Feeding of dextran sulfate sodium (DSS) causes midgut cell proliferation
When included in drinking water, DSS causes injury in the intestines of experimental mammals and complex interactions in the gut milieu lead to excessive inflammation resembling ulcerative colitis in humans (Kawada et al., 2007
; Rakoff-Nahoum et al., 2004
; Tlaskalova-Hogenova et al., 2005
). We tested the idea that DSS could cause similar tissue damage in adult Drosophila gut. When the feeding sucrose solution contained 5% of DSS (average m.w. 40 kDa), approximately 50% of the flies died within 5 days (). Together with the killing curves of 1% and 3 % DSS, we show that there is a dose dependent killing of flies by DSS. We also tested related polysaccharides and found that feeding sucrose solution containing 3% of 400 kDa DSS or dextran (a non-sulfated polysaccharide) did not change significantly the viability of the flies (). Heparin, another sulfated polysaccharide, had an intermediate effect, while 5 kDa DSS also caused significantly increased lethality.
-promoter driven Gal4 expression (esg-Gal4) (Micchelli and Perrimon, 2006
) when crossed with the UAS-CD8GFP can mark both ISC and enteroblasts in adult Drosophila midgut (). We used these flies for feeding experiments with 3% of the various polysaccharides in the sucrose solution for 2 days. We found that the 5 and 40 kDa DSS caused a significant increase in the number of GFP positive cells, while the 400 kDa DSS and dextran did not induce a gut phenotype (). The 40 kDa DSS is the most commonly used reagent in mammalian experiments and the effect in Drosophila gut was highly reproducible, as 84.3% (n=51) of the DSS fed samples and 0% (n=52) of control samples showed the cell proliferation phenotype. A correlation between the killing effect and cell proliferation phenotype by the various polysaccharides is also apparent.
DSS feeding increases the number of enteroblasts
The esg-Gal4/UAS-CD8GFP marks both ISC and enteroblasts. After co-staining with Delta (data not shown; Micchelli and Perrimon, 2006
; Ohlstein and Spradling, 2007
), we confirm that the ISC is usually the one with a lower level of GFP signal and a more rounded cell shape (, arrowhead, insert). After DSS feeding most of the GFP positive cells appeared to be of bigger size with strong GFP signal (). We next examined the maker Su(H)Gbe-lacZ, which is a Notch pathway target gene and is specific for enteroblasts (Micchelli and Perrimon, 2006
; Ohlstein and Spradling, 2007
). The staining showed that there were clearly more β-galactosidase expressing cells after DSS treatment ().
Prospero is a homeodomain protein that is expressed specifically in enteroendocrine cells (Micchelli and Perrimon, 2006
). The Prospero staining showed no apparent increase in the number of enteroendocrine cells (). Meanwhile, punctate Delta staining in cytoplasm represents the only known ISC specific marker (Ohlstein and Spradling, 2007
). In control samples, low levels of punctate Delta could be detected in cells with small nuclei, suggesting that they are ISC. In DSS treated guts, the Delta fluorescent signal became more easily detectable (). We counted the number of cells that expressed the different markers and in each case normalized with non-stained cells based on DAPI staining. The quantification indicates that the number of cells expressing esg-Gal4/UAS-CD8GFP and Su(H)Gbe-LacZ had increased by approximately 4 fold, while Delta and Prospero positive cell number had changed by less than 1.5 fold (). These results support our idea that the most prominent phenotype induced by DSS is an accumulation of enteroblasts.
Intestinal stem cell division is increased after DSS feeding
We surmised that DSS feeding caused an increase in ISC division, leading to the accumulation of enteroblasts. ISC is the only cell type in midgut that goes through division, while enteroblasts undergo endoreplication during differentiation into enterocytes, which are polyploid (). Phosphorylated-histone 3 (phospho-H3) is a highly specific marker for condensed chromosomes, and thus can be used to assess the number of mitotic ISC. The number of phospho-H3 positive cells in dissected guts from DSS fed flies was approximately 20 fold higher than in control samples (). These results support our hypothesis that DSS increases ISC division. We used 3–5 day old flies for experiments but also noted that older flies had more mitotic cells (see also Choi et al, 2008
DSS increases ISC division and does not affect cell fate determination
We performed a series of experiments to determine the cell fate and stem cell properties after DSS feeding. Double staining for Delta and Su(H)Gbe-lacZ revealed that there was no overlap before or after DSS feeding (, arrows), demonstrating that ISC and enteroblast fates are preserved. By another double staining experiment, we found that all the phospho-H3 positive cells had Delta staining (), demonstrating that after DSS feeding ISC retained the ability to proliferate and contained high level of Delta. Therefore, the asymmetry, cell fate determination and known properties of ISC are not affected during the experimental period of DSS feeding.
We then performed the mosaic analysis with repressible cell marker (MARCM), which randomly allows Gal4 driven GFP marking of individual ISC lineage due to FLP-FRT-mediated mitotic recombination that removes the repressor Gal80 (Lee and Luo, 2001
; Micchelli and Perrimon, 2006
; Ohlstein and Spradling, 2006
). In control guts of MARCM flies fed with sucrose, GFP expression was detected in a small number of isolated cells (). Under the same feeding condition the DSS treated flies had more GFP positive cells and were present in clusters (). Within each cluster only one cell exhibited punctate Delta staining (arrowhead). Each isolated cluster should represent a single lineage originated from one ISC, supporting the idea that DSS feeding increases the number of cells produced by an ISC and only one ISC is formed after each division.
DSS disrupts basement membrane organization and does not induce enteroblast differentiation
One speculation regarding stem cell-mediated repair is that stem cell division is increased and the daughter cells should differentiate to replenish lost cells. While many enteroblasts accumulated around ISC, no similar increase in mature enterocytes was observed after DSS feeding. Phalloidin stains F-actin, which is present at high levels in the apical brush border of mature enterocytes and in smooth muscle cells (Micchelli and Perrimon, 2006
). In both control and DSS fed samples (), all the MARCM GFP expressing cells (arrowheads) had smaller nuclei, stayed closer to the basal side and did not closely associate with phalloidin staining (arrows). This result suggests that DSS induces enteroblast accumulation but does not induce differentiation into mature enterocytes.
DSS disrupts basement membrane organization and causes enteroblast accumulation
To gain insights into the apparent lack of stimulated differentiation, we examined the tissue injury phenotype after DSS feeding. By tissue sectioning, we found that the overall morphology of the gut appeared intact and did not exhibit extensive ulcer or cell loss (), different from that observed in mammalian experiments (see discussion). Instead, we noticed that the basal layer underneath the enterocytes appeared broken (arrows). To further investigate this phenotype, we used a fly line expressing GFP fusion protein of collagen IV (), which is part of the basement membrane structure (Baba et al., 2007
; LeBleu et al., 2007
). Interestingly the scaffolding pattern of collagen IV-GFP became less organized in DSS fed samples (). In sagittal view of control gut, there are two layers of collagenIV-GFP signal connected by transverse filaments (). In DSS treated guts, the two layers became more rounded and separated (). We speculate that DSS disrupts basement membrane organization, which is sensed by ISC and causes increased cell division. The lack of cell loss in the epithelial layer, however, may account for the absence of a signal required to stimulate enteroblast differentiation.
Bleomycin feeding causes damage specifically in enterocytes
We examined other tissue damaging agents in order to test the idea that epithelial cell loss may stimulate both ISC division and enteroblast differentiation, hallmarks of stem cell-mediated tissue repair. Bleomycin is an anticancer drug that is also widely used as DNA damaging agent in experimental systems (Takada et al., 2003
; Morel et al., 2008
). We found that feeding of bleomycin to adult flies caused lethality in a dose dependent manner (). We then stained for phospho-histone 2A variant D (H2AvD), which accumulates in response to DNA damage, possibly through the protein kinase ATM (Clarkson et al., 1999
; Klattenhoff et al., 2007
). The control guts showed a low level nuclear punctate staining of H2AvD (). In bleomycin fed samples, we observed a much stronger staining associated with entire nuclei of most enterocytes (, arrow). More importantly, no H2AvD staining was observed with small nuclei in the basal side (marked by esg-Gal4/UAS-GFP, arrowhead). These results demonstrate that bleomycin feeding causes DNA damage specifically in enterocytes. We also stained the guts for Spectrin, a membrane skeleton protein localized at the apical side of enterocytes (). The staining revealed that after bleomycin feeding the epithelial layer became disorganized and more enterocytes were shedding into the lumen (open arrow). Plastic tissue sectioning revealed a similar epithelial cell loss phenotype (). We speculate that bleomycin may not be able to diffuse through tight junction, and thus causes damage only in enterocytes.
Bleomycin feeding induces damage specifically in epithelial layer
Bleomycin induces ISC division and enteroblast differentiation
We further show that bleomycin increases ISC division while the midgut cell fates are normal. Phospho-H3 staining revealed that there was a substantial, dose dependent increase in cell division (). We then used the Su(H)Gbe-lacZ line to perform double staining after bleomycin feeding and observed that the Delta and β-galactosidase staining were clearly distinct (, P arrowhead). Meanwhile, the MARCM experiment showed that only one Delta positive cell was present in each cluster (, R arrowhead).
One clear difference between the phenotypes induced by bleomycin versus DSS is that after bleomycin feeding there were larger cells in each cluster (, R arrow). To assess whether these bigger cells are mature enterocytes, we stained MARCM samples with fluorescent-phalloidin after bleomycin feeding. In the control guts only small cells with GFP were present (, arrowhead). In bleomycin fed samples, there were more GFP positive cells and the bigger GFP positive cells also had tightly associated phalloidin staining at the apical side, similar to other mature enterocytes (, arrow). In comparison, DSS induced only clusters of enteroblasts in the same period of time (). Double feeding experiment showed that GFP-positive mature enterocytes were formed efficiently with both DSS and bleomycin in the feeding solution (, quantification in 5M), demonstrating that DSS does not inhibit enteroblast differentiation. Overall, our data support the idea that bleomycin causes enterocyte-specific damage and cell loss, which in turn causes ISC to divide faster and facilitates enteroblast differentiation into new enterocytes.
Epithelial cell loss is coincident with ISC division and enteroblast differentiation
Epithelial expression of apoptotic proteins induces ISC division
We performed a series of experiment to further demonstrate that epithelial damage can indirectly induce ISC division. First, based on microarray expression assays for genes that have high expression in adult gut, we constructed an NPC (Niemann-Pick type C) 1b promoter-Gal4 driver line. Second, we examined the expression after crossing with UAS-GFP, and the GFP signal was clearly seen in the cytoplasm of large-sized enterocytes (arrow, ) but not in nucleus or cytoplasm of small cells (arrowhead). Thus, the NPC1b-Gal4 directs the expression only in mature enterocytes. We then expressed the viral antiapoptotic protein p35 in epithelial cells by NPC1b-Gal4 and performed bleomcyin feeding experiments. The result showed that the bleomycin-induced ISC division is suppressed by 52% ().
Because p35 inhibits caspases (Domingos and Steller, 2007
), our result suggests that bleomycin causes DNA damage and triggers caspase activation within the epithelial cells, and somehow stimulates ISC division. Thus, we expressed in epithelial cells the well-established apoptotic inducers Hid and Reaper, which would activate caspases. Such expression caused ISC division to increase by approximately 10 fold, based on phospho-H3 staining (). Overall, our results suggest that epithelial damage by feeding tissue damaging agents or by expressing apoptotic proteins can trigger a mechanism that ultimately stimulates ISC division.
The insulin receptor signaling pathway is required for induced ISC division
We used the inducible ISC division phenotype to examine mutant fly strains and found that insulin receptor (InR) mutants showed a significantly lower number of phospho-H3 positive cells (). The InR mutant also suppressed the DSS and bleomycin-induced cell organization phenotype as revealed by β-catenin staining (), probably due to lower ISC division and smaller number of cells in the mutant guts. We examined two other components of the InR signaling pathway. Chico is the Drosophila homologue of insulin receptor substrate and the chico1
mutant showed reduced ISC division (). FOXO is a transcription factor negatively regulated by insulin (Bohni et al., 1999
). The FOXO™
transgenic construct contains point mutations at 3 phosphorylation sites that render it constitutively active equal to a loss of insulin regulation (Hwangbo et al., 2004
crossed with esg-Gal4 produced no viable adults; we therefore used the Armadillo-Gal4 driver that has low level ubiquitous expression and these fly guts showed the predicted phenotype of reduced ISC division (). Thus, three components of the insulin signaling pathway are critical for ISC division.
Insulin receptor signaling is required for ISC division
The MARCM strategy coupled with mutant chromosomes allows positive marking of mutant cell lineage by GFP. Analysis of MARCM flies with InR mutant chromosomes revealed that after DSS feeding 90% of GFP-positive clusters contained only 1 or 2 cells (, blank region of the bars, and ), demonstrating a lack of proliferation. In contrast, MARCM flies with wild type chromosomes had 80% of their GFP-positive clusters contained 3 or more cells (, filled region of the bars). This demonstrates that InR function is required within ISC for proliferation.
We also used various Gal4 drivers to express a constitutively active insulin receptor (InRA1325D) to test in which cell type InR can exert its effect. The esg-Gal4 but not the NPC1b-Gal4 driver led to highly increased ISC division, suggesting that the expression of InRA1325D in precursor cells but not in epithelial cells can stimulate ISC division (). Heat shock induced expression in adult flies also caused a substantial increase in ISC division, demonstrating that InR can regulate ISC division after development has completed. Furthermore, the expression of InRA1325D by the esg-Gal4 driver caused phenotypic changes in the midgut similar to that induced by DSS, including a less organized β-catenin staining pattern (), clearly stronger Delta staining in ISC () and bigger clusters of esg-Gal4/GFP cells (). We also performed double staining of phospho-H3 and Delta in guts from flies overexpressing InRA1325D (). 93.0% (n=499) of the phospho-H3 positive cells also contained Delta staining, demonstrating that it is the stem cells but not other cells types that have increased division. Together, the results are consistent with the idea that InR activation leads to ISC proliferation.
Regulation of ISC division by systemic insulin
Drosophila has 7 insulin-like peptides (DILPs) and when expressed by transgenic constructs all can function to change the growth and metabolic state of the animal (Brogiolo et al., 2001
; Ikeya et al., 2002
). The main source of DILPs is the median neurosecretory cells in the brain. Ablation of these neurosecretory cells causes insulin depletion phenotypes such as retarded growth and elevated carbohydrate level (Ikeya et al., 2002
; Rulifson et al., 2002
). Thus, we performed this ablation experiment by crossing a DILP2-Gal4 driver with UAS-Reaper constructs. Flies containing both DILP2-Gal4 and UAS-Reaper hatched approximately 10 days later than other flies, demonstrating that the ablation caused an insulin depletion phenotype. These flies also showed approximately 90% decrease of DSS-induced and 50% decrease of bleomycin-induced ISC division (). β-catenin staining showed that these DILPs depleted flies and InR mutant flies had a similar gut phenotype, that means a better organization and smaller number of cells after DSS and bleomycin feeding (). We examined whether this ablation experiment causes a change of gut DILP2 expression. First, we checked the DILP2-Gal4 driven UAS-GFP expression and we could not observe GFP signal in gut (data not shown). Second, we examined dilp2
mRNA expression in heads vs. guts in the ablated flies and found that the dilp2
mRNA was largely reduced in heads while the expression had no change in guts, which was very low to start with (). Third, published papers also show that DILP2 has no expression in the gut (Brogiolo et al., 2001
; Ikeya et al., 2002
). All these suggest that the ablation experiment mainly affects brain expression. However, there are 7 DILPs in Drosophila and we cannot exclude the possibility that other gene expression is affected. Moreover, ablation of brain cells may indirectly affect gut physiology.
Systemic insulin level can regulate ISC division
Previous reports show that DILP2 has a consistent effect on growth and metabolism (Brogiolo et al., 2001
; Ikeya et al., 2002
; Rulifson et al., 2002
). Thus, we expressed DILP2 in neurosecretory cells, in gut cells, or ubiquitously by DILP2-Gal4, esg-Gal4 or Armadillo-Gal4, respectively, and examined the effect on ISC proliferation. All these drivers caused approximately 2–3 fold increase in phospho-H3 positive cells in the midgut (). This increase is modest but similar to previously reported 2 fold increase in overall growth (Ikeya et al., 2002
). Moreover, double staining in DILP2-Gal4/UAS-DILP2 guts showed that 94.0% (n=134) of the phospho-H3 positive cells also contained Delta staining, demonstrating that it is ISC that have increased division. We conclude that the systemic level of insulin can regulate ISC division in adult midgut.