Damage sensing by a Nox-Ask1-MKK3-p38 signaling pathway mediates regeneration in the adult Drosophila midgut

Epithelia are exposed to diverse types of stress and damage from pathogens and the environment, and respond by regenerating. Yet, the proximal mechanisms that sense epithelial damage remain poorly understood. Here we report that p38 signaling is activated in adult Drosophila midgut enterocytes in response to diverse stresses including pathogenic bacterial infection and chemical and mechanical insult. Two upstream kinases, Ask1 and Licorne (MKK3), are required for p38 activation following infection, oxidative stress, detergent exposure and wounding. Ask1-p38 signaling in enterocytes is required upon infection to promote full intestinal stem cell (ISC) activation and regeneration, partly through Upd3/Jak-Stat signaling. Furthermore, reactive oxygen species (ROS) produced by the NADPH oxidase Nox in enterocytes, are required for p38 activation in enterocytes following infection or wounding, and for ISC activation upon infection or detergent exposure. We propose that Nox-ROS-Ask1-MKK3-p38 signaling in enterocytes integrates multiple different stresses to induce regeneration.

Drosophila study found that p38a and p38b mutant flies were as sensitive to P.e. infection as control flies, and that p38a was not required for Duox expression upon infection 23 . Instead, they reported that p38c is upregulated after pathogenic bacterial infection to not only promote Duox-mediated ROS production via Atf2 but also to restrict ISC proliferation by limiting cytokine (Upd3) expression 23 . Although it is not clear which fly p38 (p38a or p38c) is involved, these data suggest that p38 signaling promotes fly intestinal immunity through Duox-mediated ROS production. However, whether p38 signaling plays an important role in Drosophila intestinal regeneration is not known.
Work in the mouse intestinal epithelium suggests a role for p38 signaling in intestinal regeneration; p38α (one of four p38s in mice and homologous to fly p38a and p38b) was required in the mouse intestinal epithelium for repair after detergent (DSS)induced damage 24,25 . Consistent with this, p38b signaling in fly ECs was also shown to partially regulate upd3 expression in Ecc15-infected flies 19 ; however, its requirement for ISC response and midgut regeneration was not directly tested. Although these studies suggest a role for p38 in intestinal regeneration, how p38 is activated upon intestinal damage, which cell types utilize it, and how p38 promotes intestinal regeneration are not well understood.
ROS are also thought to promote intestinal turnover and regeneration. The NADPH oxidase, Nox, has been reported to produce ROS (O 2 • ) upon exposure to commensal bacteria such as Lactobacillus plantarum, which also consequently promotes fly and mouse ISC proliferation 26 . Furthermore, paraquat or H 2 O 2 exposure promote fly intestinal epithelial turnover 10 . In contrast, antioxidants (i.e., N-acetyl cysteine and glutathione) feeding or Duox depletion blocked intestinal regeneration in Ecc15-infected flies 10 . Moreover, Nox1 knockout mice showed reduced ISC proliferation after DSS exposure 27 . Although ROS can stimulate JNK activity 14 , how ROS promote intestinal regeneration is not well understood. In addition, although it has been reported that intestinal ROS production is boosted by p38 signaling, it is not known whether ROS can activate p38 signaling to effect intestinal regeneration.
SAPKs (JNK, p38) are activated by the upstream SAP kinase kinases (SAP2Ks) such as MKK4/7 and MKK3/6, which activate JNK and p38, respectively 13 . SAP2Ks such as MKK3 are dualspecificity kinases that can phosphorylate the highly conserved dual-phosphorylation site (TGY) of p38s to further promote their activity 13,28 . The sole fly MKK3, Licorne, can phosphorylate both p38a and p38b but not p38c, which lacks the dualphosphorylation site 29 . SAP2Ks are activated by SAP kinase kinase kinases (SAP3Ks) (i.e., MAPK/ERK kinase kinase 1 (Mekk1) 30 , Apoptosis signal-regulating kinase 1 (Ask1), transforming growth factor-β (TGFβ)-activated kinase 1 (Tak1)), which are activated by a variety of stresses, such as heat, oxidative, osmotic, or pathogen. In flies, both Mekk1 and Tak1 signaling have been reported to play important roles in sensing pathogenic bacteria and promoting gut immunity 21,23,31 . In C. elegans, Ask1/ nsy-1 is required for pathogen resistance 20 . Ask1 is also thought to sense oxidative stress 32 and in flies it promotes wing disc regeneration via JNK and p38 33 . A role for SAP2Ks and SAP3Ks in intestinal regeneration, however, is unknown. As ROS promote tissue regeneration in other contexts [34][35][36] , SAP3Ks such as Ask1 may have an important role in coupling intestinal immunity to regeneration.
Here, we show that p38 signaling promotes intestinal regeneration by sensing a variety of stresses in ECs and promoting ISC proliferation. In addition, we find that Nox-derived ROS and Ask1 are required to fully activate MKK3-p38 signaling to promote intestinal regeneration upon pathogenic infection. Furthermore, we find that Nox-derived ROS are required to promote intestinal regeneration upon detergent exposure. Together, our data suggest that Nox-Ask1-MKK3-p38 signaling senses multiple different stresses and effectively couples these stresses with regeneration.
Results p38 in ECs promotes ISC-mediated regeneration upon damage. Previously, p38 was described to be mostly inactive in the adult fly midgut 21 . Using an anti-phospho-p38 antibody that should recognize both activated p38a and p38b, we found that in both male and female midguts p38 was activated often in the anterior midgut (R1, anterior R2) and occasionally in (and around) the mid-midgut (R3) 37 . In the posterior midgut (R4a-R5), p38 activity was typically absent and we only occasionally observed low levels of activated p38. p38a was shown to be activated in midgut ECs after feeding of soluble microbial extract or peptidoglycan from Ecc15 21 . Consistent with this result, we found that p38 was strongly activated in ECs throughout the midgut after oral infection with another Gram-negative bacterium P.e. relative to uninfected control midguts (Fig. 1a-c). Single depletion of either p38a or p38b in ECs with UAS-RNAi transgenes under the control of the EC-specific driver Myo1A-GAL4 (Myo1A ts ) reduced neither the basal active p38 levels nor the p38 activation after P.e. infection relative to controls (Supplementary Fig. 2a-d). Although depletion of both p38a and p38b simultaneously (p38a + b) using an RNAi that targets both mRNAs with Myo1A ts did not detectably reduce the basal levels of p38 activation, it did however suppress the p38 activation upon P.e. infection relative to control-infected midguts ( Supplementary  Fig. 1a, b). This suggested that the phospho-p38-specific antibody recognizes both the phosphorylatable forms of fly p38 (p38a and p38b) but not p38c, which lacks the phosphorylation site 29 Fig. 1c, d). These data suggest that both old and new ECs can sense stress during pathogenic bacterial infection and activate p38a + b signaling.
In addition to ECs, we observed strong p38 activation in the VM after P.e. infection relative to the VM of control, uninfected midguts ( Supplementary Fig. 1e-f). To determine whether p38 activity in ECs and the VM was important for the ISC response to infection, we depleted both p38a + b in the VM with how-GAL4, tubGAL80 ts (how ts ) or in ECs with Myo1A ts . Depletion of p38a + b in VM neither affected basal ISC divisions relative to uninfected control midguts nor the ISC response to P.e. infection relative to infected control midguts ( Supplementary Fig. 1g). Although depletion of p38a, p38b, or both p38a + b or overexpression of p38 DN in ECs did not affect basal ISC divisions relative to uninfected control midguts, blocking both p38a + b in ECs, but neither p38a or p38b alone, did strongly suppress (by 63.2% with RNAi (1); 57.9%, RNAi (2)) the ISC response to P.e. infection relative to infected controls ( Fig. 1d; Supplementary Figs. 1h, j and 2e-g). Interestingly, we found that the ISC response to P.e. infection was mildly suppressed in the midguts depleted for p38b in ECs (by 24.1%) or mutant for p38b (p38b ex9 ) (22.5% in p38b ex9 /+; 37.2%, p38b ex9 /p38b ex9 ) 39 ( Supplementary Fig. 2g, i). In contrast, the ISC response in p38a and p38c double-mutant (p38a 1 ) midguts 23,29 was not blocked upon P.e. infection ( Supplementary Fig. 2h). Together, these data suggest that upon P.e. infection, p38b is required in ECs for the ISC response, and that p38a may be required to sense stress in the absence of p38b or together with p38b. These data suggest that p38a + b signaling is required in ECs to promote ISC-mediated regeneration.
The SAP2K, MKK3, is known to activate p38 in mammalian cells 13,28 ; however, its ability to activate p38 in intestinal epithelial cells is not known. Moreover, stress responses known to be p38dependent (i.e., osmotic stress in the larval Drosophila hindgut) have been previously shown to not require MKK3 40 . We found that overexpression of the Drosophila MKK3, licorne (lic), with the Myo1A ts driver strongly activated p38 in ECs (Fig. 1e, f; Supplementary Fig. 3a-c). Furthermore, Lic overexpression in ECs for 2 days also increased ISC proliferation relative to basal ISC divisions in control midguts (Fig. 1g). Conversely, we found that while depletion of the fly MKK3, licorne, in ECs with Myo1A ts did not noticeably affect basal active p38 levels relative to control uninfected midguts, it inhibited p38 activation upon P.e. infection ( Fig. 1h-j, red; Supplementary Fig. 3d-i). These data suggest that p38 is activated upon infection via MKK3/Lic resulting in ISC proliferation.
The boost in ISC proliferation after P.e. infection is partly due to increased mitogen production (i.e., cytokines: Upd2, Upd3) by ECs 8 . Indeed, we found that mRNA levels encoding Upd3 and the JAK-STAT target, Socs36E, were increased after a variety of stresses, including P.e., Ecc15, H 2 O 2 , Sodium dodecyl sulfate (SDS), bleomycin, and heat shock (Fig. 2a). To determine the role of p38 in this stress-dependent cytokine induction, we tested whether upd3 mRNA was induced after P.e. infection in p38depleted midguts. Using quantitative PCR (qPCR), we found that EC-targeted RNAis against p38b and p38a + b suppressed the induction of upd3 mRNA following P.e. infection by 62 and 49%, respectively (Fig. 2b). In support of a role for p38a + b signaling in regulating cytokine production, Licorne overexpression for 24 h was sufficient to induce upd2 and upd3 expression (≈8-fold) (Fig. 2c). These data suggest that MKK3-p38 signaling promotes upd3 induction upon P.e. infection, largely through p38b. Cytokine-induced JAK-STAT activity in the midgut can be detected using the in vivo reporter, 10x STAT-GFP, which we found to be high in progenitors and a few ECs in uninfected controls and in uninfected controls depleted for p38a + b ( Fig. 2d; Supplementary Fig. 1i). However, P.e.-infected controls showed strongly increased 10x STAT-GFP in most ECs, whereas this was markedly decreased in p38a + b-depleted midguts ( Fig. 2e-g). These data suggest that p38 signaling in ECs promotes ISCmediated regeneration by boosting cytokine Upd3 expression.
We also tested if known p38 phosphorylation targets were required in ECs to promote ISC response to infection by assessing whether ISC proliferation was affected in P.e.-infected flies mutant for the transcriptional regulator Atf2 (Atf2 PB ) 41 or in flies depleted for Atf2 in ECs, or mutant for MAPK-activated protein kinase 2 (MAPKAP-K2 or MK2) 40 . We found that basal ISC division rates were not affected in Atf2 mutant midguts, or in midguts depleted of Atf2 in ECs ( Supplementary Fig. 4a, b, d). Furthermore, we did not find an effect on the ISC response to P.e. infection in Atf2 mutant midguts, or in midguts depleted for Atf2 in ECs ( Supplementary Fig. 4a, b). Similarly, the ISC response to P.e. infection was not affected in the MK2 P-element excision deletion mutant (MK2 Δ43 ) when compared with its precise excision control, MK2 Δ1A ( Supplementary Fig. 4c). These data suggest that p38 activity in stressed ECs does not promote ISC proliferation via MK2 or ATF2.
Chemical and mechanical stress activate p38 in the midgut. To test whether stresses in addition to pathogenic bacteria can activate p38, we tested whether several other types of stress could activate p38 in midgut ECs. Exposing flies or fly cultured S2 cells to a diverse array of stresses (i.e., Gram-negative pathogenic bacteria (Ecc15, P.e.), septic injury (Salmonella), heat shock, dry starvation, ROS (i.e., hydrogen peroxide (H 2 O 2 ), high salt (0.2 M NaCl), metal ions) has been shown to activate p38 21,29,38 ; however, only enteric infection had been shown to activate p38 in the adult fly midgut 21 . We found that p38 signaling was not activated in midguts after heat shock at 37°C for 30 min (Supplementary Fig. 5a, b). However, in vivo reporters for JNK activity (puc-lacZ) and cytokine upd3 expression (upd3-lacZ) were activated after heat stress (Supplementary Fig. 5c-f). We also found that feeding 0.2 M NaCl or 0.4 M NaCl for 1 week ( Supplementary Fig. 5g, h) or starvation (water only) did not activate p38 in midguts. These data suggest that heat, high salt stress, or starvation stress were not able to activate p38 signaling in the fly midgut.
We next tested mechanical stress. Strong p38 activation was observed in the VM and ECs shortly after wounding the posterior midgut with a needle (Fig. 3a, b; Supplementary Fig. 9a, b). Interestingly, p38 was activated not only near the wound site but also in the VM (and in many ECs) throughout the midgut, suggesting propagation of stress signaling ( Fig. 3b; Supplementary  Fig. 9b). Previously, we reported that EC detachment by integrin subunit depletion or by rapidly growing, differentiation-defective ISC tumors activates JNK and YAP/Yki activity 9 . Similarly, we found that ISC tumor growth 9 also activated p38 in surrounding ECs relative to control midguts bearing no tumors (Fig. 3c, d). Moreover, detaching ECs by depleting the integrin subunit βPS1 (myospheroid) or both αPS3 (scab) + αPS4 activated p38 signaling (Fig. 3e, f). Lastly, feeding flies with 1% H 2 O 2 or 0.2% sodium dodecyl sulfate (SDS) for 2 days activated p38 in the midgut relative to midguts from flies fed water alone ( Fig. 4a, b, d, e). Together these data suggest that p38 signaling responds to several different stresses, including pathogenic infection, chemical stress, and mechanical stress.
Ask1 activates p38 upon stress and promotes regeneration. Both JNK and p38 are known to be activated upon stress via upstream SAP3Ks (i.e., Mekk1, Ask1, and Tak1) and SAP2Ks (i.e., MKK4/7 and MKK3/6) 13 . Previously it was reported that p38 was activated in fly ECs via Mekk1 and MKK3 after feeding flies soluble microbial extract 21 . Surprisingly, neither p38 activation nor the ISC response to infection was inhibited upon P.e. → ±18 h P.e.    Fig. 8a-d, f, h). However, we found that these treatments all strongly suppressed p38 activation during P.e. infection (Fig. 5b, c, e, f; Supplementary Fig. 8e, g, i). Ask1 was also required for p38 activation in ECs after 1% H 2 O 2 or 0.2% SDS exposure relative to stressed control midguts (Fig. 4b, c, e, f). Next, we determined whether Ask1 was also required in ECs for the ISC response to damage. While Ask1 depletion alone did not detectably affect basal ISC divisions, Ask1 depletion in ECs impaired the ISC proliferation response to P.e. infection (Fig. 5g). Furthermore, we found that Ask1 depletion in ECs suppressed p38 activation in ECs, but not the VM, upon wounding ( Fig. 5h-k; Supplementary Fig. 9c-j). These data indicate that Ask1 is required for p38 activation upon chemical stress, mechanical stress, and pathogenic bacterial infection. Furthermore, they suggest that Ask1-p38 signaling but not Mekk1-or Tak1-p38 signaling is required in ECs for the ISC response to damage.
Nox-derived ROS activate p38 and promote midgut regeneration. Since ingested ROS could activate Ask1-p38 in midgut ECs, we next tested whether ROS were produced in response to different stresses. Using qPCR, we found that the Nrf2 target, GstD1, was induced upon a variety of stresses, suggesting an increase in ROS levels in stressed midguts relative to unstressed midguts (Fig. 6f). Since gut-produced ROS contributes to innate immunity and ISC proliferation in the fly midgut 10,26 , we next tested whether ROS were required for p38 activation after pathogenic infection. To decrease ROS in ECs, we overexpressed Nrf2 (Cnc), a transcription factor that promotes antioxidant factor expression 43 . As an alternative, we co-overexpressed the antioxidant enzymes superoxide dismutase 1 (Sod1), which converts superoxide (O 2 •) to H 2 O 2 , and Catalase, which converts H 2 O 2 to H 2 O and O 2 . While overexpression of these factors alone moderately increased active p38 levels and ISC proliferation without infection (Fig. 6a, d, e; Supplementary Fig. 10a, b, d), overexpression of these factors inhibited p38 activation during P.e. infection (Fig. 6b, c; Supplementary Fig. 10c) suppressed relative to control by Nrf2 overexpression (Fig. 6d, e), suggesting that ROS are required in ECs to promote ISCmediated regeneration in response to a variety of damage and not only to pathogens. To identify the molecular source of ROS production, we depleted Nox or Duox, which produce O 2 • and H 2 O 2 , respectively, and contribute to mucosal immunity through ROS generation 22,26,44 . Depletion of Nox or Duox in ECs of unchallenged midguts neither detectably affected basal active p38 levels nor the basal rate of ISC division (Fig. 7a, d, e; Supplementary Fig. 11a, b, d, f). However, Nox depletion was able to inhibit p38 activation during P.e. infection relative to infected controls (Fig. 7b, c). Duox depletion in ECs did not detectably affect p38 activation upon P.e. infection ( Supplementary Fig. 11c, e). These data suggest that p38 is activated by Nox-derived ROS produced during pathogenic infection. We then determined if ROS generated during pathogenic infection are required in ECs for ISC proliferation. We found that depletion of either Nox or Duox in ECs partially suppressed the normal proliferative response in ISCs relative to infected control midguts ( Fig. 7d; Supplementary Fig. 11f). These data suggest that in addition to promoting pathogen clearance, ROS produced by both Nox and Duox in ECs promote intestinal regeneration. In addition, we found that Nox was required in ECs for optimal ISC responses following detergent damage and p38 activation upon wounding (Fig. 7e-h; Supplementary Figs. 11g, 12). Nox-derived ROS appears to promote regeneration in response to a variety of stresses. Interestingly, we found that Duox was not required in ECs for intestinal regeneration following detergent damage ( Supplementary Fig. 11g), suggesting that Duox activity may be specific for pathogen clearance. Together these data suggest that Nox can sense diverse types of damage and stress in ECs, and that Nox-derived ROS promote intestinal regeneration by activating Ask1-MKK3-p38 signaling.

Discussion
A role for p38 has been reported in intestinal immunity 1 and in mouse intestinal regeneration upon detergent-induced damage 24,25 . However, how p38 senses stress and promotes intestinal regeneration is not well understood. We found that p38 signaling is activated in ECs during pathogenic bacterial (P.e.) infection, oxidative stress (H 2 O 2 ), detergent exposure (SDS), wounding, tumor growth, and EC detachment from the BM. In addition, we found that the SAP2K MKK3/Licorne and the SAP3K, Ask1, were required to activate p38 in ECs during P.e. infection. Moreover, Ask1 was required for p38 activation in ECs after oxidative stress (H 2 O 2 ), detergent (SDS) exposure, and wounding, suggesting that Ask1 may act as a node that integrates multiple types of stress inputs for p38 activation. Using Nrf2 and antioxidant (Sod1 and Catalase) overexpression, we also found that ROS are required for p38 activation during P.e. infection. Furthermore, we determined that ROS contributed by the NADPH oxidase, Nox, were required for full p38 activation upon P.e. infection and wounding. Surprisingly, Duox, previously also suggested to produce ROS to promote intestinal immunity and repair 10,22 , was not detectably required for p38 activation. Furthermore, Nox, Ask1, and p38 were all required in ECs to stimulate ISC proliferation and regeneration following P.e. infection. This stem cell response appeared to be mediated at least in part by Upd3/Jak-Stat signaling. Lastly, Nox-derived ROS was also required for ISCmediated regeneration upon detergent exposure and p38 activation upon wounding, suggesting that Nox, like Ask1, may also sense multiple different types of epithelial stresses to promote regeneration. Our work suggests that in addition to JNK and Hpo signaling, Nox-Ask1-MKK3-p38 signaling promotes intestinal regeneration in response to multiple different types of stress, thereby enhancing tissue and organismal  (Fig. 8). Moreover, since NADPH oxidases (Nox, Duox) are expressed in a variety of cell types, including macrophages 45 , and since Ask1-p38 signaling can potentially be activated in any cell type, our work suggests that Nox-Ask1-MKK3-p38 signaling may promote regeneration in a variety of contexts. Previous work in the fly midgut reported that Mekk1-p38a-Atf2 and Mekk1-p38c-Atf2 activity promote intestinal immunity by inducing Duox-mediated ROS production 22,23 , and that ROS are required for intestinal regeneration 10 . However, we found that Mekk1, p38a, p38c, and Atf2 are not required for intestinal regeneration. Instead, we did find that both p38a + b are required for intestinal regeneration, partly by regulating upd3 expression. Interestingly, Duox was also required upon P.e. infection for robust intestinal regeneration, suggesting that additional mechanisms may increase Duox activity and that its increased    (1) in ECs for 5 days with Myo1A ts > Dcr-2 (n = 19 midguts), P.e.infected midguts expressing Nox RNAi (1) in ECs for 5 days (n = 11 midguts), midguts expressing Duox RNAi (1) in ECs for 5 days with Myo1A ts > Dcr-2 (n = 11 midguts) and P.e.-infected midguts expressing Duox RNAi (1) in ECs for 5 days with Myo1A ts > Dcr-2 (n = 11 midguts). e Nox is required in ECs for ISC-mediated regeneration upon detergent exposure. The ISC response is blocked in SDS-fed (20 h) midguts expressing Nox RNAi (2) for 5 days with Myo1A ts > Dcr-2 compared with SDS-fed control midguts (unpaired t test; p < 0.0001). The mean number of phosphorylated histone H3 Ser 10positive cells per midgut with s.e.m. in control (2) midguts (n = 17 midguts from one representative experiment) (Myo1A ts > Dcr-2 alone), SDS-fed control (2) midguts (n = 29 midguts from two experiments), midguts expressing Nox RNAi (2) in ECs for 5 days with Myo1A ts > Dcr-2 (n = 6 midguts from one representative experiment), and in SDS-fed midguts expressing Nox RNAi (2) in ECs for 5 days (n = 32 midguts from two experiments). f-h Nox-derived ROS are required for p38 activation in ECs upon wounding. p38 is activated in ECs of wounded control midguts (g red) compared with unwounded control midguts (f red). p38 activation is blocked in ECs of wounded midguts expressing Nox RNAi (2) (h red) relative to ECs in wounded control midguts. DNA is in blue. Scale bars in a-c, f-h are 50 μm. Representative images in a-c from three experiments; f-h, two experiments. Source data are provided as a Source Data File from the BM. However, unlike p38, JNK can sense heat stress, and heat stress can induce upd3 expression. While JNK activation in ECs had previously been shown to be sufficient to promote cytokine upd2-3 expression and ISC proliferation 8 , whether JNK is required in ECs for intestinal regeneration had, surprisingly, not been directly demonstrated. We found that JNK signaling is indeed required in ECs for normal ISC-mediated regeneration upon P.e. infection ( Supplementary Fig. 13a). In addition to JNK, Hpo signaling is inhibited in response to a variety of stresses and required for intestinal regeneration 7,11,12 . Since depleting Nox, Ask1, or p38 in ECs only partially suppressed regeneration, it seems likely that Nox-Ask1-p38 signaling works together with JNK and/or Hpo/Yki signaling to promote mitogen production and ISC activation. Alternatively, p38 might act via Yki, activating it in ECs by antagonizing Hpo/Wts signaling 46 . To test whether p38 and JNK work together, we depleted p38a + b with RNAi and overexpressed the JNK inhibitor, puckered in ECs. We found that blocking p38a + b and JNK with these treatments did not inhibit the ISC response to P.e. infection relative to infected controls. Furthermore, we found increased basal ISC divisions without infection in midguts with ECs with reduced p38a + b and JNK activity, confounding our interpretation of this experiment. The details of how these stress-sensing pathways (Jnk, p38, and Hpo) cross-talk and integrate stress responses in the midgut remain an interesting area for future studies. We found that many different types of stress can activate p38 signaling in adult fly midgut ECs, i.e., Gram-negative pathogenic bacteria (P.e.), ROS (H 2 O 2 ), detergent (SDS), wounding, ISC tumor growth and EC detachment. Furthermore, we found that Gram-negative pathogenic bacteria, ROS, detergents, and wounding all require Ask1 to activate p38 in ECs. These data suggest that a variety of stresses can be sensed in ECs by Ask1. How Ask1 senses these stresses is not well understood. Ask1 could possibly sense each stress by distinct mechanisms. On the other hand, each stress might also trigger ROS production, possibly via Nox, and thereby activate Ask1. Consistent with this mechanism, we found that a ROS-responsive Nrf2 target, GstD1, is induced after infection, detergent, oxidative, DNA damage, and heat shock stress, suggesting that ROS are produced by all these stresses. Furthermore, Nox-derived ROS were required in ECs for p38 activation upon infection and wounding, and for the ISC response to infection and detergent exposure. Thus our data are consistent with the possibility that Ask1 senses damage through Nox-derived ROS to effect intestinal regeneration. Ask1 proteins have been proposed to oligomerize using their C-terminal coiled-coiled domains 47 . Ask1 activity is thought to be regulated by ROS at its N-terminal coil-coil domain, which binds to the redox protein thioredoxin 32 . Thioredoxin binding to the Nterminal coil-coil domain of Ask1 is thought to prevent autophosphorylation and activation of oligomerized Ask1 47 . Upon oxidative stress, ROS oxidize thioredoxin, causing thioredoxin to dissociate from Ask1 and resulting in the activation of oligomerized Ask1 proteins 47 . For this reason, an N-terminally deleted Ask1 (ΔN Ask1) is thought to be constitutively active 32 . An alternate model proposes that Ask1 oxidation results in oligomerization via intermolecular cysteine disulfide bond formation and activation through phosphorylation 48 . In addition, oligomerization has been proposed to be dependent on multiple cysteines throughout Ask1, and interestingly, ΔN Ask1 was able to oligomerize and activate JNK 48 . Furthermore, disulfide bond formation between thioredoxin and Ask1 is thought to result in rapid reduction of Ask1 oligomers to an active monomeric form 48 . However, we found that overexpression of ΔN Ask1 49 in ECs did not act as a constitutively active form of Ask1 and was unable to strongly activate p38, raising questions about these proposed mechanisms. Surprisingly, we were able to suppress p38 activation during P.e. infection by expressing the ΔN Ask1. While we show that Ask1 plays an important role in stress-triggered intestinal regeneration, precisely how these stresses activate Ask1 remains to be determined.
ROS produced by the NADPH oxidase, Duox, in ECs have been reported to promote not only intestinal immunity toward pathogenic bacteria but also intestinal regeneration 10,22 . Similarly, another NADPH oxidase, Nox, also in ECs, has been reported to produce ROS upon exposure to specific commensal bacteria, which subsequently also stimulates ISC proliferation 26 . Nevertheless, whether Nox-derived ROS provides immunity toward either commensal or pathogenic bacteria or whether it normally promotes intestinal turnover remained to be determined. We found that Nox was required in ECs even more than Duox to activate p38 and promote ISC proliferation in response to pathogenic bacteria. This is consistent with our finding that Nox, but not Duox, mRNA is strongly upregulated (18 × ) in ECs after P.e. infection 50 . Our data suggest that Nox plays an important role, possibly as important as Duox, in intestinal immunity and regeneration during pathogenic bacterial infection. It is thought that ROS can directly activate ISC proliferation upon stress 51 . However, our work suggests that a significant contributor to ISCmediated regeneration is nonautonomous activation of JAK-STAT signaling in ISCs by Nox-ROS-p38 signaling in ECs 52 . Furthermore, our data suggest that unlike Duox, Nox plays an important role in intestinal regeneration following diverse types of damage. An important question remaining is how ROS production by Nox is activated by each stress. Nox could be stimulated by a single cue, such as actin 53 or by different cues activated by each stress.
Inflammatory bowel diseases (e.g., ulcerative colitis, Crohn's disease) are characterized by chronic gastrointestinal inflammation due to damage, and have been linked to an increased risk for colorectal cancer. The cause of IBDs and how they develop is unfortunately not well understood. Work in the mouse intestine indicates that p38α signaling is required for intestinal repair after detergent (DSS)-induced damage 24,25 . Furthermore, loss of p38α in the intestinal epithelium enhances carcinogen/DSS damageinduced intestinal tumor initiation 24,54 . Together with our work, these data indicate the importance of the damage sensing and pro-regenerative roles of p38 signaling in intestinal epithelial homeostasis. Thus, a better understanding of how damage is sensed by epithelia and coupled to tissue repair will hopefully yield insights important for the treatment of inflammatory diseases and cancer prevention.

Methods
Stress experiments. P.e. infection: flies were treated for 18 h with food laced with either 500-650 μl of 5% sucrose or 20 × P.e. resuspended in 5% sucrose from an 18 h culture grown at 30°C, 130 RPM or treated on filter paper with either 450 μl of 5% sucrose or 5% sucrose containing P.e. resuspended from a 20-24 h culture grown at 29°C, 200 RPM. Erwinia caratova caratova 15 (Ecc15) infection: flies were treated on the filter paper with either 450 μl of 5% sucrose or 5% sucrose containing Ecc15 (OD = 200) resuspended from a 20-24 h culture grown at 29°C, 200 RPM. Hydrogen peroxide exposure: 1% final of stabilized H 2 O 2 (Sigma-Aldrich) or H 2 O was mixed into food or fed on filter paper soaked with 450 μl of either 5% sucrose or 0.8% H 2 O 2 in 5% sucrose. Detergent exposure: 0.2% sodium dodecyl sulfate (SDS) or H 2 O was mixed into food or fed on the filter paper soaked with 450 μl of either 5% sucrose or 0.5% SDS in 5% sucrose. DNA damage: flies were fed on filter paper soaked with 450 μl of either 5% sucrose or 500 μM bleomycin sulfate (Abcam) in 5% sucrose. Heat shock: flies were heat shocked for 30 min or 4 h at 37°C.
Drosophila genetics. Flies raised at 18°C were shifted to 29°C to induce UAS transgene expression in progenitors with esgGAL4; tubGAL80 ts UAS-GFP, ECs with Myo1AGAL4; tubGAL80 ts (Myo1A ts ), and in VM with how-GAL4; tubGAL0 ts (how ts ). No statistical method was used to predetermine sample size, but typically between 10 and 20 flies were used per experiment. When selecting animals for an experiment, the parental genotype was not concealed since it was required to select relevant progeny. Animals were first selected for their genotype and then randomly chosen to be used for an experiment. See Supplementary Methods for genotypes.
Mitotic indices were determined by counting the number of pH3-positive cells from whole female midguts from two or more independent experiments. The mean number of mitoses per midgut and s.e.m. are presented for each genotype or treatment. Before quantifying the number of mitoses per midgut, the genotype of each sample was concealed. Samples were then randomly analyzed, and the genotype was revealed only after completing analysis.
Statistical analysis. Statistical analyses were performed using GraphPad Prism 5 or 8. For data describing mitoses per midgut, the two-sided Mann-Whitney test (for non-Gaussian data) or the two-sided unpaired t test (for Gaussian data) was applied to determine statistical significance. For qPCR data, the one-way ANOVA test and the Dunnetts sub-test were used on relative mRNA levels not normalized to control. The one sample t test was used on relative mRNA levels normalized to control or treated control. For Stat-GFP or phospho-p38 levels, unpaired t test was used on relative levels not normalized to control. The significance level is indicated by an * for p ≤ 0.05, ** for p ≤ 0.01, *** for p ≤ 0.001, **** for p ≤ 0.0001, and by NS for not significant, p > 0.05.
Microscopy. Samples were analyzed using Leica M205FA, DM5000B, SP5 and SP8 microscopes. Images were processed with ImageJ (NIH) and Adobe Photoshop CS5. Confocal images are presented as maximal intensity projections of images obtained every 0.5-1.0 μm. When comparing protein levels, each z-stack was acquired with the same laser intensity and gain, maximal intensity projections of identical dimensions were created and were further similarly processed with Adobe Photoshop CS5. Representative images presented were obtained from ≥ 2 independent experiments.
Image quantitation. Maximal intensity projections (of 5-6 μm thickness) of confocal images from the EC layer of each midgut were generated with ImageJ (NIH). ImageJ was then used to measure the surface area and the signal integrated density for each midgut. The ratio of integrated density:area was determined for each midgut and then normalized to the mean integrated density/area of control midguts to determine the relative signal (phospho-p38 or GFP) levels between treatments and/or genotypes.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.