|Home | About | Journals | Submit | Contact Us | Français|
In the posterior midgut of Drosophila, homeostasis is maintained by stem cells [1, 2]. The intestinal epithelium contains two types of differentiated cells that are continually lost and replenished - secretory enteroendocrine (EE) cells and absorptive enterocytes (ECs) (Fig. 1A). Intestinal stem cells (ISCs) are the only cells in the adult midgut that proliferate [3, 4], and ISC divisions give rise to an ISC and an enteroblast (EB), which then differentiates into either an EC or an EE cell (Fig. 1A) [3–5]. If the midgut epithelium is damaged, then ISC proliferation increases [6–12]. Damaged ECs express secreted ligands (Unpaired proteins, Upd) that activate Jak-Stat signaling in ISCs and EBs to promote their proliferation and differentiation, respectively [7, 9, 13, 14]. We show that the Hippo pathway components Warts (Wts) and Yorkie (Yki) mediate a transition from low to high level ISC proliferation to facilitate intestinal regeneration. The Hippo pathway regulates growth in diverse organisms, and has been linked to cancer [15, 16]. Yki is activated in ECs in response to tissue damage, or activation of the damage-sensing Jnk pathway. Activation of Yki promotes expression of upd genes and triggers a non-autonomous increase in ISC proliferation. Our observations uncover a novel role for Hippo pathway components in regulating stem cell proliferation and intestinal regeneration.
Yki is negatively regulated by the Warts (Wts) kinase, which promotes its cytoplasmic localization [17–21]. To investigate potential roles for the Hippo pathway in the development or physiology of the intestine, we analyzed the consequences of Yki activation. Yki was activated either by mutation or RNAi-mediated depletion of wts, or by expression of an activated form of Yki (YkiS168A:V5) , which contains a mutation in a key Wts phosphorylation site. To restrict Yki activation to adult stages, we made use of the TARGET system , in which a gene or RNAi line of interest is expressed under UAS-Gal4 control, and then tight temporal control is provided by simultaneous expression of a temperature-sensitive transcriptional repressor of Gal4-driven expression (Gal80ts).
In initial experiments, endogenous Yki was activated throughout the adult fly by using a wts hairpin transgene (wts RNAi) expressed under tub-Gal4 control to deplete endogenous Wts. This resulted in a pronounced hyperplasia of the intestine, with many tightly packed cells, and a thicker intestinal epithelium (Figs 1B,C, S1A–C). This was accompanied by an apparent increase in ISC proliferation, as revealed by staining for expression of the Notch ligand Delta (Dl) (Figs 1C, S1C). Normally, Dl is specifically expressed in ISCs within the intestine, and is lost from the EB daughter cell when ISCs divide . However, toxins, infection, or Jnk pathway activation can substantially increase numbers of Dl-expressing cells [6, 8, 9, 11, 12], apparently because the loss of Dl expression in EBs is slow compared to the increased proliferation rate of ISCs. The intestinal hyperplasia and increased number of Dl-expressing cells imply that Hippo signaling normally functions to limit cell numbers in the intestine, at least in part by keeping ISC proliferation at a low level.
To confirm the role of Wts and Yki in ISC proliferation, and to identify the cell types in which they function, we employed cell-type specific Gal4 drivers. As markers of progenitor cells, we used both Dl, and escargot (esg), which is expressed specifically in ISCs and EBs . Surprisingly, when YkiS168A:V5 was expressed in the ISC and EB under esg-Gal4 control (Supplementary Fig. S1M), no effect on intestinal morphology or progenitor cell numbers was detected, even 10 days after induction of YkiS168A:V5 expression (Fig. 1D). Thus, Yki activation in ISCs does not affect their proliferation. Visceral muscle cells adjacent to the intestinal epithelium have been reported to promote ISC proliferation by expressing Wingless . To examine whether Yki acts in muscle cells to promote ISC proliferation, we expressed YkiS168A:V5 using How-Gal4, which drives expression in muscles  (Fig. S1G,H). Although at longer time points (ten days after induction), a patchy increase in progenitor cells was detected (Supplementary Fig. S1F), at shorter time points (three days after induction of YkiS168A:V5), hardly any effect was observed (Fig. S1E).
By contrast to the relatively modest or undetectable consequences of Yki activation in muscles or ISCs, Yki activation in the differentiated ECs of the intestinal epithelium resulted in a rapid and dramatic hypertrophy (Fig. 1). YkiS168A:V5 was expressed specifically in ECs using Myo-Ia-Gal4  (Fig. S1I–L). The resulting intestinal overgrowth was associated with multiple layers of intestinal cells (Fig. 1H), by contrast to the normal monolayer (Fig. 1G). Moreover, the intestinal lumen filled with cells (Fig. 1H), presumably sloughed off from the over-proliferating intestinal epithelium. The intestinal hypertrophy was accompanied by a substantial increase in numbers of Dl- and esg-expressing cells (Fig. 1I,J). esg-lacZ was no longer restricted to adjacent pairs of cells (the ISC and EB) (Figs 2F, S1M), but instead was detected in many cells that morphologically appear to be ECs (recognizable by their large size and large, polyploid nucleus) (Fig. 1I). We suggest that this reflects both the accelerated proliferation and differentiation of progenitor cells, and the stability of β-galactosidase. Confirmation that activation of endogenous Yki specifically within ECs also increases Dl-expressing progenitor cells was provided by examination of intestines in which Wts was depleted from ECs by TARGET-controlled induction of wts RNAi under MyoIa-Gal4 control, which resulted in accumulation of small Dl-expressing cells basal to the large, MyoIa-expressing ECs. (Fig. 1K,L). These observations imply that Yki activation in ECs exerts a strong, non-autonomous influence on the proliferation of ISCs.
To directly visualize this increased proliferation, we stained for cells in mitosis using anti-phospho-Ser10-Histone H3 (pH3), and labeled cells in S-phase using EdU. When Yki was activated specifically in ECs using MyoIa-Gal4 to drive expression of YkiS168A:V5 or wts RNAi, then the number of cells labeling for pH3 and EdU increased substantially (Fig 1M–Q). pH3 staining was only detected in small, scattered cells, which based on their distribution and morphology we infer are ISCs (Fig. 1P). This implies that activation of Yki does not cause the normally non-proliferating EBs, EEs, or ECs to proliferate, but rather increases the proliferation rate of ISCs. ISC proliferation is normally very low, and we detected an average of two pH3-staining mitoses per adult posterior midgut in control animals, consistent with prior studies (Fig. 1Q) . By contrast, wts-RNAi treated posterior midguts averaged fifty-five pH3-staining mitoses (Fig 1Q). EdU labeling was increased not only in ISCs, but also in ECs (Fig. 1M–P). ECs become polyploid through endoreplication; the increased DNA synthesis might reflect a more rapid differentiation of ECs , and/or an increase in polyploidy, which could enable the cells to grow larger.
To further characterize the non-autonomous influence of Yki activation on ISC proliferation, we generated clones of cells with activated Yki. Induction of YkiS168A:V5 expression in clones using the Flp-out technique, which enables expression to be induced in non-dividing cells , and thus allows induction of expression directly in ECs, resulted in a rapid, non-autonomous increase in esg-lacZ-expressing cells (Fig. 2C). Induction of clones by mitotic recombination is limited to dividing cells, which means that such clones can only be induced in the ISC. However, after waiting several days, these clones can increase in size through subsequent divisions of the ISC, and can include ECs differentiated from EB daughter cells within the clone. Mitotic recombination was coupled to the MARCM technique to make clones of cells expressing YkiS168A:V5, or mutant for wts. Both of these methods of Yki activation also non-autonomously increased ISC proliferation, based on examination of esg and Dl, as well as direct examination of proliferation using pH3 or EdU labeling (Figs 2, S2). Activation of Yki did not block differentiation, because MARCM clones with activated Yki, which must have been induced in dividing ISCs, could include cells with large nuclei, which expressed Pdm1, a marker of differentiated ECs (Fig S2E). Based on EdU labeling, pH3 staining, esg-lacZ expression, or Dl expression, the range of the non-autonomous response to Yki activation in EC cells could extend for several cells (Fig. 2).
The non-autonomous influence of Yki activation on ISC proliferation implies that activation of Yki in ECs leads to production of a secreted signal that can influence ISC proliferation. Several manipulations that can influence intestinal cell proliferation have been identified, including expression of Wg in visceral muscle cells , mutation of Notch in ISCs [3, 4], activation of the Jnk pathway in ECs [6, 7, 9, 12], and activation of the Jak-Stat pathway in ISCs [7, 9, 13, 14]. The Jak-Stat pathway was a candidate effector of the influence of activated Yki, because ligands for this pathway are expressed in ECs in response to damage, infection, or activation of Jnk signaling, and activation of the pathway within ISCs in response to secreted Upd ligands promotes their proliferation [7, 9, 13, 14]. Additionally, like Wts and Yki, Jak-Stat and Jnk signaling have rapid effects on ISC proliferation, visible in less than 2 days, whereas manipulations of Wg or Notch signaling require longer times for significant effects on proliferation to be observed. Indeed, using an upd-lacZ reporter, we found that upd was barely detected in control intestines, but was readily detected after activation of Yki in ECs (Fig. 3A,B). The upd locus includes three related genes, upd1 (os), upd2, and upd3. To examine the expression of individual upd genes, and to quantify their expression, we performed quantitative RT-PCR on intestines with activated-Yki in ECs. Expression of upd3 increased approximately 25-fold, whereas upd1 and upd2 increased 7–8-fold (Fig. 3L). This upregulation resulted in a visible increase in Jak-Stat pathway activity, as monitored by expression of a Stat-DGFP reporter  (Fig. 3C,D). Based on cellular morphology, Stat-DGFP was upregulated most strongly in ISCs and EBs, but it was also visibly upregulated in other intestinal cells. These observations suggest that activation of Yki in ECs promotes ISC proliferation and intestinal hyperplasia by inducing expression of upd genes, which are secreted from ECs and then activate the Jak-Stat pathway in ISCs and EBs to promote their proliferation and differentiation.
To test this hypothesis, RNAi was used to simultaneously reduce expression of wts and hopscotch (hop), which encodes the Drosophila Jak kinase. As wts is required in ECs, whereas Jak-Stat signaling is required in ISCs, tub-Gal4 was used to reduce the levels of both genes in both cell types. The increases in Dl-expression and pH3 staining cells normally associated with wts RNAi were reduced by RNAi of hop (Figs 3I–K, S3), whereas neutral UAS transgenes (ie UAS-GFP) did not affect wts RNAi phenotypes. In a second approach, mitotic recombination was used to make clones doubly mutant for wts and stat92E, which encodes the DNA-binding transcription factor of the Jak-Stat pathway. The influence of stat92E on ISC proliferation in these experiments was assayed by counting the number of cells within each clone. Clone size was significantly reduced using either of two different stat92E alleles (Fig 3E–H). As expected, non-autonomous effects of wts on ISC proliferation were not blocked (Fig. 3F). Thus, both experimental approaches confirm that the influence of Yki activation on Jak-Stat signaling contributes to its effect on ISC proliferation. The suppression was incomplete, which might be due to residual Hop or Stat, but the possibility that additional mitogens acting via other pathways are also produced in ECs upon Yki activation cannot be excluded.
The observation that reducing wts levels increases ISC proliferation indicates that Wts is normally active in ECs, and needs to be active there to prevent excess ISC proliferation. In order to investigate whether Yki has a normal role in promoting endogenous levels of ISC proliferation, we employed RNAi-mediated knockdown of yki in EC cells. Antibody staining confirmed that RNAi reduced levels of Yki (Fig. 4B and data not shown). However, even after seven to ten days, the overall size of the intestine and number of ISCs was similar or only modestly reduced compared to that in control animals. This suggests that endogenous levels of Yki are not required for maintaining intestinal homeostasis in healthy animals, although this is subject to the caveat that some residual Yki was detected in RNAi-treated intestines.
Expression of upd genes is normally very low in intestinal cells, but is upregulated in response to damage or infection to trigger a regenerative response that induces ISC proliferation [7, 9, 13, 14]. To investigate whether Yki has a role in damage response, we treated animals with bleomycin, a DNA-damaging agent that induces a regenerative response in the intestine . The increase in Dl-expressing progenitors associated with bleomycin treatment (Fig. 4A) was reduced by RNAi of yki (Fig. 4B). The increase in ISC mitosis associated with bleomycin (visualized by pH3 staining) was also reduced by yki RNAi, to about half the level in sibling controls (Fig. 4C–E). These observations indicate that the damage-response pathway activated by bleomycin acts at least in part through Yki. Consistent with this, Yki localization was altered in bleomycin-treated intestines. In wild-type, Yki is predominantly cytoplasmic (Fig. 4I)[19, 20], but in ECs of bleomycin-treated intestines, increased Yki could be detected within the nucleus of many cells (Fig. 4H), which is a hallmark of Yki activation [19, 20]. Additionally, a transcriptional reporter of Hpo signaling, ex-lacZ , was upregulated (Fig. 4G). Thus, a toxin that damages ECs can induce Yki activation.
Analysis of intestinal regeneration and upd expression has implied that there are multiple, parallel, mechanisms by which infection or tissue damage can upregulate upd genes [7, 9, 13]. Many of these act through the Jnk signaling pathway, which mediates responses to tissue damage in diverse contexts, including toxins, aging, and infection. Direct activation of Jnk signaling can also induce intestinal hyperplasia and expression of upd genes [6, 7, 9, 12]. Thus, we investigated whether Jnk is involved in mediating the effects of tissue damage on Yki. Indeed, direct activation of Jnk signaling in ECs, achieved by expression of an activated form of the Jnk-kinase Hemipterous (Hep), HepCA, using MyoIa-Gal4, or Gal4-expressing clones, increased nuclear Yki localization (Fig. 4J,K). Thus, Jnk activation can promote Yki activation. Consistent with this, we observed that inhibition of Jnk signaling, achieved by expressing a dominant negative form of the Drosophila Jnk Basket (Bsk), BskDN, which blocks Jnk pathway activation in the intestine [6, 9, 12], did not block the influence of YkiS168A:V5 on ISC proliferation (Fig. S4A,B). We also note that unlike Jnk pathway activation, Yki activation did not induce visible apoptosis or Caspase cleavage (Fig S2C,D).
Our results establish that two Hippo pathway components, Wts and Yki, play a crucial role in a damage-response pathway (Fig. SC). Activation of Yki promotes the expression of cytokines that promote the proliferation and differentiation of ISCs. Because Jnk signaling is generally activated by tissue damage and infection, the discovery that Yki is activated by Jnk signaling in the intestine raises the possibility that Hippo signaling might play a general role in regenerative responses to tissue damage. This possibility also suggests a convergence between the role of Hippo signaling in cancer biology [17, 29], and the role of infection and tissue damage in oncogenesis [30, 31].
Hippo signaling can have both autonomous and non-autonomous effects on growth [32, 33], and we report here that in the adult Drosophila midgut, Yki has profound non-autonomous effects on growth via the Jak-Stat pathway. Jak-Stat signaling is important for proliferation control and stem cell biology not only in the Drosophila intestine, but also in other tissues, both in Drosophila and in vertebrates [34–36]. Members of the interleukin (IL) family of cytokines are homologous to Upd ligands, and a microarray study in cultured mammalian cells found that the Yki homolog, Yap, could regulate IL cytokines , which raises the possibility that a regulatory connection between Hippo signaling and Jak-Stat signaling might be conserved. Increased levels and nuclear localization of Yap have been reported in colon cancer patient samples , and ubiquitous Yap1 over-expression causes over-proliferation of progenitor cells in the murine intestine . Our observations suggest that future considerations of the potential contributions of Hippo signaling to colon cancer should include evaluations both of its possible regulation by Jnk signaling, and of possible non-autonomous effects mediated by cytokines.
Conditional expression in adult flies using tub-Gal80ts was achieved by maintaining flies at 18C throughout their development, and then shifting 6–8 day old (at 18C) adult females to higher temperature for induction of gene expression. Initial experiments employed using a shift to 30C, but we found later that a shift to 28.5C was also effective. For wts conditional RNAi, Tub-Gal80ts/CyO; Tub-Gal4 UAS-dcr2/TM6b or MyoIA-Gal4/CyO; Tub-Gal80ts UAS-dcr2/TM6b, or MyoIA-Gal4 Tubgal80ts UAS-GFP/CyO; UAS-dcr2/TM6B flies were crossed to UAS-wts-RNAi (vdrc9928). For conditional Yki expression, esg-Gal4/CyO; Tub-Gal80ts/TM6b, esg-Gal4 UAS-GFP UAS-n.lacZ/CyO; Tub-Gal80ts/TM6b, Tub-Gal80ts/(CyO); How-Gal4/(TM6b), or MyoIA-Gal4/Cyo; Tub-Gal80ts/TM6b were crossed to esg-lacZ/(CyO); UAS-yki168A:V5/(TM6b), UAS-nls: GFP/(CyO); UAS-yki168A:V5/(TM6b), y w hs-FLP; 10XStat-DGFP (destabilized GFP)/(CyO);UAS- yki168A:V5/(TM6B), upd-lacZ;UAS- yki168A:V5/TM3, UAS-BskDN; UAS-yki168A:V5/TM3 or y w hs-FLP; IF/CyO; UAS-HepCA/TM6b. For expression of activated Yki or Hep in FLP-out clones, w tub>CD2>Gal4 UAS CD8:GFP/CyO; Tub-Gal80ts/TM6b was crossed to y w hs-FLP; IF/CyO; UAS-yki168A:V5 or y w hs-FLP; IF/CyO; UAS-HepCA/TM6b. Adult females were aged for 3–4 days and heat shocked at 37C for 1 hour and the vials placed back to 18C for 5–12 days, and then shifted to 30C for the desired time and dissected. For expression of activated Yki in MARCM clones, y+ FRT40A/CyO;UAS- yki168A:V5/(TM6B) was crossed to y w hs-FLP Tub-Gal4 UAS-GFP/FM7; Tub-Gal80 FRT40A/CyO. For wts and stat92E MARCM clones, y w hs-FLP Tub-Gal4 UAS-GFP; UAS-y+/(CyO); FRT82B tub-gal80/TM6B was crossed to w; wtsX1 FRT82B/TM6B, w; stat92E85C9 FRT82B/TM6B, w; stat92E06346 FRT82B/TM6B, w; wtsX1 stat92E85C9 FRT82B/TM6B, or w; wtsX1 stat92E06346 FRT82B/TM6B. 6–8 day old (at 18C) adult females were then heat shocked at 37C for 35 min, and kept at 25C until dissection. For hop-RNAi we used vdrc102830. A substantial increase in esg+ cells was observed in a small fraction (approximately 1/15) of intestines in both experimental and control animals. We suspect this sporadic phenotype reflects stimulation of ISC proliferation by ingested microbes [6–10, 13], and we focused only on phenotypes that consistently correlated with experimental manipulations. As negative controls we used flies lacking the relevant UAS or Gal4 transgene, and temperature shifted in parallel. In addition, we also performed control experiments for all TARGET genotypes, in which flies were not shifted to the permissive temperature for expression.
Bleomycin (Sigma) treatment was done as described previously , by mixing bleomycin into instant food (final concentration 25 µg/mL) and feeding to flies cultured at 29C; flies were transferred daily into new vials with fresh bleomycin. For RNAi experiments, MyoIA-Gal4/Cyo; Tub-Gal80ts/TM6b was crossed to y w hs-FLP; UAS-yki RNAi (vdrc104523)/CyO; UAS-dcr2/ TM6b, and 3–4 day old adult females were shifted to 29C for 5–8 days before transferring to Bleomycin at 29C for 2 more days.
Posterior midguts of adult females were dissected in PBS, fixed in 4% PFA for 45 min, and then stained with antibodies as described previously , using mouse anti-Arm (DSHB, 1:100), mouse anti-Dl (1:40, DSHB), goat anti-β-gal (1:100, Biogenesis), rabbit anti-Yki (1:400) , rabbit anti-pH3 (1:500, Cell Signaling Technology), rabbit anti-Pdm1 (1:500, Y. Xiaohang), rabbit anti-Cleaved Caspase-3 (1:200, Cell Signaling Technology), and Alexa488-conjugated phalloidin (1:100, Invitrogen), secondary antibodies were from Jackson ImmunoResearch. Confocal images were captured on a Leica SP5, images shown represent projections through a subset of sections of a confocal stack.
Additional methods are in the supplemental material.
We thank the Developmental Studies Hybridoma Bank, the Bloomington stock center, E. Bach, B. Edgar, S. Hayashi, R. Steward, Y.H. Sun, Y. Xiaohang, for antibodies and Drosophila stocks, and C. Rauskolb for comments on the manuscript. This research was supported by the HHMI.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.