The process through which multiple types of cell-lineage-restricted progenitor cells are specified from multipotent stem cells is unclear. Here we show that, in intestinal stem cell lineages in adult Drosophila, in which the Delta-Notch-signalling-guided progenitor cell differentiation into enterocytes is the default mode, the specification of enteroendocrine cells (EEs) is initiated by transient Scute activation in a process driven by transcriptional self-stimulation combined with a negative feedback regulation between Scute and Notch targets. Scute activation induces asymmetric intestinal stem cell divisions that generate EE progenitor cells. The mitosis-inducing and fate-inducing activities of Scute guide each EE progenitor cell to divide exactly once prior to its terminal differentiation, yielding a pair of EEs. The transient expression of a fate inducer therefore specifies both type and numbers of committed progenitor cells originating from stem cells, which could represent a general mechanism used for diversifying committed progenitor cells from multipotent stem cells.
Homeostatic renewal of tissues and organs commonly depends on the resident tissue-specific stem cells, which have the ability to self-renew and produce diverse committed progenitor cells that differentiate into multiple cell types. The generation of lineage-restricted progenitor cells can be actively specified by asymmetric cell division, in which cell fate determinants are specifically segregated into one of two stem cell daughter cells1. Alternatively, committed progenitor cells can be specified passively; in this case, stem cells differentiate as they physically depart from the niche environment. Examples are Drosophila germline stem cells2,3 and stem cells in the mouse small intestine4,5. However, very little is known about the molecular mechanisms by which distinct, lineage-restricted progenitor cells are generated from a common stem cell pool.
Multipotent intestinal stem cells (ISCs) in the Drosophila posterior midgut can differentiate into two distinct cell lineages: absorptive enterocytes (ECs), which have been shown to occur from ~90% of ISC divisions, and secretory enteroendocrine cells (EEs), which occur from ~10% of ISC divisions6,7,8. A typical ISC division generates a new ISC and a post-mitotic enteroblast (EB), the lineage-committed progenitor cell. ISCs produce the membrane-bound ligand Delta (Dl), which activates its receptor Notch in EBs. Notch activation results in unidirectional differentiation of EBs into ECs6,7,8. The process of EE specification, however, is much less well understood. Recent studies have suggested that EEs are directly differentiated from ISCs, implying that the decision of EE specification may occur at the stem cell level in ISCs9,10, but how this occurs remains unclear. A recent study indicates that, in ~90% of cases, EEs are differentiated directly from ISCs in a process that compromises stemness: an ISC divides once, yielding a pair of EEs10. This model is in line with the observation that EEs are typically generated in pairs6, but it implies that EE generation occurs at the expense of continued existence of ISCs. In view of stem cell maintenance, this seems to be a risky developmental strategy.
Previous studies have demonstrated that the transcription factor Prospero (Pros), which is specifically expressed in EEs, acts as an EE-fate-determination factor9,10,11. The expression of Pros is promoted by the acheate-scute complex (AS-C) genes, and the expression of AS-C genes scute (sc) and asense (ase) is normally suppressed by the transcriptional repressor Ttk6911. Loss of AS-C genes compromises EE generation without affecting EC generation from ISCs12,13. Moreover, overexpressing either Sc or Ase induces increased EE generation from ISCs12. However, how Sc is regulated to control EE specification is unclear.
Here, we investigated the regulation of Sc during the process of EE generation, and our results allow us to propose a model through which discrete committed progenitor cell types are generated from stem cells.
An EE-regeneration model reveals that ISCs self-renew during generation of EE pairs
To better understand the process of EE specification in ISCs, we set up an EE regeneration assay and examined de novo EE regeneration. Conditionally depleting sc in Esg-GAL4ts, upstream activation sequence (UAS)-sc-RNAi flies by shifting flies to restrictive temperature (RT) (29 °C) from pupal stage caused complete EE depletion in the midgut of newly eclosed flies (Fig. 1a and Supplementary Fig. 1), as previously reported13. Shifting flies to RT after eclosion also completely disrupted new EE generation, although some EEs produced from the pupal stage remained (Supplementary Fig. 1). Strikingly, three to four days after shifting EE-less flies back to 18 °C, EEs, marked by Pros, appeared concurrently in many regions of the midgut (Fig. 1b). Interestingly, many cells were co-stained with Dl (membrane) and Pros (nucleus), and this was further confirmed by separate fluorescent labelling for Dl and Pros (Fig. 1b,c). These Dl+ Pros+ cells are probably the differentiating EEs described previously in normal midgut9,10. Similar to what has been observed in normal midgut6, ~71% of Pros+ cells appeared in pairs (Fig. 1d). The EE pair is probably generated by cell division of a mother cell, as two EEs in the pair appeared simultaneously and mitotic Pros+ cells could be observed (Fig. 1c). After carefully counting the total number of Dl+ cells in the epithelium before and after EE appearance, we found that the number of Dl+ cells remained constant, although Dl expression was downregulated during EE appearance (Fig. 1b,e). In addition, there was always a Dl+ or Dllow cell next to each newly formed EE or EE pair (Fig. 1b). These observations suggest that, instead of directly differentiating into an EE as proposed previously10, an ISC actually divides first to generate a new EE progenitor cell (EEP), which then undergoes one round of cell division before its terminal differentiation, yielding an EE pair (Fig. 1f). Interestingly, like in normal midgut6,14, the two EEs in the newly generated EE pair produce different peptide hormones (Supplementary Fig. 2). The EEP is reminiscent of the EE mother cell in the pupal midgut, which is also capable of cell division prior to terminal differentiation15.
Sc is dynamically expressed in ISCs
Among AS-C complex genes, sc is both necessary and sufficient for EE specification12,13. It is therefore important to understand how sc expression is regulated. Using the CRISPR-Cas9 technique, we generated a Sc-GFP knock-in line in which green fluorescent protein (GFP) was fused to 3′ of the Sc coding region (Fig. 2a). The expression pattern of Sc-GFP in larval imaginal discs was virtually identical to the reported sc mRNA and protein expression patterns (Supplementary Fig. 3)16. In adult midgut, albeit variable in numbers among different midguts, Sc-GFP (GFP fluorescence)-expressing cells were observed only in a subset of Dl+ cells (Fig. 2b,c). By quantification, on average, ~15% of Dl+ cells had distinguishable Sc-GFP expression in both posterior (15.6 ± 2.0% n = 32 guts) and anterior (15.5 ± 5.9%, n = 12 guts) midguts. Immunostaining with anti-GFP, which greatly amplified the Sc-GFP signal, revealed that Sc-GFP could be observed in virtually all Dl+ cells (Supplementary Fig. 3). Therefore, Sc is dynamically expressed in ISCs, with a weak expression level in most ISCs, and increased expression levels in a small subset of ISCs.
Transient Sc activation induces EE generation from ISCs
From 23 (upstream and downstream) glass multiple reporter (GMR) enhancer-GAL4 lines generated for sc (ref. 17) (Fig. 2d), we identified one line, GMR14C12, that drove UAS-GFP expression in some diploid cells in the midgut epithelium (Fig. 2e,f). The density, distribution and individual variability of the GFP+ cells were largely similar to those of Sc-GFP+ cells, suggesting that this reporter line is driven by the enhancer element for sc expression in the midgut. Co-staining with cell markers revealed that GMR14C12>GFP was mainly expressed in 15 ± 0.69% Dl+ cells (n = 26 guts), but residual GFP was also often present in the newly formed EEs in the same cell nest (Fig. 2f,g). Interestingly, the GFP+ Dl+ cells commonly showed relatively low levels of Dl expression compared to the GFP− Dl+ cells (Fig. 2e,f,h). Interestingly, ISCs that are producing EEs are known to be associated with low levels of Dl expression6. We also compared the expression pattern between Sc-GFP and GMR14C12-GAL4, UAS-RFP (red fluorescent protein) (GMR14C12>RFP). Similarly, GMR14C12>RFP+ cells were found in small cell nests of two to three cells. About half of them contain Sc-GFP+ cells (Supplementary Fig. 4), suggesting that the GMR14C12->RFP reporter only partially recapitulates Sc-GFP expression in ISCs. This is probably because Sc is expressed transiently and the expression of GMR14C12->RFP is relatively delayed due to the binary expression system used and/or extra time required for RFP maturation. Similar to the GMR14C12>GFP cell nest, many GMR14C12->RFP cell nests also contained newly formed EEs (Supplementary Fig. 4). The residual GMR14C12->RFP or GMR14C12->GFP expression found only in the newly formed EEs but rarely in any ECs indicates that the ISCs with GMR14C12-GAL4 expression only generate EEs. To test this hypothesis, we performed cell lineage tracing studies (Fig. 2i) with flies carrying GMR14C12-GAL4, Tub-Gal80ts; UAS-flp, Tub-Gal80ts; and the flp-out cassette (Act<stop<lacZ)18. This analysis revealed that the immediate daughter cells (at day 2 after labelling) of GMR14C12>GAL4+ ISCs were mainly EEs. However, EC daughter cells appeared at 7 days after labelling (Fig. 2j–l), suggesting that transient Sc upregulation in ISCs primes them to generate EEs, and ISCs resume EC production once Sc expression is downregulated.
Notch mutant ISCs have increased Sc expression and only give rise to EEs
It has been shown that loss of Notch (N) in ISCs causes the formation of Dl+ and Pros+ cell tumours7,8. We found that the majority of Dl+ cells in the tumour showed increased expression of Sc-GFP (Fig. 3a). Interestingly, Sc-GFP protein level was transiently downregulated at metaphase of mitosis (Fig. 3a′). In agreement with a recent study19, the EE-like tumour cells were post-mitotic cells, in contrast to the ISC-like tumour cells (Fig. 3b,c). In addition, BrdU pulse and chase experiments revealed that Dl+ cells could be readily labelled with BrdU, but by chase, the BrdU signal in Dl+ cells quickly disappeared and reappeared in Pros+ cells (Fig. 3d,e), suggesting that the clustered Pros+ cells are derived from Dl+ cells. Consistent with this notion, there are many Dl+ Pros+ cells at the boundary between Dl+ and Pros+ cell clusters (Fig. 3f), and these are probably the cells in the process of EE differentiation. The Dl+ cells in N mutant tumour are therefore unipotent: they only give rise to EEs (Fig. 3g). This property of N mutant Dl+ cells further supports the notion that increased Sc expression primes ISCs to produce EE progenies.
Sc functions as both a mitogenic factor and a cell-fate inducer
To further understand the function of sc, we conditionally induced sc expression in ISCs and monitored the cellular events in a time-lapse. Transient sc induction caused a rapid cell division response (Fig. 4a,b), revealing that Sc is a potent inducer of cell division. As Sc is known as an EE fate regulator by inducing Pros expression, we examined whether continuous expression of sc could cause ISC differentiation into EE. Continuous sc expression for a week resulted in dramatic accumulation of Pros+ EEs in the epithelium (Fig. 4c and Supplementary Fig. 5). Cell lineage tracing studies confirmed that sc-overexpressing ISCs only generate EEs (Supplementary Fig. 5). After continuous sc overexpression in ISCs for 3 weeks, we began to observe regional ISC loss at the anterior midgut (Fig. 4d). Some residual Esg>GFP+ cells co-expressed Pros, indicating that the ISC loss is due to differentiation into EEs. Therefore, although short-time expression of sc is not, prolonged activation of sc is detrimental to ISC self-renewal/maintenance.
Pros is also known as a potent cell cycle inhibitor20,21. Indeed, in contrast to the effect of Sc, conditionally inducing Pros expression immediately caused ISCs to exit the cell cycle (Fig. 4b). Sc is thus both a mitogenic factor and an EE-cell-fate inducer, whereas Pros is both a cell cycle inhibitor and an EE-cell-fate determination factor. These properties underline a precisely regulated circuitry that directs formation of a pair of EEs from each EEP, which will be described below.
Process of sc-overexpression-induced EE specification in ISCs
To understand how EEs are generated from ISCs, we examined the cellular events in a time window (24–72 h post sc induction) when new EEs just begin to be generated. To do this, we expressed a UAS-RedStinger reporter in the overexpression system. RedStinger is relatively stable and can serve as a lineage marker to trace the progeny of the originally marked ISCs. The number of cell divisions of the initially labelled ISCs could be deduced based on the total number of cells present within each clone. The first cell division following sc overexpression occurred in ISCs (PH3+ in a one-cell clone, Fig. 5a), and during mitosis, Pros expression was marginally detectable on the ISC membrane (Fig. 5a). Differential Pros expression was first observed in cells at telophase, where only one of the two daughter cells retained membrane and cytoplasmic Pros expression (Fig. 5b,f), indicating that the cell division is asymmetric. By contrast, Dl was still evenly expressed in the two daughter cells (Fig. 5f). The Dl+ Prosweak daughter is probably the EEP, which seems to enter mitosis rapidly, as the second cell division (one PH3+ cell in a two-cell clone, Fig. 5c) following sc overexpression always happened in these cells, before the original ISC re-enters mitosis (in a three-cell clone, Fig. 5e). During mitosis, EEPs began to have strong and punctate nuclear Pros expression (Fig. 5c,d,g), indicative of the initiation of EE commitment.
Collectively, these observations suggest a stepwise model for EE regeneration from ISCs: transient activation of sc expression induces asymmetric cell division, which generates a new ISC and an EEP. Sc also induces Pros expression, but the nuclear translocation and transcriptional activity of Pros can only occur in EEPs, not ISCs. Residual Sc activity in the newly formed EEP then induces one round of cell division before the accumulation of nuclear Pros that causes the cell cycle exit and EE commitment. In this manner, precisely a pair of EEs is generated from each EEP (Fig. 5h).
Two regulatory feedback loops control transient Sc activation in ISCs
It is intriguing how this transient activation of Sc is achieved. A feedback mechanism via Slit-Robo2 signalling has been implicated in regulating EE generation9,22, but we found that the Slit-Robo2 signalling does not affect Sc expression in ISCs (Supplementary Fig. 3). Previous studies on early Drosophila development have suggested reciprocal regulatory relationships between AS-C genes and the enhancer of split complex (E(spl)) genes, which are known as the Notch target genes23,24,25,26,27. To test whether similar mechanisms occur in ISCs, we screened a number of lacZ, GAL4 or GFP reporters or candidate reporters for individual E(spl) genes28. m3>GFP and mβ-lacZ showed patterns similar to the general Notch activation reporter NRE-lacZ, which is specifically expressed in EBs. Interestingly, m8-lacZ showed a very weak, but similar expression pattern to Sc (Fig. 6a–c): it was weakly detectable in a small subset of diploid cells. Moreover, m8-lacZ+ cells were largely overlapped with GMR14C12>GFP+ cells (Fig. 6a,b). In addition, the majority of E(spl)m8-lacZ+ ISCs were Dllow (Fig. 6b), again similar to the profile of GMR14C12>GFP+ cells (Fig. 2h). This indicates a direct regulatory relationship between Sc and E(spl)m8. Indeed, transient overexpression of sc in ISCs led to robust upregulation of m8-lacZ expression in all progenitor cells (Fig. 6c–e). In addition, ectopic expression of sc in ECs using MyoIA-GAL4 also led to ectopic m8-lacZ expression in ECs (Fig. 6f). Notably, co-expressing Notch-RNAi did not prevent the upregulation of m8-lacZ expression caused by sc overexpression (Fig. 6g,h), suggesting that E(spl)m8 expression is independent of Notch activity in ISCs.
To further characterize the regulatory relationship between sc and E(spl) genes, we transiently overexpressed sc in ISCs by Dl-GAL4 and sorted out Dl>GFP+ cells for RNA-seq analysis. From this analysis, we found that, in addition to m8, many other E(spl) genes, such as m4, m6, m7, mγ and mδ, were significantly upregulated as well (>100-fold) upon sc overexpression (Fig. 6i). The analysis also revealed that sc overexpression strongly induced the expression of multiple cell cycle-promoting genes, such as stg, CycA, CycB and CycE (Fig. 6i), further supporting that Sc is a potent mitogenic factor. By analysing previously reported ChIP-seq data for Sc in ISCs29, we found that Sc could bind to the enhancer regions of multiple E(spl) genes, including mγ, m6, m7 and m8 (Supplementary Fig. 6), suggesting that Sc directly regulates their transcription. In the enhancer sequence for m8, we identified two consensus E-box sites (CANNTG) for proneural factors, named here E-box #1 (E1) and E-box #2(E2) (Supplementary Fig. 6). Mutation of E1 severely dampened lacZ reporter expression in response to sc induction, while mutation of E2 only mildly (if at all) reduced the reporter expression (Supplementary Fig. 6), suggesting that the E1 site is a major Sc-binding site in the E(spl)m8 enhancer.
To study the effect of E(spl) genes on sc, we found that transient overexpression of E(spl)m8 rapidly eliminated sc-GFP expression in all ISCs (Fig. 7a,b), and also rapidly suppressed Dl expression in ISCs (Fig. 7a,b). This explains the decreased Dl expression in ISCs during EE regeneration as well as in GMR14C12>GFP+ cells and m8-lacZ+ cells. To determine whether or not E(spl)m8 suppresses sc expression directly, we expressed E(spl)m8-Dam fusion gene in ISCs, and performed DNA adenine methyltransferase identification (DamID) analysis. Interestingly, we observed a strong and specific enrichment peak at the GMR14C12 region (Fig. 7e), suggesting that E(spl)m8 is able to directly bind to the enhancer of sc in ISCs. The direct two-way regulation between Sc and E(spl)m8 further supports the notion that Sc and E(spl)m8 (plus other E(spl) proteins) form a negative feedback regulatory loop in ISCs, which may explain the transient activation pattern of Sc in ISCs.
To test the functional requirement of E(spl) genes in suppressing sc expression, we generated MARCM clones mutant for Df(3R) E(spl) b32.2 gro+, a deletion and transgene combination in which all of the E(spl) genes have been removed30. Consistent with previous observations12, the mutant clones contained increased number of Dl+ cells, and further properly differentiated ECs, but not EEs (Supplementary Fig. 3). We found that there was no obvious upregulation of sc-GFP expression in any of the ISC-like cells within the Df(3R) E(spl) b32.2 gro + mutant clones (25 clones examined, corresponding to 416 Dl+ cells). One likely possibility is that, in addition to the E(spl) genes, there are other Notch downstream target genes that function to promote EC differentiation, and that participate in negative feedback regulation with sc. Conceivably, the loss of E(spl) genes may possibly cause compensatory upregulation of such unidentified genes in a manner that prevents the activation of AS-C genes, and that primes all of the undifferentiated cells to adopt an EC fate.
Sc has been reported to transcriptionally self-stimulate itself, which acts as an essential mechanism for proneural protein accumulation during sensory organ development25,31. To test whether this also occurs in ISCs, we constructed LacZ reporter for sc using the GMR14C12 enhancer fragment. GMR14C12-lacZ was barely detectable in WT midgut epithelium. However, overexpression of sc (2 days at RT) by Dl-GAL4 effectively induced lacZ reporter expression in many diploid cells in the epithelium, including ISCs, presumptive EEPs, and newly formed EEs (Fig. 7c,d). The Sc ChIP-seq data analysis also revealed two Sc binding peaks within the GMR14C12 region (Fig. 7e). Together, these data suggest that Sc is able to stimulate its own transcription directly by binding to the sc enhancer, and this self-stimulation may allow Sc to gradually accumulate to a relatively high level before being shut down via feedback regulation.
Collectively, these data suggest a model to explain the dynamic Sc expression in ISCs. Sc is usually expressed at low levels, and E(spl) genes are not expressed. Through transcriptional self-stimulation, Sc gradually builds up and eventually reaches a level that is sufficient to induce EEP generation. Concomitantly, the high level of Sc also induces E(spl) genes. These induced E(spl) proteins and other unidentified Notch targets then function as a negative-feedback mechanism that rapidly turns sc expression back to the baseline level. The expression of E(spl) genes is consequently downregulated, and the expression of both E(spl) proteins and Sc then returns to the default states (Fig. 7f).
The production of EEs from pupal ISCs (pISCs) occurs at ~48–72 h after pupal formation (APF) (Supplementary Fig. 7)15,32,33,34,35, and we found that transient expression of Sc directs EE generation in pupal ISCs as well.
This study suggests a simple model to describe how ECs and EEs are respectively generated from ISCs: Delta-Notch-signalling-guided EC generation from ISCs acts as the default mode, while transient Sc expression triggers EE generation from ISCs (Fig. 7f).
The finding that a cell-fate inducer is expressed in stem cells to induce progenitor cell specification is unexpected, as expression of a cell-fate inducer could potentially compromise stem cell maintenance by promoting differentiation. As we have demonstrated, overexpressing sc in ISCs for a prolonged time is able to deplete ISCs by forcing their differentiation. Under normal conditions, however, Sc expression is only transiently increased. In addition, although Sc is sufficient to induce Pros expression in ISCs, Pros protein is only marginally detectable, strictly localized to the cell membrane and/or cytoplasm of ISCs. This indicates that there are additional mechanisms through which Pros activation is prevented in ISCs, such as by destabilizing Pros or preventing Pros from entering the nucleus, mechanisms that are utilized in neural stem cell lineages36,37. Therefore, transient expression of a cell-fate inducer is a seemingly risky but rather safeguarded mechanism by which lineage-committed progenitors are generated from stem cells.
Sc is dynamically expressed in ISCs, and our study suggests that this dynamic expression pattern is driven by a coordinated action of two feedback loops: a transcriptionally self-stimulatory loop of Sc and a negative feedback loop between Sc and E(spl) proteins. As the depletion of the entire E(spl) genes does not lead to continuous activation of Sc, it is likely that other Notch downstream targets or other mechanisms could also participate in this negative feedback loop. For instance, a Sina-Phyl-Ttk69 complex has recently been implicated in a feedback loop with Sc to regulate EE differentiation38. Given that the negative feedback is a common mechanism underlying biochemical oscillations39, these feedback interplays could provide a mechanism that drives oscillatory expression of Sc in ISCs. This hypothesis is attractive, because Sc oscillation could potentially serve as an internal timer for periodic production of EEs from ISCs. Future cellular and molecular analysis combined with in vivo live imaging should help to further test and refine this oscillation model, and to understand whether and how this internal timer is regulated with environmental cues.
Lineage-committed progenitor cells usually transiently amplify themselves prior to terminal differentiation. For instance, the transit-amplifying progenitor cells in mammalian small intestine usually divide four to five times before terminal differentiation40. Similarly, each ganglion mother cell in Drosophila neuroblasts divides exactly once prior to terminal differentiation, which yields a pair of neurons or glial cells41, akin to the EE pair generation from an EEP reported here. We find that Sc is both a mitogenic factor and a cell-fate inducer. Pros, on the other hand, is a known inhibitor of cell division and a cell differentiation factor42. Because of these properties, the net result of the diminishing and accumulating activity of Sc and Pros, respectively, guides each EEP to divide exactly once prior to terminal differentiation, yielding an EE pair (Fig. 7f). Our study therefore also demonstrates a mechanism that determines the number of transient amplifications of committed progenitor cells.
In summary, our study demonstrates that the transient expression of a fate inducer specifies both the type and the cell division number of committed progenitor cells from multipotent stem cells. Recent studies on progenitor cells in mammalian tissues have found that the expression of fate-inducing transcription factors is frequently detected in self-renewing stem cells43,44,45, indicating that this seemingly risky but safely guarded mechanism reported here could represent a general principle for producing several distinct types of committed progenitor cell from the same pool of multipotent stem cells.
Female flies aged 4–7 days were selected for the experiments, unless otherwise noted. Animal care and use followed the institutional guidelines of the National Institute of Biological Sciences (NIBS), Beijing, and the authors affirm that the study is compliant with all relevant ethical regulations regarding animal research.
The following stocks were used in this study: UAS-sc-RNAi (BDSC, #26206)46; Esg-GAL4,UAS-GFP (gift from S. Hayashi); UAS-RedStinger (BDSC, #8547); Dl-GAL4 (gift from X. Zeng and S. Hou)47; GMR14C12-GAL4 (BDSC, #48607)48; UAS-RFP; UAS-Notch-RNAi (BDSC, #7078); UAS-Sc (BDSC, #26687); lea2 (BDSC, #3102); m8-lacZ (BDSC, #26786); UAS-E(spl)m8-DamID (gift from F. Schweisguth)49; Df(3R) E(spl) b32.2 gro+50. The Scute-GFP knock-in line was generated by Cas9- mediated gene knock-in using a previously described protocol51. The GMR14C12-LacZ reporter line was generated by inserting the GMR14C12 DNA fragment into the multiple cloning site of C4PLZ vector, followed by P-element-mediated transformation. The m8-nlacZ reporter construct was generated by inserting the enhancer region of E(spl)m8 into the multiple cloning site of C4PLZ vector. The E1 or E2 site mutant m8-nlacZ reporter constructs were generated by site-directed mutagenesis, and in both cases, the E-box sequence ‘CANNTG’ was mutated to ‘AANNGG’.
Midgut samples were fixed and immunostained as previously described52. Briefly, midguts were dissected into cold Grace’s media, and fixed in a mixture of 4% formaldehyde in PBS and n-heptane in an equal volume for 30 min. The lower phase of the mixture was then replaced with methanol, followed by vigorous shaking for 30 s. The samples were then washed once with 100% methanol for 5 min, followed by gradual rehydration in PBT (PBS with 0.1% Triton X-100). For staining with anti-GFP anybody, the fixation was done by incubating midguts with 4% formaldehyde in PBS for 30 min. Samples were then incubated in block (5% normal goat serum in PBT) for 1 h. Samples were stained with primary antibodies overnight at 4 °C, washed with PBT, and incubated with secondary antibodies for 2 h at room temperature or overnight at 4 °C. Samples were washed with PBT and mounted in 70% glycerol. The primary antibodies used in this study include mouse anti-Dl (Developmental Studies Hybridoma Bank, Iowa (DSHB), 1:100 dilution); mouse anti-Pros (DSHB, 1:300); rabbit anti-phospho-Histone H3 antibody (Cell Signaling Technology, 1:500); rabbit anti-GFP (Cell Signaling Technology, 1:200); rabbit polyclonal anti-lacZ antibody (Cappel, 1:6,000); rabbit anti-Pros (gift from Y.-N. Jan, 1:1,000); mouse anti-Allatostatin A (DSHB, 1:300); rabbit anti-tachykinin antibody (gift from D. Nassel, 1:300); rat anti-BrdU (Abcam, 1:200); and mouse anti-Cut (DSHB, 1:20). The secondary antibodies used in the study were goat anti-rabbit, anti-rat or anti-mouse IgGs-conjugated to Alexa (568 or Cy5) (Molecular Probes, 1:300). Images were acquired on a Zeiss LSM510 or LSM880 inverted confocal microscope. All images were assembled in Adobe Photoshop and Illustrator.
Binary GAL4/UAS system and mosaic analysis
The temporal and regional gene expression targeting (TARGET) system, which utilizes the binary GAL4/UAS system and a temperature-sensitive GAL80, was used for spatial and temporal control of transgene expression in fly midgut53,54. Crosses were carried out at 18 °C unless otherwise stated. Two- to three-day-old female progenies were then cultured at 29 °C with regular corn meal topped with fresh yeast paste. Flies were provided with new food every two days until analysis. For EE depletion and regeneration experiment, flies were transferred to 29 °C at L3 stage until eclosion. Female progenies with the desired genotype were cultured with regular corn meal with yeast paste for an additional 3 days. Half of the flies were dissected for EE depletion analysis, and the other half were shifted back to 18 °C and cultured with regular corn meal with yeast paste for another 3 days before analysis. Mosaic analysis with a repressible cell marker (MARCM) was used to generate GFP- or RFP-marked wild-type or mutant clones in the midgut epithelium55. Clones were induced by a 1 h heat-shock treatment in a 37 °C running water bath.
Flies were fed on standard corn meal supplemented with 2 mg ml−1 BrdU (Sigma-Aldrich) for 2 h. Half of the flies were immediately dissected and fixed for analysis (pulse BrdU), and another half were transferred to standard corn meal without BrdU supplement and cultured for an additional 4 days before analysis (BrdU chase). Midguts were dissected in Grace’s medium and fixed in 4% formaldehyde in PBT for 30 min, followed by DNase I treatment for 15 min at 37 °C. The subsequent antibody staining steps were as described above.
Fluorescence-activated cell sorting (FACS) and RNA extraction
RNA-seq analysis of ISCs was performed according to a previously decribed method56,57,58. Briefly, 50 female guts were dissected into ice-cold diethyl pyrocarbonate (DEPC)-treated water-PBS, and incubated with 1 mg ml−1 Elastase (Sigma, cat. no. E0258) for 1 h at 25 °C, during which the sample was softly mixed every 15 min by pipetting and inverting four to six times. Dissociated samples were pelleted at 400g for 20 min at 25 °C, resuspended in 0.5 ml ice-old DEPC-PBS, filtered with 40 μm filters (BD Falcon) and sorted using a FACS Aria II sorter (BD Biosciences). GFP+ cells in the midgut of Dl-GAL4,UAS-GFP flies were sorted out, using w1118 midgut to set the fluorescence gate. For each of the three biological replicates, about 10,000 ISCs labelled with GFP were sorted, and RNA was harvested using the Arcturus PicoPure RNA isolation kit (Applied Biosystems) based on the manufacturer’s protocol. The RNA was then amplified using an Arcturus RiboAmp HS PLUS RNA amplification kit (Applied Biosystems).
Single-end reads were mapped to the Drosophila melanogaster genome (Release 6) using STAR (v020201). Each sequencing experiment generated an average of 34.3 million raw reads, and 92% was uniquely mapped for each experiment. Gene expression was quantified, normalized and analysed as previously described58.
In total, 1 µg RNA was used for cDNA construction for each sample. cDNA libraries were constructed using NEBNext DNA library prep master mix set (New England Biolabs, cat. no. E6040L) and NEBNext multiplex oligos (New England Biolabs, cat. nos. E7335S/E7500S). Libraries were qualified on an Agilent 2100 Bioanalyzer using an Agilent high-sensitivity DNA kit (Agilent Technologies, cat. no. 5067-1513), and then quantified using a Qubit dsDNA HS assay kit (Thermo Fisher Scientific, cat. no. Q32851) and an Illumina library quantification kit (Kapa Biosystems, cat. no. KK4824). All steps were performed according to the manufacturer’s protocols. Sequencing was performed on an Illumina Hiseq-2500 sequencing system with 50 bp read length.
Bioinformatics analysis for RNA-seq data
Single-end reads were aligned against the D. melanogaster genome (Release 6) using STAR (v020201). Each sequencing experiment generated an average of 34.3 million raw reads, and 92% was uniquely mapped for each experiment. Gene expression was quantified by the number of reads that fall into the exons. The results were normalized to RPKM (reads per kilobase of exon model per million mapped reads) using Cufflinks (v2.2.1). Differentially expressed genes were identified by cuffdiff, and significantly differentially expressed genes were filtered with P values ≤0.05 and fold change of ≥2.
DamID was carried out according to a previously described method59. Briefly, the lines of Dl-GAL4, Gal80ts, UAS-Dam, and Dl-GAL4, Gal80ts, UAS-Dam-E(spl)m8, were used to harvest midgut two days after transgene expression at the adult stage. About 60 midguts were collected for gDNA isolation, then followed by DpnII and alkaline phosphatase treatment. The treated samples were digested with DpnI and ligated with adaptor for ligation-mediated polymerase chain reaction (LM-PCR). Following PCR, the amplified DNAs were treated with T7 exonuclease and purified for sequencing following fragmentation.
ChIP-seq reads were aligned using Bowtie (version 1.1.2) to build version BDGP6 of the D. melanogaster genome. MACS (version 1.4.1) was used to identify regions of ChIP-seq enrichment. The density of reads in each region was normalized to the total number of 10 million mapped reads. BigWig files were generated for visualization using the Homer package.
Statistics and reproducibility
Sample size was determined by previous experience with similar experimental designs. Typically, 10–30 flies were scored for each experiment, and all experiments were repeated at least three times unless otherwise indicated. For DamID experiments, one replicate was performed in which the relative enrichment peaks were determined by log(fold change) between Sc-Dam-expressed (experiment) and Dam-expressed (control) samples. Mitotic index was determined by counting the PH3+ cells of whole midguts. The proportion of Sc+ ISCs was determined as the number of Sc-GFP+ cells in the total Dl+ cells within a microscopic field, with the mean number, n number and standard error of the mean (s.e.m.) presented for each genotype. To analyse the first few rounds of cell division in sc-overexpressed ISCs, RedStinger+ clones with three or fewer cells were taken for analysis. The number of cell divisions was deduced based on the total number of cells in each clone. The first cell division features one PH3+ cell in a one-cell clone, the second division features one PH3+ cell in a two-cell clone, and the third division features one PH3+ cell in a three-cell clone. No other exclusion criteria were applied. No sample randomization or blinding was performed. Statistical analysis was performed using Graphpad Prism 6. All significance tests were carried out with unpaired two-tailed t-tests. n is as indicated in each figure. Significance values: NS, not significant (P > 0.05), *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.
Life Sciences Reporting Summary
Further information on experimental design is available in the Life Sciences Reporting Summary.
DamID and RNA-seq data that support the findings of this study have been deposited in the Gene Expression Omnibus (GEO) under accession codes GSE102568 and GSE102569, respectively. Previously published ChIP-seq data that were reanalysed here are available under accession code GSE84283. Source data are provided in Supplementary Table 1. All other data supporting the findings of this study are available from the corresponding author upon reasonable request.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The authors thank the members of the fly community, as identified in the Methods, for providing fly stocks and antibodies, the Bloomington Drosophila Stock Center, the Tsinghua Fly Center and the Developmental Studies Hybridoma Bank (DSHB) for reagents, and A. Spradling, J. Rajagopal and members of the Xi laboratory for critical reading, and J. Snyder for proofreading the manuscript. This work was supported by the National Key Research and Development Program of China (2017YFA0103602 to R.X), the National Basic Research Program of China (2014CB850002 and 2011CB812700 to R.X.) and the National Natural Science Foundation of China (31501105 to N.X.).