Cell signalling mechanisms for insect stress tolerance

Insects are exposed to multiple environmental stressors across a variety of habitats. In particular, insects are routinely exposed to desiccation, osmotic and xenobiotic stress and so have evolved highly successful strategies to combat these. There are multiple mechanisms for stress tolerance, thus allowing insects to successfully occupy virtually all ecological niches. It is becoming apparent that the molecular mechanisms for stress resistance and/or tolerance occur in key tissues, specifically epithelia. Epithelial tissue such as salivary glands, crop, gut and Malpighian tubules are barriers between the external and internal environments and so perform crucial roles in stress sensing and response. Moreover, organismal homeostasis depends on epithelial tissue, in particular, the osmoregulatory system. This comprises the fluid-secreting Malpighian tubules (in most insect species that have these) and the fluid-absorbing hindgut, which together maintain organismal ion and water homeostasis (Dow, 2013). Drosophila melanogaster has been used for studies in genetics and as a model insect for more than 100 years because of the molecular genetic tools available. This species is now increasingly used for work in biomedicine as well as in fundamental and applied biology, to understand mechanisms of function from molecule, cell and tissue to organism (Bellen et al., 2010; Dow, 2012a; Schneider, 2000).

As insect Malpighian tubules are fluid-secreting tissues, tubule ion transport pathways have been extensively mapped (Beyenbach, 2003; O'Donnell et al., 2003; Spring et al., 2009; Wieczorek et al., 2009), with tubules from D. melanogaster holding the distinction of being the fastest fluid-transporting epithelia known in biology (Maddrell, 2009). However, much more remains to be discovered, including neuroendocrine control of ion transport and fluid secretion (Coast, 2007). It is now also known that insect tubules perform many more functions than just osmoregulation. Tissue-specific transcriptomics of D. melanogaster (Chintapalli et al., 2007; Wang et al., 2004) have led the way in assigning novel functions to D. melanogaster tubules (Dow, 2009). Importantly, assignation of novel functions such as detoxification and xenobiotic handling, and stress sensing of oxidative, osmotic (ionic/salt) and immune challenges by transcriptomics analysis (Chintapalli et al., 2007; Dow, 2009; Wang et al., 2004), has been underpinned by functional and physiological analysis (Chahine and O'Donnell, 2011; Daborn et al., 2012; Davies et al., 2012; Naikkhwah and O'Donnell, 2011; Torrie et al., 2004; Yang et al., 2007). Thus, the tubules are mission-critical tissues for insect survival.

Drosophila melanogaster tubules emerge from the hindgut, just behind the junction with the midgut, and constitute a pair of anterior and posterior tubules (Beyenbach et al., 2010). Recent work indicates that the anterior and posterior tubules exhibit transcriptome and functional asymmetry (Chintapalli et al., 2012), suggesting further intriguing possibilities of anterior- and posterior-specific roles for each pair of tubules. Tubules consist of two major cell types, the principal and stellate cells (Dow, 2009), which allow functional separation of ion transport and cell signalling pathways (Dow and Davies, 2003) for physiological function. Principal cells contain the large ion transport complexes, e.g. the vacuolar H⁺-ATPase (V-ATPase) (Allan et al., 2005; Wieczorek et al., 2009) and the Na⁺/K⁺ exchanger (Torrie et al., 2004). Stellate cells, by contrast, express water channels (aquaporins) (Dow et al., 1995; Kaufmann et al., 2005) and control chloride flux (Denholm et al., 2013; Dow, 2012b; O'Donnell et al., 1998). Functional analysis so far suggests that the signalling pathways which mediate stress responses occur in the principal cell; however, the role of the stellate cell cannot be excluded.

Signalling pathways, specifically those for calcium and cyclic nucleotides, were initially investigated in tubules from large and physiologically amenable insects, e.g. locusts and crickets (Anstee et al., 1980; Morgan and Mordue, 1985; Phillips, 1982), as well as medically relevant insects, e.g. Rhodnius prolixus (Maddrell et al., 1971). Work on the development of D. melanogaster tubules as a genetic model for fluid-transporting epithelia subsequently indicated

signalling pathways that modulated tubule fluid secretion (Dow et al., 1994a; Dow et al., 1994b). These signalling pathways were shown to be 3′,5′-cyclic adenosine monophosphate (cAMP) and, later, 3′,5′-cyclic guanosine monophosphate (cGMP). Moreover, work in D. melanogaster tubules showed for the first time that the gaseous second messenger, nitric oxide (NO), modulated renal function (Dow et al., 1994a). More recent work has shown very complex regulation of cAMP, cGMP and calcium signalling in tubule principal cells (Davies, 2006; Dow and Davies, 2003) (Fig. 1), especially in relation to signalling cascades initiated by neuropeptides (Davies et al., 2013). Mosquito tubules also contain principal and stellate cells, and research into cell signalling mechanisms in mosquito tubules has revealed important insights into control of epithelial function in blood-feeding insects via both cell types (Coast et al., 2005; Kersch and Pietrantonio, 2011; Pollock et al., 2004; Radford et al., 2004; Schepel et al., 2010).

The current state of knowledge for cyclic nucleotide and calcium signalling in the tubule principal cell will be discussed in the following sections. This will provide some insight into the complexity of these signalling pathways in vivo; the newly discovered role(s) of some of these signalling pathways in organismal stress tolerance will then be described.

In D. melanogaster tubules, fluid secretion into the tubule lumen is energised by the V-ATPase located on the tubule principal cell apical membrane (Allan et al., 2005; Davies et al., 1996; Dow, 1999). Transepithelial fluid secretion rates in the tubule main segment are stimulated by cAMP or cGMP (Dow et al., 1994b) and the V-ATPase is also thought to be the ultimate target of cyclic nucleotide signalling in the tubule principal cell, because of increased transepithelial potential difference in intact tubules treated with either cAMP or cGMP (Bijelic and O'Donnell, 2005; Davies et al., 1995) (Fig. 1). Recent work in mosquito tubules has demonstrated that cAMP stimulates both V-ATPase activity and assembly of the holoenzyme (Tiburcy et al., 2013), which supports the findings in other insects [e.g. the blowfly (Baumann and Bauer, 2013)] and in other systems (Bond and Forgac, 2008). Although it is also possible that cyclic nucleotide kinases such as protein kinase A (PKA) may act to phosphorylate modulatory proteins for the V-ATPase, how may cyclic nucleotides increase the transepithelial potential difference across tubule principal cells? Recent evidence shows that cGMP directly increases ATP concentration in tubules. cGMP is transported into principal, but not stellate, cells via transporters (Riegel et al., 1999), and application of exogenous cGMP to intact tubules results in increased ATP concentration (Davies et al., 2013). Such increased availability of ATP substrate for the V-ATPase may ultimately increase V-ATPase activity.

Neuropeptide signalling via the corticotropin-releasing factor (CRF)-related diuretic hormone (CRF-related DH), also known as DH-44, activates cAMP signalling in tubule principal cells (Cabrero et al., 2002). DH-44 is a ligand for two receptors, DH-44 R1 and R2 (Hector et al., 2009; Johnson et al., 2005). DH-44 R2 is encoded by CG12370 (Hector et al., 2009), and is expressed in epithelial tissue but most highly in tubules in the adult fly (Chintapalli et al., 2007; Robinson et al., 2013). cAMP signalling, induced by exogenous cAMP (Riegel et al., 1998) as well as DH-44, increases fluid secretion rates by acting on the tubule principal cell (Cabrero et al., 2002; Dow et al., 1994b). Interestingly, DH-44 also increases total cAMP hydrolysing activity by cAMP phosphodiesterases (PDEs) (Cabrero et al., 2002), suggesting that targeted breakdown of cAMP is required for DH-44 action in addition to generation of cAMP.

Calcitonin-like neuropeptide, DH-31, also raises cAMP in D. melanogaster tubules (Coast et al., 2001), but it is possible that DH-44 and DH-31 target different downstream effectors. In tubules of the malarial mosquito Anopheles gambiae, DH-31, but not DH-44, increases basolateral Na⁺ conductance (Coast et al., 2005).

Recent use of targeted genetically encoded optogenetic probes has allowed further unique insights into cAMP signalling in the tubule (Efetova et al., 2013). A photoactive adenylate cyclase transgene (bPAC) can be activated to rapidly and reversibly generate cAMP pulses in a cell-type-specific manner using the GAL4/UAS binary expression system (Elliott and Brand, 2008). The GAL4/UAS targeted expression system utilises transgene expression directed by the binding of yeast transcription factor GAL4 (for which there are no endogenous targets in the fly) to its upstream activation sequence (UAS), allowing expression of any transgene cloned downstream of UAS. Cell- and tissue-specific GAL4 lines (‘driver’ lines) are available for many cells and tissues in D. melanogaster (Duffy, 2002; Sözen et al., 1997; Yang et al., 1995), allowing highly specific, targeted expression of transgenes in vivo. In particular, the GAL4 lines for tubule cell types and regions are highly specific (Rosay et al., 1997; Sözen et al., 1997; Terhzaz et al., 2012; Terhzaz et al., 2010b). Use of the targeted photoactive adenylate cyclase transgene bPAC in either tubule principal or stellate cells showed that PKA is necessary for basal fluid secretion rates in principal cells only, but is required for stimulated fluid secretion in stellate cells. Thus, in D. melanogaster tubule at least, PKA does not increase the activity of the V-ATPase under stimulated conditions. This work also demonstrated a novel role for the cAMP exchange protein, EPAC (Borland et al., 2009), in stimulated fluid secretion rates. PKA and EPAC have key roles in compartmentalised cAMP signalling (Houslay, 2010), and the work with the bPAC transgene provides the first evidence for insect tubules that compartmentalised cAMP signalling is essential for both basal and stimulated fluid secretion rates, and may also explain different downstream signalling by different neuropeptides utilising the same second messenger, e.g. DH-44 and DH-31.

In addition to stimulation of fluid secretion, cAMP (and cGMP) has also been shown to increase transepithelial cation transport across the main segment of the D. melanogaster tubules (Bijelic and O'Donnell, 2005). This suggests that multiple ion transport events can be modulated by each cyclic nucleotide.

cGMP signalling in the tubule principal cell relies on generation of the signal via either receptor or soluble guanylate cyclases (GCs) and breakdown via PDEs (Fig. 1). Several receptor GCs (rGCs) are expressed in tubules, including Gyc76c, CG33958 (also known as CG5719) and CG34357 (previously CG9783) (Davies, 2006). cGMP is also generated via NO-stimulated soluble GC (Davies et al., 1997; Dow et al., 1994a) in only tubule principal cells (Broderick et al., 2003). Neuroendocrine stimulation of principal cell cGMP occurs via the capa peptide family, which stimulate NO and the soluble GC (Davies et al., 2013; Kean et al., 2002), and also by NPLP1-4, which causes a rise in cGMP via the Gyc76c rGC (Overend et al., 2012). cGMP stimulates tubule fluid secretion and increases transepithelial potential (Bijelic and O'Donnell, 2005; Dow et al., 1994b), but also activates the cognate serine/threonine cGMP-dependent kinases (cGKs) DG1 and DG2 (Kalderon and Rubin, 1989; MacPherson et al., 2004b; Osborne et al., 1997) at micromolar concentration (MacPherson et al., 2004b). DG1 is localised to the cytosol; by contrast, DG2 is membrane localised. Using transgenic flies bearing principal cell-targeted gain-of-function dg1 and dg2 constructs, it was demonstrated that DG1 transduces a cytosolic cGMP signal, whereas DG2 transduces the cGMP signal generated at the basolateral membrane (MacPherson et al., 2004b). Thus, each kinase, although with similar EC₅₀ values for cGMP (MacPherson et al., 2004b), transduces a localised cGMP signal. Moreover, dg1 is almost exclusively expressed in tubules and in hindgut (Chintapalli et al., 2007; Robinson et al., 2013) whereas dg2 is expressed throughout the fly and also regulates behaviour (Reaume and Sokolowski, 2009), so it is possible that these two cGKs have entirely different roles in tubule principal cells.

cAMP and cGMP have been shown to play a role in stellate cells (Kerr et al., 2004), although the endogenous pathways for cAMP and cGMP are not known for this tubule cell type. Recent research, however, has demonstrated that cGMP acting through DG1 (but not DG2) can inhibit transepithelial responses induced by both tyramine and D. melanogaster leucokinin (Ruka et al., 2013), both of which increase calcium signalling and chloride conductance (Blumenthal, 2003; Cabrero et al., 2013; O'Donnell et al., 1996; Radford et al., 2004; Terhzaz et al., 1999). Thus, a yet-unidentified inhibitory process for tyramine and D. melanogaster leucokinin signalling in stellate cells is cGMP/DG1-mediated.

Degradation of cGMP occurs by the cGMP-hydrolysing PDEs (cG-PDEs), whose activity can be regulated to maintain cGMP levels. For example, cG-PDE activity in tubules is depressed by Manduca sexta CAP2b (MacPherson et al., 2004a), a member of the capa neuropeptide family (Tublitz and Truman, 1985a). Thus, capa peptides regulate cGMP concentration in the principal cells by both generation and breakdown of cGMP.

Further regulation of cGMP signalling can occur as a result of interactions between the cGKs and the cG-PDEs. cGKs phosphorylate cG-PDEs (Francis et al., 2011), e.g. PDE5, thus modifying their activity. The D. melanogaster PDE5 orthologue (DmPDE5/6) is a cGMP-specific PDE (Day et al., 2005), although it is not known whether it is phosphorylated by either DG1 or DG2. However, DmPDE5/6 contains two consensus serine/threonine phosphorylation sites for cGK/PKA-KKRS and KRPS, as well as the regulatory GAF domains present in PDE5 (Davies and Day, 2007), so it is likely that DmPDE5/6 is regulated by cGK (and possibly PKA). Cross-talk between the cAMP and cGMP pathways may also occur as both DG1 and DG2 can be activated by 20 μmol l⁻¹ cAMP (MacPherson et al., 2004b). Also, some of the PDEs are dual specificity enzymes, capable of hydrolysing both cAMP and cGMP (e.g. PDE1 and PDE11), so this allows for further cross-talk between cAMP and cGMP under physiological conditions.

Although the only route for degradation of cyclic nucleotides is via the PDEs (Bender and Beavo, 2006), extrusion of cyclic nucleotides in some cell types also contributes to reduction of cyclic nucleotide levels. In tubule principal cells, DmPDE5/6 is expressed at the apical membrane, and controls cGMP efflux into the lumen (Day et al., 2006), suggesting that control of PDE levels in tubule principal cells is not only due to breakdown by PDEs. The transporters for cGMP are not yet known, although White has been shown to encode an ABC transporter for cGMP in tubule principal cells (Evans et al., 2008).

Calcium (Ca²⁺) is a ubiquitous second messenger molecule in all cell types and tissues, and in the tubule, it modulates fluid secretion rates, V-ATPase activity, and downstream signalling and ion transport events (Davies and Terhzaz, 2009).

Use of genetically encoded aequorin-based luminescent Ca²⁺ reporters specifically targeted to only tubule principal cells using the binary GAL4/UAS expression system (Brand and Perrimon, 1993) was first used to demonstrate in vivo Ca²⁺ signalling in defined populations of tubule cells (Kean et al., 2002; Rosay et al., 1997; Terhzaz et al., 2012). This work demonstrated that the Manduca sexta neuropeptide CAP2b, a member of the capa peptide family (Davies et al., 1995; Loi and Tublitz, 2004), stimulated a rise in intracellular (Ca²⁺) ([Ca²⁺]_i) in only principal cells, comprising a fast [Ca²⁺]_i spike followed by a slow [Ca²⁺] rise. The fast [Ca²⁺]_i rise was shown to occur via phospholipase Cβ and inositol 1,4,5-trisphosphate receptor (Pollock et al., 2003), and also involves Ca²⁺ release via the Golgi sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA) channel (Davies and Terhzaz, 2009). CAP2b was also shown to stimulate Ca²⁺ influx through principal cell plasma membrane Ca²⁺ channels – transient receptor-like (TRPL), L-type and cyclic nucleotide-gated (CNG) channels (Fig. 1) – which contributes to the slow CAP2b-stimulated [Ca²⁺]_i rise (Broderick et al., 2003; MacPherson et al., 2001; MacPherson et al., 2005). The physiological effects of CAP2b and endogenous D. melanogaster capa peptides, Drome-capa-1 and -2 (Davies et al., 2013), act similarly on D. melanogaster principal cells, and Drome-capa-1 was subsequently shown to modulate organellar Ca²⁺ signalling in the mitochondria (Terhzaz et al., 2006), and also the Golgi- and peroxisome-localised secretory pathway Ca²⁺/Mn²⁺-ATPases (Southall et al., 2006).

Luminescent and fluorescent Ca²⁺ reporters (Davies and Terhzaz, 2009) targeted to mitochondria of tubule principal cells also showed that Drome-capa-1 triggered mitochondrial Ca²⁺ uptake via the mitochondrial calcium uniporter at the apical membrane, resulting in activation of these mitochondria in the vicinity of the V-ATPase (Terhzaz et al., 2006). Drome-capa-1-activated mitochondria increase ATP production, driving proton pumping of the V-ATPase complex (Terhzaz et al., 2006). However, the Drome-capa-1 receptor resides on the principal cell basolateral membrane, so Drome-capa-induced activation of the apical mitochondria demonstrates a novel spatio-temporal mode of action.

Overall, this highly complex repertoire of intracellular and plasma membrane calcium channels, ATPases and intracellular organelles (reviewed in Davies and Terhzaz, 2009) results in increased [Ca²⁺]_i by capa peptides (Kean et al., 2002). This [Ca²⁺]_i rise can trigger the activation of Drosophila calcium/calmodulin-sensitive nitric oxide synthase (DNOS) (Regulski and Tully, 1995), as Drome-capa-1 increases principal cell [Ca²⁺]_i from 87 to 255 nmol l⁻¹, which is close to the EC₅₀ for calcium activation of DNOS. This results in the production of NO and subsequent increased cytoplasmic [cGMP] (Kean et al., 2002) due to NO-induced activity of the soluble GC (Davies et al., 1997) (Fig. 1).

Further cross-talk can occur between the cGMP and Ca²⁺ signalling pathways, as it has been demonstrated that exogenous cGMP induces [Ca²⁺]_i in tubule principal cells, which is abolished by plasma membrane Ca²⁺ channel blockers (MacPherson et al., 2001). Also, targeting a DNOS transgene to only principal cells shows potentiation of capa-peptide-induced [Ca²⁺]_i signals, potentially via Ca²⁺ influx through plasma membrane CNG channels (Broderick et al., 2003). Thus there is positive feedback regulation of Ca²⁺ signalling by NO/cGMP, to further increase [Ca²⁺]_i, which then activates apical mitochondria and, subsequently, the V-ATPase. This may further explain the activation of mitochondrial ATP production by cGMP (potentially by increasing Ca²⁺ influx through CNG channels), thus, activating the V-ATPase and driving fluid secretion (Fig. 1).

Tubules are immune tissues and express all components of the Imd (McGettigan et al., 2005) and Toll (Chintapalli et al., 2007; Robinson et al., 2013) innate immunity pathways. Acutely dissected tubules can bind and internalise lipopolysaccharide, a component of gram-negative bacterial coat, and can mount a significant bacterial killing response (McGettigan et al., 2005) via production of antimicrobial peptides. Antimicrobial peptides produced by the tubules are derived from both the Imd and Toll signalling pathways, and include diptericin, attacin, cecropin, metchnikowin, defensin and drosomycin (McGettigan et al., 2005; Tzou et al., 2000), which may be secreted into either the haemolymph or into the gut. Tubules sense immune challenge by expressed peptidoglycan recognition proteins (PGRPs) (Charroux et al., 2009; Kurata, 2010). For example, PGRP-LE, which binds diaminopimelic-acid-containing peptidoglycan from gram-negative bacteria, is localised to the tubule principal cells (Kaneko et al., 2006). Interestingly, a fragment of PGRP-LE consisting only the PGRP domain acts as an extracellular receptor, possibly allowing tubules to sense pathogen invasion from the haemocoel. Tubules may also express transporters that can transport peptidoglycan fragments across the membrane (Kaneko et al., 2006).

Activation of Imd-associated antimicrobial peptide gene transcription occurs via the action of the NF-kB orthologue, Relish. Relish expression is enriched in epithelia and in fat body, both in adult and larval stages (Chintapalli et al., 2007; Robinson et al., 2013). Immune challenge causes phosphorylation and endoproteolytic cleavage of Relish, resulting in a nuclear-translocated Rel homology domain, and the Iκb-like domain, which remains cytoplasmic (Stoven et al., 2003). Nuclear translocation of Relish is utilised to assess Relish activity in vivo, in intact tubules. Under resting conditions, Relish is mainly localised to the tubule basolateral membrane, with some localisation to the large nuclei of principal cells (Fig. 2A). Treatment of tubules with gram-negative bacterial coat peptidoglycan and subsequent activation of the Imd pathway promotes complete nuclear translocation of Relish (Fig. 2A). Nitric oxide activates the Imd pathway and antimicrobial peptide expression in tissues including the tubules (McGettigan et al., 2005). Moreover, NO-stimulated diptericin expression in the principal cells is dependent on functional soluble GC (Davies and Dow, 2009). Here, we show that cGMP regulates Relish translocation in a dose-dependent manner (Fig. 2C). Nanomolar concentrations of cGMP (1–100 nmol l⁻¹) induce Relish translocation to the nucleus. By contrast, higher (μmol l⁻¹) concentrations of cGMP block nuclear localization of Relish, and tubules pre-treated with 100 μmol l⁻¹ cGMP prior to peptidoglycan treatment show reduced translocation of Relish to the nucleus. Thus, a saturating concentration of cGMP prevents nuclear translocation of Relish, even upon immune challenge. cGMP thus modulates Imd pathway signalling. It remains to be resolved whether the cGKs modulate Imd activity downstream of NO and cGMP (Davies et al., 2009).

Salt, or osmotic (ionic), stress on the whole organism is sensed and transduced by epithelial tissue, notably the tubule and gut. A microarray study on whole-fly responses to salt (NaCl) stress revealed gene changes across distinct functional gene groups, where the most significantly altered gene groups were those most highly expressed in tubules or in gut (Stergiopoulos et al., 2009). The transcription factor NFAT (nuclear factor of activated T-cells), which is enriched in tubules (Chintapalli et al., 2007; Robinson et al., 2013), also plays a role in salt tolerance (Keyser et al., 2007). Thus, the tubule is a key tissue for salt tolerance in the fly. Interestingly, salt stress induces an early immune gene transcriptional programme in epithelial tissue (Stergiopoulos et al., 2009).

More recently, we have shown that induction of salt stress in whole adult flies causes nuclear translocation of Relish in tubule principal cells (Overend et al., 2012). Salt stress also induces significantly increased expression of Imd pathway antimicrobial peptides diptericin and cecropin, and of the Toll pathway antimicrobial peptide drosomycin, in isolated tubules, thus validating the whole-fly salt stress microarray dataset.

Furthermore, the endogenous D. melanogaster peptide NPLP1-VQQ (NLGALKSSPVHGVQQ) (Baggerman et al., 2005) was shown to be a ligand for the tubule-enriched Gyc76c rGC, where a small but significant increase in cGMP levels was demonstrated using a peptide library screen in D. melanogaster S2 cells for activators of Gyc76c. Drosophila melanogaster rGCs have remained without identified ligands until now, and NPLP1-VQQ is the first putative ligand identified for a Drosophila rGC. NPLP1-VQQ also has a physiological role: it stimulates fluid secretion rates by intact tubules. NPLP1-VQQ/Gyc76c activation also results in Relish nuclear translocation and increased diptericin expression in tubule principal cells. It is possible that only a small increase in cGMP is observed upon NPLP1-VQQ stimulation of Gyc76c (Overend et al., 2012), as this is required to be within the stimulatory [cGMP] range for Relish translocation (Fig. 2). Finally, this work showed that targeted knockdown of Gyc76c in only tubule principal cells prevents cGMP production, localisation of Relish to the nucleus and induction of diptericin expression. Thus, Gyc76c and cGMP signalling modulates Relish/Imd pathway activation in tubule principal cells.

As Gyc76c/cGMP is a modulator of the Imd pathway (Overend et al., 2012), it was likely that NPLP1-VQQ/Gyc76c activation would enhance survival of immune challenge. However, targeted knockdown of Gyc76c to tubule principal cell showed that flies are not compromised for immune challenge, but rather for salt stress. cGMP has been shown to have both immune and stress-associated transcriptional targets by microarray (transcriptome) analysis of cGMP-treated tubule samples (Davies et al., 2012). Thus, cGMP signalling, and, therefore, pathway components (e.g. rGCs), modulate both immune and stress-responsive genes.

It has been recently demonstrated that the Malpighian tubule is a major sensor for oxidative stress for the whole fly, as it is enriched for antioxidant genes (Terhzaz et al., 2010b). Production of reactive oxygen species (ROS) as a byproduct of ATP production by a very metabolically active tissue that is packed with mitochondria means that the tubule must be able to detect ROS and oxidative stress. Mitochondria in tubule principal cells are placed in either an apical or basal membrane location, and are differentially activated to produce ATP in response to a neuropeptide stimulus, i.e. Drome-capa-1 (Terhzaz et al., 2006). Mitochondria are thus critical determinants of tubule function, and therefore of organismal survival. Mitochondria are also key organelles for Ca²⁺ homeostasis in tubules and in all other biological systems (Rizzuto et al., 2012). As mitochondria are so important for cellular function, dysregulation of mitochondrial Ca²⁺ in humans leads to disease states (Duchen et al., 2008), and D. melanogaster is successfully being utilised to produce new leads for diseases of oxidative stress, including neurodegeneration (Jaiswal et al., 2012).

Interestingly, work in D. melanogaster tubules has demonstrated that specific signalling genes expressed in only tubule principal cells can alter organismal susceptibility to oxidative stress. Modulation of principal cell inositol phosphate signalling, specifically via inositol 1,4,5 trisphosphate 3-kinase (IP₃K), using the GAL4/UAS system has been shown to increase ROS production. Precise targeting of either gain-of-function or loss-of-function (RNAi) D. melanogaster IP₃K-1 constructs to only tubule principal cells showed that IP₃K-1 increased H₂O₂ production, pro-apoptotic caspase-9 activity and mitochondrial membrane potential. IP₃K-1 also significantly increased mitochondrial Ca²⁺ under oxidative stress conditions, leading to apoptosis. Flies in which IP₃K-1 is overexpressed in only tubule principal cells are also significantly more susceptible to oxidative stress. Intriguingly, IP₃K-1 modulates epithelial cell apoptosis without involvement of bcl-2-type proteins.

Recently, other tubule principal cell candidate genes for oxidative stress tolerance have been discovered. Signalling by D. melanogaster insulin-like peptides (DILPs) has previously been successfully investigated in terms of neurobiology, nutritional status and ageing (Birse et al., 2011; Nassel, 2012; Partridge et al., 2011; Söderberg et al., 2012). There are seven DILPs, and DILP-5 has been shown to be expressed in tubule principal cells, together with the receptor for DILPs, Drosophila Insulin Receptor, dINR (Söderberg et al., 2011). Knockdown of DILP5 in tubule principal cells increases survival to oxidative stress induced by paraquat feeding. Interestingly, knockdown of DILP5 in larval principal cells results in increased lifespan. It is possible that DILP signalling modulates activity of the mitochondrial Mn²⁺ superoxide dismutase (SOD), as knockdown of sod2, encoding this SOD in only principal cells, reduces oxidative stress tolerance.

Other work involving mitochondrial function demonstrated that defects in mitochondrial proteins, e.g. the ADP/ATP translocase (encoded by sesB, which is notably enriched in tubule), cause reduced cytosolic and mitochondrial calcium signals, as well as reduced fluid secretion by intact tubules (Terhzaz et al., 2010a). Furthermore, sesB mutant tubules contain only 10% of the ATP levels but five times the ROS levels of control tubules, and sesB mutant flies display reduced resistance to oxidative stress. Thus, mitochondrial function in tubule principal cells is crucial for oxidative stress tolerance.

Most recently, cAMP signalling has been implicated in oxidative stress tolerance in Drosophila. Mammalian PDE8 is a cAMP PDE that is phosphorylated by PKA, resulting in enhanced enzyme activity (Brown et al., 2012). The PDE8A isoform is localised to mitochondria (Tsai and Beavo, 2011), and in D. melanogaster, DmPDE8 is the orthologue of mammalian PDE8A (Davies and Day, 2007). In a recent study, PDE8A was shown to have a novel binding partner, Raf-1, which allows modulation of downstream extracellular signal-regulated kinase (ERK) signalling via phosphorylation of ERK by Raf-1 kinase (Brown et al., 2013). DmPDE8 is abundantly expressed in epithelia, including the tubules, and PDE8 deletion mutants show reduced phospho-ERK signals. Critically, this biochemical interaction was shown to be relevant to oxidative stress tolerance in the organism, especially given the putative localisation of PDE8 in mitochondria as well as the role of the tubules in stress resistance. PDE8 deletion mutants are significantly more susceptible to oxidative stress imposed by either H₂O₂ or paraquat feeding, compared with parental controls (Brown et al., 2013). This work demonstrates for the first time in insects that cAMP and ERK signalling mediates oxidative stress tolerance.

Capa peptides stimulate both cGMP and Ca²⁺ signalling, with several points for cross-regulation between the signalling pathways (Davies et al., 2013). Capa peptides modulate diuresis in all insects tested, and thus such complexity in downstream signalling may be necessary for maintenance of fluid homeostasis. Fluid homeostasis is critical for survival, as in many environments, desiccation is a major threat to terrestrial organisms. This is particularly relevant to insects, which have a small size and thus a large surface area to volume ratio. However, insect osmoregulatory systems may be adapted for water conservation, and insects may survive desiccation by regulating fluid secretion (excreted water loss) by the tubules, an energetically efficient strategy. Drosophila melanogaster has been used effectively in studies of insect desiccation tolerance: excreted water loss rates are reduced in desiccated D. melanogaster (Folk and Bradley, 2003), and work with wild desiccation-resistant Drosophila populations showed single-feature polymorphisms in several gene groups. These included tubule ion transport genes, e.g. sodium and potassium channels/transporters and chloride transporters. Intriguingly, genes associated with Drome-capa-1 signalling pathways in tubule principal cells were also implicated, including 1,4,5-trisphosphate receptor, TRPL and DNOS (Telonis-Scott et al., 2012).

The D. melanogaster capa receptor, capaR, is a G-protein coupled receptor and is encoded by gene CG14575 (Iversen et al., 2002; Park et al., 2002; Terhzaz et al., 2012). CapaR has been identified in other insect species, including A. gambiae (Olsen et al., 2007; Pollock et al., 2004) and R. prolixus (for RhoprCAPA-α2) (Paluzzi et al., 2010). In the lepidopteran species M. sexta, the Drome-capa-1 orthologue MAS-capa-1 stimulates heart rate (Tublitz and Truman, 1985a; Tublitz and Truman, 1985b) and also hindgut contraction (Tublitz et al., 1992), suggesting that M. sexta capaR is expressed in heart and in hindgut. Furthermore, in R. prolixus, capaR is expressed in anterior midgut as well as the tubules (Paluzzi et al., 2010). This is not the case in D. melanogaster, where the capaR gene is almost uniquely expressed in both the adult and larval tubules (Chintapalli et al., 2007; Robinson et al., 2013; Terhzaz et al., 2012) (Fig. 3A). Furthermore, Drome-capa-1 does not modulate heart rate in D. melanogaster (Loi and Tublitz, 2004), nor does it affect calcium signalling in adult midgut (S.T. and S.A.D., unpublished). Thus in D. melanogaster (and perhaps in Diptera), capa signalling only affects the tubules, which are key tissues for fluid homeostasis.

CapaR is localised at the tubule principal cell basolateral membrane (Terhzaz et al., 2012) (Fig. 3B) and activated by both Drome-capa-1 and -2 to elevate [Ca²⁺]_i (Terhzaz et al., 2012). Tubules from transgenic principal cell-specific RNAi capaR knockdowns showed that both Drome-capa-1-stimulated [Ca²⁺]_i and fluid secretion are abolished (Terhzaz et al., 2012), (Fig. 3C,D). Thus capaR modulates Drome-capa-1-induced [Ca²⁺]_i and fluid secretion – and, therefore, fluid homeostasis – so there was a possibility that capaR modulates organismal responses to desiccation stress. The capaR RNAi D. melanogaster lines were assessed for desiccation tolerance using survival assays, in comparison with parental lines. The data clearly showed that significantly reduced expression of capaR in only tubule principal cells was sufficient to prolong survival to desiccation compared with control lines, suggesting that reduced Drome-capa-modulated diuresis by the tubule, and associated reduction of fluid loss, is crucial for desiccation tolerance (Fig. 3E).

Other neuropeptides, such as D. melanogaster Short neuropeptide F and tachykinin, also have roles in desiccation tolerance. Interestingly, downstream components of tachkykinin and insulin signalling in the tubule principal cells also modulate desiccation tolerance. Overexpression of ribosomal S6 kinase in only principal cells reduces survival under desiccation conditions, whereas targeted expression of a dominant negative S6 kinase results in increased desiccation tolerance (Söderberg et al., 2011).

Thus there are multiple, complex signalling pathways in tubule principal cells that operate for this stress response alone.

There is still much to be learned about the regulation of cell signalling pathways by individual signalling components in insect epithelia. Furthermore, there are multiple layers of control, including localisation and compartmentalisation (Ahmad et al., 2012; Wong and Scott, 2004), post-translational modification, and transcriptional regulation and control. For cGMP signalling, in particular, even in mammalian systems, these are recently discovered processes (Francis et al., 2011), and so further understanding of cGMP signalling in insect stress responses will have new and wide-ranging implications.

New roles in stress biology are also being discovered for Ca²⁺ signalling pathways, especially mitochondrial Ca²⁺ (Mammucari and Rizzuto, 2010). Performing such fundamental work in insects will also reveal new mechanisms in human stress signalling (Becker et al., 2010; Jaiswal et al., 2012). However, the impact of environmental stress on agriculturally friendly insects, as well as on insect pests, is very much on the current agenda worldwide, and so understanding the molecular mechanisms of insect stress tolerance will be invaluable for deciphering insect survival in response to environmental stressors.

We thank Dr A. J. Dornan for image generation.