MK2-Dependent p38b Signalling Protects Drosophila Hindgut Enterocytes against JNK-Induced Apoptosis under Chronic Stress

The integrity of the intestinal epithelium is crucial for the barrier function of the gut. Replenishment of the gut epithelium by intestinal stem cells contributes to gut homeostasis, but how the differentiated enterocytes are protected against stressors is less well understood. Here we use the Drosophila larval hindgut as a model system in which damaged enterocytes are not replaced by stem cell descendants. By performing a thorough genetic analysis, we demonstrate that a signalling complex consisting of p38b and MK2 forms a branch of SAPK signalling that is required in the larval hindgut to prevent stress-dependent damage to the enterocytes. Impaired p38b/MK2 signalling leads to apoptosis of the enterocytes and a subsequent loss of hindgut epithelial integrity, as manifested by the deterioration of the overlaying muscle layer. Damaged hindguts show increased JNK activity, and removing upstream activators of JNK suppresses the loss of hindgut homeostasis. Thus, the p38/MK2 complex ensures homeostasis of the hindgut epithelium by counteracting JNK-mediated apoptosis of the enterocytes upon chronic stress.


Introduction
In its function as a protective barrier, the intestinal epithelium is constantly exposed to stressors from the outside [1]. It acts as a mediator between the bacterial flora and the host's immune system, and the intestinal epithelial cells have to respond to extrinsic and intrinsic factors to ensure their own survival and proper gut homeostasis. Two populations of cells have to orchestrate different aspects of intestinal epithelium survival. While the intestinal stem cells (ISCs) are essential for the proliferative aspects of intestinal homeostasis [2][3][4], the enterocytes (ECs) form the first line of defence against pathogens and stressors. Signalling cascades that are modulated by external signals and by cellular stress are crucial regulators of intestinal epithelial survival. For example, mice with gut-specific knockout of NEMO spontaneously develop intestinal lesions reminiscent of those in inflammatory bowel diseases (IBDs), indicating an essential role for NFkB signalling in EC survival [5,6]. Recently, ER stress in the ECs has also been found to influence epithelial homeostasis, and mutations in XBP1 are sufficient to trigger an IBD-like phenotype [7]. The p38 stress-activated protein kinase pathway has also been implicated in intestinal disorders [8] but its role in intestinal diseases is still controversially discussed [9,10].
The p38 SAPK belongs to the MAPK family and is conserved from yeast to humans. In higher eukaryotes, p38 associates with its major target, the MAPK-activated protein kinase MK2. This complex resides in the nucleus in the resting state. Upon stress, p38 is activated by MKK3/MKK6 and phosphorylates MK2, which results in an exposure of the nuclear export signal of MK2 and a subsequent nuclear export of the complex [11]. Another consequence of the p38/MK2 complex formation is the stabilisation of p38 protein. Interestingly, the kinase activity of MK2 is required neither for the nucleo-cytoplasmic shuttling nor for the p38 protein stabilisation [11][12][13]. Conversely, MK2 kinase activity is crucial to phosphorylate small heat-shock proteins, transcription factors (e.g., SRF and HSF-1), and TTP [14][15][16]. The inhibitory phosphorylation of TTP by the p38/MK2 complex has been shown to increase the translation of AU-rich elements (ARE)-containing mRNAs including the mRNA encoding the proinflammatory cytokine TNFa [17]. Furthermore, p38 and MK2 have been shown to act as cytoplasmic checkpoint kinases in parallel to CHK1 [18,19]. Due to the plethora of p38/MK2 functions, targeting the p38 SAPK branch with inhibitors might lead to harmful side effects. Thus, it is important to understand the roles of p38 signalling in a tissue-specific context. The availability of mouse models has helped to decipher some in vivo roles of p38 SAPK signalling [20,21] but the complex nature of the intestinal system has hampered a detailed analysis. Studies in the model organism Drosophila have provided new insights into intestinal maintenance and how different signalling pathways are employed to ensure proper gut homeostasis. The Drosophila gut consists of the fore-, the mid-, and the hindgut. The larval and adult midgut displays a regional specification along the antero-posterior axis and fulfils vital functions such as nutrient absorption [22]. The main part of the embryonic and larval hindgut, the large intestine, is subdivided into a ventral (hv, positive for Delta expression) and a dorsal domain (hd, positive for engrailed expression) [23] that are separated by a single row of boundary cells [24]. The adult hindgut shows a similar yet more complex organisation [25]. The function of the hindgut, however, remains largely unknown. Ultrastructurally, the hd domain is marked by deep infoldings and enriched in elongated mitochondria, resembling the rectal papillae of other insects. Based on these similarities, it has been speculated that the Drosophila large intestine plays a role in ion and/or water resorption [23,24].
At least four distinct intestinal epithelial cell types are found in Drosophila: intestinal stem cells (ISCs), the hormone-producing enteroendocrine cells (EEs), enterocytes (ECs) and the transient enteroblasts (EBs; progenitors of EEs and ECs). In the adult midgut, ISCs are required for normal gut homeostasis, but in aged and/or stressed individuals the number of midgut ISCs is increased and differentiation is disturbed [26,27]. The orchestrated activation of Hippo, JNK, JAK/STAT, Notch and EGFR signalling within and between the ISCs, the ECs and the visceral muscles is required to coordinate proliferation, differentiation and cellular turnover in the midgut epithelium [26,[28][29][30][31][32][33][34][35][36]. In contrast to the situation in the midgut, the hindgut stem cells are not required for hindgut homeostasis in the larva and in the adult fly. The hindgut ISCs are rather needed during the shift from larval to adult hindgut, and stress induces proliferation and cell migration in the pylorus region [37]. Thus, especially the adult midgut serves as a good experimental model for the analysis of EC replenishment upon damage, but the mechanisms governing proper stress response in the ECs have remained elusive.
p38 SAPK signalling also plays a role in gut homeostasis in Drosophila. In aged adult midguts, an increase in p38b expression has been observed in Delta-positive stem cells, which appears to be partly due to DREF-mediated transcriptional activation [38,39]. Knockdown of p38b in the ISCs prevents age-and stress-induced ISC overproliferation and accumulation of aberrant Delta-positive cells, implying a role for p38b in regulating intestinal regeneration [38]. p38 signalling has been shown to be required for DUOX expression in differentiated ECs and for normal differentiation in the ISCs [38,40]. Consistently, a recent study showed that the larval intestinal epithelium is more susceptible to damage by pathogens in the absence of p38 function [41]. However, the mechanism of p38 action within the ECs remains unclear. In Drosophila, p38 signalling can antagonize the closely related JNK SAPK branch [42]. JNK signalling has also been shown to regulate several aspects of intestinal function. It is required in the midgut ECs to induce autophagy and thereby ensure their survival during oxidative stress [43]. In the ISCs, JNK is needed for proper stress response but strong activation of JNK leads to differentiation defects and loss of gut homeostasis [26]. Whether p38 and JNK influence each other in their intestinal function has not been addressed so far.
In this study, we investigate how the Drosophila larval hindgut is enabled to maintain homeostasis under stress conditions. Using deletion mutants for MK2, p38a and p38b, we show that MK2 and p38b form a complex that is specifically required to protect the dorsal hindgut ECs against chronic stress. In the absence of this p38b/MK2 complex, JNK is activated in patches of hindgut ECs, resulting in JNK-dependent apoptosis, loss of epithelial organisation, and melanisation of hindgut regions. This melanisation of the ECs does not require the recruitment of hemocytes, indicating an epithelial response that might also precede immune activation in mammalian intestinal diseases. Thus, we identify a specific SAPK signalling module required to maintain hindgut epithelial integrity upon stress.

Drosophila MK2 Is Dispensable for Normal Development and Survival at Non-Stress Conditions
The mammalian MAPKAP-K2 is known to be a downstream kinase of the p38 branch of SAPKs [44]. To generate deletion mutants for Drosophila MK2, we mobilised a P-element insertion located in the MK2 locus ( Figure 1A). Whereas D43 is a null allele as judged by the absence of MK2 protein and by the failure of D43 larval lysates to phosphorylate mammalian small heat shock protein 25 in a kinase assay (Figure S1A), the alleles D41 and D12 are likely to represent hypomorphic alleles. D38 is probably also a null allele although the generation of a truncated protein (initiated from alternative Methionine codons) cannot be excluded ( Figure 1A). A precise excision allele of the same P-element (D1A) was used as control throughout this study. All of the generated alleles show no obvious phenotypic alteration and can be kept as homozygous lines at normal conditions. Since several mutants of the SAPK pathway are sensitive to stresses including high osmolarity [45,46], we tested the MK2 mutants in various stress assays including oxidative stress (paraquat feeding), UV exposure during late embryonic/early larval development, heavy metals (0.5 mM copper sulphate), high osmolarity (0.2 M NaCl) and SDS (0.2%) ( Figure S1B and data not shown). Interestingly, only salt and SDS feeding resulted in a melanotic phenotype in 35% to 45% of the mutant larvae, characterized by the appearance of a ''black dot'' (BD) in the posterior part of the body ( Figure 1B and Figure S1C-S1E). Closer examination revealed that the BD localised to the posterior hindgut, and the affected hindgut epithelium appeared pathologically altered ( Figure 1C). All MK2 alleles were analysed for the appearance of BDs and survival at three conditions: normal food, weak salt stress (0.1 M NaCl) and strong salt stress (0.2 M NaCl) ( Figure 1D). The behaviour of MK2 hypomorphic larvae indicated that the levels of

Author Summary
The gut epithelium forms the first barrier against pathogens and stressors in the gut lumen, and a loss of this defence function can result in intestinal diseases. Damage in the gut epithelium triggers the proliferation of intestinal stem cells to replenish the epithelium. However, little is known about how the enterocytes are protecting themselves against stressors. We addressed the function of stress-activated protein kinase (SAPK) signal transduction pathways in the larval gut of Drosophila. Our study revealed that a particular module of the p38 SAPK signal cascade is required to protect the larval hindgut epithelium against chronic salt stress. We identified the two kinases, p38b and MK2, as key components of this protective signal. In the absence of p38b or MK2, the stress-inducible JNK cascade is locally upregulated and eventually induces apoptosis. Although the function of the p38b/MK2 module is only required in the enterocytes, the elimination of the affected cells results in atrophy of the overlaying muscle layer and subsequent systemic defects in the larvae (e.g., induction of antimicrobial peptides). We hope that our findings will contribute to a better understanding of early (i.e., pre-inflammatory) events in the development of human intestinal diseases.
MK2 get more important with increasing osmolarity. Introducing a genomic rescue construct completely rescued both the BD phenotype and the lethality of D43 mutants at 0.2 M NaCl ( Figure 1D). Thus, MK2 is not essential at normal conditions but is required when larvae are reared on a high sodium chloride diet.

Drosophila MK2 Is Required to Protect the Hindgut from Stress-Induced Apoptosis
We speculated that MK2 is specifically required to protect the hindgut epithelium. In sections of the hindgut, dorsal ECs with BDs were apically ruptured (Figure 2Aii). In wild-type hindguts, ECs of the dorsal domain were not damaged under salt stress conditions ( Figure 2Ai and Figure S2C). Similarly, most MK2 mutant hindguts without BDs did not display morphological alterations ( Figure S2C). However, a blistering of the apical surface without damaging the apical membrane was occasionally observed (Figure 2Aiii and Figure S2C). Blistering of the apical surface was also observed in MK2 mutant hindguts with BDs in regions away from the BD ( Figure S2C). The melanisations occurred at the apical surface within the ECs, probably preceded by the blistering (Figure 2Aiii and Figure S2C). Next, we tested the hindgut tissue for the presence of dying cells. TUNEL staining revealed local clusters of apoptotic hindgut ECs in MK2 mutant larvae (Figure 2Bi and 2Bii) but not in wild-type larvae reared on 0.2 M NaCl (Figure 2Biii). The visceral muscles surrounding the hindgut ECs appeared unaffected in MK2 mutant larvae. A disturbed hindgut musculature was only observed in hindguts with large BDs ( Figure 2C). Furthermore, loss of epithelial integrity was evident from the mislocalisation of Neuroglian (Nrg) ( Figure 2D). Nrg-GFP was localised laterally in cell-cell junctions in hindguts of wild-type and MK2 mutant larvae (Figure 2Di). In contrast, in MK2 mutant larvae with BDs, Nrg-GFP was normally localised in unaffected regions but displayed a more diffuse pattern close to the BD (Figure 2Dii). This mislocalisation was more pronounced at 0.2 M NaCl (Figure 2Dii and 2Diii). Interestingly, BDs were only found in the dorsal hindgut (hd) compartment, as judged by cell and nuclear size (hd is composed of smaller cells) (Figure 2Dii and 2Diii).
We performed a series of rescue experiments to determine where MK2 function is required. The BD phenotype was rescued by ubiquitous and hindgut-specific but not by midgut-specific expression of MK2 ( Figure S2D). Moreover, only wild-type MK2 but not a kinase-dead version of MK2 rescued the BD phenotype. The largest domains of the larval hindgut are the dorsal engrailedpositive (hd) and the ventral Delta-positive (hv) domain. BDs were only observed in the hd domain ( Figure 2E). Consistently, engrailed-GAL4 driven expression of a wild-type Drosophila MK2 or a wildtype murine MK2, but not of a kinase-dead Drosophila MK2, rescued the BD phenotype ( Figure 2F).
Melanisation in insects can be regarded as a wound healing or defence response and has been shown to be either hemocytedependent or hemocyte-independent [47]. We thus attempted to clarify whether BD formation in MK2 mutant larvae was dependent on the recruitment of hemocytes. Staining for the blood cell marker Hemese -either direct (using antibodies) or indirect (using He-GAL4 UAS-GFP) -revealed the absence of blood cells at or within the hindgut epithelium in both wild-type and MK2 mutant hindguts ( Figure S2A and data not shown). Consistently, the formation of BDs was still observed in hemocyte-ablated MK2 mutants ( Figure S2B).
Together, those results show that Drosophila MK2 kinase function is required in the dorsal hindgut compartment to protect larval hindgut ECs from stress-induced apoptosis upon salt stress.

MK2 Genetically Interacts with p38a and p38b
Mutations in MEKK1 and p38b but not in MKK3/lic and p38a resulted in a strong BD phenotype even at normal food conditions ( Figure 3A). To define the roles of p38a and p38b with respect to MK2, we next tested these kinases genetically for the behaviour at normal conditions and under salt stress ( Figure 3E). At unstressed conditions, MK2 was required neither for hindgut homeostasis nor for survival. Similarly, p38a mutants did not display BDs or elevated mortality rates. In contrast, MK2; p38a double mutants developed a weak BD phenotype, indicating that p38a and MK2 act in two parallel stress-signalling pathways. p38b mutants had a decreased survival rate, consistent with published findings [46]. p38b is required for hindgut homeostasis even under normal conditions, because larvae lacking p38b function developed BDs on normal food. Interestingly, MK2; p38b double mutants displayed a slight increase in BDs but no increase in lethality rate compared to p38b single mutants. Thus, p38b and MK2 are likely to function in the same pathway but both kinases may have additional independent functions in the hindgut or in other tissues.
Consistent with a common MK2/p38b pathway, p38b single mutants and MK2; p38b double mutants displayed the same phenotype at both weak (0.1 M NaCl) and strong (0.2 M NaCl) stress conditions. At weak stress conditions, p38a was not crucial for survival and hindgut homeostasis. In contrast, MK2 mutants displayed hindgut defects but no increase in mortality. In agreement with MK2 and p38a acting in two parallel stresssignalling pathways, BD formation was only slightly increased in the MK2; p38a double mutants but the absence of MK2 significantly enhanced the mortality of p38a mutants. This lethality increase of MK2; p38a double mutants was also observed at strong salt stress. At this condition, both MK2 and p38a mutants resulted in increased mortality rates, indicating that both branches of stress signalling are required for survival. Whereas MK2 is specifically required in the hindgut, p38a might be needed in other organs since p38a mutants hardly developed BDs.
Taken together, our genetic data suggest the existence of two major p38 branches in Drosophila that are required to varying extents during normal and different salt stress conditions. The p38a branch is not essential at normal or weak salt stress conditions but required at strong salt stress conditions. In contrast, the p38b branch is required at both normal and salt stress conditions to protect the hindgut. Moreover, the lethality of p38a; p38b double mutants [46] suggests that, besides their specific functions, both p38 kinases engage in a common essential function. MK2 is involved in a sub-branch of the p38b branch and appears to be a key effector of p38b in the hindgut. MK2 becomes more important with increasing stress conditions specifically in this tissue.

p38b Phosphorylation Is Increased in Melanised Larval Hindguts
We next checked the activation status of p38 by Western analysis ( Figure 3B). Chronic exposure (from L1 to L3) to 0.2 M NaCl did not increase p38 phosphorylation in total and hindgut lysates of wild-type larvae. In contrast, stronger p38 activation was observed in total larval lysates of MK2 mutants under stress conditions. This strong activation was also apparent in the hindgut but only in larvae with BDs. Since we were not able to distinguish the endogenous p38a and p38b, we overexpressed GST-tagged versions of p38 and analysed their activation status in the hindgut ( Figure 3C). GST-p38a was strongly activated even under normal conditions, and the activation was slightly increased under stress conditions. In MK2 mutants, GST-p38a was more strongly activated under all conditions and especially in hindguts with BDs. In contrast, GST-p38b was only weakly phosphorylated in wild-type and MK2 mutant hindguts at all conditions, except in hindguts of MK2 mutant larvae with BDs where a boost in GST-p38b phosphorylation was seen. Thus, the increase in endogenous p38 phosphorylation observed in MK2 mutant larval hindguts is probably due to p38b phosphorylation, suggesting that a negative feedback loop operates from MK2 to the upstream signalling components. Alternatively, the lack of a functional stress-protective MK2/p38b function may increase the stress in the ECs and lead to a vicious cycle that boosts p38b activation.

MK2 Protein Abundance Depends on p38b
In mammalian cells, MK2 is needed to stabilise p38 protein levels [13], which does not appear to be the case in Drosophila ( Figure 3B and 3C). We wondered whether the protein levels of MK2 would depend on the presence of the upstream components. No change in MK2 expression was observed in p38a mutants, and a slight increase was detected in MEKK1 mutants. In sharp contrast, MK2 protein levels were reduced in p38b mutants ( Figure 3D). Using a genomic MK2 rescue construct and a genomic MK2-GFP reporter in an MK2 null mutant background, we found that transcription from the MK2 locus was unchanged but protein levels were reduced, probably due to destabilisation of MK2 in the absence of p38b ( Figure 3D).

MK2 Physically Interacts with p38b
The facts that MK2 and p38b genetically interact and that MK2 protein levels depend on the presence of p38b suggest a close physical interaction of the two kinases. Therefore, we expressed tagged versions of p38a, p38b and MK2 in S2 cells and checked for co-localization of these kinases. In mammalian cells, MK2 is nuclear under normal conditions and translocates to the cytoplasm upon stress [48]. Similarly, in Drosophila S2 cells, GFP-MK2 was mainly found in the nucleus, whereas overexpressed p38a and p38b were largely cytoplasmic ( Figure 4A). Co-overexpression of p38b and MK2 (but not of p38a and MK2) resulted in a nuclearto-cytoplasmic translocation of GFP-tagged MK2 ( Figure 4A). Consistently, MK2 was found to bind p38b but not p38a in pull down experiments ( Figure 4B). This interaction as well as the nuclear-to-cytoplasmic shuttling was dependent on a four amino acid motif (DPTD) in p38b resembling the common docking motif (CD domain) that is critical for docking interactions in MAPKs [49]. In p38a, the respective four amino acids are EPSV. Exchanging these motifs revealed that the DPTD motif is necessary and sufficient to dock MK2 to p38 proteins ( Figure 4A and 4B).
The expression of GST-p38b completely rescued the BD phenotype of p38b deficient flies, whereas expression of GST-p38a had no influence on the BD phenotype ( Figure 4C and Figure S3). Expressing either GST-p38aRb (p38a with docking motif of p38b) or GST-p38bRa (p38b with respective sequences of p38a) resulted in a partial rescue of the BD phenotype. On normal food, both versions rescued partially, indicating that although not essential, binding to MK2 is required for a complete rescue under normal conditions. On 0.1 M and 0.2 M NaCl, GST-p38aRb resulted in substantial but incomplete rescue, suggesting that other aspects of p38b function not mediated by binding to MK2 are required for a complete rescue. Consistently, a GST-p38bRa protein that is not able to bind MK2 also partially rescued the sequence encoding GFP) is unaltered in a p38b mutant background. In contrast, MK2 expressed from the same genomic region (MK2-MK2, MK2 coding sequence not replaced) does not result in restoration of MK2 protein levels (lower blot). (E) p38a and p38b deletion alleles (in wild-type and MK2 mutant background, respectively) were assessed for the appearance of BDs (upper panel) and survival (lower panel) on normal food, 0.1 M NaCl and 0.2 M NaCl, respectively. doi:10.1371/journal.pgen.1002168.g003 p38b phenotype but to a lesser extent than p38aRb ( Figure 4C and Figure S3). MK2 protein levels in the hindgut were restored by expressing p38b or p38aRb but not by p38a or p38bRa ( Figure 4D). Thus, binding of p38 to MK2 is required to localise and stabilise MK2 and is important for the stress-protective function in the hindgut.
The Catalytic Activity of p38b Is Required to Localise MK2 The catalytic activity of MK2 and MK2 binding to p38b are required to protect the hindgut epithelial cells upon salt stress. To address whether the catalytic activity of p38b is also necessary, we used GST-tagged non-activatable p38b AGF and kinase-dead p38b KR protein mutants. Whereas co-expression of wild-type GST-p38b and GFP-MK2 led to a nuclear export (.70%) of MK2, the localisation of GFP-MK2 was random when GST-p38b AGF or GST-p38b KR were co-expressed ( Figure 5A and 5B). Consistently, the BD phenotype of p38b null mutants was not rescued by re-expression of the kinase-dead or of the non-activatable p38b protein version ( Figure 5C). Moreover, at high NaCl stress (0.2 M NaCl), the p38b AGF and p38b KR expressing larvae died. This could be explained by the titration of an upstream kinase of p38b, which might impinge on the activation of other downstream effectors. Thus, the catalytic activity of p38b is required to impact on the subcellular localisation and thereby the proper function of MK2.

JNK Signalling Is Induced in MK2 Mutants and Is Linked to Apoptosis
JNK signalling has been implicated in triggering apoptosis [50]. Furthermore, a JNK antagonizing activity of p38 signalling has been observed in developmental processes [42] and at the systemic level during infection [41]. Thus, we tested whether the cell death observed in MK2 mutant larvae reared on high salt correlated with JNK activation. Indeed, elevated levels of phosphorylated JNK were detected in hindgut lysates of MK2 mutant larvae with BDs ( Figure 6A). As in vivo readouts for JNK signalling activity, we used reporters for misshapen and puckered [51,52]. An induction of both msn.lacZ and puc.lacZ was observed in MK2 mutant larval hindguts, with highest levels adjacent to the BDs ( Figure 6B). To exclude that the induction of JNK signalling is a secondary consequence of wound healing or the melanisation process, we checked for puc-GFP induction in stress-challenged MK2 mutant larvae before BD formation. Interestingly, patches with puc-GFP signal were readily detected upon stress in larvae devoid of BDs, and the area of those patches correlated with the strength of the stress ( Figure 6C). We noted that puc-GFP was activated in a graded fashion, with highest activity where ECs undergo apoptosis and a BD will ultimately form ( Figure 6D). Strong puc-GFP reporter activity co-localised with TUNEL positive cells close to the BDs ( Figure 6E). Removing the JNK upstream components TAK1 and MKK4, respectively, in an MK2 mutant background partially suppressed the BD phenotype, indicating that the hindgut epithelial cells are dying due to JNKinduced apoptosis ( Figure 6F). Expressing a dominant-negative version of JNK (Bsk DN ) in the dorsal domain using engrailed-GAL4 resulted in a suppression of the BD phenotype ( Figure 6G). Furthermore, expression of Bsk DN specifically in cells with high JNK activity (using puc-GAL4) substantially suppressed BD formation. In contrast, re-expression of MK2 at this stage reduced the number of larvae with BDs only mildly ( Figure 6G). Thus, deregulated JNK Figure 4. A p38b/MK2 complex is required for hindgut EC stress protection. (A) GFP-MK2 (green) localises to the nucleus. GST-p38a and GST-p38b show a broad, more cytoplasmic distribution (red). GST-p38b co-expression shuttles GFP-MK2 to the cytoplasm in a DPTD-motif dependent manner. GST-p38a or GST-p38bRa (DPTD changed to EPSV) do not change the localisation of MK2, but a GST-p38aRb protein mutant can export GFP-MK2 from the nucleus. (B) In pull down experiments, GST-p38b co-precipitates MK2, whereas p38a and the docking mutant p38bRa do not bind MK2. The DPTD motif introduced into p38 (p38aRb) is sufficient for binding to MK2. (C) Whereas overexpression of GST-p38b rescues completely, GST-p38a fails to rescue the BD phenotype of p38b mutants. GST-p38aRb and GST-p38bRa partially rescue the BD phenotype with p38aRb performing better. (D) Docking of MK2 to p38 that harbours a DPTD motif is sufficient to restore wild-type levels of MK2 protein in a p38b mutant background. doi:10.1371/journal.pgen.1002168.g004 activation in the hindgut of MK2 mutants precedes and probably causes cell death and BD formation.

Discussion
Gut homeostasis-under normal and stress conditions-is ensured by complex interactions between the intestinal epithelium, the immune system and the gut flora. Drosophila has been used as a simple model organism to address different aspects of intestinal homeostasis. Replenishment of the gut epithelium by ISCs clearly contributes to epithelial homeostasis but how the differentiated ECs are protected against stressors has remained largely unknown. We used the larval hindgut of Drosophila as a simple intestinal model organ to address how stress signalling in the hindgut ECs ensures intestinal homeostasis in the absence of proliferative cells. Our analysis identifies the p38b/MK2 signalling module as a We propose a model that puts a p38b/MK2 complex in the centre of stress-protection of the hindgut ECs (Figure 7). In the absence of this signalling module, cells are undergoing JNKdependent apoptosis upon stress. The lesion in the EC monolayer results in the damage of the overlying hindgut musculature. This regional loss of the barrier function leads to systemic defects in the larvae ( Figure S4), further weakening the larvae and impairing growth under stress conditions. As a consequence, pathogens and toxic substances might enter the body cavity, eventually resulting in the melanisation of pericardial cells and the induction of cecropin in the midgut ( Figure S4).
Interestingly, JNK activation in MK2 mutant hindguts precedes the melanisation, and it consistently occurs in patches. Within these areas, some cells acquire highest amounts of JNK activity and eventually undergo apoptosis. The surrounding cells maintain JNK activity, forming a rim around the scar in the tissue. The number of affected ECs remains roughly constant for a given stress. Presently we do not know what determines the patches with high JNK activity within the tissue. Although the hd domain ECs of the hindgut form a homogeneous epithelium and are facing the same stressor, JNK signalling is only induced in clusters of a certain size but not in surrounding cells. Increased JNK activation was also observed in the p38a deleted intestinal epithelium of a mouse model for IBDs [53]. Furthermore, ulcerations occur in similar patchy patterns in IBDs [54]. Thus, the MK2 mutant phenotype may be useful to decipher how a group of cells within a tissue of genetically identical cells transforms into the weakest link in the chain upon stressful conditions. Several lines of evidence support the notion that the p38b/MK2 signalling complex is key to EC protection against chronic salt stress. (1) p38b and MK2 mutant larvae both develop BDs upon stress conditions. (2) The severity of the p38b; MK2 double mutant phenotype upon stress is similar to the p38b single mutant phenotype, suggesting that they act in the same signalling pathway. (3) p38b but not p38a physically associates with MK2 via its C-terminal CD domain (DPTD motif). (4) The binding of p38b to MK2 stabilises MK2. (5) Upon co-expression, p38b but not p38a redirects MK2 to the cytoplasm. (6) Both the activation and the catalytic activity of p38b are required to efficiently relocalise MK2. (7) The binding of MK2 to p38 and the catalytic activities of both kinases are essential to protect the ECs of the larval hindgut upon salt stress. Taken together, stabilisation, localisation and activation of MK2 by p38b are required for a proper stress response.
Our genetic analysis also revealed that p38 SAPK signalling is contributing to stress protection in different ways in addition to the pivotal role of the p38b/MK2 complex. First, p38b impacts on hindgut homeostasis in an MK2-independent manner. This is apparent from the p38b mutant larvae that, in contrast to MK2 mutant larvae, develop BDs even at normal conditions. Consistently, a p38b protein version that no longer binds MK2 partially rescues the p38b mutant phenotype. Second, p38a is also required for full stress protection. The strong phenotype of MK2; p38a double mutants underscores the importance of the p38a SAPK pathway upon severe salt stress. However, the double mutants do not display an increase in BD formation but rather a decrease in viability. Thus, p38a may be involved in general stress protection that is not specific to the hindgut tissue. Third, a common p38 SAPK branch, encompassing p38a, p38b and potentially also p38c, is essential as the p38a; p38b double mutants die. Since the p38a 1 allele affects the coding sequences of p38a and p38c, the p38a 1 ; p38b d27 double mutants may even represent p38a; p38b; p38c triple mutants. However, it is unclear to date whether p38c is a pseudogene. Recent studies have suggested an involvement of p38c function in immune gene regulation, early larval survival, and fertility [41,55]. Thus, further studies will be required to clarify whether p38c contributes to p38 signalling subbranches. Finally, the slight increase in BDs seen in MK2; p38b double mutants under normal conditions suggests that MK2 also performs a p38b-independent function in the hindgut. p38b mutants always impact on MK2 signalling since the MK2 protein is not stable and probably not correctly localised if not bound to p38b. Negative feedback regulation acting from MK2 on the activation of p38b further complicates the SAPK signalling network.
What are the upstream components regulating the p38b/MK2 complex? To our surprise, MKK3/Lic does not appear to play a role in the hindgut function of the p38b/MK2 branch. MKK3 but not MKK4 can activate p38 proteins in cell culture [56]. On the other hand, siRNA mediated knockdown of both MKK4 and MKK3 is required to fully block the activation of p38 under certain stresses in S2 cells [57]. Both p38b and MKK3/lic mutants show a strong reduction in p38 activation but no BDs are observed in MKK3/lic mutants. In contrast, mutants for MEKK1, which acts upstream of MKK3, do develop BDs similar to p38b and MK2 mutants. In mammalian cells, it has been shown that p38 can be activated independently of a MAP2 kinase [58][59][60]. However, no activation of p38 occurred in fibroblasts of MKK3 MKK6 double mutant mice [61]. Since MKK4 is a suppressor of the BD phenotype and MKK7 most likely does not activate p38 in Drosophila, a scenario of MAP2K-independent p38 activation could also apply for the p38b/MK2 signalling branch in the larval hindgut ( Figure 7).
Interestingly, overexpression of the kinase-dead or of the nonactivatable p38b protein version results in an even stronger phenotype than the deletion of p38b, probably by titrating upstream partners that would have additional functions besides activating p38b. A common p38a/p38b activator would be a strong candidate for such an upstream component. Saturating this common p38 activator with p38b KR or p38b AGF would essentially result in a p38a/p38b double mutant situation and therefore would explain the strong phenotype, especially at high salt conditions.
Our analysis of the p38b/MK2 signalling module in hindgut ECs reveals how deletion of SAPK members results in increased sensitivity towards a particular stressor from the molecular level to the level of the whole organism. These findings provide a new model of how hindgut homeostasis is maintained and how different SAPK branches act together in vivo to ensure cellular survival upon stress. The p38 SAPK pathway efficiently protects hindgut ECs over a wide range of stress conditions and for an astonishingly long time period of at least five days without cell replenishment. In this light, a comparative analysis of the larval and the adult hindgut (which has to be maintained for a time period of up to forty days) will be of great interest.
How does p38 interfere with JNK activation in the hindgut epithelium? In mammalian cells, p38a binds to and phosphorylates TAB1 (and potentially TAB2). As a consequence, the activities of TAK1 and thereby JNK are reduced [62]. TAB1 is not conserved in Drosophila but TAB2 might be involved in a similar negative feedback loop. TAB2 has been implicated in JNK activation in response to peptidoglycans and lipopolysaccharides. However, in the absence of TAB2, no change in JNK activation in response to Sorbitol or NaCl has been observed [63,64]. Alternatively, p38 might induce a JNK phosphatase. p38a has been shown to impact on JNK activation by inducing DUSP1/ MKP-1 in mammalian cells. Sustained JNK activation due to a loss of DUSP1/MKP-1 resulted in increased cell death in response to UV stress [65]. Interestingly, p38 activation was also increased in DUSP1/MKP-1 mutant cells but p38 activity was not linked to cell death induction. Similarly, MK2 mutant larvae bearing BDs display an overactivation of both p38 and JNK, and blocking JNK only is sufficient to prevent the BD phenotype.
Despite the more complex nature of mammalian guts, the strong conservation of stress-signalling pathways and the similar demands of ECs make it likely that our results will also be important in the context of various diseases of the intestinal system. A variety of different signalling pathways have been implicated in IBDs, underscoring the complex nature of these diseases. p38 and MK2 are critical regulators of TNFa production and are thereby associated with IBDs [8] but the role of p38 SAPK in IBDs has remained controversial [9,10]. Our study identifies a crucial role of p38b/MK2 signalling in the first line of defence against a particular stressor in a model system devoid of an adaptive immune system. The consequences of lacking this immune system-independent protective function of a SAPK branch might parallel early steps of IBD development in intestinal epithelial cells. Indeed, a recent study has revealed tissue-specific effects of p38a in a DSS-induced mouse model for IBDs [53]. Deleting p38a in the myeloid lineage had beneficial effects, consistent with the inflammatory nature of IBDs. In contrast, p38a deletion in the intestinal epithelium resulted in a loss of gut homeostasis, marked by increased proliferation and by a reduction in goblet cells. Thus, our studies on Drosophila p38 signalling and its role in the larval hindgut provide a basis to specifically address the role of ECs in the maintenance of an intestinal epithelium in the absence of proliferation and immune response.

Materials and Methods
Fly Media and Stock Keeping 1 litre Drosophila media contains 100 g fresh yeast, 55 g cornmeal, 10 g wheat flour, 75 g sugar, and 8 g bacto-agar. For stress medium, Drosophila media were boiled and sodium chloride was added from a 5 M stock solution. 15 ml/l of a stock solution containing 33 g/L nipagin and 66 g/L nipasol in 96% EtOH was added to prevent growth of mould and bacteria. All crosses and experiments were performed at 25uC.

Stress Experiments
Females were allowed to lay eggs overnight on apple agar plates. Eggs were collected and 80-120 eggs were transferred to the different food vials. For BD quantification, larvae were analysed before reaching L3 wandering stage. For survival quantification, dead embryos were counted 24 h after seeding to the food and survival to pupae was recorded, respectively.

Plasmid Constructs and Transgenic Animals
For the MK2 genomic rescue construct, the genomic region between CG15771 and CG15770 was PCR-amplified and cloned into pCasper3. For the genomic MK2-GFP reporter, the same region was used but the MK2 coding sequence was replaced by the GFP coding sequence.
For overexpression constructs, the MK2 coding sequence was amplified and cloned into pENTR/D-TOPO (Invitrogen). For the kinase-dead MK2 protein, the mutation leading to the K49A substitution was introduced by PCR mutagenesis. The inserts were shuttled into the destination vector pTGW (http://www.ciwemb. edu/labs/murphy/Gateway%20vectors.html) for N-terminal GFP tagging. To express untagged MK2, the MK2 coding sequence was cut from the pENTR/D-TOPO and ligated into a pUAST-attB vector.
GST-tagged p38a and p38b overexpression constructs were generated by ligating the GST coding sequence in frame to the p38a or p38b coding sequence, and the resulting fusion sequences were cloned into a pUAST-attB vector. The constructs encoding the p38a or p38b protein mutants were generated by PCR mutagenesis. In the p38aRb and the p38bRa protein mutants, the EPSV motif was changed to DPTD and vice versa. The kinase-dead or non-activatable p38b protein mutants were generated by introducing mutations in the coding sequence that result in the K53R substitution and in the replacement of the TGY dual phosphorylation motif by AGF, respectively.

Cell Transfection and Pull Downs
Transfection of S2 cells was done using the Effectene Transfection Reagent (Qiagen) according to the manufacturer's protocol. After four days of protein expression, cells were harvested and lysed. Pull down of GST-p38 proteins was performed using glutathione sepharose beads (Pharmacia Biotech AB). 10% of the lysates was loaded as input and the complete pull down sample was loaded onto SDS PAGE.

Dissection and Fixation of Larval Tissues
Larvae were collected in PBS on ice. Three to five larvae were transferred to new vials containing 400 ml ice-cold PBS. Larvae were dissected and the desired organs were transferred into a microfuge tube containing 500 ml of ice-cold 4% paraformaldehyde in PBS. Hindguts and midguts were fixed for 40-509 at 4uC. Subsequently, the fixative was removed by three washing steps with cold PBS. Fixed preparations were stored in PBS at 4uC until furthre use.

Preparation of Cells for IHC
Cover slips were washed by dipping into 100% EtOH and airdried. They were then incubated in 0.15% ConA solution (in ddH 2 O) for 1-2 h, washed with ddH 2 O and air-dried overnight.
ConA slides were placed into a small Petri dish and covered with Schneider's medium (approx. 1 ml). 200 ml S2 cells were added and allowed to settle onto the discs for 459 to max. 909. Cover slips were then washed once with ice-cold PBS, and 1 ml 4%PFA was added for 59 fixation on ice followed by 109 fixation at room temperature. Cover slips were washed three times with PBS. Cells were then permeabilised with PBT for 109 and stored in PBS until further processing.

Antibody Stainings
Specimens were blocked by incubating in 2% NDS in PBS with 0.2% Triton X-100 (or 0.3% Triton X-100 for hindguts) for 1 h (larval tissues) or 309 (S2 cells), respectively. Primary antibodies were added in PBS with 2% NDS and 0.2% Triton X-100 for 1 h (S2 cells) or overnight (larval tissues), respectively. Before secondary antibodies were added, samples were washed three times in PBS with 0.2% Triton X-100. Secondary antibodies were added in PBS with 2% NDS and 0.2% Triton X-100 for 1 h (S2 cells and larval tissues).
Western blot (WB) membranes were blocked in 3% membrane blocking agent (GE Healthcare) for one hour. Membranes were incubated with the primary antibodies overnight and one hour with the secondary antibodies in 3% membrane blocking agent.

Other Histological Stainings
Larvae were dissected and fixed by standard procedures. After washing with PBS, 500 ml X-gal staining solution was added, and the samples were incubated at 37uC in the dark. The staining progress was observed every 109, and the staining reaction was stopped by two washes with PBT.
Alexa Fluor 594-conjugated phalloidin (Molecular Probes) was used to stain muscles. For apoptosis detection, the TUNEL assay kit ApopTag RED In Situ Detection Kit (Millipore S7165) was used.

Hindgut Sections
Larvae were dissected on ice and hindguts were immediately fixed in 2.5% glutaraldehyde, 1% paraformaldehyde, 1% potassium ferrocyanide, 0.1 M cacodylate buffer for 809. After washing three times in 0.1 M cacodylate buffer, hindguts were postfixed in 1% osmium tetroxide, 1% potassium ferrocyanide, 0.1 M cacodylate buffer (pH 7.4) for 609. Hindguts were then dehydrated in an ascending acetone series (30%.50%.70%.90%.100% 39 each and 59 100%). The samples were incubated overnight in a 1:1 acetone:Spurr solution. After equilibration in Spurr solution for 4 h, samples were embedded in Spurr solution and hardened at 65uC overnight. 2 mm sections were made with a Supercut Reichert-Jung 2050 microtome, and sections were mounted in DPX Mountant for histology (Fluka). Figure S1 MK2 null mutants and development of BDs. (A) Larval extracts of wild-type (D1A) larvae but not of MK2 mutants (D43) have in vitro kinase activity towards small heat shock protein 25 (hsp25). Overexpression of wild-type MK2, but not of kinasedead MK2, boosts hsp25 phosphorylation. Antibodies against Drosophila MK2 do not recognize a band in Western analysis on MK2 mutant total larval lysates. (B) MK2 mutants were tested for the appearance of BDs by feeding different stressors. A rough classification reveals that only high salt and SDS feeding induce a BD phenotype. (C) Bright field pictures of hindgut sections of wild-type (D1A) and MK2 mutant (D43) larvae reared on 0.1 M NaCl. In MK2 mutants without BDs, the hindgut structure appears either undamaged (second panel) or displays blistering of ECs in the dorsal hd domain (black asterisk in third panel). EC blistering is also observed in MK2 mutants with BDs at a distance of the BD (black asterisk in fourth panel). Panels four and five show sections of the same hindgut ahead of and at the BD lesion site, respectively. White asterisks mark the gut content; black arrows point to undamaged apical membranes; white arrow indicates BD. (D) Various GAL4 lines were used to drive MK2 expression from a wild-type UAS-MK2 cDNA construct in an MK2 mutant background (D43), and the ability to rescue the BD phenotype was scored. Only ubiquitous and hindgut-specific expression of catalytically active but not of a kinase-dead MK2 rescues the BD phenotype. (TIF) Figure S3 p38b BD phenotype rescued by p38 expression. Homozygous p38b mutants were reared on 0.2 M NaCl food. Rescue of the BD phenotype by p38a and p38b was quantified ( Figure 4C). Here we show representative examples of larvae that were quantified for their BD appearance. (TIF) Figure S4 Systemic effects observed in MK2 mutants. (A) MK2 mutants reared on 0.2 M NaCl food often display a severely ruptured hindgut musculature (white arrows), resulting in a local gut barrier breakdown. The BD can be recognized based on its autofluorescence (yellow). (B) In such strongly affected larvae, the antimicrobial peptide CecA1 is induced in the midgut, indicative of a systemic response. The black bar (labelled with hg) indicates the hindgut; the black arrow marks the BD; the white arrows point to the CecA1-lacZ induction (blue). (C) The systemic disturbance in larvae with large BDs (asterisk) is underscored by the appearance of melanised pericardial cells (arrow). (TIF)