In holometabolous insects, the adult appendages and internal organs form anew from larval progenitor cells during metamorphosis. As described here, the adult Drosophila midgut, including intestinal stem cells (ISCs),develops from adult midgut progenitor cells (AMPs) that proliferate during larval development in two phases. Dividing AMPs first disperse, but later proliferate within distinct islands, forming large cell clusters that eventually fuse during metamorphosis to make the adult midgut epithelium. We find that signaling through the EGFR/RAS/MAPK pathway is necessary and limiting for AMP proliferation. Midgut visceral muscle produces a weak EGFR ligand, Vein, which is required for early AMP proliferation. Two stronger EGFR ligands, Spitz and Keren, are expressed by the AMPs themselves and provide an additional, autocrine mitogenic stimulus to the AMPs during late larval stages.
INTRODUCTION
The insect midgut, like the vertebrate intestine, is an endoderm-derived organ. Both the larval and adult Drosophila midguts are composed of a single layer of epithelial cells with two layers of visceral muscle (VM)wrapped outside. Inside the gut lumen, a peritrophic membrane separates the food from the intestinal epithelium. During both mammalian and insect embryonic development, Forkhead and GATA transcription factors play evolutionary conserved roles in the specification and subsequent morphogenesis of the digestive tract (Stainier,2005). Similarly, multiple signaling pathways, including the EGF,Wingless (Wnt), Dpp (TGFβ), Notch and Hedgehog pathways, are involved in the embryonic development of the Drosophila midgut and mammalian intestine (Sancho et al.,2004). In both systems, cross-talk between mesodermal cells and endoderm-derived epithelial cells in the gut primordium plays important roles during embryonic gut development(Stainier, 2005; Szuts et al., 1998).
Starting from embryonic development stage 11, the Drosophilamidgut epithelium consists of two distinct cell populations: differentiating midgut epithelial cells (larval enterocytes, ECs) and undifferentiated adult midgut progenitors (AMPs, also referred to as midgut histoblast islets or midgut imaginal islets) (Hartenstein et al., 1992). In Drosophila embryos, AMPs can be marked by expression of asense or by one of several lacZ- or Gal4-expressing enhancer-trap insertions(Brand et al., 1993; Hartenstein et al., 1992; Hartenstein and Jan, 1992). AMPs first appear as spindle-shaped cells localized to the apical surface of the midgut epithelium, but later migrate to the basal surface of the epithelium where they remain throughout larval development(Hartenstein and Jan, 1992; Technau and Campos-Ortega,1986). Notch signaling has been shown to be involved in the development of Drosophila AMPs. In Notch mutant embryos, the number of AMPs in the midgut rudiment is strongly increased at the expense of differentiated larval ECs (Hartenstein et al., 1992). During larval development, the ECs grow in both size and ploidy by undergoing several endocycles, reaching 64C (DNA content) by the wandering L3 stage (Lamb,1982). The AMPs remain diploid throughout larval development and appear as scattered islets of cells (hence the term `midgut histoblast islets') in late-stage larval midguts. During pupal development, the ECs histolyze and a new adult midgut epithelium forms from the AMPs(Bender et al., 1997; Jiang et al., 1997; Li and White, 2003). Similar midgut progenitor cells have also been found in other insect species(Corley and Lavine, 2006).
Recently, the adult Drosophila midgut has been shown to undergo dynamic self-renewal, a process similar to that found in the mammalian intestine/colon. Fly and mammalian gut homeostasis are both powered by intestinal stem cells (ISCs), and Notch signaling plays similar roles in regulating their differentiation into mature gut cells(Fre et al., 2005; Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006; Ohlstein and Spradling, 2007; van Es et al., 2005). Thus,the Drosophila midgut may serve as a model to study gut homeostasis and the development of cancers, such as colorectal carcinoma, that are directly associated with this dynamic process in humans.
Here we describe the development of the AMPs in Drosophila larvae and pupae. We discovered that Drosophila AMPs divide extensively throughout larval development, and that their proliferation can be separated into two distinct phases. During early larval stages, the AMPs divide and disperse to form islets throughout the midgut, but during late larval development the dividing AMPs are contained within these islets. Furthermore,our study revealed that Drosophila EGFR signaling is both necessary and sufficient to induce the proliferation of AMPs during larval development.
MATERIALS AND METHODS
Fly stocks
UAS transgenes
The following were used: UAS-RasV12,UAS-RasV12S35, UAS-RasV12G37, UAS-Rafgof,UAS-λTOP, UAS-SEM, UAS-RafDN, UAS-Mkp3, UAS-sSpi,UAS-sKrn, UAS-Krn, UAS-grkΔTC and UAS-Vn1.2. UAS-RNAi transgenes were obtained from the Bloomington Stock Center (Bloomington, IN, USA), the National Institute of Genetics Fly Stock Center (NIG, Japan) or the Vienna Drosophila RNAi Center (VDRC,Austria). According to information from NIG and VDRC, all the RNAi lines used are specific to the genes targeted (NIG, http://www.shigen.nig.ac.jp/fly/nigfly/index.jsp;VDRC, http://stockcenter.vdrc.at/control/main).
Mutants
FRT42D Egfrf1, FRT42D Egfr[CO], FRT82B Ras1Δc40b, spiA14 FRT40A,FRT42D shot[65-2], FRT42D shot[V104], vnP1749FRT80B, rhodel1 FRT80B, Krn27-3-4, vnP1749,vnγ7, stet871 and ru1 were used (see FlyBase for further information: http://flybase.org).
Gal4/lacZ reporters
esgGal4NP7397, spiGal4NP0261,MyoIAGal4NP0001 (NIG, Japan), rholacZAA69, rholacZX81, howGal424B and esglacZK00606 were used (Bloomington Stock Center).
Lineage analysis
MARCM lineage analysis
Newly hatched first instar [24 hours after egg deposition (AED)] or mid-third instar (96 hours AED) larvae of the correct genotype were heat shocked for 45 minutes at 37°C. The midguts were then dissected from wandering L3 larvae (120 hours AED) and analyzed.
Flp/Gal4 lineage analysis
Newly hatched first instar larvae (24 hours AED) of the correct genotype were heat shocked for 20 minutes at 37°C to induce clones and then dissected at various developmental stages and analyzed.
Enhancer traps
P-element enhancer traps with midgut expression were obtained from several sources, including FlyView (University of Münster, Germany; http://flyview.uni-muenster.de)and GETDB (Gal4 Enhancer-Trap Insertion Database, NIG, Japan). We identified a number of enhancer traps showing reporter expression specifically in the AMPs,including one insertion in spi (NP0261) and several insertions in esg (NP0726, 7397 and 7399). esgGal4NP7397-driven GFP expression was used to mark the AMPs. We also identified an enhancer trap in brush border Myosin IA (MyoIAGal4, NP0001) that drives GFP expression specifically in midgut ECs(Morgan et al., 1995).
Ectopic gene expression
We generated inducible AMP-, EC- and VM-specific expression systems(esgGal4ts, MyoIAGal4ts and howGal4ts) by combining esgGal4NP7397,MyoIAGal4NP0001 or howGal424B(Hartenstein and Jan, 1992)with ubiquitously expressed temperature-sensitive alleles of the Gal4 inhibitor, Gal80 (tubGal80ts; Bloomington Stock Center)and UAS-GFP.
Quantification of AMP clusters
We counted AMPs or AMP clusters marked by esgGal4-driven GFP expression throughout the entire midgut during larval and pupal development. UAS-GFP, UAS-sSpi, UAS-sKrn or UAS-Krn were induced in the AMPs starting from first instar larvae (24 hours AED) using the esgGal4ts system and the midguts were dissected from wandering L3 larvae and the number of AMP clusters counted.
Generation of mutant AMP clones
Clones of AMPs homozygous for Egfrf1, Egfr[CO],Ras1Δc40b, spiA14,vnP1749, rhodel1, shot[V104] or shot[65-2] were generated using the MARCM system(Lee and Luo, 2001). First instar larvae (24 hours AED) of the correct genotype were heat shocked for 45 minutes at 37°C to induce clones. Larvae were then dissected at 120 hours AED. The number of GFP-positive clusters in each clone was quantified; in most cases, clones from at least ten midguts were counted.
RNA in situ hybridization and immunofluorescence
RNA in situ hybridization was performed as described(O'Neill and Bier, 1994). Rabbit anti-dpERK (Cell Signaling) was used to detect MAP kinase activity in the midgut. Anti-Delta and anti-Prospero were obtained from the Developmental Studies Hybridoma Bank and used to mark ISCs and enteroendocrine cells in the midgut. Rabbit anti-β-galactosidase (Cappel) was used to identify the esg-positive cells in an esglacZ background. Rabbit anti-phospho-histone H3 (PH3, Upstate) was used to identify dividing cells.
Quantitative real-time PCR (qRT-PCR)
We used qRT-PCR to quantify levels of vn mRNA from midgut cDNA. For mRpL30 (reference gene) primers see Buttitta et al.(Buttitta et al., 2007); for vn: vn 5′ primer,5′-TCACACATTTAGTGGTGGAAG-3′; vn 3′ primer,5′-TCACACATTTAGTGGTGGAAG-3′. The relative expression of vn was analyzed on the Bio-Rad iQ5 system.
Sectioning
Wandering L3 midguts were dissected in PBS and fixed in half-strength Karnovsky's fixative. Following dehydration, the tissues were embedded in Epon and sectioned at 1 μm. The sections were stained with Toluidine Blue.
RESULTS
Drosophila AMPs divide extensively during larval development
To study the development of AMPs during larval development, we first looked for AMP markers. From existing collections of Drosophila enhancer traps we identified Gal4 or lacZ enhancer traps that are expressed specifically in the AMPs. Among these are Gal4 enhancer traps inserted in the escargot (esg) locus, which encodes a member of the Snail family of transcription factors. esg has previously been shown to be expressed in imaginal discs and abdominal histoblast nests and is required there for maintaining cells in the diploid state during larval development (Hayashi et al., 1993). When combined with UAS-GFP, esgGal4enhancer trap NP7397 drove GFP expression specifically in the larval AMPs(Fig. 1). Similar esgenhancer traps have been used to mark the adult ISCs and their daughter enteroblasts (Micchelli and Perrimon,2006).
GFP expression driven by esgGal4 was detected in the AMPs scattered throughout the midgut of newly hatched larvae (24 hours AED)(Fig. 1A, arrows). AMPs appeared as small diploid cells, and were easily distinguishable from the large polyploid midgut enterocytes (ECs). The number of GFP-positive AMPs increased during early larval development (24-72 hours AED)(Fig. 1A,B); however, they remained dispersed. Cell contacts between paired AMPs were readily observed in the early larval midgut (Fig. 1B, inset) and are likely to represent two daughter AMPs from the previous division migrating away from each other. By mid-third instar (96 hours AED), AMPs formed discrete 2- to 3-cell clusters(Fig. 1C), suggesting that they proliferate within individual islets instead of migrating away from each other. The AMPs continued to proliferate within these clusters(Fig. 1D), undergoing several rounds of rapid proliferation to enlarge each cluster to 8-30 cells by the onset of metamorphosis [0 hours after pupae formation (APF), ∼130 hours AED] (Fig. 1E).
These results do not support the idea that Drosophila AMPs are quiescent during larval development(Bodenstein, 1994). Instead, we observed that the AMPs proliferate extensively during larval development,resulting in large increases in both the number (early larval stages) and size(late larval stages) of the AMP clusters. To further document this process, we analyzed AMP lineages by positively marking individual AMPs with GFP using the MARCM system (Lee and Luo,2001). When clones were induced in first or second instar (24-48 hours AED), they all contained multiple AMP clusters by the wandering L3 stage(120 hours AED), and all cells in any GFP-positive cluster were GFP-positive(Fig. 2A-A″; Fig. 4A). However, when clones were induced in mid-third instar (96 hours AED), clusters mosaic for GFP were observed by the wandering L3 stage (120 hours AED)(Fig. 2B-B″). These results confirmed that the AMPs switch to proliferating within islets to form clusters by mid-third instar. To quantify the number of divisions during the early proliferative phase, we counted the number of the marked AMP clusters encompassed by each clone. When induced in the newly hatched first instar larvae (24 hours AED), the clones contained, on average, 7.5 GFP-positive clusters at the wandering L3 stage (120 hours AED) (see Table S1 in the supplementary material). This suggests that the AMPs divide about four times during the early larval stages (note that only half of all the clusters generated by each AMP were marked in the MARCM system). Since no mosaic AMP clusters were found in the late larval midgut when clones were induced at first or second instar, we propose that the majority, if not all, of the early larval AMPs disperse after each cell division. We then counted the number of cells in each cluster at white prepupa formation (0 hours APF), when most of the AMP clusters have stopped proliferating. Each AMP cluster contained 8 to greater than 30 cells. This indicates that the AMPs divide an additional three to five times within a cluster, after the clusters are established. In total,the AMPs appear to divide seven to ten times throughout larval development.
Development of Drosophila adult midgut progenitors (AMPs).AMPs were marked by GFP expression (green) driven by esgGal4NP7397. The numbers of GFP-positive AMPs, AMP clusters or adult intestinal stem cells (ISCs) are indicated in the appropriate panels. DNA is stained with DAPI (blue). (A) First instar larval midgut (24 hours AED). GFP was detected in the AMPs as individual diploid cells (arrows). (B) Early third instar larval midgut (72 hours AED). Larval enterocytes (ECs) undergo several rounds of endoreplication,enlarging the larval midgut. AMPs remain diploid and their numbers increase during the first two larval stages. However, they remain mostly dispersed as individual cells. Inset shows GFP expression that is overexposed to show cell contacts between two neighboring AMPs. (C) Mid-third instar larval midgut (96 hours AED). AMPs form distinctive 2- to 3-cell clusters. (D)Late third instar larval midgut (120 hours AED). AMPs continue to proliferate and enlarge the clusters. (E) White prepupa stage (0 hours APF,∼130 hours AED). The size of the AMP clusters has increased further.(F) Prepupa stage (4 hours APF). The AMP clusters fuse to form a new midgut epithelium. Larval ECs (out of focal plane) are sloughed into the lumen and histolyze. (G,H) Early pupa stage (8 and 12 hours APF). The majority of the cells in the new midgut epithelium gradually lose GFP expression, except for a few scattered cells that maintain strong GFP expression. (I,J) Pupa stage (24 hours APF). The future adult ISCs are clearly identifiable by strong GFP expression (asterisks in I) and basal localization in the epithelium (asterisks in J, cross-sectional view). GFP expression is lost in the rest of the cells in the new epithelium. Scale bars: 20 μm.
Development of Drosophila adult midgut progenitors (AMPs).AMPs were marked by GFP expression (green) driven by esgGal4NP7397. The numbers of GFP-positive AMPs, AMP clusters or adult intestinal stem cells (ISCs) are indicated in the appropriate panels. DNA is stained with DAPI (blue). (A) First instar larval midgut (24 hours AED). GFP was detected in the AMPs as individual diploid cells (arrows). (B) Early third instar larval midgut (72 hours AED). Larval enterocytes (ECs) undergo several rounds of endoreplication,enlarging the larval midgut. AMPs remain diploid and their numbers increase during the first two larval stages. However, they remain mostly dispersed as individual cells. Inset shows GFP expression that is overexposed to show cell contacts between two neighboring AMPs. (C) Mid-third instar larval midgut (96 hours AED). AMPs form distinctive 2- to 3-cell clusters. (D)Late third instar larval midgut (120 hours AED). AMPs continue to proliferate and enlarge the clusters. (E) White prepupa stage (0 hours APF,∼130 hours AED). The size of the AMP clusters has increased further.(F) Prepupa stage (4 hours APF). The AMP clusters fuse to form a new midgut epithelium. Larval ECs (out of focal plane) are sloughed into the lumen and histolyze. (G,H) Early pupa stage (8 and 12 hours APF). The majority of the cells in the new midgut epithelium gradually lose GFP expression, except for a few scattered cells that maintain strong GFP expression. (I,J) Pupa stage (24 hours APF). The future adult ISCs are clearly identifiable by strong GFP expression (asterisks in I) and basal localization in the epithelium (asterisks in J, cross-sectional view). GFP expression is lost in the rest of the cells in the new epithelium. Scale bars: 20 μm.
Midgut development during early metamorphosis
Staining for the division marker phospho-histone H3 (PH3) indicated that by white prepupa formation (0 hours APF), the majority of the AMPs had ceased their proliferation. Some AMPs in the posterior midgut, however, did not cease proliferation until 4 hours APF (data not shown). Meanwhile, the visceral muscles (VMs) contracted, and the larval midgut shortened itself dramatically. At the same time, the AMP clusters fused to form a new midgut epithelium(Fig. 1F), while the larval midgut epithelium became extremely compacted and was sloughed into the intestinal lumen (Li and White,2003). The midgut continued to contract and shorten, and by 12 hours APF it became a sac-like structure containing histolyzing larval epithelium inside the newly formed midgut(Juhasz and Neufeld, 2008). Visceral muscles undergo a process termed `de-differentiation', in which the muscle fibers histolyze; however, the muscle cells themselves do not die and will redifferentiate to form the adult midgut VM during late metamorphosis(Klapper, 2000). During early metamorphosis (0-24 hours APF), the majority of cells in the newly formed midgut epithelium gradually lost esgGal4-driven GFP expression, with the exception of a few scattered cells that maintained strong GFP expression(Fig. 1F-I, asterisks). By 24 hours APF, these GFP-positive cells became basally localized in the new epithelium (Fig. 1J,asterisks). As described below, we believe that these cells are the future adult midgut ISCs (see Discussion).
Lineage analysis of the AMPs. (A-B″) Drosophila AMP clones induced using the MARCM system. Clones were induced at either first instar (24 hours AED, A-A″) or third instar (96 hours AED, B-B″) and analyzed at the wandering L3 stage (120 hours AED). When induced at first instar, the clones appear as multiple marked clusters with all cells labeled (A-A″), whereas clones induced late (at 96 hours AED) were all confined to a single cluster that is mosaic for GFP(B-B″). (C-E″) Pupal or adult AMP clones induced using the Flp/Gal4 system. Flp/Gal4 AMP clones were induced at first instar larval stage (24 hours AED) and analyzed at 24 hours APF (C-C″)or from newly eclosed adults (D-E″). At 24 hours APF, each AMP clone contains 0-2 esg-positive cells (C, arrows); the asterisk marks the histolyzing larval midgut. In newly eclosed adults, the midgut contains enteroendocrine cells and ISCs; arrows indicate cells within the clone that are positive for Prospero (D) and Delta (E).
Lineage analysis of the AMPs. (A-B″) Drosophila AMP clones induced using the MARCM system. Clones were induced at either first instar (24 hours AED, A-A″) or third instar (96 hours AED, B-B″) and analyzed at the wandering L3 stage (120 hours AED). When induced at first instar, the clones appear as multiple marked clusters with all cells labeled (A-A″), whereas clones induced late (at 96 hours AED) were all confined to a single cluster that is mosaic for GFP(B-B″). (C-E″) Pupal or adult AMP clones induced using the Flp/Gal4 system. Flp/Gal4 AMP clones were induced at first instar larval stage (24 hours AED) and analyzed at 24 hours APF (C-C″)or from newly eclosed adults (D-E″). At 24 hours APF, each AMP clone contains 0-2 esg-positive cells (C, arrows); the asterisk marks the histolyzing larval midgut. In newly eclosed adults, the midgut contains enteroendocrine cells and ISCs; arrows indicate cells within the clone that are positive for Prospero (D) and Delta (E).
Around 24 hours APF, the esg-positive cells in the new adult midgut epithelium started to proliferate (see Fig. S1A-A‴ in the supplementary material). They continued to divide at 48 hours APF and increased in number (see Fig. S1B-B‴ in the supplementary material). At 72 hours APF, the esg-positive cells continued to divide and some of them also expressed the enteroendocrine cell marker Prospero (see Fig. S1C-C‴ in the supplementary material), indicating their capacity to differentiate. Whether these cells already behaved as stem cells, which both self-renew and differentiate, was not determined. GFP-marked AMP Flp/Gal4 clones induced at early larval stages contained only 0-2 esg-positive cells at 24 hours APF(Fig. 2C-C″), but when these clones were scored later, in newly eclosed adults, they contained both large apically localized ECs and smaller, basally localized cells positive for the enteroendocrine cell marker Prospero or the ISC marker Delta(Fig. 2D-D″,E-E″). These results indicate that some of the AMPs become adult ISCs(Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006),and that this transition occurs in the pupa. This was further supported by our observation that AMP clones induced in the larva persisted in the adult midgut for at least 2 months (data not shown).
EGFR signaling stimulates AMP proliferation
Using the esgGal4ts system, we manipulated the activity of several known Drosophila signaling pathways specifically in the AMPs. Our tests included Wingless, Dpp, Hedgehog, Notch and EGFR signaling components (see Table S2 in the supplementary material). Activation of EGFR/RAS/MAPK signaling in the AMPs was able to drive their overproliferation during larval development. Compared with control midguts(Fig. 3A), in which the AMPs appeared as 2- to 3-cell clusters by mid-third instar (96 hours AED), the induction of activated Ras (Ras oncogene at 85D - FlyBase)(RasV12) in the AMPs led to the formation of much larger AMP clusters (Fig. 3B-D). Ectopic expression of RasV12 in the AMPs throughout larval development resulted, by the wandering stage (120 hours AED), in a midgut comprising mostly esg-positive AMPs(Fig. 3F) in which the intestinal lumen was occluded. By contrast, wild-type AMPs appeared as basally localized cell clusters in the midgut epithelium(Fig. 3E). The following evidence suggested that EGFR signaling promoted AMP proliferation through activating the MAPK pathway. First, induction of RasV12S35, which preferentially activates the MAPK pathway, drove similar ectopic proliferation of the AMPs as did RasV12 (see Table S2 in the supplementary material),whereas expression of RasV12G37, which preferentially activates the Phosphotidylinositol 3 kinase (PI3K) or Ral guanine nucleotide exchange factor 2 (RalGDS) pathway (Karim and Rubin, 1998; Prober and Edgar, 2002), had little effect on their proliferation (see Table S2 in the supplementary material). Second, ectopic expression of Dp110 (Pi3K92E - FlyBase; PI3K) had no detectable effect on AMP proliferation (see Table S2 in the supplementary material). Third,increased proliferation of the AMPs was observed when activated Egfr(λTOP) (Queenan et al.,1997), gain-of-function Raf (Rafgof)(Brand and Perrimon, 1994) or activated MAPK [sevenmaker (sem); rolled - FlyBase](Martin-Blanco, 1998) was induced in these cells (see Table S2 in the supplementary material). Fourth,expression of a dominant-negative form of Raf(RafDN) (Roch et al.,1998) together with RasV12 gave a phenotype similar to that of RafDN alone (see Fig. S2E,F in the supplementary material), and thus Raf is epistatic to Ras in regulating AMP proliferation. Fifth, expression of Mkp3, a negative regulator of MAPK (Rintelen et al.,2003), did not affect AMP proliferation (see Fig. S2G in the supplementary material). Interestingly, however, Mkp3 did significantly suppress the AMP overproliferation phenotype induced by RasV12 expression (see Fig. S2H in the supplementary material; compare with Fig. 3D).
Activated Ras (RasV12) stimulates AMP proliferation. UAS-transgenes were induced in the Drosophila AMPs using the esgGal4ts system. Larvae were shifted to 29°C at the indicated times and dissected at 96 hours AED. (A) GFP (24-96 hours AED, control). (B) RasV12(72-96 hours AED). (C) RasV12 (48-96 hours AED).(D) RasV12 (24-96 hours AED). (E,F)Cross-sections of posterior midguts from wandering L3 larvae expressing ectopic GFP (E, wild type, WT) or RasV12 (F)throughout larval development (24-120 hours AED). The control AMP clusters are basally localized in the epithelium (E, arrows). PM, peritrophic membrane. The samples in E and F were stained with Toluidine Blue.
Activated Ras (RasV12) stimulates AMP proliferation. UAS-transgenes were induced in the Drosophila AMPs using the esgGal4ts system. Larvae were shifted to 29°C at the indicated times and dissected at 96 hours AED. (A) GFP (24-96 hours AED, control). (B) RasV12(72-96 hours AED). (C) RasV12 (48-96 hours AED).(D) RasV12 (24-96 hours AED). (E,F)Cross-sections of posterior midguts from wandering L3 larvae expressing ectopic GFP (E, wild type, WT) or RasV12 (F)throughout larval development (24-120 hours AED). The control AMP clusters are basally localized in the epithelium (E, arrows). PM, peritrophic membrane. The samples in E and F were stained with Toluidine Blue.
EGFR signaling is required for AMP proliferation
Next, we tested whether the EGFR/RAS pathway is required for the normal proliferation of AMPs during larval development. Using the same esgGal4ts system, we depleted crucial components of the pathway by expressing RNA inverted repeats (IR, RNAi) specific to Drosophila Egfr, Ras and Raf (pole hole - FlyBase)in the AMPs. As a control, UAS-driven RNAi directed against GFP was induced in the AMPs using the esgGal4ts system. This treatment did not affect AMP development (data not shown). When compared with control midguts from white prepupa (0 hours APF), RNAi-mediated depletion of each of these gene products in the AMPs throughout larval development significantly decreased both the number and size of the AMP clusters (see Fig. S2A-D and Table S2 in the supplementary material). This indicates that both phases of AMP proliferation were affected when EGFR signaling was downregulated. Similar results were observed when the dominant-negative form of Raf(RafDN) was induced in the AMPs (see Fig. S2E and Table S2 in the supplementary material).
In further tests we generated AMP clones defective in EGFR signaling using the MARCM system. AMPs mutant for Egfr (Egfr[CO])or Ras (Ras1Δc40b) did not proliferate during larval development(Fig. 4; see Table S1 in the supplementary material). Instead of forming multiple GFP-positive clusters as in controls (Fig. 4A), these mutant clones appeared as single GFP-positive cells(Fig. 4B,C). We conclude that EGFR/RAS/MAPK signaling is required for AMP proliferation during both early and late larval development.
Drosophila MAPK is activated in the AMPs
To examine whether downstream components of EGFR signaling are activated in the AMPs, we stained the larval midgut with antibodies against diphospho-extracellular signal-regulated kinase (dpERK; Rolled - FlyBase), the level of which is a direct measurement of the activated form of Drosophila MAPK (Gabay et al.,1997). dpERK staining was indeed detected in the AMP clusters, but not in the larval gut epithelial cells(Fig. 5A-A″), indicating activation of MAPK in these cells. This result is consistent with our genetic results and supports the notion that EGFR signaling regulates AMP proliferation through activating the MAPK pathway.
Next, we examined the expression patterns of several EGFR ligands in the larval midgut. We identified a Gal4 enhancer trap in spi (NP0261, see Materials and methods) that drove UAS-GFP expression specifically in the AMPs (Fig. 5B-B″). RNA in situ hybridization confirmed that spi was specifically expressed in the AMP clusters (Fig. 5D,D′). Krn was also specifically expressed in the AMPs as shown by RNA in situ hybridization(Fig. 5E,E′).
EGFR signaling mutant AMPs fail to proliferate.Egfr-/- or Ras-/- Drosophila AMP clones were generated using the MARCM system, which positively marks mutant cells with GFP expression. (A) FRT42D only (control).(B) FRT42D Egfr[CO]. (C) FRT82B RasΔc40b. The boxed regions in A-C are shown to the right as GFP (A′-C′), DNA(A″-C″) and merged(A‴-C‴) images. Unlike in the control(A-A‴), where GFP-positive clones form multiple clusters, clones of Egfr[CO] and RasΔc40b AMPs (B-B‴;C-C‴) appear as single GFP-positive cells. Arrowheads indicate the positions of Egfr-/- or Ras-/- AMPs. The asterisk in C indicates one larval EC with non-specific GFP expression from the MARCM system (FRT82B).
EGFR signaling mutant AMPs fail to proliferate.Egfr-/- or Ras-/- Drosophila AMP clones were generated using the MARCM system, which positively marks mutant cells with GFP expression. (A) FRT42D only (control).(B) FRT42D Egfr[CO]. (C) FRT82B RasΔc40b. The boxed regions in A-C are shown to the right as GFP (A′-C′), DNA(A″-C″) and merged(A‴-C‴) images. Unlike in the control(A-A‴), where GFP-positive clones form multiple clusters, clones of Egfr[CO] and RasΔc40b AMPs (B-B‴;C-C‴) appear as single GFP-positive cells. Arrowheads indicate the positions of Egfr-/- or Ras-/- AMPs. The asterisk in C indicates one larval EC with non-specific GFP expression from the MARCM system (FRT82B).
Multiple EGFR ligands are involved in AMP proliferation
To investigate which EGFR ligands regulate the proliferation of the AMPs,we expressed each of the four known EGFR activating ligands (gurken,spitz, Keren and vein) in the AMPs using the esgGal4ts system. Induction of activated gurken(grkΔTC), a strong EGFR ligand,the function of which is believed to be exclusively in female oogenesis(Nilson and Schupbach, 1999),did not affect the proliferation of the AMPs (see Table S2 in the supplementary material). However, induction of activated (secreted) spitz or Keren (sSpi or sKrn), two other strong EGFR activating ligands (Reich and Shilo, 2002; Schweitzer et al., 1995), promoted extensive overproliferation of the AMPs(Fig. 6A-C; see Table S2 in the supplementary material). Furthermore, induction of wild-type Krn,which requires cleavage by rhomboid family proteases to become fully active,similarly promoted AMP proliferation (Fig. 6D-D′; see Table S2 in the supplementary material). In addition, induction of sSpi or sKrn limited the dispersal of the AMPs, thus reducing the number of the clusters in the wandering L3 larval midguts (Fig. 6E). Induction of the weak EGFR ligand vein (vn)(Schnepp et al., 1998) with esgGal4ts, however, had little effect on AMP proliferation(see Table S2 in the supplementary material).
To determine whether spi or Krn are required for AMP proliferation, we downregulated the levels of these EGFR ligands in the AMPs by RNAi. Ectopic expression of UAS-RNAi directed at spiand/or Krn using esgGal4 had no effect on AMP proliferation(see Table S2 in the supplementary material). Consistent with this, the proliferation of the AMPs in Krn27-3-4 (null allele)mutant larvae was normal (see Table S1 in the supplementary material). The same was also found for spiA14 mutant AMP clones generated in a Krn27-3-4 homozygous mutant background (see Table S1 in the supplementary material). The proliferation of mutant AMPs lacking rhomboid (rhodel1, null allele) or spi(spiA14, null allele) function, generated using the MARCM system, was also normal (see Table S1 in the supplementary material). We examined lacZ expression from two rholacZ reporters in the larval midgut (rhoAA69 and rhoX81) and found that neither were expressed in the AMPs. Since the Drosophilagenome encodes multiple rhomboid-like genes, we also generated MARCM clones in the mutant background of rho-2(stet871) and rho-3 (ru1). These mutant clones also contained normal numbers of AMP clusters (see Table S1 in the supplementary material). These results suggest that either multiple,redundant rhomboid-like genes are utilized in the AMPs, or (less likely) that rhomboid-like function is dispensable in the larval midgut. Furthermore, we conclude that spi and Krn are likely to be dispensable for AMP proliferation (see Discussion).
Surprisingly, in several vn mutants (vnP1749/P1749,vnγ7/γ7and vnP1749/γ7), few AMP clusters were found in the late larval midgut(Fig. 7B,C), whereas the AMPs in wild-type controls formed many large clusters(Fig. 7A). This suggests that vn is required for normal AMP development. To further study the function of vn in AMP development, we carried out lineage analysis of the AMPs in the vn mutant animals. We induced GFP-marked AMP clones in first instar larvae using the Flp/Gal4 system. Compared with control midguts, which contained on average 15.2 marked AMP clusters per clone(n=50 clones) (see Fig. S3A-A″ in the supplementary material),we consistently observed only a single GFP-positive AMP cluster in the midguts of weak vn mutants(vnγ7/P1749; animals of this genotype are not developmentally delayed during larval development and most die as pharate adults) (see Fig. S3B-B″ in the supplementary material). Furthermore, we counted the number of esg-positive cells (marked by esglacZ) in newly hatched larval midguts. Control larval midguts(vnP1749/+) contained on average 121 AMPs per gut, whereas there were on average 137 AMPs per midgut in vnγ7/P1749mutants (ten midguts for each genotype were scored). This indicates that the reduction in the number of AMP clusters in the late larval midgut of vnγ7/P1749 mutants was not due to the production of fewer AMPs during embryogenesis. Taken together, these results suggest that the proliferation of AMPs during the early larval stages is completely inhibited in vn mutants. However, the size of the few remaining AMP clusters in the vnγ7/P1749 mutant midguts was relatively normal (see Fig. S3B-B″ in the supplementary material),suggesting that the late phase of AMP proliferation is largely unaffected in vn mutants. We speculate that the reason vn becomes dispensable for AMP proliferation during late larval development is that Krn and spi expression in the AMPs supplies a redundant function.
Expression and activity of the EGFR ligands spitz, Keren and vein in the larval midgut. (A-A″)MAPK activity (dpERK staining) in the midgut of late third instar Drosophila larvae. (B-B″) The expression of UAS-GFP driven by spiGal4NP0261 in the midgut of L3 wandering larvae. (C-C″) vnlacZ reporter expression pattern in the midgut. Large arrows indicate the circular visceral muscle cells, which form four distinct rows (two are shown). Small arrows indicate the longitudinal visceral muscle cells. (D,D′) Krn RNA in situ hybridization in L3 wandering larval midgut.(E,E′) spi RNA in situ hybridization in L3 wandering larval midgut. Arrowheads indicate the positions of the AMP clusters in all panels.
Expression and activity of the EGFR ligands spitz, Keren and vein in the larval midgut. (A-A″)MAPK activity (dpERK staining) in the midgut of late third instar Drosophila larvae. (B-B″) The expression of UAS-GFP driven by spiGal4NP0261 in the midgut of L3 wandering larvae. (C-C″) vnlacZ reporter expression pattern in the midgut. Large arrows indicate the circular visceral muscle cells, which form four distinct rows (two are shown). Small arrows indicate the longitudinal visceral muscle cells. (D,D′) Krn RNA in situ hybridization in L3 wandering larval midgut.(E,E′) spi RNA in situ hybridization in L3 wandering larval midgut. Arrowheads indicate the positions of the AMP clusters in all panels.
Interestingly, we found that vn is specifically expressed in VM cells throughout larval development, as revealed by the expression of a well-characterized lacZ enhancer-trap insertion, vnlacZP1749 (Fig. 5C-C″) (Kiger et al.,2000). The Drosophila midgut VM comprises an outer layer of 21 longitudinal muscle strips and an inner layer of circular muscle that forms four distinctive rows (Klapper,2000), as revealed by a muscle-specific driver, howGal424B, which drives UAS-GFP expression in both types of VM cells (Fig. 5C′) (Brand and Perrimon,1993). vnlacZ expression was stronger in the inner,circular VM cells than in the outer, longitudinal muscle(Fig. 5C).
To test the importance of Vn signaling from VM, we specifically depleted vn from VM cells by expressing UAS-Vn RNAi using an inducible muscle-specific driver, howGal4ts. Induction of vn RNAi throughout larval development (24-120 hours AED) resulted in late larval midguts with very few AMP clusters, as in vn mutants(Fig. 7D-D″). However,induction of vn RNAi starting at early third instar (72 hours AED)had no effect on the AMPs (Fig. 7E-E″). Furthermore, induction of UAS-Vn RNAi in the AMPs or in the larval ECs using the esgGal4ts or MyoIAGal4ts (MyoIA is also known as Myo31DF - FlyBase) drivers had no effect on AMP proliferation (see Fig. S4A,B in the supplementary material), suggesting that the principal source of Vn is VM. This was confirmed by quantitative real-time PCR showing that induction of UAS-Vn RNAi in VM significantly reduced vnmRNA levels in whole midguts, whereas induction of vn RNAi in ECs or AMPs did not (Fig. 7G). In further tests, we attempted to rescue the AMP phenotype of vnP1749 mutants by expressing UAS-Vn in AMPs, ECs or VM, using the esgGal4ts, MyoIAGal4ts or howGal4ts drivers. Induction of UAS-Vn in the AMPs or VM completely rescued the phenotype of vnP1749mutants (Fig. 7F-F″; see Fig. S5A-A″ in the supplementary material). Induction of vn in the larval ECs, which constitute the bulk of the midgut mass, not only rescued the proliferative defects of the AMPs, but also caused ectopic AMP proliferation (see Fig. S5B-B″ in the supplementary material). We conclude that Vn, expressed in VM, is the principal mitogen for AMPs during early larval development. Later, autocrine Spi and Krn might complement this function.
This scenario, in which VM-derived Vn activates EGFR signaling in AMPs, is reminiscent of the role of Vn in muscle/tendon development during embryogenesis. In this case, muscle-derived Vn is specifically concentrated on tendon cells and activates EGFR there(Strumpf and Volk, 1998). The concentration of Vn is highly dependent upon the activity of the short stop (shot, also called kakapo) gene in the tendon cells (Strumpf and Volk,1998). We tested whether shot is also required for VM-derived Vn to activate EGFR signaling in the AMPs by quantifying AMP clusters in shot mutant MARCM clones. These clones all contained normal numbers of AMP clusters (see Table S1 in the supplementary material),and thus the role of shot in transducing the Vn signal is uncertain.
Expression of sSpi, Krn or sKrn in the AMPs induces their proliferation. The ligands were induced in the Drosophila AMPs using the esgGal4ts system starting at 24 hours AED, and larvae were dissected at 96 hours AED. (A,A′) GFP(control). (B,B′) Activated (secreted) Spi (sSpi).(C,C′) Activated (secreted) Krn (sKrn).(D,D′)Krn. (A-D) GFP marks the AMP clusters.(A′-D′) Merged images of GFP (green) and DNA (DAPI, blue).(E) The ectopic expression of strong EGF ligands in the AMPs dramatically reduces the total number of AMP clusters in the midgut. WT,wild-type.
Expression of sSpi, Krn or sKrn in the AMPs induces their proliferation. The ligands were induced in the Drosophila AMPs using the esgGal4ts system starting at 24 hours AED, and larvae were dissected at 96 hours AED. (A,A′) GFP(control). (B,B′) Activated (secreted) Spi (sSpi).(C,C′) Activated (secreted) Krn (sKrn).(D,D′)Krn. (A-D) GFP marks the AMP clusters.(A′-D′) Merged images of GFP (green) and DNA (DAPI, blue).(E) The ectopic expression of strong EGF ligands in the AMPs dramatically reduces the total number of AMP clusters in the midgut. WT,wild-type.
DISCUSSION
Drosophila AMPs undergo extensive proliferation during larval development
Drosophila AMPs were previously thought to be relatively quiescent during larval development, dividing just once or twice, and not initiating rapid proliferation until the onset of metamorphosis(Bodenstein, 1994). This is the case for several other larval progenitor/imaginal cell types, such as the abdominal histoblasts and cells in the salivary gland, foregut and hindgut imaginal rings (Bodenstein,1994). More recent studies have suggested that AMP proliferation might precede the onset of metamorphosis(Hall and Thummel, 1998; Jiang et al., 1997; Li and White, 2003). However,these studies did not report the extensive proliferation of the AMPs that we describe here, and failed to recognize the early larval proliferative phase when the AMPs divide and disperse (Figs 1 and 8). The extensive proliferation of the AMPs is similar to that of the larval imaginal disc cells, which also proliferate throughout larval development, dividing about ten times.
Lineage analysis revealed that the proliferation of the DrosophilaAMPs occurs in two distinct phases (Fig. 8). In early larvae, the AMPs divide and disperse throughout the midgut to form individual islets. During later larval development, the AMPs continue to divide but do so within these islets, forming large cell clusters. We speculate that in the early larva, secretion of Vn from the midgut visceral muscle (VM) cells results in low-level activation of EGFR signaling in the AMPs, which is sufficient for their proliferation and might also promote their dispersal. We did not observe any proliferation defects in AMPs defective in shot function, suggesting that the mechanism of EGFR activation used by tendon cells during muscle/tendon development is probably not the same as in the larval midgut. Specifically, it is unlikely that the Shot-mediated concentration of Vn on AMPs activates EGFR signaling in the AMPs during early larval development. Consistent with this, we only observed dpERK staining in AMP clusters (Fig. 5A-A″)and not in the isolated AMPs present at early larval stages (24-72 hours AED;data not shown).
The mechanisms that regulate the transition between these two proliferation phases remain unclear. We observed fewer AMP clusters when sSpi,sKrn, λTOP (activated Egfr) or RasV12 were induced in the AMPs starting from early larval stages (Fig. 6E; see Table S2 in the supplementary material), suggesting that EGFR signaling, in addition to its crucial role as an AMP mitogen, might also play a role in AMP cluster formation. In the late larval midgut (96-120 hours AED), high-level EGFR activation, resulting from expression of spi and Krn in the AMPs themselves, might not only promote AMP proliferation, but might also suppress AMP dispersal and thus promote formation of the AMP clusters. How the timing and location of Spi- or Krn-mediated EGFR activation are regulated during larval development is also unclear. We note, however, that the pro-ligand form of Krn acted similarly to sKrn(Fig. 6), and that we failed to uncover any functions for the Rho-like gene products that regulate Spi and Krn function by proteolytic cleavage in other tissues (see Tables S1 and S2 in the supplementary material). This suggests that the localized expression of these ligands in the AMP clusters might be the critical parameter that controls their effects. Consistent with this, Rho-independent cleavage and function of Krn have been documented (Reich and Shilo,2002).
Vein is required for AMP proliferation. (A-C) Posterior midguts from white prepupa (0 hours APF) of wild-type (WT) Drosophilacontain multiple AMP clusters (A), which are missing from the midguts of vn mutants (B, vnP1749; C, vnγ7). Midguts are outlined with dashed lines. (D-E′) Posterior midguts from wandering L3 larvae in which vn was specifically knocked down in the visceral muscle cells throughout larval development (24-120 hours AED, D) or only during late larval development (72-120 hours AED, E). Most of the remaining small cells in B-D are visceral muscle cells. Arrows in D point to the few AMP clusters in the midgut. (F,F′) Induction of UAS-Vn expression throughout larval development (24-120 hours AED) using the muscle-specific driver howGal4ts rescued the vn mutant phenotype. D′-F′ show merged images of DNA (DAPI, blue) and howGal4ts-driven GFP expression (green) in the visceral muscle and trachea (asterisks in D′,F′) cells. (G)Knockdown of vn mRNA in the midgut by vn RNAi. Relative levels of vn mRNA in the larval midgut were quantified by qRT-PCR. Only UAS-Vn RNAi expression driven by the muscle-specific Gal4 driver, howGal4ts, knocked down vn significantly in the larval midgut.
Vein is required for AMP proliferation. (A-C) Posterior midguts from white prepupa (0 hours APF) of wild-type (WT) Drosophilacontain multiple AMP clusters (A), which are missing from the midguts of vn mutants (B, vnP1749; C, vnγ7). Midguts are outlined with dashed lines. (D-E′) Posterior midguts from wandering L3 larvae in which vn was specifically knocked down in the visceral muscle cells throughout larval development (24-120 hours AED, D) or only during late larval development (72-120 hours AED, E). Most of the remaining small cells in B-D are visceral muscle cells. Arrows in D point to the few AMP clusters in the midgut. (F,F′) Induction of UAS-Vn expression throughout larval development (24-120 hours AED) using the muscle-specific driver howGal4ts rescued the vn mutant phenotype. D′-F′ show merged images of DNA (DAPI, blue) and howGal4ts-driven GFP expression (green) in the visceral muscle and trachea (asterisks in D′,F′) cells. (G)Knockdown of vn mRNA in the midgut by vn RNAi. Relative levels of vn mRNA in the larval midgut were quantified by qRT-PCR. Only UAS-Vn RNAi expression driven by the muscle-specific Gal4 driver, howGal4ts, knocked down vn significantly in the larval midgut.
In the developing Drosophila wing, EGFR/RAS/MAPK signaling promotes the expression and controls the localization of the cell adhesion molecule Shotgun (Shg, Drosophila DE-cadherin)(O'Keefe et al., 2007). RasV12-expressing clones generated in the wing imaginal disc are round (Prober and Edgar,2002), much like the AMP clusters described here, owing to increased adhesive junctions. In developing Drosophila trachea, EGFR activity upregulates shg expression to maintain epithelial integrity in the elongating tracheal tubes (Cela and Llimargas, 2006). In the eye, EGFR activity leads to increased levels of Shg and adhesion between photoreceptors(Brown et al., 2006; Mirkovic and Mlodzik, 2006). Given these precedents, it seems reasonable to suggest that high-level EGFR activity in the AMP islets upregulates Shg and promotes the homotypic adhesion of the AMPs. Alternatively, changes in the differentiated cells of the midgut epithelium might promote AMP clustering. In either case, the dispersal of early AMPs and subsequent formation of late AMP clusters facilitate the formation of the adult midgut epithelium during metamorphosis.
AMPs give rise to adult intestinal stem cells during metamorphosis
Our study confirms previous reports that Drosophila AMPs replace larval midgut epithelial cells to form the adult midgut epithelium during metamorphosis (Figs 1 and 8)(Bender et al., 1997; Jiang et al., 1997; Li and White, 2003). Furthermore, we show that the majority of AMPs lose esgGal4-driven GFP expression as they differentiate to form the new adult midgut epithelium(Fig. 1F-J). These cells lacked Prospero, which marks enteroendocrine cells in both the larval and adult midgut (Micchelli and Perrimon,2006; Ohlstein and Spradling,2006). They went through several rounds of endoreplication during late pupal development (not shown), and thus probably all differentiated into adult enterocytes (ECs). During early metamorphosis, some cells in the new midgut epithelium remained small and diploid and maintained strong esgGal4 expression (Fig. 1I,J; Fig. 8). For several reasons, we believe that these esg-positive cells are the future adult intestinal stem cells (ISCs). First, esgGal4 expression marks AMPs, including adult ISCs and enteroblasts(Micchelli and Perrimon,2006). Second, mitoses in the adult midgut are only observed in ISCs (Micchelli and Perrimon,2006; Ohlstein and Spradling,2006), and we observed mitoses only in the esg-positive cells during metamorphosis (see Fig. S1 in the supplementary material). Third, esg-positive cells migrated to the basal side of the midgut epithelium (Fig. 1J), the location of adult ISCs (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006). Fourth, AMP clones generated during early larval development contained just a few esg-positive cells when the new adult midgut first formed (24 hours APF) (see Fig. S1C-C″ in the supplementary material), but when such clones were scored in newly eclosed adults, they contained large numbers of ECs, as well as cells positive for the enteroendocrine marker Prospero and the ISC marker Delta(Fig. 2D,E). This suggests that a small fraction of AMPs differentiate into adult ISCs. However, esg-positive cells in the new pupal midgut lacked Delta expression until eclosion (Fig. 2E-E‴; data not shown), suggesting that they are probably not mature adult ISCs.
Postembryonic development of the Drosophila midgut epithelium. AMPs (green) proliferate in two phases and several EGFR ligands are involved in each phase. Also note the specification of future adult intestinal stem cells during early metamorphosis and the reappearance of enteroendocrine cells (red) at a late stage of metamorphosis (72 hours APF). See text for details.
Postembryonic development of the Drosophila midgut epithelium. AMPs (green) proliferate in two phases and several EGFR ligands are involved in each phase. Also note the specification of future adult intestinal stem cells during early metamorphosis and the reappearance of enteroendocrine cells (red) at a late stage of metamorphosis (72 hours APF). See text for details.
How a small fraction of AMPs are selected to become adult ISCs in the newly formed pupal midgut epithelium is not known. One possibility is that the adult ISCs are determined during larval development, long before the formation of the adult midgut. Another is that they are specified during early metamorphosis. We prefer this second hypothesis for several reasons. First, in our lineage analysis, we found that all AMP clones induced during early larval stages formed multiple clusters (Fig. 2A-A″; see Fig. S3A-A″ in the supplementary material). This suggests that there are no quiescent AMPs in the larval midgut. Second,when AMP clones were induced at mid-third instar, the mosaic clusters always contained multiple GFP-positive cells, suggesting that all AMPs in the mid-third instar midgut remain equally proliferative(Fig. 2B-B″). Third,during larval development, we never observed differentiation of the AMPs, as judged by their ploidy (diploid) and lack of expression of the enteroendocrine marker Prospero (not shown). Fourth, all AMPs appeared to express esgGal4 throughout larval development. Given the crucial role that Notch signaling plays in regulating AMPs during embryonic midgut development(Hartenstein et al., 1992) and ISCs in adult midgut homeostasis(Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006; Ohlstein and Spradling, 2007),we expect that Notch might also function to specify adult ISCs during metamorphosis.
Implications for EGFR/RAS signaling in insect midgut development
EGFR signaling is both required and sufficient to promote AMP proliferation(Figs 3, 4, 6 and 7; see Fig. S2 in the supplementary material). Hyperactivation of EGFR signaling, such as by expression of activated Ras (RasV12), promoted massive AMP overproliferation and generated hyperplastic midguts that were clearly dysfunctional (Fig. 3F). On the other hand, inhibiting EGFR/RAS/MAPK signaling dramatically reduced AMP proliferation(Fig. 4; see Fig. S2 in the supplementary material). Furthermore, the ability of EGFR signaling to induce ectopic AMP proliferation is almost unique. With the exception of larval hemocytes (Zettervall et al.,2004), activated EGFR signaling does not promote cell proliferation in the imaginal discs, salivary gland imaginal rings, abdominal histoblasts, foregut and hindgut imaginal rings. This suggests that the regulation of AMP proliferation is different from that in other imaginal cells.
Regulation of AMP proliferation by non-epithelial muscle cells
Despite the obvious differences between adult ISCs and their larval progenitors, the AMPs, there are also similarities. First, when the new adult midgut epithelium forms, larval AMPs give rise to the new adult midgut including the adult ISCs. Many genes, such as esg, that are specifically expressed in the larval AMPs are also expressed in the adult ISCs(our unpublished data). Second, the structure of the midgut epithelium with basal AMPs or ISCs is similar in larval and adult stages. Third, vnexpression in larval VM persists in the adult midgut (our unpublished data),suggesting that Vn from the adult VM might also regulate the ISCs.
In two Drosophila stem cell models, the testis and ovary, stem cells reside in special niches comprising other supporting cell types. These niches maintain the stem cells and provide them with proliferative cues(Ohlstein et al., 2004). For example, in the testis, germ stem cells attach to the niche that comprises cap cells. The cap cells release Jak/Stat and BMP ligands [Upd (Os) and Gbb/Dpp],which maintain the stem cells and induce their proliferation. Whether Drosophila ISCs utilize supporting cells that constitute a niche remains unclear. Here we show that multiple EGFR ligands are involved in the regulation of Drosophila AMP proliferation. During early larval development, the midgut VM expresses the EGFR ligand vn(Fig. 5C-C″), which is required for AMP proliferation (Fig. 7; see Fig. S3B-B″ in the supplementary material). Thus, the early AMPs might be considered to require a niche comprising non-epithelial VM. Later in larval development, however, the AMPs express two other EGFR ligands, spi and Krn(Fig. 5E,F), which are capable of autonomously promoting their proliferation(Fig. 6) and may render vn dispensable (Fig. 7E,E′; see Fig. S3B-B″ in the supplementary material). We found, however, that depleting spi and Krn in the AMPs did not affect AMP proliferation, suggesting that vn or another trigger of EGFR/RAS/MAPK activity might complement spi and Krn in late-stage larvae.
We thank Amanda Simcox, Matthew Freeman, Jocelyn McDonald, Ryu Ueda,Celeste Berg, Laura Johnston, Andrew Dingwall, Margaret Fuller and the NIG(Japan), VDRC (Vienna) and Bloomington (USA) Stock Centers for providing flies; the FHCRC EM Laboratory for preparing larval midgut sections; the Moen's lab for their help with confocal imaging; two anonymous reviewers for their helpful comments; and members of Edgar lab for their support, especially Dr Tao Wang, Dr Parthive Patel and Aida de la Cruz for critical reading of the manuscript. This work was supported by pilot funds from the UW/FHCRC Cancer Consortium and NIH grant R01 GM51186 to B.A.E. Deposited in PMC for release after 12 months.
References
Supplementary information
Adobe PDF
Adobe PDF