger transcription factor (Jafar-Nejad et al., 2006). Da/E47 members have a broad expression pattern, such that class II bHLH partners with a more restricted expression pattern often confer cell- or tissue-specificity to the heterodimeric complex. In the Drosophila embryo, Da is ubiquitous and acts with proneural bHLH factors of the achaete-scute complex (AS-C) that are expressed in adult midgut precursor cells to regulate their fate (Tepass and Hartenstein, 1995). Since Da expression was fairly general in the intestine, we similarly reasoned that a putative Da bHLH class II protein might be specifically expressed in the ISC to promote E-box-dependent ISC-specific expression, as seen for mira. To identify ISC-specific bHLH genes, we compared the transcriptional profile of wild-type posterior midguts with those mutant for Su(H) that contained a large excess of ISC-like cells and a mild increase in Pros+ ee-like cells (data to be presented elsewhere). From this analysis, two class II bHLH family genes were highly upregulated: scute (60-fold) and asense (22-fold). The Asense protein was detected in a subset of Pros+ ee cells (Fig. 6A-A″). Since Scute antibodies are no longer available, we used in situ hybridization to detect the scute RNA. scute transcripts were detected in Su(H) mutant clones that contain many ISC- and ee-like cells (Fig. 6B,B′), as well as in single basal cells with small nuclei in the heterozygous tissue outside of clones (Fig. 6C), suggesting that scute is expressed in the ISC, EB and/or ee.
We therefore investigated the role of scute and asense in ISC maintenance and ee differentiation using Df(1)scB57, which lacks all four AS-C genes (achaete, scute, lethal of scute and asense). No effect on clone growth or on Delta+ expression was observed (Fig. 6D-I). We conclude that the scute and asense genes are not required for ISC maintenance. The bHLH genes amos and atonal were similarly not essential for ISC maintenance (data not shown). Whether proneural proteins act redundantly in the maintenance of ISCs or, alternatively, whether Da acts as a heterodimer with a non-proneural bHLH protein or as a homodimer, remains to be studied. We next studied ee differentiation and found that Df(1)scB57 clones are devoid of Pros+ cells (Fig. 6F,I,J). These expression and genetic data indicate that asense and/or scute are necessary for ee differentiation. Conversely, overexpression in ISCs of asense or scute using esgGAL4 GAL80ts increased the number of Pros+ cells (Fig. 6J-O), indicating that asense and scute are sufficient to promote ee differentiation.
Adult stem cells self-renew and, at the same time, give rise to progeny that eventually differentiate. Our work provides evidence that one of the strategies used to maintain the identity of ISCs in Drosophila is to repress the expression of Notch target genes. Consistent with our finding, the loss of a general regulator of transcriptional repression, the Histone H2B ubiquitin protease Scrawny, gives a similar phenotype to Hairless (Buszczak et al., 2009). Additionally, several recent studies indicate that transcriptional repression of differentiation genes may be a central hallmark of stem cells in general (Dejosez et al., 2008; Jepsen et al., 2007; Liang et al., 2008; Maines et al., 2007; Pietersen and van Lohuizen, 2008).
Two models have been proposed for Hairless activity. One proposes that Hairless competes with NICD for interaction with Su(H), thereby preventing transcriptional activation of Notch target genes by low-level Notch receptor activation (Bang et al., 1995; Morel et al., 2001). A second, non-exclusive, model proposes that Hairless antagonizes the transcriptional activation of Notch target genes by tissue-specific transcription factors other than Notch (Barolo et al., 2002; Barolo et al., 2000; Castro et al., 2005). Since the loss of Su(H) can suppress the phenotype of Hairless on ISC clone growth, we propose that Hairless promotes ISC maintenance by repressing the transcription of genes that would otherwise be activated by Notch signaling in ISCs (Fig. 7). Thus, Hairless appears to set a threshold level to buffer Notch signaling in ISCs. In the absence of this repression, the expression of E(spl)-C genes and other Notch targets would lead to loss of the ISC fate. Importantly, our findings suggest a mechanism for how the transcriptionally repressed state is turned off and activation of the differentiation program is initiated: high activation of Notch in EBs displaces Hairless from Su(H) and leads to expression of the E(spl)-C genes (Fig. 7).
E(spl)-C bHLH repressors act in part through their ability to inhibit bHLH activators (Kageyama et al., 2007). Our data demonstrate that Da is also essential to maintain ISC fate and that E-box Da-binding sites are required to promote ISC-specific enhancer activity. Thus, we propose that activation of E(spl)-C genes by Notch in EBs downregulates Da bHLH activity and thereby contributes to turning off ISC identity in the differentiating cell (Fig. 7). The specificity of ISC-specific E-box expression might be due to the ISC-specific expression of a bHLH family member. Although our array analysis raised the possibility that Scute may be specifically expressed in ISCs, our genetic analysis indicates that scute function is not essential for ISC maintenance. Alternatively, specificity of gene expression might result from inhibition of bHLH activity in the EB and differentiating daughters, possibly by E(spl)-bHLH factors, rather than by the ISC-specific expression of a Da partner. It is also possible that a non-bHLH, ISC-specific factor restricts the Da-dependent bHLH activity to ISCs in a manner similar to the synergism observed in wing margin sensory organ precursors (SOPs) between the Zn-finger transcription factor Senseless and Da (Acar et al., 2006; Jafar-Nejad et al., 2006).
Recently, a role for the Da homologs E2A (Tcf3) and HEB (Tcf12) has been found in mammalian ISCs marked by the expression of Lgr5 and, in this context, E2A and HEB are thought to heterodimerize with achaete-scute like 2 (Ascl2), which is essential for the maintenance and/or identity of Lgr5+ ISCs (van der Flier et al., 2009). In Drosophila, however, AS-C genes are not essential for ISC maintenance, but appear to play a role in enteroendocrine fate specification. The observation that Da bHLH activity is required for the identity of both Drosophila ISCs and mammalian Lgr5+ ISCs suggests that there might be conservation at the level of the gene expression program. Additionally, the bHLH genes Atoh1 (Math1) and Neurog3 are both important for differentiation of secretory cells in the mammalian intestine (Lee et al., 2002; Yang et al., 2001). Clearly, further analysis of the control of Da/E2A bHLH activity, as well as of the gene networks downstream of Da/E2A, will be of great interest.
Our data suggest that ISC fate is promoted both by inhibition of Notch target genes through Hairless/Su(H) repression and by activation of ISC-specific genes through bHLH activity. How then is asymmetry in Notch activity eventually established between the two ISC daughters to allow one cell to remain an ISC and one cell to differentiate? We can envisage three types of mechanism that would allow for asymmetry of Notch signaling.
First, the binary decision between the ISC and EB fates might result from a competition process akin to lateral inhibition for the selection of SOPs (Heitzler and Simpson, 1991). In this process, feedback loops establish directionality by amplifying stochastic fluctuations in signaling between equivalent cells into a robust unidirectional signal. Our finding that the Da activator and E(spl)-bHLH repressors are important to properly resolve ISC/EB fate is consistent with this type of model. Activation of the Notch pathway in one of the daughter cells could then lead to the changes in nuclear position previously noted (Ohlstein and Spradling, 2007).
Second, the asymmetric segregation of determinants could bias Notch-mediated cell fate decisions. The cell fate determinants Numb and Neur are asymmetrically segregated in neural progenitor cells to control Notch signaling (Bardin et al., 2004; Knoblich, 2008). However, we find no evidence for the asymmetric segregation of these proteins in dividing ISCs (A.J.B., unpublished). Additionally, our data indicate that Numb is not important to maintain ISC fate. We cannot exclude, however, the possibility that another, unknown Notch regulator is asymmetrically segregated to regulate the fate of the two ISC daughters.
A third possibility is that after ISC division, one of the two daughter cells receives a signal that promotes differential regulation of Notch. Indeed, it has been noted that the axis of ISC division is tilted relative to the basement membrane, resulting in one of the progeny maintaining greater basal contact than the other (Ohlstein and Spradling, 2007). An extracellular signal coming either basally or apically could bias the Notch-mediated ISC versus EB fate decision. For instance, Wg secreted by muscle cells could act as a basal signal to counteract Notch receptor signaling activity in presumptive ISCs (Lin et al., 2008). This could be accomplished by Wg promoting bHLH activity or gene expression. Indeed, Wg has been demonstrated to promote proneural bHLH activity in Drosophila (Couso et al., 1994; Phillips and Whittle, 1993; Tomoyasu et al., 1998).
These models are not mutually exclusive, however, and proper control of ISC and differentiated cell fates during tissue homeostasis might involve multiple mechanisms.
We thank S. Artavanis-Tsakonas, S. Bray, C. Cronmiller, C. Delidakis, S. Hayashi, P. Heitzler, J.-R. Huyhn, D. Ish-Horowicz, Y.-N. Jan, A. Jarman, T. Klein, A. Preiss, J. Skeath, S. Younger, the Developmental Studies Hybridoma Bank (DSHB, under the auspices of NICHD, University of Iowa, Iowa City) and the Bloomington Stock Center for providing flies and antibodies; K. Mazouni and L. Couturier for the initial characterization of mira expression in ISCs; and C. Goridis, C. Brou, P.-A. Defossez, J.-E. Gomes, J.-R. Huyhn, P. Mourikis and laboratory members for helpful discussion and comments on the manuscript. This work was funded by the CNRS, the Ecole Normale Supérieure, the Pasteur Institute, the ARC (grant #3856 to F.S.), the ANR (grant BLAN08-2_371836 to F.S.), a LNC Comité de Paris grant to A.J.B. and by a Wellcome Trust Programme Grant (068055) to A.H.B. C.N.P. was supported by a FCT fellowship; T.D.S was supported by a Herchel Smith Fellowship. Deposited in PMC for release after 6 months.
Competing interests statement
The authors declare no competing financial interests.
Supplementary material for this article is available at http://dev.biologists.org/lookup/suppl/doi:10.1242/dev.039404/-/DC1
- Accepted December 30, 2009.
- © 2010.