The origin of intestinal stem cells in Drosophila


  • Craig A. Micchelli

    Corresponding author
    1. Department of Developmental Biology, Washington University School of Medicine, St. Louis, Missouri
    • Department of Developmental Biology, Washington University School of Medicine, McDonnell Sciences Building, Rm. 328, Campus Box 8103, 660 South Euclid Avenue, St. Louis, MO 63110.

    Search for more papers by this author


Renewing tissues in the adult organism such as the gastrointestinal (GI) epithelium depend on stem cells for epithelial maintenance and repair. Yet, little is known about the developmental origins of adult stem cells and their niches. Studies of Drosophila adult midgut precursors (AMPs), a population of endodermal progenitors, demonstrate that adult intestinal stem cells (ISCs) arise from the AMP lineage and provide insight into the stepwise process by which the adult midgut epithelium is established during development. Here, I review the current literature on AMPs, where local, inductive and long-range humoral signals have been found to control progenitor cell behavior. Future studies will be necessary to determine the precise mechanism by which adult intestinal stem cells are established in the endodermal lineage. Developmental Dynamics 241:85–91, 2012. © 2011 Wiley Periodicals, Inc.


Stem cells are among the most primitive cells of a lineage and are distinguished by the properties of self-renewal and multipotency. Stem cells are, thus, ideally suited for a central role in tissue establishment and homeostasis where they often function to replenish the differentiated cells of an organism that are lost through turnover or acute injury. Not surprisingly, defects in stem cell biology underlie the processes of aging and disease, biological states characterized by homeostatic imbalance. Therefore, uncovering the factors controlling stem cell specification and behavior will ultimately serve to unify our understanding of development and disease.

The Drosophila midgut has recently emerged as a genetic model system to study fundamental principles of epithelial stem cell biology. Striking parallels between the insect and mammalian gut have served to stimulate research in this field. The Drosophila gastrointestinal (GI) tract, like its mammalian counterpart, is lined by an epithelial monolayer, which is specialized for nutritional uptake and protection of the organism from the environment. The midgut contains both absorptive and secretory cells, which are maintained by a hierarchically organized intestinal stem cell lineage (ISC; Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006). Moreover, the ISC lineage is controlled by a number of conserved signal transduction pathways including Notch (N), Epidermal growth factor (EGF), Bone morphogenetic protein (BMP), Hedgehog, (Hh) Wnt, and Hippo. Recent reviews have highlighted the progress currently being made in understanding homeostasis in the adult (Wilson and Kotton, 2008; Lin and Xi, 2008; Casali and Batlle, 2009; Wang and Hou, 2010; Hou, 2010; Apidianakis and Rahme 2010; Karpowics and Perrimon, 2010; Hartenstein et al., 2010).

The fact that certain adult tissues rely on stem cells for renewal raises the questions of where, when, and how a stem cell first arises within a tissue and the related question of how sufficient numbers of stem cells and differentiated daughters are generated, to keep pace with the rapid growth experienced during development. Addressing these issues is challenging in vertebrate tissues due, in part, to the difficulty of identifying and manipulating stem cells in their native microenvironments during embryogenesis. However, the Drosophila midgut with its accessibility throughout development and excellent genetic tools provides a tractable experimental model to dissect stem cell dynamics at spatio-temporal resolution.

Several recent studies have exploited these advantages and focused on the analysis of adult midgut precursor cells (AMPs), endodermally derived progenitors, which give rise to adult ISCs. Cells similar to AMPs have been widely documented in other insect species where they are called stem cells, regenerative cells, or replacement cells; all share the ability to proliferate extensively and generate differentiated daughters in response to the demands of growth or injury. While there is now clear evidence to support the idea that ISCs descend lineally from AMPs, it remains unclear how different these progenitor cells are at both the genetic and epigenetic levels and the process by which any differences might arise. Here, I review advances that have begun to shed light on the molecular mechanism controlling AMP behavior during Drosophila development.


In Drosophila, the GI tract is comprised of three distinct anatomical regions: foregut, midgut, and hindgut. While the foregut and hindgut are ectodermally derived, the midgut is of endodermal origin. During embryogenesis, endoderm is first specified as two distinct primordia positioned at the embryo's termini (Fig. 1). Cells of the endodermal rudiment first invaginate and then undergo an epithelial-mesenchymal transition (EMT), to produce anterior and posteriorly localized mesenchymal cell masses. Mesenchymal cells migrate towards each other and eventually meet in the middle of the embryo. Midgut cells are then deposited on the surface of two lateral sheets of visceral muscle (VM) flanking the embryo. These cellular sheets close over at their dorsal and ventral aspects to form the embryonic midgut by the end of embryogenesis.

Figure 1.

Overview of embryonic midgut development. The embryonic endoderm is specified as two regions at the embryonic termini (dashed regions). During gastrulation, the endoderm invaginates to form an anterior and posterior rudiment. Endodermal cells undergo EMT to create loosely organized mesenchyme, which contains distinct midgut cell types including AMPs. Following contact with flanking tracts of visceral muscle, endodermal cells undergo MET using the visceral muscle as substrate. Following embryogenesis, AMPs are present in close association with the visceral muscle in first instar larvae. At the end of three larval stages during the process of metamorphosis, the midgut will be shed from its visceral mesoderm scaffold and replaced by AMP daughters to form a new midgut epithelium.

The embryonic midgut is retained through the course of larval development. However, the differentiated cells of the epithelium and the discrete AMPs contained within it have two very different fates. At the end of larval development, during metamorphosis the midgut delaminates from the visceral mesoderm and basement membrane, which serves as a scaffold and is shed into the gut lumen. Coordinately, AMPs and their daughters fuse to form the presumptive adult midgut epithelium. The molecular control of this step-wise process is considered in greater detail below.


The earliest specification events in the endodermal lineage depend on several transcription factors including those encoded by huckebein (hkb), fork head (fkd), and serpent (srp) genes. Hkb encodes a zinc finger transcription factor and zygotic hkb is the earliest known transcription factor required in the endodermal lineage (Bronner et al., 1994). Loss of hkb leads to a loss of endoderm while misexpression blocks mesodermal and ectoderm formation. Fork head, the founding member of the FOX family of winged-helix transcription factors, is expressed during the blastoderm stage at the embryonic termini (Weigel et al., 1989). Presumptive endoderm successfully invaginates in fkd mutants, but fails to subsequently be maintained. Finally, serpent, the GATA-type transcription factor, is expressed in the endodermal rudiment, but the absence of srp results in a failure of epithelial-mesenchymal transition, and subsequent endodermal differentiation (Reuter, 1994). Studies in vertebrates show that both GATA and FOX family transcription factors are required for endoderm specification (Zaret, 1999; Stainier, 2002). Thus, there is molecular conservation in the transcriptional regulators required for endoderm development between invertebrates and vertebrates.

escargot (esg), a member of the snail/slug superfamily of transcription factors, is expressed in the early endoderm as it invaginates during embryogenesis (Takashima et al., 2011). This early expression of esg in the endodermal rudiment is, perhaps, noteworthy, as esg is the only transcription factor known to be expressed in adult ISCs (Micchelli and Perrimon, 2006). In addition, esg expression is retained in adult midgut precursors (AMPs) during both larvae and pupal stages (Jiang and Edgar, 2009; Micchelli et al., 2011; see AMP CELL LINEAGE section below). Currently, it is unclear which, if any, of the other transcription factors known to characterize the early endodermal rudiment are expressed and function in AMPs.


A number of molecular and morphological criteria have been employed to identify AMPs during both embryonic and larval stages. For example, homotypic transplantation experiments of histologically labeled cells demonstrate the presence of single “spindle cells” interspersed among cells of the embryonic midgut by stage 16 (Technau and Campos-Ortega, 1986). A relatively small cell size has also been used to follow AMPs (Hartenstein et al., 1992; Tepass and Hartenstein, 1994, 1995). The first putative molecular markers emerged from early enhancer trapping studies and include the B11-2-2 and A490.2M3 lines, which are expressed in embryonic stage 13 in a distinct group of apically localized cells in the midgut (Bier et al., 1989; Hartenstein and Jan, 1992). Finally, expression of prospero (pros) at embryonic stage 11 (Oliver et al., 1993; Spana and Doe, 1995; Hirata et al., 1995) and scute (sc) at stage 11–12 (Brand et al., 1993; Tepass and Hartenstein, 1995) have also been used as AMP markers.

Early observations in the blowfly suggested that AMPs could be identified at early larval stages, interspersed among large enterocytes based on morphology alone (Kowalevsky, 1887; Perez, 1910). Recent studies demonstrate that esg and the N ligand encoded by Delta (Dl) are both expressed in individual AMPs in the first instar larvae (Jiang and Edgar, 2009; Mathur et al., 2010; Micchelli et al., 2011, Takashima et al., 2011). Estimates of the number of AMPs generated during embyrogenesis range from 45–121 cells, depending on the marker gene(s) used.


AMP cell fate, as well as the fate of other midgut cells types, is first determined in the invaginated endodermal rudiment (Fig. 2; Hartenstein et al., 1992, Tepass and Hartenstein, 1994, 1995). Between embryonic stage 10–11, cells delaminate in waves from the rudiment to form a mesenchymal cell mass consisting of at least three cell types: AMPs, enterocytes, and interstitial cell precursor (ICP) cells. Then, in a defined order, these mesenchymal cells coalesce on visceral muscle to form the embryonic midgut epithelium. Enterocytes are the first cell type to undergo this mesenchymal-epithelial transition (MET) and form a monolayer adherent to the VM. However, AMPs (and ICPs) remain temporarily associated with the apical surface of the newly formed epithelium. The exact time at which AMPs invade the epithelium is still not defined. However, by the embryonic/larval transition AMPs are clearly present in association with the VM (Micchelli et al., 2011). This anatomical position is similar to the intestinal stem cell niche of the adult midgut.

Figure 2.

AMP segregation from the endodermal rudiment during embryogenesis. Expression of AS-C and E(spl) genes (shaded cells) defines subdomains within the invaginated endoderm. Notch signaling is necessary to restrict the number of AMPs produced by the endoderm. AMPs undergo epithelial to mesenchymal transition (EMT) together with other cells of the presumptive midgut. In defined sequence, endodermal cells undergo a mesenchymal to epithelial transition (MET), once endoderm and visceral mesoderm come into contact. AMPs remain apically localized during embryogenesis; contact with the basement membrane is established by the beginning of the first larval instar and AMPs exhibit a distinct pyramidal shape.

AMPs are initially determined within the invaginated rudiment through the combined activity of Enhancer of split complex (E(spl)-C) and achaete-scute complex (AS-C) genes, as well as the Notch signaling pathway. Both E(spl)-C and AS-C gene expression is dynamic and marks spatially delimited expression patterns within the endoderm during stages 9–11 (Tepass and Hartenstein, 1995). Genetic analysis shows that reduction of AS-C function in the embryo leads to a loss of AMPs (and ICPs), without otherwise disrupting midgut epithelium. These observations suggest that AS-C is necessary for AMP specification within the endodermal rudiment. In contrast, loss of N pathway components leads to an increased number of AMPs (and ICPs) at the expense of enterocytes (Hartenstein et al., 1992; Tepass and Hartenstein, 1995), a phenotype similar to what has been observed in adult ISCs (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006). The precise timing of this N requirement is not known, nor is the location of the sending and receiving cells, issues complicated by an earlier requirement for N in endoderm invagination (Hartenstein et al., 1992). Visceral mesoderm is unlikely to be a source of the signal, as AMPs are still detected in embryos lacking mesoderm (twist, snail double mutant). Thus, competence to become an AMP requires AS-C genes, while AMP number is controlled by N-mediated signaling events. This two-step process occurs autonomously within the midgut rudiment and does not appear to involve signals from neighboring mesodermal tissues.


Narrative accounts of cell fate for a specific marker gene are inherently limited by restricted spatial and temporal resolution. In the case of the early midgut rudiment, this problem is compounded by the tissues' mesenchymal organization, extensive cell migration, and dynamic gene expression (Tepass and Hartenstein, 1994, 1995). Thus, the extent to which reported AMP markers label equivalent or distinct cell populations or whether they in fact correspond to adult midgut progenitors has remained unclear.

Recently, cell lineage–tracing experiments have begun to clarify our understanding of the AMP cell lineage (Fig. 3). First, lineage-tracing experiments conducted during larval development demonstrate that AMPs are present in first instar larvae (Jiang and Edgar, 2009; Micchelli et al., 2011). Second, cell lineages labeled in larvae give rise to all differentiated cells of the adult midgut including ISCs, enterocytes, and enteroendocrine cells (Jiang and Edgar, 2009; Micchelli et al., 2011). Third, directed cell-labeling experiments demonstrate that esg+ cells present in the first instar give rise to the adult midgut (Micchelli et al., 2011). Taken together these lineage studies support the view that esg+ cells present at the embryonic/larval transition define the AMP population.

Figure 3.

AMP lineage during larval and pupal development. AMPs give rise to cells of the adult midgut in a step-wise process. AMPs lie in contact with the basement membrane (BM) and visceral muscle that surround the gut epithelium at the beginning of larval development. AMP numbers increase by symmetric expansion and progenitors disperse throughout the rapidly growing tissue remaining in close contact with the basement membrane. During third instar, AMPs undergo at least one asymmetric division to produce a peripheral cell (PC). PC cells surrounding an expanded cluster of AMPs form a transient niche by white prepupal (WPP) stage. Reorganization of the midgut during metamorphosis yields differentiated enterocytes (EC) and single AMPs, which comprise the pupal midgut. AMP number again increases through symmetric division, with enteroendocrine (ee) daughters produced just prior to eclosion.

On average, AMPs give rise to the adult midgut through a series of seven to ten divisions, most of which occur during larval stages (Fig. 3; Jiang and Edgar, 2009; Mathur et al., 2010; Micchelli et al., 2011; Takashima et al., 2011). This process appears to involve two types of divisions: symmetric divisions, which increase AMP number, and asymmetric divisions, which produce lineage-restricted daughters. For example, during larval stages, AMPs initially expand through a series of symmetric divisions to increase their numbers as the midgut rapidly increases its size. AMPs then undergo at least one asymmetric division during third instar to initiate AMP cluster formation; subsequent larval divisions are symmetric. During pupal stages, symmetric division again serves to increase the number of AMPs as the midgut increases its size following metamorphosis. AMPs switch to an asymmetric mode of division in late pupal stages to generate the endocrine cells of the adult midgut prior to eclosion.

The number of AMP divisions during development is also tightly controlled. For example, during early larval stages, AMPs undergo 3–4 symmetric divisions. However, on average only one symmetric AMP division occurs during pupal stages (Micchelli et al., 2011). Thus, the pattern and extent of AMP divisions are both tightly regulated during development.


Genetic screens have aimed to identify key regulators of AMP proliferation. Several lines of evidence suggest that EGFR/RAS/MAPK signaling functions as a powerful AMP mitogen (Jiang and Edgar, 2009). Expression studies show that, of the four activating EGF ligands (gurken, spitz, Keren, vein), spitz and Keren are expressed at high levels within the AMP clusters, as is diphospho-extracellular signal-regulated kinase (dpERK) staining, all consistent with a role for MAPK pathway in AMPs. Genetic analysis has shown that blocking EGFR signaling in larvae either conditionally using RNAi or through loss of function mutants, leads to a decrease in both the number and size of the AMP clusters in late third instar larvae. Conversely, expression of activated forms of Egfr, ras, and raf in AMPs each lead to a striking increase in AMP cell number; similar results were also observed following misexpression of the ligands spitz (spi), Keren (Krn) and vein (vn). Together, these observations indicate that EGFR signaling plays a role in both the early and late phases of larval AMP expansion, although direct tests of EGF function on cell proliferation has not been performed. Previous studies indicate that EGF signaling regulates cell adhesion in diverse tissue types (Cela and Llimargas, 2006), so reduced AMP number might also be expected if EGF signaling affected AMP cluster aggregation, maintenance, or retention of AMPs within the epithelium. Thus, EGFR may have additional roles in controlling AMP behavior.


There is evidence that two distinct neighboring midgut cell types influence AMP behavior during larval development. The first is the visceral muscle surrounding the midgut epithelium. The EGFR ligand encoded by vn is first expressed in the visceral muscle during embryogenesis and can be detected in both layers of visceral muscle throughout larval development (Szuts et al., 1998; Jiang and Edgar, 2009). Viable vn mutant combinations that survive through larval development are associated with a failure of AMP maintenance (Jiang and Edgar, 2009). A similar outcome was observed when AMP lineages were labeled in first instar in a vn mutant background demonstrating that vn is required for AMP maintenance. Interestingly, those AMPs that persist in the midgut go on to form clusters suggesting that the vn requirement may be temporally limited and not required during late third instar. Furthermore, vn is required specifically in the visceral muscle; no effects were observed when vn function was blocked in either AMPs or larval enterocytes. Finally, while expression of vn in the muscle was sufficient to rescue vn mutants, expression in AMP did not have a phenotype (Jiang and Edgar, 2009). Thus, signaling between the visceral mesoderm and endoderm is necessary for the maintenance of the AMP population during larval development.

Elegant cell lineage–tracing experiments have demonstrated that the visceral muscle undergoes extensive remodeling during pupal stages (Klapper, 2000). Beginning at the white prepupal (WPP) stage, the muscle surrounding the midgut begins to dedifferentiate and the contractile system starts to break down. From 40–70 hr after puparium formation (APF), the fibers begin to grow and reestablish contractile potential such that by 96 hr APF, visceral muscle is again fully differentiated. Temporal reorganization of VM appears to correlate with newly described changes in AMP behavior. It has been suggested that pupal visceral muscle may signal to the underlying midgut epithelum in an instructive manner (Klapper, 2000). Given the role of visceral mesoderm as a source of niche signals controlling AMPs during larval stages and ISCs in the adult (Jiang and Edgar, 2009, 2010), this hypothesis seems quite plausible.

The second cell type influencing the behavior of AMPs is daughter cells produced through asymmetric division (Fig. 3). At the beginning of third instar larvae, AMPs transition from a symmetric mode of division to increase the total number of AMPs dispersed throughout the midgut to an asymmetric division to form larval AMP clusters (Jiang and Edgar, 2009, Mathur et al., 2010; Micchelli et al., 2011). It has long been known that one or more peripheral cells (PC) temporarily surround each AMP cluster before being “thrown-off” into the gut lumen (El Shatoury and Waddington, 1957). However the mechanism of PC specification and function has only recently been tested. New experiments provide insights into this question and demonstrate that Notch signaling functions to distinguish the AMP and PC cell fate (Mathur et al., 2010). Expression studies show that AMPs express elevated levels of Dl ligand, while PC daughters exhibit high levels of N pathway activation. In addition, N activity is both necessary and sufficient for the specification of PC cells.

Once specified, one or more outer PC cells surround a growing cluster of encapsulated AMPs. PC cells function as a “transient niche” during the remainder of third larval instar promoting AMP self-renewal and preventing newly formed stem cell daughters from differentiating (Mathur et al., 2010). At least part of the mechanism by which PC cells regulate AMPs and their daughters involves the conserved BMP pathway, as abrogating Decapentapalegic (Dpp) signaling leads to premature differentiation of stem cell daughters, as does PC cell ablation (Mathur et al., 2010).

While existing studies suggest that a single founder cell forms each AMP cluster and a single AMP is subsequently liberated following metamorphosis, the mechanism controlling this process is unknown. Single AMPs express high levels of esg and Dl. However, AMPs within a cluster cannot yet be distinguished from one another on the basis of marker gene expression. Therefore, it remains unclear if the cluster forming AMP is retained in the cluster and emerges following metamorphosis, if equipotential AMPs within each cluster signal to determine which cell will remain an AMP, or if AMPs' fate is determined following metamorphosis through interactions with others cells or tissues such as the muscle. How single AMPs arise within the pupal midgut and how niche number is regulated remains an open question.


Do AMPs only respond to local niche signals within the midgut or can the physiological status of the organism also affect AMP behavior? The steroid hormone ecdysone is known to act broadly on different larval tissues to coordinate the otherwise disparate events of insect morphogenesis. In the case of the midgut, function was first suggested by expression studies showing differential ecdysone receptor (EcR) expression in the tissue (Talbot et al., 1993). A number of subsequent studies provided functional evidence that ecdysone is required for normal midgut development (Jiang et al., 1997; Hall and Thummel, 1998; Li and Bender, 2000; Li and White, 2003). However, it remained unclear how direct these effects are on cells of the midgut. Recently, experiments suggest that ecdysone signaling is directly required in developing AMPs. Analysis of dominant-negative EcR transgenes using conditional expression and mosaic analysis both revealed a reduction in the number of cells within AMP clusters by late third instar (Micchelli et al., 2011). It is not clear, for example, if ecdysone affects EGFR signaling, the specification or maintenance of PC cells, or functions by some other mechanism. Thus, AMP expansion is controlled by signals within the midgut as well as by humoral factors.


A number of insect species studied have been found contain to AMPs, including for example, aedes egepti (mosquito); Bombyx mori (silkworm); Manduca sexta (tobacco hornworm); and Melipona quadrifasciata anthidioides (bee), suggesting that AMPs are a general feature of the insect midgut. Nonetheless, the local tissue architecture surrounding AMPs can be captivatingly varied (Shinoda, 1927). These studies are rich in descriptive content, yet cell lineage–tracing and genetic methodologies are often not available to manipulate AMP function in these systems. Notably, the isolation and long-term culture of midgut stem cells have been achieved in a number of insects, including manduca (Loeb, 2010). Studies of cultured AMPs have led to the identification of factors regulating stem cell proliferation and differentiation, such as ecdysone and EGF, whose mechanism of action may be conserved. Recent in vivo analysis suggests that coordinate expansion of oxygen supplying trachael cells surrounding the midgut epithelium may also be limiting for AMP growth in manduca (Nardi et al., 2011). Such studies exemplify the value of studying AMPs in a variety of different organisms, as they serve to uncover the range of mechanisms controlling stem cell behavior, thus broadening our concept of the niche.


Lineage-tracing experiments demonstrate that AMPs, which are established during embryogenesis and reside adjacent to the visceral muscle by the embryonic/larval transition, give rise to ISCs of the adult. This organ-generating lineage undergoes a series of expansionary divisions that amplify AMP number and asymmetric divisions that produce the distinct daughter cells necessary to establish the adult midgut. Studies indicate that AMP behavior is controlled through both local niche signals as well as long-range humoral factors. Interactions between visceral mesoderm and endoderm may emerge as an important regulatory interaction during all post-embryonic stages of development, as has the iterative use of conserved signaling pathways such as EGF and Notch signaling to control AMP proliferation and cell fate.

And yet, exactly where and when intestinal stem cells are specified within the AMP lineage remains unclear. In general, two simple models could theoretically explain the process of stem cell specification (Fig. 4). In one view, the broad developmental potential of the germ layer or embryonic rudiment is retained into adulthood by early sequestration of stem cells into a tissue-specific niche. The stem cell, thus, retains a more primitive developmental state, as cells of the embryo undergo a program of progressive determination and lineage restriction. One prediction of this model is that stem cells should retain a genetic/epigenetic program that is very similar to the embryonic rudiment from which they are derived. This would minimize the need to evolve an independent mechanism to confer cells with stem cell characteristics later in development. An alternative model is that the stem cell state is acquired during development, through one or more signaling events acting on either lineage-specific progenitor cells, the presumptive niche, or both to permit successful stem cell/niche engraftment in the mature host tissue.

Figure 4.

Two different hypothetical models could explain the process by which stem cells arise in adult tissues. In the retention model, a small number of stem cells derived from the embryonic tissue rudiment are sequestered in niches at an early stage of development, and are subsequently maintained in an undifferentiated cell state. In the acquisition model, progenitor cells first expand during development, then through a series of one or more signaling events (dashed arrows), which act on the progenitor and/or the niche, to program engraftment competence.

Further studies will be needed to distinguish these models in the case of the midgut. If experiments support the sequestration model, it will be valuable to further investigate the mechanisms that coordinate the scaling of stem cell/niche number in the midgut with overall tissue growth. What are the mechanisms mediating the switch between symmetric and asymmetric stem cell divisions patterns? In contrast, if future studies support the acquisition model, it will be critical to determine the signals necessary for selection of stem cells from progenitors and/or the establishment of competent niches. And what environmental and physiological factors mediate the switch of stem cells from developmental to homeostatic modes of behavior? Finally, such models may not be mutually exclusive. For example, it is possible that stem cells are sequestered early in tissue-specific niches, but produce daughter cells that retain the capacity to acquire stem cell character, buffering the tissue against developmental stress. Answers to these questions will contribute significantly to our understanding of the factors necessary to specify the stem cell state and role that stem cells play in the establishment and maintenance of epithelial tissues.