Drosophila Sulf1 is required for the termination of intestinal stem cell division during regeneration Free

Stem cell activity changes over time to meet variable tissue demands during normal homeostasis and regeneration. When a tissue is severely damaged, resident stem cells transiently activate their division and produce differentiated cells to compensate for damaged and lost cells. Once the tissue recovers, the stem cell division rate returns to its baseline level. Improper control of stem cells can lead to impaired regeneration or tumorigenesis. Thus, it is important to understand the molecular mechanisms by which stem cell division rates are precisely regulated during regeneration.

The Drosophila adult midgut, the equivalent of the mammalian small intestine, offers an excellent model system to study the molecular mechanisms underlying tissue turnover and regeneration. Midgut homeostasis is maintained by intestinal stem cells (ISCs), which are scattered along the basement membrane and visceral muscles surrounding the midgut epithelium. ISCs divide to self-renew and produce non-dividing enteroblasts, which differentiate into either absorptive enterocytes or secretory enteroendocrine cells (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006). It has been shown that ISC division is stimulated by tissue damage (Amcheslavsky et al., 2009). When the midgut is exposed to stress, such as enteric infection or oxidative stress, the ISC division rate dramatically increases and then the midgut recovers its normal morphology within a few days (Jiang et al., 2009).

It has been shown that ISC division and differentiation are regulated by several signaling pathways, including Janus kinase and signal transducer and activator of transcription (JAK-STAT), epidermal growth factor receptor (EGFR), Hedgehog (Hh), bone morphogenetic protein (BMP) and Wingless (Wg) signaling (Ayyaz et al., 2015; Beebe et al., 2010; Biteau and Jasper, 2011; Buchon et al., 2009a,b, 2010; Cordero et al., 2007, 2012; Cronin et al., 2009; Guo et al., 2013; Jiang et al., 2009, 2011; Lee et al., 2009; Li et al., 2014, 2013b; Lin et al., 2008, 2010; Liu et al., 2010; Osman et al., 2013; Tian and Jiang, 2014; Tian et al., 2015; Xu et al., 2011; Zhou et al., 2013, 2015). In response to tissue damage, the ligands of these pathways are transcriptionally upregulated in cells surrounding ISCs, and activate signal transduction in ISCs to promote their division. However, compared to the initiation of regeneration, the mechanism for how regeneration termination is controlled is poorly understood. Although previous studies have shown that Decapentaplegic (Dpp; a Drosophila BMP homolog) can negatively regulate ISC division and thus can contribute to the termination of midgut regeneration (Ayyaz et al., 2015; Guo et al., 2013), it is still largely unknown how stem cells rapidly and precisely change their mode from a mitotically active state to a homeostatic state at the end of regeneration.

Heparan sulfate proteoglycans (HSPGs) are one class of candidate molecules that might regulate ISC activity during regeneration. HSPGs are present on the cell surface and in the extracellular matrix, and regulate the activity and spatial distribution of a wide variety of growth factors, cytokines and morphogens (Filmus et al., 2008; Sarrazin et al., 2011). A HSPG is composed of a core protein and heparan sulfate chains, which are long linear carbohydrate polymers. Heparan sulfate is synthesized by a series of heparan sulfate biosynthetic and modifying enzymes in the Golgi (Esko and Lindahl, 2001). During this process, sulfate groups are added to specific ring positions of heparan sulfate. The amount and patterns of sulfation greatly affect the affinity of heparan sulfate for signaling molecules, and thus modulate the function of HSPGs. Recent studies have shown that sulfation at the 6-O position of glucosamine residues is a key modification for heparan sulfate function (Kamimura et al., 2006; Pye et al., 1998). After heparan sulfate 6-O sulfation is catalyzed by a 6-O sulfotransferase (Hs6st) in the Golgi, the levels of 6-O sulfation can be further modified by the extracellular endosulfatases called Sulfs (Ai et al., 2003; Dhoot et al., 2001; Uchimura et al., 2006). Sulfs specifically remove 6-O sulfate groups from highly sulfated regions of heparan sulfate. In Drosophila, the single Sulf gene, Sulf1, modulates FGF, Wg, Hh and EGFR signaling during development (Butchar et al., 2012; Dani et al., 2012; Kamimura et al., 2006; Kleinschmit et al., 2013, 2010; Wojcinski et al., 2011; You et al., 2011).

A previous study has shown that the Drosophila perlecan Trol is required for ISC–basement-membrane attachment (You et al., 2014). In addition, loss of heparan sulfate 3-O sulfotransferase (Hs3st) leads to activated ISC division that is mediated by upregulated EGFR signaling during normal homeostasis (Guo et al., 2014). These studies suggest that HSPGs and their sulfation status are important regulators of ISC activity. However, important unanswered questions remain, including: (1) is heparan sulfate required for control of stem cell activity during regeneration, (2) is a specific modification of heparan sulfate critical for this control, and (3) does such a modification contribute to regeneration termination? Here, we show that heparan sulfate 6-O sulfation plays a crucial role in activation and inactivation of ISC mitosis during both normal homeostasis and regeneration. Decreasing heparan sulfate 6-O sulfation by performing Hs6st knockdown resulted in a reduced activity of ISC division during regeneration. Loss of Sulf1 led to increased ISC mitotic activity and aberrant activation of the JAK-STAT, EGFR and Hh signaling pathways. Furthermore, we found that ISC division was continuously activated even at the later stages of regeneration in Sulf1 mutants. Taken together, we propose that heparan sulfate modification by Sulf1 is required for terminating ISC division promptly at the end of regeneration. Our study provides a new regulatory mechanism by which a specific modification of heparan sulfate regulates stem cell activity during regeneration through post-transcriptional control of mitogen signaling.

To determine whether HSPGs are required for the ISC damage response, we examined the effect of RNA interference (RNAi)-mediated knockdown of sulfateless (sfl) during midgut regeneration. sfl encodes the only Drosophila heparan sulfate N-deacetylase/N-sulfotransferase, which catalyzes the first step of heparan sulfate modification in the Golgi. Given that this reaction is essential for subsequent steps of heparan sulfate modification, blocking this enzyme results in the loss of heparan sulfate activity in a cell autonomous manner (Lin and Perrimon, 1999; Lin et al., 1999; Toyoda et al., 2000). We downregulated sfl in progenitor cells (ISC and enteroblast) using the escargot-GAL4 tubulin-GAL80^ts (esg^ts) system (Micchelli and Perrimon, 2006) during regeneration. Briefly, newly eclosed females carrying UAS-sfl RNAi and esg^ts transgenes were raised at 19°C for 3–4 days and shifted to 30°C for 3 days to allow UAS transgene expression. The flies were then infected with Pseudomonas entomophila (Vodovar et al., 2005) to induce midgut regeneration and examined for ISC division by performing anti-phospho-histone-H3 (pH3) antibody labeling. As previously reported (Jiang et al., 2009), a dramatic increase in the number of pH3⁺ cells in the posterior midgut from control flies was observed in response to P. entomophila infection (Fig. 1A,D). When sfl was downregulated by UAS-sfl RNAi in progenitor cells, the number of pH3⁺ cells was significantly decreased (Fig. 1B,D). Silencing sfl with another RNAi line similarly resulted in reduced ISC mitotic activity during regeneration (Fig. 1E), which allowed us to exclude potential off-target effects of RNAi on ISC division. Taken together, these results demonstrate that heparan sulfate is necessary for ISC activation during regeneration.

Perlecan, an evolutionarily conserved secreted HSPG, is a major constituent of the basement membrane and plays crucial roles in mammalian development (Poulain and Yost, 2015). Expression of Trol was detected in the basement membrane using a Trol protein-trap line (Morin et al., 2001) (Fig. 1F), consistent with a previous study (You et al., 2014). It has been reported that Trol functions in maintenance of ISCs during normal homeostasis (You et al., 2014). To determine whether trol is required for midgut regeneration, we performed an RNAi knockdown experiment using the esg^ts system. We found that knockdown of trol in progenitor cells using two independent UAS-trol RNAi transgenes significantly disrupted bacteria-induced ISC division (Fig. 1C–E). These results indicate that, consistent with its role in normal homeostasis, trol is required for ISC activation during regeneration. RNAi knockdown of trol in progenitor cells did not cause detachment of progenitor cells from the basement membrane (Fig. 1G,H), suggesting that Trol is involved in regulation of ISC division independently of its function in ISC–basement-membrane attachment.

An increasing amount of evidence shows that 6-O sulfation plays a key role in regulating heparan sulfate function. Specifically, the level of 6-O sulfation modulates the activities of various signaling molecules, such as FGF, Wnt and Hh (Ai et al., 2003; Danesin et al., 2006; Dhoot et al., 2001; Ma et al., 2011; Seffouh et al., 2013; Uchimura et al., 2006). To determine whether 6-O sulfation is key in the regulation of stem cell activity during midgut regeneration, we first determined expression patterns of Hs6st in the midgut using a Hs6st::EGFP protein-trap allele (Nagarkar-Jaiswal et al., 2015). We detected the GFP signal in progenitor cells marked by esg-lacZ (an ISC and enteroblast marker) (Wilk et al., 2000) as well as in cells stained by anti-Prospero (Pros) antibody (an enteroendocrine marker) (Fig. 2A,B), indicating that Hs6st is expressed in ISCs, enteroblasts and enteroendocrine cells.

To determine whether heparan sulfate 6-O sulfation is necessary for ISC division activation during regeneration, we downregulated Hs6st by expressing UAS-Hs6st RNAi (Kamimura et al., 2006) in the progenitor cells using the esg^ts system. We found that this treatment resulted in a significant reduction in the number of pH3⁺ cells (Fig. 2D,F) compared to the control regenerating midgut (Fig. 2C,F). Knockdown of Hs6st using another RNAi line similarly led to a reduced number of dividing ISCs during regeneration (Fig. 2E,F). These results show that 6-O sulfation of heparan sulfate by Hs6st promotes ISC division during regeneration.

6-O sulfation is a unique modification in that its status can be modulated in a post-biosynthetic manner by the extracellular heparan sulfate 6-O endosulfatase Sulf1. This fine-tuning of heparan sulfate sulfation plays an important role in modulating the function of HSPGs (Ai et al., 2003; Dhoot et al., 2001; Lamanna et al., 2006; Morimoto-Tomita et al., 2002; Wang et al., 2004). Previous studies have shown that Sulf1-catalyzed decrease in 6-O sulfation lowers the affinity between heparan sulfate and ligand proteins (Kleinschmit et al., 2013). This can positively or negatively regulate signal transduction in a context-dependent manner (Ai et al., 2003; Kleinschmit et al., 2013; Wojcinski et al., 2011). In the Drosophila wing disc, Sulf1 negatively regulates the FGF, Wg, EGFR and Hh pathways (Butchar et al., 2012; Kamimura et al., 2006; Kleinschmit et al., 2013, 2010; Wojcinski et al., 2011; You et al., 2011). Given that some of the signaling ligands known to regulate ISC division in the midgut are heparan-sulfate-dependent factors, including Vein (Vn; an EGFR ligand), Hh, and Wg, we hypothesized that dynamic regulation of 6-O sulfation might be important for controlling ISC mitotic activity during regeneration. To examine the role of Sulf1 in the midgut, we generated Sulf1 knockout (KO) flies using the clustered regularly interspaced short palindromic repeats (CRISPR)-Cas9 system (Fig. 3A,B) (Gratz et al., 2014; Jinek et al., 2012; Ren et al., 2014). In these Sulf1{KO; mCherry} flies, most of the Sulf1 coding sequence is replaced with a red fluorescent protein mCherry gene by homology-directed repair, which was confirmed by Sanger sequencing (Fig. 3B,C). Given that the Sulf1{KO; mCherry} allele retains Sulf1 endogenous regulatory sequences, it can be used as not only a molecular null mutant but also as a highly reliable Sulf1 transcriptional reporter. We examined mCherry expression in the wing imaginal disc where Sulf1 is expressed at high levels along the anteroposterior and dorsoventral boundaries (Butchar et al., 2012; Kleinschmit et al., 2010; Wojcinski et al., 2011; You et al., 2011). The mCherry signal was detected in the previously reported pattern (Fig. 3D), confirming that mCherry expression of Sulf1{KO; mCherry} recapitulates endogenous Sulf1 expression.

We next analyzed expression patterns of Sulf1 in the adult posterior midgut using Sulf1{KO; mCherry}. We did not observe mCherry expression in progenitor cells (Fig. 4A,B). It was detected in enterocytes, which have polyploid large nuclei and are marked by MyoIA-lacZ (Jiang et al., 2009) (Fig. 4C). We also detected the mCherry signal in visceral muscles (Fig. 4D). Taken together, we conclude that Sulf1 is expressed in enterocytes and visceral muscles in the posterior midgut.

To determine whether Sulf1 plays a role in controlling ISC mitotic activity during normal midgut homeostasis, we compared the number of mitotic ISCs in the posterior midgut from wild-type and Sulf1 mutant flies. We found that midgut samples from Sulf1{KO; mCherry} homozygotes exhibited a significant increase in ISC mitotic activity compared to wild-type samples (Fig. 5A,C,E). Similarly, the ISC mitotic activity was increased in midguts transheterozygous for Sulf1{KO; mCherry} and Sulf1^P1, which is another null allele of Sulf1 previously generated by imprecise P-element excision (Kleinschmit et al., 2010) (Fig. 5D,E). Interestingly, a modest but significant increase in the ISC mitotic index was also observed in Sulf1{KO; mCherry}/+ heterozygotes (Fig. 5B,E).

We also found that esg>GFP⁺ polyploid cells (expressing GFP under the control of esg-GAL4) were observed in the Sulf1 mutant midgut (Fig. 5G′; arrows). These cells appear to represent differentiating enterocytes. In contrast, esg>GFP usually marks only progenitor cells in the wild-type midgut (Fig. 5F). This phenotype has been observed in the aged or regenerating midgut, where ISC division and differentiation are promoted (Amcheslavsky et al., 2009; Biteau et al., 2008; Buchon et al., 2009b; Choi et al., 2008).

ISCs can undergo symmetric cell division to increase the stem cell population size (de Navascués et al., 2012; Jiang et al., 2009; O'Brien et al., 2011), which might contribute to an increase in the number of mitotic ISCs in the midgut. To examine whether loss of Sulf1 affects the number of ISCs, we quantified the proportion of ISCs (calculated by dividing the number of ISCs by the total number of cells per unit area) in the posterior midgut from wild-type and Sulf1 mutant flies. We observed no significant difference in the proportion of ISCs, which are labeled by Dl::EGFP (Nagarkar-Jaiswal et al., 2015; Ohlstein and Spradling, 2007), between these genotypes (Fig. 5H–J). This result showed that loss of Sulf1 does not affect the number of ISCs. Thus, the observed increase in pH3⁺ cells in the Sulf1 mutant midgut is due to increased mitotic activity of ISCs rather than an increase in ISC population size. Taken together, our results demonstrate that Sulf1 negatively regulates ISC division during normal midgut turnover.

ISC division is regulated by multiple signaling pathways, including JAK-STAT, EGFR, Hh and BMP signaling. Given that ISC mitotic activity increases in Sulf1 mutants, we asked whether any of these signaling pathways are altered by loss of Sulf1. We performed quantitative reverse transcription PCR (RT-qPCR) using whole midgut samples from wild-type and Sulf1 mutant flies to determine mRNA levels of ligands and target genes for these pathways. The RT-qPCR analysis revealed that among the genes we tested, unpaired 2 (upd2), upd3 (genes encoding JAK-STAT ligands), Socs36E (a target gene of JAK-STAT signaling), vn, hh and patched (ptc; a target gene of Hh signaling) were upregulated more than twofold in the Sulf1 mutant compared to control (Fig. 6A).

To confirm this RT-qPCR result, we performed immunohistochemical analyses using antibodies and reporter lines for each pathway. The JAK-STAT ligand Upd3 is expressed in enterocytes and activates signal transduction in progenitor cells to promote ISC division and differentiation (Beebe et al., 2010; Buchon et al., 2009a,b; Cronin et al., 2009; Jiang et al., 2009; Lin et al., 2010; Liu et al., 2010; Osman et al., 2013; Zhou et al., 2013). We examined upd3 expression using an upd3-GAL4 UAS-GFP (upd3>GFP) reporter strain (Agaisse et al., 2003). The GFP signal was detected in a few enterocytes in the control midgut (Fig. 6B). By contrast, a substantially larger fraction of enterocytes expressed upd3>GFP in the Sulf1 mutant midgut (Fig. 6C). We also monitored JAK-STAT activity by using the reporter line 10×STAT92E-DGFP, in which destabilized GFP downstream of STAT92E-binding sites reports JAK-STAT signaling (Bach et al., 2007). We detected a substantial enhancement of 10×STAT92E-DGFP signals in progenitor cells in the Sulf1 mutant midgut compared to control (Fig. 6D,E). Thus, consistent with the RT-qPCR analysis, these results confirmed that JAK-STAT signaling is hyper-activated in the absence of Sulf1 activity.

Both EGFR signaling (Biteau and Jasper, 2011; Buchon et al., 2010; Jiang et al., 2011; Xu et al., 2011) and Hh signaling (Han et al., 2015; Li et al., 2014; Tian et al., 2015) have been shown to promote ISC division during normal midgut homeostasis and regeneration. In addition to the JAK-STAT pathway, we also confirmed the activation of EGFR and Hh signaling in the Sulf1 mutant midgut as follows: (1) upregulation of the vn-lacZ reporter (Yarnitzky et al., 1997) in visceral muscles (Fig. 6F,G), (2) an increased signal intensity of anti-phospho-ERK (pERK, phosphorylated Drosophila Rolled) antibody (Gabay et al., 1997) in progenitor cells (Fig. 6H,I), and (3) an elevated expression of Hh as determined by using anti-Hh antibody (Fig. 6J,K). Blocking JAK-STAT, EGFR and Hh signaling independently each resulted in reduced activity of ISC division in the Sulf1 mutant midgut (Fig. S1), indicating that these pathways are responsible for activated ISC division in the mutant.

Previous studies have reported that BMP signaling is activated in ISCs, enteroblasts and enterocytes, and that it plays several roles, including regulation of ISC division and differentiation, enterocyte survival and copper cell differentiation (Ayyaz et al., 2015; Guo et al., 2013; Li et al., 2013a,b; Tian and Jiang, 2014; Zhou et al., 2015). Our RT-qPCR analysis did not detect any difference in the mRNA levels of dpp and glass bottom boat (gbb), which encode ligands of the BMP pathway, between the wild-type and Sulf1 mutant midgut samples (Fig. 6A). In addition, no difference was observed in the level of Daughters against dpp (Dad), a target gene of the BMP pathway (Fig. 6A). Thus, BMP signaling is not significantly affected by loss of Sulf1 in the midgut.

It is known that the stress-responsive JNK pathway acts upstream of JAK-STAT and EGFR signaling in the regenerating or aged midgut: JNK signaling promotes ISC division through inducing expression of JAK-STAT and EGFR ligands (Biteau et al., 2008; Buchon et al., 2009a,b; Jiang et al., 2009). To determine whether the JNK pathway is activated in the Sulf1 mutant midgut, we monitored the expression level of puckered (puc), a target gene of JNK signaling (Martín-Blanco et al., 1998), by RT-qPCR. Interestingly, we observed no obvious difference in puc mRNA levels between the wild-type and Sulf1 mutant samples (Fig. 6A). Moreover, we did not observe apoptosis in the Sulf1 mutant midgut, as monitored by use of anti-cleaved caspase-3 antibody (Fig. 6L,M). This suggests that loss of Sulf1 leads to elevated activity of the JAK-STAT, EGFR and Hh pathways without activating tissue stress signaling.

Our results indicate that Sulf1 negatively regulates ISC division by modulating JAK-STAT, EGFR and Hh signaling during normal homeostasis. These pathways are known to be highly activated in response to stress and tissue damage, and are responsible for transient activation of ISC division during regeneration (Buchon et al., 2009a,b, 2010; Cronin et al., 2009; Jiang et al., 2009, 2011; Osman et al., 2013; Tian et al., 2015; Zhou et al., 2013). Importantly, ISC mitotic activity reverts to a baseline level once regeneration is completed (Jiang et al., 2009). The inhibitory effect of Sulf1 on ISC division during normal turnover prompted us to hypothesize that Sulf1 contributes to terminating midgut regeneration by limiting ISC division. To test this idea, we measured the number of mitotic ISCs in wild-type and Sulf1 mutant midgut samples at three time points (day 0, day 3 and day 5) after P. entomophila infection (Fig. 7A) or mock infection (sucrose feeding; Fig. S2A). The wild-type midgut showed an increased number of pH3⁺ cells right after P. entomophila infection, or 0 days post-infection (referred as to 0 dpi; Fig. 7B,H). The ISC mitotic activity decreased at 3 dpi and returned to near the baseline level at 5 dpi (Fig. 7C,D,H). In the Sulf1 mutant midgut, a comparable number of mitotic cells was observed in response to P. entomophila infection at 0 dpi (Fig. 7E,H), suggesting that a dramatically increased level of mitogens can sufficiently induce the same levels of ISC mitoses in the presence or absence of Sulf1 during early regeneration. At 3 and 5 dpi, however, significantly increased numbers of mitotic cells were detected in the Sulf1 mutant midgut compared to wild-type (Fig. 7F,G,H). This finding demonstrated that ISC division remains active in Sulf1 mutants at later stages of midgut regeneration.

Given that we detected Sulf1 expression in enterocytes and visceral muscles (Fig. 4), we asked whether Sulf1 secreted from these cells is responsible for limiting the number of mitotic ISCs at the later stages of regeneration. We found that knockdown of Sulf1 in enterocytes using MyoIA-GAL4 significantly increased the number of mitotic cells at 3 dpi (Fig. S2B). A similar phenotype was observed upon Sulf1 knockdown in visceral muscles achieved by using Mef2-GAL4 (Fig. S2C). These results both confirmed the result of Sulf1 mutant phenotypes, and furthermore demonstrate that Sulf1 derived from enterocytes and visceral muscles is indeed responsible for the proper termination of ISC division during regeneration.

We next asked whether Sulf1 expression changes during midgut regeneration. To this end, we monitored expression of the transcriptional reporter Sulf1{KO; mCherry} in the posterior midgut at 0 or 5 dpi, representing actively regenerating and termination stages, respectively. In the midgut of mock-infected animals, an mCherry signal was detected in enterocytes but not in progenitor cells labeled by esg>GFP (Fig. 7I,K). In the regenerating midgut, the GFP signal marks not only ISCs and enteroblasts but also differentiating or newly formed enterocytes (Fig. 7J). This allowed us to compare Sulf1 expression between differentiated enterocytes and newly differentiating cells. At 0 dpi, we detected the mCherry signal at much lower levels in the esg>GFP⁺ large polyploid cells (differentiating cells) compared with that in surrounding esg>GFP⁻ enterocytes (Fig. 7J). This result indicates that Sulf1 expression is low in differentiating or newly formed enterocytes. In contrast, the patterns of esg>GFP and mCherry in the bacteria-infected midgut became comparable to those of the mock-infected midgut at 5 dpi (Fig. 7K,L). We next quantified the mCherry signal intensity per unit area (103.6×103.6 µm²) in the posterior midgut under each condition. At 0 dpi, we found a significant reduction in the average mCherry intensity in the bacteria-infected midgut compared to the mock-infected midgut (Fig. 7M). At 5 dpi, however, no significant difference in mCherry signal intensity was detected between both conditions (Fig. 7M). These results show that the Sulf1 expression level decreases at early stages of midgut regeneration, but once regeneration is completed Sulf1 expression returns to a normal level. Taken together, we propose that Sulf1 acts as a crucial regulator of regeneration termination in the Drosophila adult midgut by rapidly shutting down several ISC mitogen pathways (Fig. 8), providing an additional post-transcriptional step to control stem cell division.

The failure of proper regeneration termination leads to abnormal organ size and a high risk of cancer. Despite its biological importance, the mechanism by which stem cell mitotic activity is downregulated at the end of regeneration is poorly understood. This is mainly due to the lack of suitable model systems to systematically monitor stem cell behavior during the termination of regeneration. Our study shows that the Drosophila adult midgut system, with its stereotypical time course of regeneration after damage, offers a powerful model to study this understudied phenomenon. Historically, the midgut has been used to study normal turnover and regeneration of epithelial cells. Previous studies have identified that several signaling pathways, including JAK-STAT, EGFR and Hh signaling, promote ISC division during regeneration. In this work, we report that the Drosophila extracellular heparan sulfate endosulfatase Sulf1 downregulates ISC mitotic activity during normal homeostasis and is required to properly shut down ISC division at regeneration termination.

We showed that fine-tuning of heparan sulfate 6-O sulfation levels adds a new regulatory step for control of ISC mitotic activity during normal turnover. In the absence of Sulf1, ISC division is increased during homeostasis. Loss of Sulf1 also led to elevated activities of JAK-STAT, EGFR and Hh signaling. These results are consistent with previously established roles of Sulf1 in regulating EGFR and Hh signaling in the developing wing (Butchar et al., 2012; Wojcinski et al., 2011). In addition, it has been shown that JAK-STAT signaling is regulated by HSPGs in the ovary (Hayashi et al., 2012). Therefore, it is reasonable to speculate that Sulf1 also regulates this pathway. It is likely that Sulf1 directly regulates these signaling pathways by modulating the fine heparan sulfate structure, which affects the affinity of heparan sulfate to these ligands, as has been shown for Wg signaling (Kleinschmit et al., 2013, 2010). However, our results do not exclude the possibility that some of the pathways are also affected indirectly by Sulf1 through crosstalk between signaling pathways in the midgut. For instance, activation of the Hh pathway is known to induce the expression of ligands for the JAK-STAT (Upd2 and Upd3) and EGFR pathways (Vn and Keren) (Tian et al., 2015). Furthermore, activation of EGFR signaling induces expression of JAK-STAT ligands (Upd, Upd2 and Upd3). Conversely, activation of JAK-STAT signaling upregulates vn expression in visceral muscles (Jiang et al., 2011). Our results show that Sulf1 is widely expressed in enterocytes and visceral muscles (Fig. 4). Sulf1 secreted from these cells is likely to modulate the activities of HSPGs in the surrounding tissues, including Trol (Drosophila Perlecan) in the basement membrane and, possibly, glypicans and syndecan.

In response to damage or stress, transcription of ISC mitogens is transiently activated to trigger regeneration. As regeneration progresses, ISC mitotic activity declines and eventually returns to a normal level. We showed that in the absence of Sulf1, ISCs failed to properly halt division at the regeneration termination stage. Given that loss of Sulf1 results in heparan sulfate oversulfation (Kleinschmit et al., 2010), our results suggest that the hyper-activation of ISC division is caused by an upregulated ligand–heparan-sulfate interaction. We propose that structural modification of heparan sulfate is a necessary step for rapid shutdown of stem cell division, in addition to transcriptional inactivation of mitogens (Fig. 8). For example, even when the transcription of mitogens is turned off at late stages of regeneration, remaining mRNAs and newly synthesized proteins might continue to activate ISC division for a time. Thus, Sulf1 could act as an additional brake on ISC division immediately after regeneration is completed, allowing for prompt and precise temporal control of ISC division.

We also analyzed the effect of overexpression of Sulf1 on ISC division at 0 dpi, when ISC division is active (Fig. S4). When we overexpressed Sulf1 using a ubiquitous tubulin-GAL4, we observed a partial but significant reduction in ISC mitotic activity in the regenerating midgut (Fig. S3A–E). By contrast, overexpression of Sulf1 using the enterocyte-specific GAL4 (MyoIA-GAL4) or muscle-specific GAL4 (Mef2-GAL4) did not affect the ISC division rate during regeneration (Fig. S3F,G). At this point, we do not know the reason for the inconsistent results from these experiments. It is possible that it is due to different levels of Sulf1 expression induced by the different drivers. Alternatively, the observed decrease in mitotic activity might be a secondary effect of systemic Sulf1 expression, and an excess level of local Sulf1 might not inhibit ISC division during regeneration. Further study is required to address this point.

Given that Sulfs are evolutionarily conserved between Drosophila and mammals, our study provides a new insight into the roles of mammalian Sulfs in stem cell control during homeostasis and regeneration. Indeed, it has been shown that Sulfs are required for maintaining spermatogonial stem cells in mice (Langsdorf et al., 2011). Our study also implies additional roles for Sulfs in cancer formation. Excess activation of stem cell division caused by loss of Sulf1 could increase the risk of tumorigenesis. In fact, mammalian Sulfs are dysregulated in ovarian, breast, lung, pancreatic and hepatocellular cancer cells (Khurana et al., 2013; Lai et al., 2003, 2008; Lemjabbar-Alaoui et al., 2010; Nawroth et al., 2007). Thus, it is important to understand how expression of mammalian and Drosophila Sulfs is regulated. In the Drosophila wing disc, EGFR and Wg pathways induce expression of Sulf1, which in turn negatively regulates these pathways, indicating that Sulf1 forms negative-feedback loops for EGFR and Wg signaling (Butchar et al., 2012; Kleinschmit et al., 2010). Further study is required to reveal how Sulf1 transcription is precisely regulated during midgut homeostasis and regeneration, and how Sulf1-mediated feedback systems contribute to stabilizing stem cell systems and/or preventing cancer formation.

w¹¹¹⁸ was used as a wild-type stock. The following stocks were obtained from the Bloomington Drosophila Stock Center (BDSC, Indiana University, Bloomington, IN), the KYOTO Stock Center (DGRC) in the Kyoto Institute of Technology (Ukyo-ku, Kyoto, Japan), the National Institute of Genetics Fly Stock Center (Mishima, Shizuoka, Japan) or the VDRC (Vienna, Austria): UAS-GFP (BDSC #1521 and BDSC #1522), UAS-tdTomato (BDSC #36327), tub-GAL80^ts (BDSC #7018), vasa-Cas9 (BDSC #55821), esg-GAL4 (DGRC#113886) (Hayashi et al., 2002), tub^ts (a gift from Michael O'Connor, University of Minnesota, MN), MyoIA-GAL4^NP0001 (DGRC #112001) (Hayashi et al., 2002), Mef2-GAL4 (BDSC #27390), Trol::GFP (DGRC #110807) (Morin et al., 2001), Hs6st::EGFP (BDSC #60217) (Nagarkar-Jaiswal et al., 2015), Dl::EGFP (BDSC #59819) (Nagarkar-Jaiswal et al., 2015), esg-lacZ (DGRC #108851) (Wilk et al., 2000), MyoIA-lacZ (a gift from Huaqi Jiang, UT Southwestern Medical Center, TX; Jiang et al., 2009), MyoIA-GAL4 (DGRC #112001) (Jiang et al., 2009), upd3-GAL4 (a gift from Huagi Jiang; Agaisse et al., 2003), 10×STAT92E-DGFP (BDSC #26199) (Bach et al., 2007), vn-lacZ (BDSC #11749), Sulf1^P1 (Kleinschmit et al., 2010), UAS-Sulf1 #13 (Kamimura et al., 2006), UAS-Egfr^DN 29-77-1 (BDSC #5364), UAS-ptc (BDSC #44614), UAS-dicer2 (VDRC #60008), UAS-trol RNAi (BDSC #29440), UAS-trol RNAi (VDRC #110494), UAS-sfl RNAi (BDSC #34601), UAS-sfl RNAi (VDRC #5070), UAS-Hs6st RNAi (VDRC#110424), UAS-Hs6st RNAi F6M1 (Kamimura et al., 2006), UAS-STAT92E RNAi (BDSC #26899) (Kim et al., 2007) and UAS-Sulf1 RNAi (VDRC #45954). The effectiveness of RNAi knockdown using some of these lines was examined in Fig. S4. The details of UAS-RNAi transgenic flies from BDSC and VDRC are described elsewhere (Dietzl et al., 2007; Perkins et al., 2015). Sulf1{KO; mCherry} was generated in this study, as described below. Detailed genotypes used in individual experiments are listed in Table S2.

Fly stocks were reared on a standard cornmeal fly medium at 25°C except for those containing tub-GAL80^ts, which prevents a GAL4-mediated UAS transgene expression at 19°C but activates it at 30°C (McGuire et al., 2003). For experiments using esg^ts and UAS-RNAi transgenes, female flies were raised at 19°C and aged for 3–4 days after eclosion. The flies were then raised at 30°C for 3 days to activate the UAS-RNAi transgene expression. For experiments using tub^ts and UAS-Sulf1, female flies were raised at 19°C and aged for 5–6 days after eclosion, and then raised at 30°C for 1 day to induce UAS-Sulf1 expression. The 6–7-day-old females were starved for 2 h, and placed in empty vials containing a piece of Whatman filter paper (1001-110, GE Healthcare Life Sciences, Pittsburgh, PA) soaked in 5% sucrose solution (mock infection) or soaked in P. entomophila culture (a gift from Huaqi Jiang) for 22 h. Flies were transferred to new food every 2 days.

Sulf1{KO; mCherry} was generated by CRISPR/Cas9-mediated homology-directed repair. The synthesized sequences were annealed and ligated into the BbsI-digested pU6-BbsI-chiRNA plasmid (a gift from Melissa Harrison, Kate O'Connor-Giles and Jill Wildonger, University of Wisconsin–Madison, Madison, WI; Addgene #45946, Cambridge, MA) (Gratz et al., 2013). To generate the repair template, an mCherry-coding sequence and the 1.2-kb homologous sequences on either side of the predicted DSBs were cloned into the pHD-DsRed-attP backbone (a gift from Melissa Harrison, Kate O'Connor-Giles and Jill Wildonger; Addgene #51019) (Gratz et al., 2014) using Gibson Assembly Master Mix (E2611S, New England Biolabs, Ipswich, MA). The mCherry-coding sequence was amplified by PCR from pUAST-mCherry (a gift from Thomas Neufeld, University of Minnesota, MN) (Chang and Neufeld, 2009). The 1.2-kb Sulf1 homology arms were amplified from genomic DNA of the vasa-Cas9 strain. A mixture of pU6-sgRNAs and the repair template was injected into the vasa-Cas9 embryos by BestGene Inc. (Chino Hills, CA). The homologous recombinants were screened by PCR and verified by Sanger sequencing.

Female midgut samples were dissected in phosphate-buffered saline (PBS, pH 7.4) and subsequently fixed in 3.7% formaldehyde in PBS (BP531500, Fisher Scientific) for 1 h. For pERK staining, midgut samples were fixed in 7.4% formaldehyde instead of 3.7%. After three 10-min washes with PBST (PBS with 0.1% Triton X-100), the samples were incubated with primary antibodies overnight at 4°C. After three PBST washes, the samples were incubated with Alexa-Fluor-conjugated secondary antibodies (1:500, Thermo Fisher Scientific, Waltham, MA) for 2 h at room temperature or overnight at 4°C, washed in PBST three times, and subsequently mounted in VECTASHIELD Antifade Mounting Medium (H-1000, Vector Laboratories, Burlingame, CA) or VECTASHIELD Antifade Mounting Medium with DAPI (H-1200, Vector Laboratories). Nuclei were stained with TO-PRO-3 (T3605, Thermo Fisher Scientific), YO-PRO-3 (Y3607, Thermo Fisher Scientific) or DAPI. The following primary antibodies were used: rabbit anti-pH3 (1:1000, 06-570, Millipore, Darmstadt, Germany), rabbit anti-mCherry (1:1000, NBP2-25157, Novus Biologicals, Littleton, CO), anti-β-galactosidase (1:200, Z3781, Promega, Madison, WI), mouse anti-Pros [1:50, MR1A, Developmental Studies Hybridoma Bank (DSHB), Iowa City, IA], rabbit anti-GFP (1:1000, A-11122, Thermo Fisher Scientific), rabbit anti-pERK (1:100, 4370, Cell Signaling, Danvers, MA), rabbit anti-Hh (1:1000, a gift from Tetsuya Tabata, University of Tokyo, Japan), and rabbit anti-cleaved caspase-3 (1:1000, 9661, Cell Signaling). F-actin was stained with Alexa-Fluor-633-conjugated phalloidin (1:500, A22284, Thermo Fisher Scientific). Images were acquired on a LSM710 (Carl Zeiss, Oberkochen, Germany) and an ECLIPSE C1 (Nikon, Minato-ku, Tokyo, Japan) confocal microscope and processed with Fiji (Schindelin et al., 2012) and Adobe Photoshop CS4 (Adobe Systems, San Jose, CA) software. The number of pH3⁺ cells in the posterior midgut, the number of Dl⁺ cells and nuclei per unit area (124.32×124.32 µm²), and average mCherry intensities per unit area (103.6×103.6 µm²) were measured using Fiji software.

Total RNA was extracted from 20 midgut samples using TRIzol (15596026, Thermo Fisher Scientific), treated with RNase-Free DNase I (79254, QIAGEN, Hilden, Germany) and purified using an RNeasy MinElute Cleanup Kit (74204, QIAGEN). cDNA was synthesized from 500 ng of total RNA using oligo dT primers and SuperScript III reverse transcriptase (18080-051, Thermo Fisher Scientific). cDNA samples were diluted 20-fold for RT-qPCR. qPCR assays were run in duplicate on each of three independent biological replicates in a LightCycler 480 Instrument II (Roche, Basel, Switzerland) using Bullseye EvaGreen qPCR 2× Mastermix (BEQPCR-S, MIDSCI, Valley Park, MO). Act5C expression was used for normalization. Fold changes were calculated using the ΔΔCt method. Primers used in this study were mostly chosen from FlyPrimerBank (Hu et al., 2013) and are listed in Table S1.

Statistical analyses were carried out using R (http://cran.r-project.org). Statistical significance was assessed using a Wilcoxon rank-sum test.

We thank Melissa Harrison, Kate O'Connor-Giles, Jill Wildonger, Michael O'Connor, Huaqi Jiang, Tetsuya Tabata, Thomas Neufeld, Takashi Adachi-Yamada, BDSC (NIH P40OD018537), DGRC, the TRiP at Harvard Medical School (NIH R01 GM084947), and VDRC for reagents. We are grateful to Katsufumi Dejima and Emiko Tomiyasu for technical support. We also thank Daniel Levings and Andy Toth for helpful comments on the manuscript.