Multipotent Drosophila Intestinal Stem Cells Specify Daughter Cell Fates by Differential Notch Signaling

See allHide authors and affiliations

Science  16 Feb 2007:
Vol. 315, Issue 5814, pp. 988-992
DOI: 10.1126/science.1136606


The adult Drosophila midgut contains multipotent intestinal stem cells (ISCs) scattered along its basement membrane that have been shown by lineage analysis to generate both enterocytes and enteroendocrine cells. ISCs containing high levels of cytoplasmic Delta-rich vesicles activate the canonical Notch pathway and down-regulate Delta within their daughters, a process that programs these daughters to become enterocytes. ISCs that express little vesiculate Delta, or are genetically impaired in Notch signaling, specify their daughters to become enteroendocrine cells. Thus, ISCs control daughter cell fate by modulating Notch signaling over time. Our studies suggest that ISCs actively coordinate cell production with local tissue requirements by this mechanism.

Stem cells in adult tissues frequently reside in specific anatomical positions known as niches, whose microenvironment represses premature differentiation and controls proliferation (1, 2). Several different signal transduction pathways—including BMP (bone morphogenetic protein), JAK/STAT (Janus kinase/signal transducer and activator of transcription), Wnt, and Notch—function in well-characterized niches such as those at the apex of the Drosophila ovariole (35) and testis (6, 7), in the Caenorhabditis elegans gonad (8), or within vertebrate bone marrow (9, 10). Most of these stem cells are known to passively receive signals from surrounding stromal cells that inhibit differentiation, and to generate a single type of daughter cell (which may specialize further after subsequent divisions). Recently, intestinal stem cells (ISCs) that require Notch signaling for normal function were described in the Drosophila adult posterior midgut (Fig. 1A) (11, 12). In contrast to previously studied Drosophila stem cells, ISCs generate two different cell types, enterocytes (ECs) and enteroendocrine (ee) cells, without intervening divisions and are not associated with specific anatomical sites containing candidate stromal partners (11).

Fig. 1.

ISCs usually divide asymmetrically with respect to Delta expression and Notch pathway activation. (A) Lineage analysis (11) shows that individual ISCs give rise to daughters (EB) that become enterocytes (EC) and less frequently enteroendocrine cells (ee). (B) Midgut cells express Delta (red) at high levels [arrow, also (C)] or low levels [arrowhead, also (D)] within one cell of each diploid cell nest. Here and in images in later figures, DNA is shown in blue. (E) 3D reconstruction of a 4-day ISC clone (green) showing that the Delta-expressing cell (red) is basal, the known location of ISCs (arrow). Dashed arrow indicates order of clone growth. (F) Notch (red) is expressed widely, except in mature ECs or ee cells. (G) Notch reception (green) is up-regulated in EBs but not in Delta-positive ISCs (arrow). (G′) Notch reception reporter channel (white) alone. (H and H′) A rare nest with two Delta-positive cells; colors as in (G). (I) Model: A Delta-rich ISC activates Notch signaling (arrow) in its daughter EB. Scale barin (B), 20 μm.

To investigate the molecular regulation of ISC multipotency, we stained midguts with antibodies specific for the Notch ligand Delta. Under steady-state nutritional conditions (13), ISCs are found primarily within small clusters of two or three diploid cells (“cell nests”; Fig. 1B) dispersed among the monolayer of polyploid ECs lining the gut. Each cell nest contains a single cell (or very rarely two cells) that expresses Delta (Fig. 1B), although the Delta levels in the positive cell vary (compare Fig. 1, C and D). Labeling is prominent in large punctate structures resembling endocytic vesicles (Fig. 1, C and D) similar to those observed in other cells with active Delta-Notch signaling (14, 15).

The cell expressing Delta was shown to be the ISC by generating marked stem cell clones. At the desired time, flies of the appropriate genotype were heat-shocked to initiate a recombination, mediated by the enzyme FLP, that activates heritable lacZ expression in rare, random dividing cells (13). On the basis of previous studies (11), ISCs can be identified in such clones as the initial LacZ-positive cell, and the ISC nucleus is known to be located closest to the basement membrane. Guts were isolated at various times after clone induction, stained for LacZ and Delta, and analyzed as three-dimensional (3D) optical stacks (Fig. 1E). We observed that the ISC was always Delta-positive, whereas prospective ECs or ee cells lacked Delta. Thus, in the adult midgut, Delta expression serves as a highly specific stem cell marker.

All cells within the nests express Notch (11) (Fig. 1F). Consequently, to determine which gut cells actually receive Notch signals, we used a transgene containing tandem binding sites for the transcription factors Grainyhead and Suppressor of Hairless [Su(H)] that acts as a sensitive reporter for Notch activation (16). When guts from adults bearing this reporter were counterstained for Delta, we observed that the Notch reporter was highly expressed in the daughter cell [the enteroblast (EB)] of ISCs with high Delta levels (Fig. 1G) but was at background levels in the ISC itself (Fig. 1G′). In rare cell nests with two Delta-positive cells, neither one strongly expressed the Notch reporter (Fig. 1, H and H′). The strong inverse relationship between Delta levels and Notch activation suggested that ISCs signal via Delta to activate Notch in their daughters (Fig. 1I).

We generated MARCM (mosaic analysis with a repressible cell marker) clones (13) of several different Notch pathway genes to test this hypothesis. The green fluorescent protein (GFP) marker gene and mutant gene produced in such experiments segregate initially into just the ISC or just its daughter, yielding two possible outcomes (Fig. 2A). ISC clones of the null allele DlrevF10 (17) produced the results expected if Delta is an essential ligand of intestinal Notch signaling. Many Delta mutant clones grew into tumors (Fig. 2B) indistinguishable from those generated previously by loss of Notch function (11), or into one EC and a tumor (Fig. 2C), because of transient perdurance of Delta protein. Perdurance probably also explains why, like N tumors (11), the Dl tumors are frequently mosaic for ee-like cells expressing Prospero and for ISC-like cells that lack Prospero (Fig. 2B). In many other Dl clones the marked cell differentiated as a single EC (Fig. 2D), presumably because only the EB was mutated and it could still receive a Notch signal from the adjacent wild-type ISC (Fig. 2A). This implies that Delta is not needed outside the ISC for EC differentiation, consistent with the rapid down-regulation of Delta protein levels in ISC daughter cells. In contrast, clones mutant for a null allele of Serrate (SerRX106), encoding the other major Notch ligand, showed wild-type development (Fig. 2E). These experiments show that a Delta signal from the ISC is functionally required for daughter cells to exit the mitotic cell cycle and differentiate into ECs.

Fig. 2.

Asymmetric Delta signaling from ISC to EB is functionally required for EC production and mitotic cell cycle exit. (A) Expected outcomes of ISC clones (green) lacking Dl (or neu); Notch signal (N) is shown by an arrow. Four-day ISC clones (green) of DlrevF10 generate (B) a tumor mosaic for Prospero (red), (C) a tumor plus a single EC, or (D) single ECs. (E) Four-day clone (green) of the SerrateRX106 null allele showing normal ISC (arrowhead) and EC production (arrows). Armadillo (red) highlights cell boundaries. (F) Seven-day ISC clones (green) of neu11 form tumors (arrow) mosaic for Prospero expression (red, nuclear), or single mature ECs (arrowhead). Delta (red, cytoplasmic). (G) The red channel from (F); cells lacking Prospero contain plasma-membrane Delta. Cells expressing Prospero lack Delta. (H) Twelve-day ISC clone (green) of the Su(H)047 null allele forms a tumor mosaic for Prospero (nuclear, red). (I) Expected outcomes of ISC clones (green) lacking Su(H) (or N). Impaired Notch signal reception in EB is denoted by a blocked arrow. (J) Notch clone (green) not expressing Prospero (nuclear, red) expresses Delta (cytoplasmic, red) like ISCs. (K) Delta (red) and DNA (blue) from (J). (L) Notch clone (green) expressing Prospero (nuclear, red) lacks Delta (cytoplasmic, red).

Other Notch pathway genes were removed in MARCM clones within ISCs to further analyze ISC signaling. Neuralized (neu) encodes an E3 ubiquitin ligase that is required in the signaling cell for Delta endocytosis and activation (15). ISC clones of neu generated tumors mosaic for Prospero expression (Fig. 2F, arrow) or else single ECs (Fig. 2F, arrowhead), the expected result for a gene required in the ISC (Fig. 2A). Moreover, Delta protein in such tumors was predominantly adjacent to the plasma membrane (Fig. 2G), consistent with the known requirement for neu in Delta endocytosis (18, 19). Clonally removing Su(H), the CSL transducer of the canonical Notch signaling pathway (15), also generated tumors mosaic for Prospero (Fig. 2H). Unlike clones of the signaling cell genes, about half of which produce single ECs, Su(H) ISC clones, like Notch ISC clones (11), generated only tumors, indicating that these genes are required in the EB (Fig. 2I). Tumor cells lacking Prospero contained high levels of Delta (Fig. 2, J and K), which supports the idea that they are stem cell–like. Prospero-expressing tumor cells contained very little Delta (Fig. 2, G and L). Thus, ISCs signal their pre-EC daughters via the canonical Notch pathway.

The striking asymmetry in Delta content between ISCs and their pre-EC daughters might arise by asymmetric Delta segregation. Consequently, we identified mitotic ISCs at various stages by staining with Anillin, and analyzed the distribution of Delta between the daughter cells. Similar amounts of Delta vesicles were observed in both the newly forming ISC and the EB, even in telophase (Fig. 3A). However, in cell pairs that had just completed mitosis, as evidenced by the reuptake of Anillin by daughter cell nuclei, Delta-rich vesicles clustered in the ISC near the former cleavage furrow, whereas they were distant from the furrow and weaker in the EB (Fig. 3B). These experiments show that Delta asymmetry does not arise simply by differential segregation of Delta protein; rather, Delta appears to be rapidly down-regulated in the EB after mitosis.

Fig. 3.

Delta vesicles are distributed asymmetrically shortly after mitosis. (A) Equal distribution of Delta vesicles (red) during ISC anaphase. Cytokinesis furrow is shown by Anillin stain (green). (B) Down-regulation of Delta (red) in the newly formed EB and transient asymmetric positioning of vesicles within the ISC (arrow); Anillin (green) shows nuclei. (C and D) Mitotic ISCs in 3D reconstruction. Phosphohistone H3 (green) and γ-tubulin (red) staining define the division angle (ϕ) relative to the basement membrane (blue and red axes). The ISC in (D) was also stained for α-tubulin (red). (E and F) An ISC-EB pair stained for Delta (red) and Notch activation reporter (green) reveals that ISCs have greater contact (dashed line) with the basement membrane (bm).

To investigate the mechanism of ISC multipotency, we first considered a role for mitotic spindle orientation. Mitotic chromosomes were stained with phosphohistone H3 and centrosomes with γ-tubulin (Fig. 3, C and D) and cells were analyzed in 3D reconstructions to determine the angle (ϕ) between the spindle and the basement membrane. These experiments revealed that ISCs divide nonrandomly, with values of ϕ ranging between 3° and 47° (mean = 29° ± 14°; N = 50). Using cytoplasmic markers, we found that the ISC (Fig. 3E, red) typically contains a much greater surface area of contact with the basement membrane than does its daughter (Fig. 3, E and F, green), presumably as a result of this divisional orientation. Division away from the basement membrane also explains why the nucleus of the ISC was previously observed to lie more basally than other cell nuclei (11). No correlation between spindle orientation and daughter cell fate was discovered, however.

We next considered whether Notch signal reception might differ in EBs fated to become ECs and those fated to become ee cells. We generated marked ISC clones of two to four cells and stained them for Prospero and Delta (Fig. 4). The ISC in these clones could be determined by its basal location; hence, the fate of its last one to three daughters (in two- to four-cell clones, respectively) could be deduced on the basis of Prospero expression. ISCs that had just produced a pre-EC invariably expressed Delta in cytoplasmic vesicles, usually at high levels (Fig. 4A, N > 100). In striking contrast, ISCs that had just produced an ee precursor contained few or no cytoplasmic Delta vesicles (Fig. 4, B to D, N = 20). This was true for ISCs that had just switched from EC to ee production (Fig. 4B) as well as those that had just completed a second consecutive ee (Fig. 4, C and D). Consecutive production of exactly two ee daughters appears to be the rule, as we observed no counterexamples in these experiments. However, ISCs that had recently completed an ee pair and had just returned to EC production contained multiple Delta vesicles (Fig. 4E).

Fig. 4.

(A to E) Delta expression levels in ISCs determine EC versus ee fate. Examples of two- to four-cell ISC clones (green) stained for Prospero (nuclear, red) and Delta (cytoplasmic, red) are shown. Arrowheads denote ISCs; arrows denote prospective ECs. (A) ISCs producing a prospective EC (Prospero absent) always express Delta. (B) ISCs that have just switched to ee cell (Prospero present) production express little or no Delta. (C and D) ISCs producing a second consecutive ee cell express little or no Delta. Inset: Clonal marker for (C). (E) ISCs that have just switched from ee to EC production express Delta. (F) ISC clones (green) expressing Nact generate one or two ECs, indicating that such ISCs are not maintained. Scale bar, 20 μm.

These results suggest a simple model for the determination of ISC daughter cell fate. Daughters of ISCs with high Delta levels would receive a strong Notch signal and become ECs. Daughters of ISCs with few or no detectable cytoplasmic Delta vesicles would receive a much weaker Notch signal and become ee cells. If Notch signaling were disrupted entirely, as in Notch pathway null mutant clones after one or two divisions, only ISC-like daughters would be produced. Neither ee precursors nor ISC-like cells would cease division when Notch signaling is impaired. We tested this model by expressing an activated Notch receptor (Nact) in MARCM ISC clones (Fig. 4F). Nact-expressing cells, which should contain a high level of Notch signaling, all became ECs (Fig. 4F). In addition, none of these clones grew above two cells in size, which shows that Notch activation within the ISC is sufficient to force its differentiation as an EC.

These experiments provide insight into several long-standing issues in stem cell biology. First, they illustrate that highly specific stem cell markers can be obtained, at least for a particular stem cell under particular conditions. Vertebrate ISCs cannot currently be identified with certainty [reviewed in (20)], but Notch signaling is similarly involved in secretory cell versus EC production. Consequently, it might be possible to identify vertebrate ISCs as a subpopulation of cells enriched in Notch ligands within regions of the vertebrate gut that are known to contain ISCs. Second, our results suggest how stem cell activity may be coordinated with tissue requirements (fig. S1A). Local feedback signals from existing ee cells and other cells may influence an ISC's Notch signaling capacity and hence the type of cell produced in that tissue region.

Drosophila neuroblasts and sensory organ precursors are programmed by the asymmetric segregation of molecules that bias the subsequent transmission of Notch signals [reviewed in (21)]. Differential Notch signaling requires the asymmetric endocytic trafficking of Delta and Notch after precursor cell division [reviewed in (15)]. Recycling endosomes, an essential component of Notch signaling, localize to pericentromeric regions and have been reported to segregate differentially in Drosophila sensory organ precursor divisions (22). Our studies suggest that endocytic trafficking of Delta may also be involved in ISC function. We observed that Delta-rich vesicles accumulate nonrandomly in the ISC shortly after mitosis near the division plane. Such a process might result from the inheritance of an endocytic subcompartment by the ISC but not its daughter, leading to differences in Delta stability between the two cells.

Finally, our experiments suggest that Drosophila ISCs play a role different from that of stem cells in previously characterized niches. In the ovarian germline stem cell (GSC) niche (1), the GSC adheres to adjacent stromal cells at an anatomically fixed location and responds to local signals that block differentiation (fig. S1B). GSC daughters differentiate because they are forced outside this zone of inhibition. In contrast, the ISC plays an active role in signaling via Notch to induce either EC or ee differentiation. It remains to be determined whether external signals are also important in maintaining ISCs, in a manner analogous to GSCs. In any case, stem cells such as ISCs, which sense local cellular requirements and produce appropriate daughter cells in response, likely play a central role in the physiology, longevity, and pathology of the tissues they maintain.

Supporting Online Material

Materials and Methods

Fig. S1


References and Notes

View Abstract

Navigate This Article