Tuberous sclerosis complex and Myc coordinate the growth and division of Drosophila intestinal stem cells

Stem cell–mediated repair is a promising approach for treating a variety of pathological disorders. Many adult tissues contain stem cells, and tissue homeostasis requires replenishment of lost cells by these adult stem cells. An imbalance between the removal of dead cells and the production of new cells can lead to tissue overgrowth, tissue damage, inflammation, and cancer (Niemeyer et al., 2006; Nystul and Spradling, 2006; Metcalfe and Ferguson, 2008).

In the adult mammalian intestine, stem cells are located near the base of each crypt (Crosnier et al., 2006; Yen and Wright, 2006; Walker and Stappenbeck, 2008; Barker et al., 2009). Two groups of cells, called label retention cells and columnar base cells, have stem cell properties but express completely different markers (Barker et al., 2007; Montgomery and Breault, 2008; Sangiorgi and Capecchi, 2008; Zhu et al., 2009; Li and Clevers, 2010). These intestinal stem cells (ISCs) give rise to progenitor cells in the transit-amplifying zone and provide a large number of precursor cells that can replenish cells of various lineages along the crypt-villus axis. However, the mechanism by which these different ISCs and progenitor cells mediate intestinal repair remains to be investigated (Barker et al., 2008; Batlle, 2008; Scoville et al., 2008; Casali and Batlle, 2009).

In the adult Drosophila melanogaster midgut, ISCs are present individually and distributed evenly underneath the epithelium (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006). When an ISC divides, it gives rise to a renewed stem cell and an enteroblast (Fig. 1 A). Immediately after division, a higher level of active cytoplasmic Delta is retained in the cell that remains as an ISC, whereas the neighboring enteroblast quickly loses the active form of Delta (Bray, 2006; Ohlstein and Spradling, 2007). This asymmetric level of active Delta in ISCs stimulates the Notch signaling pathway in the newly formed enteroblast (Bardin et al., 2010), which ceases division and starts to differentiate. Depending on the strength of Notch pathway stimulation, the enteroblast may differentiate to become an enterocyte or enteroendocrine cell (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006, 2007).

In addition to the Delta–Notch pathway, recent studies demonstrate that the EGF receptor pathway, Wingless pathway, Decapentaplegic pathway, and intrinsic chromatin modification by the deubiquitinase Scrawny are required for the development and maintenance of ISCs (Lin et al., 2008; Buszczak et al., 2009; Jiang and Edgar, 2009; Lee et al., 2009; Buchon et al., 2010; Mathur et al., 2010; Biteau and Jasper, 2011; Jiang et al., 2011). JNK, p38, and, possibly, PVF2 are required for the regulation of ISCs in aging flies (Biteau et al., 2008; Choi et al., 2008; Park et al., 2009). The insulin receptor (InR), Janus kinase–signal transducer and activator of transcription, Hippo, and JNK signaling are essential for ISC division during homeostasis and pathogenic stimulation (Maeda et al., 2008; Amcheslavsky et al., 2009; Apidianakis et al., 2009; Buchon et al., 2009a,b; Chatterjee and Ip, 2009; Cronin et al., 2009; Jiang et al., 2009; Beebe et al., 2010; Karpowicz et al., 2010; Lin et al., 2010; Ren et al., 2010; Shaw et al., 2010; Staley and Irvine, 2010). Therefore, conserved regulatory pathways are involved in ISC-mediated homeostasis.

In this study, we used an RNAi-based genetic screen to search for important stem cell regulators and have identified tuberous sclerosis complex (TSC) as an essential regulator of midgut ISC growth (in this study, cell growth is measured as an increase in cell size). The human disease TSC is characterized by the appearance of benign tumors in multiple tissues, as the result of mutations in either the TSC1 or TSC2 gene (Crino, 2008; Huang and Manning, 2008). The TSC gene products form a complex that negatively regulates Rheb and target of rapamycin (TOR). We show here that loss of TSC in midgut ISCs leads to excessive cell growth, which in turn causes defects in ISC division. These stem cell defects lead to a thinner gut epithelium, and the flies have increased susceptibility to feeding of tissue-damaging agents.

To search for essential regulators of ISCs, we used the escargot (esg) Gal4–upstream activation sequence (UAS) system to knock down specific targets by RNAi in adult midgut precursor cells (Fig. 1 A). We included the tubulin-Gal80^ts to control temporally double-stranded RNA (dsRNA) expression; at 29°C, the repressor Gal80^ts will not be active and will thus allow Gal4 to drive the expression of UAS constructs. The UAS-mCD8GFP construct was included to illuminate the affected cells. In control guts, the GFP-positive (GFP+) cells were smaller and frequently in pairs, representing the ISC and enteroblast cell nests in young flies (Fig. 1, B′ and D′, arrow pairs). In TSC2 dsRNA (Vienna Drosophila RNAi Center [VDRC] no. 6313)–expressing guts, the GFP+ cells were much larger and present as individual cells (Fig. 1, C and E). We measured the size of ISCs (Delta+ and GFP+ cells) based on the area in confocal images. The size of the ISCs increased progressively after RNAi initiation to ∼10-fold in 10 d (Fig. 1 F). The surrounding mature enterocytes had no GFP expression (Fig. 1 E′) and did not show any size increase (Fig. 1 G). The enteroendocrine cells were also not affected, as revealed by nuclear staining of Prospero and cell membrane staining of β-catenin/Armadillo (Fig. 1, D′ and E′, thick arrows). Although the esgGal4 driver is expressed in both ISC and enteroblast together with the Delta staining, we demonstrate that loss of TSC2 function induces a substantial increase in cell size in ISCs.

We performed a series of experiments to demonstrate that the well-studied TSC–TOR pathway regulates adult ISCs. TSC is a protein complex that consists of TSC1 and TSC2, and they negatively regulate TOR to control translation and cell growth. Two TSC2 RNAi lines (VDRC no. 6313 and 6314) and one TSC1 RNAi line (VDRC no. 22252) all showed a similar size increase (Fig. 1 H). Next, we examined the requirement of TOR by feeding the specific inhibitor rapamycin to the RNAi flies. Administration of rapamycin efficiently suppressed the increase in cell size during RNAi induction (Fig. 1 H). We then used a TOR dominant-negative construct as well as TOR RNAi to show that the TSC2 RNAi-induced size increase is TOR dependent (Fig. 1 I). Therefore, we have demonstrated that the well-studied TSC–TOR pathway regulates cell growth in adult midgut ISCs.

Real-time RT-PCR was used to measure the knockdown efficiency by RNAi. The result revealed an ∼60% decrease of TSC2 RNA in the whole gut under the tubulin-Gal4 ubiquitous driver (Fig. 1 J). The use of esgGal4 for this experiment would not be as accurate because only the precursor cells in the dissected gut would have the RNAi effect. Although the knockdown appeared to be specific, and the RNAi-induced cell growth phenotype was consistent with the known function of TSC, we confirmed the phenotypes by mosaic analysis with repressible cell marker (MARCM; Lee and Luo, 2001; Amcheslavsky et al., 2009). Three TSC2 mutant alleles were tested: TSC2⁵⁶, which is a weak allele, and TSC2¹⁹² and TSC2¹⁹³, which are null alleles (Ito and Rubin, 1999). The staining of these guts with Delta together with the GFP marking showed that all the TSC2 mutants caused an increased ISC size between day 4 and 12 after the heat shock regimen (Fig. 1 M, TSC2¹⁹³ allele). This increase in ISC size was also observed with TSC1¹² and TSC1²⁹ alleles (Fig. 1 L, TSC1²⁹ allele) but not with wild-type chromosomes (Fig. 1 K). We conclude that TSC loss of function causes cell-autonomous growth of ISCs in the adult midgut.

TSC has been dubbed a tumor suppressor because previous studies have shown that TSC mutant cells have increased cell division in addition to cell growth (Ito and Rubin, 1999; Gao and Pan, 2001; Tapon et al., 2001; Rosner et al., 2006; Huang and Manning, 2008). However, our observations suggest a major deviation from the current paradigm, such that loss of TSC in adult Drosophila ISCs leads to a halt in cell division. First, MARCM analysis revealed that most TSC2 mutant cells that were also Delta+ grew larger but remained as individual cells without clonal expansion (Fig. 2 A and Fig. 1 M), whereas wild-type ISCs formed GFP clones with many more cells (Fig. 1 K). Similar results were observed with the TSC1 mutant clones (Fig. 1 L). Second, the TSC2 RNAi guts had a reduction in phosphorylated histone 3–positive (pH3+) cells (Fig. 2 B, white bars). ISCs are the only mitotic cells within the adult midgut, and pH3 staining reveals ISCs that are in mitosis. Third, when the flies were fed with tissue-damaging agents dextran sulfate sodium (DSS) and bleomycin, the number of pH3+ cells increased in the wild-type guts but not in TSC2 RNAi guts (Fig. 2 B, black and gray bars). Together, these results demonstrate that TSC mutant ISCs do not divide.

We further analyzed the cell division phenotype by examining the expression of specific regulators. The kinase Polo is essential for mitosis and meiosis and is localized in different compartments at different phases of the cell cycle (Llamazares et al., 1991). By using a fusion protein trap line (FlyTrap no. CC01326), we found that Polo-GFP was detectable in Delta+ cells. Similar to a previous study (Llamazares et al., 1991), the Polo-GFP fusion protein was, at times, highly concentrated at the ISC metaphase plate (Fig. 2, C and C′). In wild-type guts, 3.0% (16/538) of the Polo-GFP–positive cells had this metaphase pattern, which likely represents the number of ISCs undergoing mitosis. In the TSC2 RNAi gut, the Polo-GFP was still clearly detectable as cytoplasmic staining (Fig. 2 D), but only 0.2% (2/843) of the cells showed the metaphase staining pattern.

Another essential component for mitotic entry is the kinase Cdc2 (Royou et al., 2008). Our staining data show that in wild-type guts, Cdc2 is expressed in all Delta+,esg>GFP+ (> is Gal4-UAS; Fig. 2, E–E‴, arrows) but only in some Delta−,esg>GFP+ cells. Therefore, Cdc2 is most likely expressed in ISCs and segregated into newly formed enteroblasts. In TSC2 RNAi guts, we could not detect any Cdc2 staining in any of the cell types (Fig. 2, F–F‴). The loss of Cdc2 supports the observation that TSC2 mutant ISCs have division defects.

The aforementioned results show that the mutant cells grow bigger and do not divide, raising the possibility that these cells may undergo endoreplication if DNA synthesis is normal. We performed 5-ethynyl-2’-deoxyuridine (EdU) feeding and incorporation experiments to examine whether the TSC2 RNAi or mutant cells are normal in S phase. The traditional BrdU incorporation/detection is not as sensitive as EdU (Staley and Irvine, 2010) and requires feeding time of a few days, which leads to a difficult interpretation of results. Therefore, we developed an EdU feeding protocol that allowed us to perform a much shorter feeding time of 6–24 h and gave better snapshots of the replication activity at various time points after RNAi initiation. We show in Fig. 3 the 24 h feeding results, which clearly indicate that the number of Delta+ cells that had detectable EdU incorporation was significantly reduced, especially in day 6 and day 8 (Fig. 3 A). We also performed MARCM clonal experiments using a null TSC2 mutant allele, and the result showed that most of these null cells had no detectable incorporation (Fig. 3 B).

To investigate the cell cycle defects further, we performed a tissue dissociation and cell-sorting experiment. Dissected guts were incubated with trypsin to release gut cells into suspension, which were stained with Hoechst 33342 DNA dye and analyzed by FACS. The esg>GFP control guts gave 11.3% of cells that had a strong GFP signal (Fig. 4 A, boxed area), whereas the TSC2 RNAi guts gave a lower percentage (6.86%) in the same gated area (Fig. 4 B), which is consistent with the mutant ISCs being unable to divide. The wild-type GFP-positive cells showed tight DNA dye profiling (Fig. 4 C), and computer modeling suggested that 82% of the cells were 2N or in G1 (Fig. 4 C′). As a comparison, the overexpression of Myc drives more cells to completion of S phase (Johnston et al., 1999; Wu and Johnston, 2010), resulting in a profile that had clearly distinguishable 2N and 4N peaks (Fig. 4, D and D′). Meanwhile, cells from TSC2 RNAi guts showed a broader profile but did not form another peak (Fig. 4 E). Computer modeling suggested that ∼65% of these cells had DNA content between 2N and 4N (Fig. 4 E′). This result suggests that many TSC2 RNAi cells can enter S phase but either stop or slow down, leading to the accumulation of different amounts of DNA in the population.

Based on both the EdU incorporation and FACS analysis, we conclude that TSC2 RNAi midgut precursor cells have significantly slower DNA synthesis and do not have synchronized S phase. However, because mitosis is abolished, even slow DNA replication over time will cause polyploidy in these large mutant cells. This will be different from mutant imaginal disc cells, which continue to divide and, therefore, have a modest size increase and remain diploid (Ito and Rubin, 1999; Gao and Pan, 2001; Tapon et al., 2001). Our FACS analysis of TSC2 RNAi cells at later time points did show an even broader profile. However, because of the extreme cell size increase, the DNA signal comparison between small control cells and large mutant cells becomes unreliable. Therefore, we speculate that the large mutant cells in the adult midgut are not in active endoreplication, although they have DNA content larger than 2N because of the lack of cell division and slow DNA synthesis.

When an ISC divides, two cells are formed with one becoming a renewed ISC and the other becoming an enteroblast. The enteroblast ceases to divide and starts to differentiate, which is associated with substantial growth if committing to the enterocyte lineage (Fig. 1 A). Therefore, the aforementioned TSC mutant phenotypes can be a result of directed differentiation. However, the detection of both Delta and esg>GFP (Fig. 5 A, arrows), two major stem cell markers in adult Drosophila ISCs, in many of the large TSC mutant cells suggests that they have retained at least some stem cell properties. Therefore, we next performed analyses of various makers to determine whether these mutant ISCs are largely undifferentiated. As shown in Fig. 5 B, staining of the large GFP+ cells with Prospero was negative, whereas the small enteroendocrine cells stained positive for this marker (Fig. 5 B, thick arrows). In addition, the Delta and the Notch pathway target gene Su(H)-lacZ stained separate cells when two occasional neighboring cells both had esg>GFP expression (Fig. 5, C–C″‴). Therefore, the ISC versus enteroblast fate has been maintained after TSC2 RNAi. We also quantified the relative number of Delta+ cells over the experimental period. In wild-type flies, progressively older fly guts had a decreasing Delta+ to esg>GFP+ cell ratio, from ∼40 to 20% (Fig. 5 D, white bars). This reflects the accumulation of more enteroblasts after divisions during which the ISCs are renewed but not increased. If there is ISC loss, we expect this ratio will decline faster. In TSC2 RNAi guts, the ratio declined from ∼60 to 50% (Fig. 5 D, black bars). This is consistent with no cell division, and there is also no indication of stem cell loss after TSC2 RNAi.

The stem cell state and subsequent differentiation are often correlated with specific chromatin modifications (Boheler, 2009; Buszczak et al., 2009; Sang et al., 2009). In wild-type guts, we observed that precursor cells marked by esg>GFP exhibited a more prominent nuclear staining of acetylated histone 3 (AcH3; Fig. 5 E, arrows). Importantly, the large TSC2 RNAi cells also exhibited this stronger AcH3 nuclear staining (Fig. 5 F, arrowheads), suggesting that they were precursor-like cells. Fluorescent phalloidin illuminates F-actin present at the apical brush border of mature enterocytes and in smooth muscle cells (Micchelli and Perrimon, 2006; Amcheslavsky et al., 2009). We found that >90% of the large GFP+ cells examined after 8 d of RNAi treatment showed no direct overlap with phalloidin staining (Fig. 5 G, arrowheads). The POU domain protein Pdm1 is a newly identified mature enterocyte marker (Bardin et al., 2010). We stained for Pdm1 and found that it was not expressed in the big TSC2 RNAi cells but was clearly present in surrounding enterocytes (Fig. 5 F). Therefore, ISCs that have lost TSC2 function remain undifferentiated.

We conducted additional experiments to gain further insight into the mechanism by which TSC regulates ISC division. In cases in which InR acts directly upstream, it usually suppresses TSC function (Pan et al., 2004; Avruch et al., 2006), and thus, InR and TSC have opposite functions and should exhibit opposite phenotypes. However, in ISCs, loss of InR and loss of TSC both abolish division, which is also observed in germline stem cells (Amcheslavsky et al., 2009; Hsu and Drummond-Barbosa, 2009). Furthermore, administration of rapamycin only modestly suppressed InR-stimulated ISC division (Fig. 6 A). Therefore, it is unlikely that InR signaling simply represses TSC to allow TOR to stimulate division in ISCs. We next simultaneously inactivated TSC by TSC2 RNAi and activated the InR pathway either by expressing an activated InR or by RNAi knockdown of PTEN, a negative regulator of the InR pathway. As shown in Fig. 6 B, ∼80% of the InR-stimulated or PTEN RNAi–induced mitotic activity was suppressed by loss of TSC function. This genetic suppression implies that TSC acts in parallel or downstream of InR/PTEN but is a positive regulator of ISC division. In consideration of all the results, we surmise that InR and TSC act in parallel to regulate ISC division (see Fig. 8 E).

Recent evidence suggests that TSC can regulate two different TOR complexes: TORC1 and TORC2 (Huang and Manning, 2009; Inoki and Guan, 2009). In the canonical pathway, TSC represses TORC1 by acting as a GTPase-activating protein for the GTPase Rheb. In the second pathway, TSC may activate TORC2 independently of the GTPase-activating protein activity (Huang et al., 2009). TORC2 contains two distinct components called Sin1 and Rictor (Sarbassov et al., 2005; Guertin et al., 2006). Therefore, we examined ISC division in the genetic mutants of sin1 and rictor, both of which are viable strains (Hietakangas and Cohen, 2007). We found that these mutants behaved similarly to wild-type flies in basal or damage-stimulated ISC division (Fig. 6 C). One observed defect in the sin1 and rictor mutants is a lack of AKT phosphorylation at the C terminus, which enhances AKT activity in conjunction with InR signaling (Sarbassov et al., 2005; Guertin et al., 2006; Hietakangas and Cohen, 2007). This lack of C-terminal AKT phosphorylation was readily detectable in whole fly and gut extracts of the sin1 and rictor mutants (Fig. 6 D). However, the TSC2 RNAi guts showed a normal level of AKT phosphorylation. Therefore, loss of ISC division in TSC mutant guts is not caused by deregulation of the TORC2 pathway.

Two other possible mechanisms are that TSC2 regulates ISC division through an unknown pathway or through TORC1, which causes excessive growth that may block cell division. Rapamycin is a potent inhibitor of TORC1. We therefore used this chemical to examine the effect of TORC1 inhibition on cell division in TSC2 RNAi guts. Rapamycin was fed either at the beginning of RNAi induction (day 0) or at the time 5 d after RNAi initiation (day 5). The experiment was continued for a total of 10 d, and DSS was added during the last 2 d to stimulate cell division. When rapamycin was given at the beginning of the experiment, the TSC2 RNAi–induced growth increase and division defect were both suppressed (Fig. 6 E). However, if TSC2 RNAi was allowed to take effect for 5 d, which causes the ISCs to grow by approximately fourfold (Fig. 1), rapamycin administration could not rescue the ISC division defect (Fig. 6 F), despite suppressing further growth. Therefore, by inhibiting growth right at the beginning, the size of the ISCs remained normal, and they were able to divide even in the absence of TSC2. These results suggest that the ISC division defects observed in the TSC2 mutant are not because of another pathway but are caused by the excessive growth mediated by TORC1.

To gain further support for the model that excessive growth in ISCs is the reason for the observed inhibition of cell division, we sought other regulators that could inhibit ISC growth. Myc has been shown to regulate cell growth during Drosophila development (Johnston et al., 1999; Wu and Johnston, 2010). Overexpression of Drosophila Myc in the midgut by the esgGal4 driver did not cause a detectable change in cell growth even after 2 or 6 d at a permissive temperature (Fig. 7 A). Nonetheless, the higher level of Myc significantly enhanced the cell growth phenotype induced by TSC2 RNAi within 2 d (Fig. 7 A), demonstrating that Myc is indeed a growth regulator in the midgut. We then tested the loss of function of Myc by esgGal4-driven RNAi. Over a short duration of 2 d, the Myc RNAi did not cause a significant change in ISC growth or division (Fig. 7, A and B). However, the Myc RNAi, for 6 d or longer, did cause a halt of division (unpublished data). We cannot distinguish whether prolonged loss of Myc function affects ISC division directly or the lack of cell growth affects division indirectly. Nevertheless, we used the short duration protocol to inhibit the cell growth phenotype induced by TSC2 RNAi. Flies that had both TSC2 RNAi and Myc RNAi showed a cell size more similar to that of the control, which was smaller than TSC2 RNAi samples (Fig. 7, A [right-most sample] and F). These double RNAi flies also exhibited a mitotic index similar to that of the control (Fig. 7 B). Moreover, cell division in these flies in response to DSS feeding was equal to the control (Fig. 7 B, black bars). The GFP-positive cell number and size were also consistent with the mitotic index (Fig. 7, C–J). Therefore, a reduction in growth by reducing Myc activity can restore ISC division. Together, these results support the hypothesis that the excessive growth in the TSC mutant blocks cell division and that, in normal guts, TSC and Myc function to coordinate ISC growth and division.

To assess the biological outcome in the absence of TSC function within ISCs, we examined the midgut organization by tissue sectioning. The wild-type midgut showed protrusions of mature enterocytes lining the lumen, which was consistent with their normal absorptive functions (Fig. 8 A). However, in TSC2 RNAi guts, the epithelial layer appeared to be thinner and smoother (Fig. 8 B). Nevertheless, the epithelium still appeared as a continuous sheet.

We then fed the TSC2 RNAi flies with tissue-damaging agents. DSS and bleomycin feeding causes different damages in the midgut and can kill flies in a dose-dependent manner (Amcheslavsky et al., 2009). When a low dose of 1% DSS in sucrose solution was given, ∼78% of the wild-type flies were still alive at the end of 7 d. However, under the same conditions, only 30% of the TSC2 RNAi flies were alive at the end of 7 d (Fig. 8 C). Similarly, when a low dose of 2 µg/ml bleomycin was administered for 7 d, 80% of the wild-type flies survived, whereas only 40% of the TSC2 RNAi flies survived under the same conditions (Fig. 8 D). The increased susceptibility to tissue damage observed in these TSC2 RNAi flies is consistent with the loss of ISC division, which would lead to fewer precursor cells being available for repair.

In this study, we have provided evidence demonstrating that TSC is an essential regulator of ISC growth and division. In the absence of TSC function, ISCs have unrestricted cell growth, which halts cell division and leads to the formation of extremely large cells. Although stem cell markers are still expressed, these ISC-like cells are nonfunctional and can no longer divide or differentiate. As a consequence, the TSC mutant gut has a thinner epithelium and the mutant fly is more susceptible to tissue-damaging agents. Our study has uncovered a tissue context–dependent phenotype of TSC mutants, such that unrestricted cell growth can lead to a stop of cell division, and thus, TSC does not function all the time as a classical tumor suppressor.

The TSC–TOR and other growth regulatory pathways, such as InR and Myc, have intricate interactions. Some suggest that the InR pathway directly represses TSC, whereas others and our study here suggest that the two pathways act in parallel (Fig. 7 E; Pan et al., 2004; Avruch et al., 2006). The TSC–TOR pathway also has a negative feedback into upstream components of the InR pathway (Huang and Manning, 2009; Inoki and Guan, 2009). Recent identification of TORC2 in addition to the original TORC1 further complicates these pathways (Hietakangas and Cohen, 2007; Guertin et al., 2009). However, our results clearly show that TORC2 mutants and TSC mutants have different phenotypes in the adult Drosophila midgut, suggesting that TSC does not function through TORC2 to regulate ISC division. Previous studies have demonstrated that Myc can modulate TSC–TOR in controlling the growth of mammalian and fly cells (Tapon et al., 2001; Teleman et al., 2008; Schmidt et al., 2009), which is consistent with what we have observed.

In normal development and adult tissue homeostasis, cells need to grow in size by approximately twofold before they divide to maintain the original cell size. Reduction in cell growth below a certain threshold can lead to a halt of division (Grewal and Edgar, 2003; Jorgensen and Tyers, 2004; Leevers and McNeill, 2005). Therefore, the balance between cell growth versus division is a complex process requiring delicate coordination (Kohlmaier and Edgar, 2008). Here, we have shown that, in TSC mutants, the increase in midgut ISC size is >10-fold (Fig. 1 and Fig. 2), whereas the increase in larval disc cell size is less than twofold (Ito and Rubin, 1999; Gao and Pan, 2001; Tapon et al., 2001). A possible reason for this difference is that the mutant larval disc cells continue to divide, thereby maintaining a moderate cell size. One key question that remains is why the larval disc cells that contain a TSC mutation have somewhat coordinated growth and division, whereas the adult mutant ISCs have completely stopped their division. It is possible that because imaginal discs are developing organs, they are designed to have faster growth and division. Adult midgut ISCs have a slower intrinsic cell cycle of >24 h (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006), and adult cells have differences in checkpoint controls (Su et al., 2000; Walworth, 2000; Song, 2005). These may allow the excessive growth to take place until it passes a critical point that blocks division.

Phenotypes manifested in TSC patients are mostly benign tumors that rarely progress into higher-grade cancers (Crino, 2008). TSC1 and 2 have expression in the intestine, and adult patients have occasional intestinal polyps (Hizawa et al., 1994; Fukuda et al., 2000; Johnson et al., 2001). Mouse embryonic fibroblasts from mutant TSC animals can also enter senescence, which is equivalent to a cessation of cell division (Zhang et al., 2007). The adult midgut ISC phenotypes shown in this study are consistent with these phenotypes. We speculate that excessive cell growth leading to a block in cell division is a common phenotype in slowly dividing adult tissues when TSC is mutated. The phenotype of increased cell growth and increased cell division may be applicable to rapidly dividing cells, including developing Drosophila disc cells, mammalian hematopoietic stem cells, and tumor cells (Ito and Rubin, 1999; Gao and Pan, 2001; Tapon et al., 2001; Rosner et al., 2006; Gan et al., 2008). A recent study demonstrates that in TSC mutants, there is loss of adult female germline stem cells because of differentiation (Sun et al., 2010). The ISCs and germline stem cells have different niche compositions that may contribute to the observed differences in the mutant phenotype. Moreover, it underscores the idea of a tissue context–dependent phenotype exhibited in TSC mutants.

Most stocks have been previously described in Amcheslavsky et al. (2009). Myc RNAi lines were obtained from VDRC, and the Transgenic RNAi Project, UAS-Myc, and PTEN RNAi lines were obtained from the Bloomington stock center (no. 8549). Polo-GFP (no. CC01326) was obtained from FlyTrap, and the sin1 and rictor mutants were obtained from S. Cohen (Temasek Life Sciences Laboratory, Singapore; Hietakangas and Cohen, 2007), and the TOR flies were obtained from E. Baehrecke (University of Massachusetts, Worcester, MA) and T. Neufeld (University of Minnesota, Minneapolis, MN). Viability tests and feeding experiments were previously described in Amcheslavsky et al. (2009). In brief, 50–100 flies of 3–5 d old were used per vial. The vial contained a piece of 2.5 × 3.75–cm chromatography paper (Thermo Fisher Scientific). 500 µl of 5% sucrose solution was used to wet the paper as a feeding medium. The chemicals included in the feeding medium were 1% DSS (MP Biomedicals) or 2 µg/ml bleomycin (Sigma-Aldrich). Flies that were still alive were transferred to new vials with fresh feeding media every day. The MARCM clones were generated as previously described (Lee and Luo, 2001; Amcheslavsky et al., 2009). Flies stocks were crossed to generate the following genotype: hsFLP,UAS-CD8GFP;esgGal4/+;FRT80B tubGal80/FRT80B TSC¹⁹². The final cross and offspring were maintained at 18°C. To induce MARCM clones, flies were heat shocked in a 37°C water bath for 30 min twice a day for 3 d. The flies were kept at 29°C and incubated for an additional 4–12 d before dissection.

For rapamycin feeding, 1 µl of 5.5-mM rapamycin (Sigma-Aldrich) dissolved in DMSO was added to 500 µl of a 20% sucrose solution. This mixture was added to standard fly food vials. The effective concentration in the food medium was ∼0.5 µM. 1-d-old flies were cultured on the rapamycin-augmented food, incubated at 29°C, and transferred to new media with rapamycin every 2 d over a period of 6–10 d. For EdU labeling, wild-type and TSC2 mutant flies were fed on 100 µM EdU (Invitrogen) in a 5% sucrose solution for 24 h. Guts were dissected and subjected to Delta antibody staining as described previously (Amcheslavsky et al, 2009). EdU incorporation was detected using EdU Alexa Fluor 555 heat shock assay (Click-iT; Invitrogen) following the manufacturer’s instructions.

Most gut dissection, fixation, antibody staining, and confocal microscopy procedures were previously described in Amcheslavsky et al. (2009). The entire gastrointestinal tract was pulled from the posterior end directly into fixation medium containing PBS and 4% formaldehyde (Mallinckrodt Chemicals). Guts were fixed in this medium for 3 h, except for Delta staining in which the fixation was for 0.5 h. Subsequent rinses, washes, and incubations with primary and secondary antibodies were performed in a solution containing PBS, 0.5% BSA, and 0.1% Triton X-100. Anti-Cdc2 (rabbit polyclonal, 1:500 dilution; Santa Cruz Biotechnology, Inc.; a gift from W. Theurkauf, University of Massachusetts, Worcester, MA), anti-Pdm1 (rabbit polyclonal, 1:100 dilution; a gift from X. Yang, University of Singapore, Singapore), anti-AcH3 (rabbit polyclonal, 1:500 dilution; Millipore), and anti-GFP (rabbit polyclonal, 1:1,000 dilution; Invitrogen) were used for immunofluorescent staining. For Western blotting, the anti–C-terminal phospho-AKT antibody (rabbit polyclonal, 1:1,000 dilution; Cell Signaling Technology) and the antitubulin antibody (1:100 dilution; Hybridoma bank) were used.

Microscope image acquisition was performed in the Digital Light Microscopy Core Facility at the University of Massachusetts Medical School using a spinning-disk confocal microscope (Nikon). The main components are an inverted microscope (TE-2000E2; Nikon) with a spinning-disk confocal attachment (CSU10; Yokogawa) and a 40× Plan Apochromat oil objective (Nikon) with a numerical aperture of 1.0. The imaging temperature was room temperature, and the medium was oil. The fluorochromes used were DAPI, Alexa 488, Alexa 568, and Alexa 633. The camera make was a Rolera MGi EM charge-coupled device (QImaging). The acquisition and processing software was MetaMorph (Molecular Devices) with no deconvolution or γ adjustment and used an 8-bit export file format.

Guts dissection and sectioning were previously described in Amcheslavsky et al. (2009). For tissue dissociation, the malpighian tubules, the esophagus, and the rectum were removed, leaving only the midgut. Approximately 30 midguts were put into a 1.5-ml microfuge tube on ice containing a 0.5-ml solution of PBS, 1 mM EDTA, and 0.5% trypsin (Invitrogen). The tissues were incubated at room temperature for 2.5 h in the trypsin solution on a Nutator (Adams; BD) and vortexed vigorously every 30 min during the incubation. At the end of the incubation, the tissues were disrupted by a hand-held homogenizer. The dissociation of cells was confirmed by examining a small fraction of the suspension under a fluorescent microscope (TE-2000E2). DNA dye Hoechst 33342 (Roche) was added to the trypsin solution during cell dissociation at a final concentration of 10 µg/ml. The samples were stored on ice in the dark until analyzed. The cell suspensions were filtered through nylon mesh to remove large debris before passing through the cell sorter (LSR II; BD). S2 cells were used for preruns of the cell sorter. These cells were washed once with PBS, resuspended in PBS, 5 mM EDTA, and 10 µg/ml Hoechst 33342, and incubated for 2.5 h at room temperature. The cells were analyzed by a cell analyzer (LSR; BD; Flow Cytometry Core Facility, University of Massachusetts Medical School).

For real-time quantitative PCR, total RNA was isolated from 10 dissected guts and used to prepare cDNA. PCR was performed using a real-time PCR detection system (iQ5; Bio-Rad Laboratories) with the following primers: TSC2 forward, 5′-ATCGTTGAGCCACTTGACCT-3′; TSC2 reverse, 5′-TGCGTGGCACAGGAATTT-3′; TSC1 forward, 5′-GCAAGGAGCAAAGAATCG-3′; TSC1 reverse, 5′-ACTCATTGGCCTTGATCG-3′; extra (CG6634) forward, 5′-ACATTCAGCACCACAGCA-3′; extra reverse, 5′-TGATTGCTGCCCTGATCGTA-3′; actin forward, 5′-AGTGTGTGCAGCGGATAACT-3′; actin reverse, 5′-AAGCTGCAACCTCTTCGTCA-3′; rp49 forward, 5′-CGGATGGATATGCTAAGCTGT-3′; and rp49 reverse, 5′-GCGCTTGTTCGATCCGTA-3′. rp49 was used as a normalization standard. RT–quantitative PCR was performed in duplicate on each of three independent biological replicates.

We thank M. Chatterjee, N. Mohammad, and N. Wakabayashi-Ito for help with cell dissociation experiments and David Guertin for comments on the manuscript. We thank S. Cohen for the rictor mutant flies, FlyTrap for the Polo-GFP line originated from the A. Spradling laboratory, E. Baehrecke and T. Neufeld for TOR transgenic lines, W. Theurkauf for the Cdc2 antibody, and X. Yang for the Pdm1 antibody.

The work in the Y.T. Ip laboratory was supported by grants from the National Institutes of Health (DK83450) and the Worcester Foundation for Biomedical Research. Y.T. Ip is a member of the University of Massachusetts Diabetes Endocrinology Research Center, and core resources supported by the center grant (DK32520) were also used. The work in the J. Jiang laboratory was supported by grants from the National Institutes of Health (GM61269 and GM67045), the Welch Foundation (I-1603), and the Cancer Prevention Research Institute of Texas (RP100561).