Organellar calcium signalling mechanisms in Drosophilaepithelial function

Calcium (Ca²⁺) is a ubiquitous second messenger molecule in all cell types and tissues, and plays a direct role in health and disease(Berridge et al., 2000). As such, research into calcium signalling has been an intensive area of investigation for some decades, and current known mechanisms of calcium signalling in many cell types, notably neuronal, endothelial and epithelial cells, include the calcium signals associated with spatial and temporal resolution within the cell (Knot et al.,2005; Petersen et al.,2005). Although much of biology is concerned with excitable cells,epithelial cells are critically important to the maintenance of life, and the understanding of the fundamental aspects of calcium signalling in relation to epithelial tissue function is of significant interest. Calcium signalling is well understood in the pancreatic acinar cell(Petersen and Tepikin, 2008),and the discovery of the family of transient receptor protein (TRP) channels in the context of renal function has widened our understanding of the functional role of calcium handling and signalling in tissue and organismal function (Hoenderop et al.,2003; Hsu et al.,2007; Mensenkamp et al.,2006). Use of transgenic mouse models has demonstrated physiological roles for calcium and phosphate regulation of intestinal and renal function via the transcription factor, klotho(Renkema et al., 2008), as well as the importance of the TRP channels in physiological function and disease (Nilius et al.,2007).

In an organotypic context, the use of genetic model organisms is necessary,as the physiological assay of cell or tissue `output' in transgenic or mutant animals allows precise analysis of gene and protein function in vivo. Whilst vertebrate models are clearly relevant in the biomedical arena, the use of Drosophila melanogaster allows greater scope for cell- and tissue-targeted genetic intervention. Also, signalling genes and proteins are structurally and functionally conserved across evolution, so the invertebrate context is not irrelevant to studies of general mechanisms of cell signalling. Furthermore, although Drosophila provides an excellent developmental model, it also comes with several functional physiological phenotypes in both adult and larval stages, including that of epithelial transport. The most well-studied tissue in this regard is the Drosophila Malpighian tubule, a fluid-transporting organ that is critical in osmoregulation and ion homeostasis (Coast, 2007),detoxification (Torrie et al.,2004; Yang et al.,2007) and immunity of the fly(Dow and Davies, 2006; Kaneko et al., 2006).

Fluorescent calcium-binding dyes, e.g. the ratiometric Fura-2 and single-wavelength fluors like fluor-3, Oregon Green and Calcium Green, have been used extensively to measure cytosolic calcium signals in vertebrate cells, with much success (Meldolesi,2004). However, there are drawbacks to the use of fluorescent calcium-binding dyes, including the effects of pH on such compounds, cell damage induced by delivery of the fluorophore (although membrane-permeant variants of these compounds now exist so as to allow for non-invasive delivery) and non-retention of fluorophores by transporting epithelia, e.g. application of Fura-2 to Drosophila Malpighian tubules results in rapid extrusion of the fluorophore by organic anion transporters (J. A. T. Dow and T. Cheek, unpublished). Finally, accurate measurements of calcium in intracellular compartments are extremely limited using fluorescent dyes. So although such dyes are still very much used in calcium signalling research,other calcium-binding probes, based on calcium-binding proteins, are preferable.

The development of protein-based calcium probes has been such an important leap forward for the life sciences that the 2008 Nobel Prize for Chemistry was awarded for the discovery of Green Fluorescent Protein (GFP) to Osamu Shimomura, Martin Chalfie and Roger Tsien. The first protein indicator for calcium measurements was calcium-binding aequorin purified from the jellyfish, Aequorea victoria (Shimomura et al., 1962; Shimomura et al.,1963). Aequorin is stable and non-toxic in cells and reversibly binds calcium ions in a linear concentration range in the presence of a chromophoric ligand, coelenterazine and molecular oxygen. Binding of calcium causes de-stabilisation of the aequorin–coelenterazine complex and results in the production of CO₂ and emission of light in the blue,UV range as luminescence, allowing quantitative measurements of calcium concentration (Fig. 1). Aequorin is then re-generated in a slow reaction upon dissociation of calcium and coelenterazine binding. Until the discovery of the aequorin gene, aequorin protein was used in cell extracts or was microinjected into cells. The discovery of the coding sequence for aequorin(Inouye et al., 1985) allowed aequorin to be expressed in cells and in tissues from different organisms,including plants (Knight et al.,1993). Following this, the first transgenic animals for apoaequorin were generated, which were transgenic Drosophila in which cell-specific calcium measurements were made in intact tissue, the Malpighian tubule (Rosay et al., 1997). Aequorin is now also used in stable vertebrate and insect cell lines as a functional read-out for screening and high-throughput applications for proteins of biomedical interest and in drug discovery, e.g. G-protein coupled receptor and tyrosine kinases (Dupriez et al., 2002; Le Poul et al.,2002; Torfs et al.,2002).

A further development of the aequorin technology resulted in targeted aequorin to intracellular compartments(Pinton et al., 2007; Robert et al., 2000). For this purpose, aequorin constructs were modified with targeting sequences for:endoplasmic reticulum [(ER) the ER-retention signal KDEL or C_H1 domain of Igγ2b heavy chain](Kendall et al., 1992a; Montero et al., 1995);mitochondria (subunit VIII of human cytochrome c oxidase)(Rutter et al., 1993); Golgi[transmembrane portion of sialyltransferase (ST), a resident protein of the lumen of the trans-Golgi and trans-Golgi network (TGN)](Pinton et al., 1998); and for the nucleus (nuclear localisation signal of glucocorticoid receptor)(Brini et al., 1994). Due to the very high calcium concentration in the ER, the aequorin for the ER needs to be of lower calcium sensitivity and so is engineered with a point mutation of Asp119-Ala (Kendall et al.,1992b). All of these targeted aequorins have been used very successfully in cell lines to reveal that all cellular organelles thus examined can act as dynamic calcium stores. Most recently, a role for peroxisomes in calcium signalling has been demonstrated using both cytosolic aequorin (Southall et al.,2006) and aequorin targeted to peroxisomes(Lasorsa et al., 2008) using the yeast peroxisomal targeting sequence 1 (PTS1)(Wanders and Waterham, 2006). Some of these targeted aequorin reporters have been successfully used in an organismal context: mitochondrial and nuclear aequorin in transgenic Drosophila (Terhzaz et al.,2006); and cytosolic aequorin in transgenic mice(Yamano et al., 2007).

In A. victoria, aequorin is associated with GFP, and green bioluminescence is emitted from GFP upon calcium binding by aequorin, due to bioluminescence resonance energy transfer (BRET) between the aequorin and GFP. Thus, expression of a GFP:aequorin fusion via a transgene results in increased light emission upon calcium binding compared with aequorin alone,and this has been utilised to generate a novel aequorin variant,GFP–aequorin (Baubet et al.,2000). GFP–aequorin has been successfully targeted to mitochondrial matrix, ER, synaptic vesicles and to the postsynaptic density of mammalian neurons (Rogers et al.,2005); and has now been used in transgenic Drosophila to demonstrate calcium transients in mushroom bodies of adult brain(Martin et al., 2007). Transgenic mice with the GFP–aequorin construct expressed in the mitochondrial matrix have also been generated(Rogers et al., 2007), which allow the analysis of calcium signals during various physiological states,including waking/sleeping. In order to improve detection of light from deep areas in mammalian tissues, the GFP–aequorin probe has also been engineered by fusing either enhanced yellow fluorescent protein (Venus) or monomeric red fluorescent protein (mRFP1) to aequorin. Light transmission through skin and thoracic cage is enhanced by using the Venus-aequorin probe whereas the emission spectrum by mRFP1-aequorin allows the detection of calcium signals in brain tissue (Curie et al., 2007).

Use of calmodulin-based fluorescent reporters as calcium probes has also been of immense value in calcium signalling research and the extremely rapid pace of development of novel probes over the last decade in the fields of cell biology, signalling, biophysics and bioengineering has meant that this area is currently very well reviewed (Chudakov et al., 2005; Mank and Griesbeck,2008; Miyawaki et al.,2005; Rudolf et al.,2003). The majority of such protein-based fluorescent reporters involve GFP from A. victoria. However, fluorescent reporters are also being developed from endogenous fluorescent molecules from other organisms including Anthozoa, allowing monitoring of red fluorescence, e.g. dsRed(Verkhusha and Lukyanov,2004).

Engineering of GFP-derivatives, e.g. cyan fluorescent protein (CFP) or yellow fluorescent protein (YFP), fused with calmodulin and other calcium-binding proteins, i.e. M13, allows a correlation between calcium binding and changes in FRET (fluorescence resonance energy transfer)(Fig. 2) [reproduced with permission from Rudolf et al. (Rudolf et al., 2003)]. The cameleons(Truong et al., 2001; Truong et al., 2007)(Fig. 2Ba) are composed of CFP–calmodulin fused with YFP–M13 peptide. Calcium binding causes a conformation change in the calcium-binding proteins, with a subsequent change in FRET. Camgaroos are based on YFP–calmodulin fusions(Baird et al., 1999; Griesbeck et al., 2001), where calcium binding leads to increased YFP fluorescence. The cameleons and camgaroos have been successfully used in vivo, notably in transgenic flies in the monitoring of neural activity(Fiala et al., 2002; Reiff et al., 2005; Yu et al., 2003). Camgaroo2(Griesbeck et al., 2001) has also been used successfully in hippocampal slices(Pologruto et al., 2004) and also in transgenic mice under tetracycline (TET) control(Hasan et al., 2004). Pericams based on calmodulin/M13 fusion with circularly permuted GFP derivatives [YFP or enhanced green fluorescent protein (EGFP)] are activated by calcium-binding, which leads to changes in fluorescence(Nagai et al., 2001)(Fig. 2Bc). Proteins that are circularly permuted have their natural termini joined, resulting in a circular protein, which can be reconfigured to create new C- and N-termini. The permuted protein may exhibit useful altered characteristics such as reduced substrate binding or can be used to generate fusion proteins by attaching a second polypeptide to the newly created termini. Pericams include inverse pericam (becomes dimmer in the presence of calcium), flash pericam (becomes brighter with calcium) and ratiometric pericam (exhibits calcium-dependent changes in excitation wavelength and allows quantitative measurements of calcium). In Drosophila, mitochondrial-targeted pericam (`Mitycam')has been used to image mitochondrial calcium in the Malpighian tubule principal cells (Terhzaz et al.,2006) whereas inverse pericam has been successfully used to image calcium changes in mice (Hasan et al.,2004). Most recently, improved pericams have been developed, which provide brighter fluorescence on calcium binding(Souslova et al., 2007).

In addition to the calmodulin/GFP probes described, newly developed calcium-binding fluorescent probes that show very fast kinetics and big fluorescent changes on calcium binding include one based on troponin C(Mank et al., 2006), which has also been used in vivo, in both mice and Drosophila(Garaschuk et al., 2007; Heim et al., 2007; Reiff et al., 2005). Interestingly, while ALL the protein-based calcium probes described here have been used in cell lines, few (as already noted) have been used in organisms. Furthermore, it has been difficult to compare the efficacy of different calcium reporters in vivo, as all such studies are carried out in different organisms, utilising different agonists, with different phenotypes as end-points. In a heroic study which compared the efficiencies of 10 protein-based reporters in transgenic Drosophila based on the fluorescence changes in larval pre-synaptic boutons(Reiff et al., 2005), it was found that flash pericam and the camgaroos did not report a change in neural activity in the larvae although Yellow Cameleon 3(Griesbeck et al., 2001) and pericam with EGFP (Nagai et al.,2001) showed linear changes that correlated with neural activity with a good level above background. Thus, although there are many protein-based calcium indicators now available, the real utility of these lies in their use in vivo, which ultimately depends on how such indicators work in the organismal, and not cell line, context.

The use of protein-based calcium probes allows expression in specific cells and tissues in vivo via targeted expression of transgenes. In Drosophila, the GAL4-UAS system(Brand and Perrimon, 1993),which has been further developed to allow conditional expression(McGuire et al., 2004), is the system of choice for targeted expression of transgenes via cell- or tissue-specific GAL4 drivers. Genes of interest are cloned downstream of the GAL4 promoter, Upstream Activating Sequence (UAS) and transgenic lines generated. The transgenes are expressed in the progeny of crosses between the GAL4 lines and the UAS lines, which can be maintained as stable lines if the transgenes have no deleterious effects on the flies. In order to investigate calcium signalling using aequorin in Drosophila,UAS–apoaequorin flies were generated(Rosay et al., 1997). These flies, the first transgenic animals for a calcium reporter, were used to monitor calcium signals in live intact tissue and provided the first in vivo measurements of cytosolic calcium concentration([Ca²⁺]cyt) in Drosophila Malpighian tubule(Rosay et al., 1997) and brain(Rosay et al., 2001). Malpighian tubules are free-floating, single-cell thick organs connected to the hindgut (see Fig. 3F). Each tubule is 2 mm×35 μm and is composed of two main cell types: Type I(principal cells) and Type II (stellate cells). Other cell types also exist,including bar-shaped cells and tiny endocrine cells(Sozen et al., 1997). Targeting of apoaequorin to different cells and regions of the tubule using different GAL4 drivers (Sozen et al.,1997) (Fig. 3A–J), reconstitution with coelenterazine, and measurement of luminescence in intact tubules showed that only principal cells in the main, fluid-transporting segment responded with increased [Ca²⁺]cyt to the neuropeptide CAP_2b (Fig. 3K–P). CAP_2b is a member of the capa family of peptides, including capa-1 and capa-2(Predel and Wegener, 2006),which all mobilise calcium, nitric oxide, the cyclic nucleotide cGMP and stimulate fluid transport in tubules with similar kinetics(Davies et al., 1995; Kean et al., 2002). This work also demonstrated that Drosophila leucokinin(Terhzaz et al., 1999)increased [Ca²⁺]cyt in only stellate cells. Leucokinins modulate fluid transport via calcium signalling in many insects(Gade, 2004), including the disease vectors Aedes aegypti(Beyenbach, 2003), Anopheles stephensi and Anopheles gambiae(Radford et al., 2004). Therefore, peptidase-resistant leucokinins are potential lead compounds for diptericides (Teal et al.,1999).

The cell-specific stimulation of the tubule by the capa peptides and the leucokinins is due to localisation of the respective receptors: the Drosophila leucokinin receptor is expressed only in tubule stellate cells (Radford et al., 2002),and the capa receptor is localised to tubule principal cells (S.T.,unpublished). The use of targeted aequorin for calcium measurements thus allowed `cell-identified' approach to calcium signalling studies.

CAP_2b acts via phospholipase Cβ (PLCβ),inositol 1,4,5-trisphosphate (IP₃) and inositol 1,4,5-trisphosphate receptor (IP₃R) to generate [Ca²⁺]cyt and increased fluid transport (Pollock et al.,2003). Drosophila leucokinin also acts viaIP₃ (Radford et al.,2002) and IP₃R(Pollock et al., 2003). As IP₃R is associated with the ER(Mikoshiba, 2007), this suggests that calcium mobilisation from the ER can modulate calcium homeostasis in principal and stellate cells.

The aequorin work also demonstrated that principal and stellate cells utilise ER-associated calcium differently from each other (see Fig. 4). This was the first demonstration that different cells in the same intact tissue respond differentially to the same pharmacological compound, suggesting necessary caution in interpretation of data from whole tissue pharmacology, whether in insects or vertebrates. Fig. 4shows that thapsigargin, which causes increased cytosolic calcium viaan inhibition of the ER Ca²⁺-ATPase and therefore prevents uptake of calcium into the ER, does cause a rise in cytosolic calcium in both principal and stellate cells, as well as increased fluid transport(Rosay et al., 1997). However,in the absence of extracellular calcium, which unmasks the `run-down' of[Ca²⁺]cyt due to extrusion of calcium from the ER in the absence of re-uptake, only the principal cells show sensitivity to this process. Thus,the principal and stellate cells show differentiated responses to thapsigargin and may utilise different calcium stores to maintain calcium homeostasis.

Therefore, although both principal and stellate cells utilise ER calcium,it is clear that the increased IP₃-induced [Ca²⁺]cyt in principal cells cannot be due solely to ER, as there is a dependence on calcium influx from the extracellular medium; as well as the existence of non-ER, hormone-sensitive organellar calcium pools.

The impact of extracellular calcium on [Ca²⁺]cyt and fluid transport by the Malpighian tubule is dramatic. Reduction of external calcium from 4 mmol l^–1 to 0 mmol l^–1 abolishes CAP_2b-stimulated [Ca²⁺]cyt and completely prevents CAP_2b-stimulated fluid transport; as well as abolishing the thapsigargin response in principal cells(Rosay et al., 1997). Thus,tubule function requires extracellular calcium influx into principal cells via plasma membrane channels.

Voltage-gated calcium channels, including l-type channels(Calin-Jageman and Lee, 2008),were once considered unique to excitable cells; however, studies in vertebrates have provided evidence for functional l-type calcium channels in non-excitable cells, including renal cells(Hayashi et al., 2007). The Drosophila orthologues of vertebrate l-type calcium channel α subunit are Dmca1A (cacophony, previously nightblind, CG1522)(Smith et al., 1996) and Dmca1D (CG4894) (Zheng et al.,1995). Tubules express both these channel subunits and l-type calcium channel antagonists, i.e. verapamil (a phenylalkylamine) and nifedipine (a dihydropyridine) inhibit CAP_2b-stimulated [Ca²⁺]cyt and fluid transport(MacPherson et al., 2001). Verapamil also partially abolishes the thapsigargin-stimulated[Ca²⁺]cyt in principal cells, suggesting that the thapsigargin-induced increase in [Ca²⁺]cyt is not necessarily ER-associated but rather is due to a phenylalkylamine-sensitive calcium channel. This supports the data in Rosay et al.(Rosay et al., 1997) and Fig. 4, which showed abolition of the thapsigargin-induced [Ca²⁺]cyt response when extracellular calcium was removed. Immunolocalisation studies indicate that α1 subunits of the L-type channels are localised to the basolateral and apical membranes of the principal cells in tubule main segment; the spatial distribution of the α1 subunits are associated with different affinities for the inhibitors of these calcium channels: high affinity binding of verapamil occurs at the basolateral surface whereas low affinity verapamil binding and binding of dihydropyridine occurs at the apical membrane. Interestingly, dihydropyridine also binds to the apical membrane of the initial segment in Malpighian tubule. This segment, especially in anterior tubules, is known to be a calcium store for the fly; and anterior tubules store 25–30% more calcium than posterior tubules(Dube et al., 2000b; Dube et al., 2000a); so the dihydropyridine-sensitive L-type channel may be involved in transepithelial calcium transport.

The entry of calcium is also modulated by a cyclic nucleotide-gated channel(CNG), encoded by cng, expressed in head and in tubules. Calcium entry into principal cells is enhanced by the cyclic nucleotide cGMP,resulting in increased [Ca²⁺]cyt and fluid transport. Furthermore,both the [Ca²⁺]cyt and fluid transport response induced by cGMP are verapamil- and nifediline-sensitive, and are abolished by a reduction of calcium in the extracellular medium(MacPherson et al., 2001). Treatment of tubules with 1 μmol cGMP also causes slow rise in mitochondrial calcium concentration, with a peak of 100–150 nmol l^–1 [Ca²⁺]_mitochondria occurring at 200 s post-stimulation (S.T., unpublished), suggesting that[Ca²⁺]_mitochondria accurately reflect changes in[Ca²⁺]cyt due to activity of CNG channels.

Although expression of cng is widespread throughout the adult fly, cng expression is 2–4-fold increased in Malpighian tubule,midgut and hindgut compared with the whole fly(www.flyatlas.org). This suggests that CNG channels contribute significantly to epithelial tissue function.

The TRP ion channel family constitutes 29 vertebrate TRP members,classified into seven subfamilies (TRPC, TRPV, TRPM, TRPML, TRPP, TRPA, TRPN). TRP was originally identified in D. melanogaster, from screens for phototransduction mutants (Montell et al.,1985). Drosophila TRP is classified within the C subfamily, which also includes the closely-related related TRPL(Xu et al., 1997) and TRPgamma channels (Xu et al.,2000), which can form heteromultimeric channels with TRP. TRP is activated via a Gq-coupled PLC mechanism in Drosophilaphotoreceptors and the TRPC, V and M subfamilies are regulated by lipid messengers including diacylglycerol (DAG) metabolites, polyunsaturated fatty acids (PUFAs) (Hardie, 2007),which are produced from DAG by putative DAG lipase. Recent work has identified the Drosophila DAG lipase, encoded by inaE, which regulates TRP/TRPL function in photoreceptors (Leung et al., 2008).

Whilst almost all work to date on Drosophila TRP channels has been confined to photoreceptors, investigation of TRP channel function in Drosophila Malpighian tubule provided the first evidence of TRP channel function in fluid-secreting epithelia, and also showed that the established model of TRP/TRPL interaction based on photoreceptor work, was not the sole model for TRP channel activation(MacPherson et al., 2005). Tubules express all known Drosophila TRPC channels, i.e. TRP, TRPL and TRPgamma, which are localised to the principal cells of the main segment. In contrast to photoreceptors, a trp null had no effect on CAP_2b-stimulated fluid transport, although CAP_2b-stimulated fluid transport rates were reduced in tubules from a trpl null, as well as a trpl;trp double mutant. Fluid transport rates reverted to normal on genetic rescue of the trpl null with a trpl transgene, and introduction of the trp null into the rescue background did not compromise the normal secretion rates.

Use of targeted aequorin to tubule principal cell using the c42 GAL driver showed a reduction of [Ca²⁺]cyt in the trpl mutant but not in a trp null. Furthermore, a correlation between TRPL protein levels and CAP_2b-stimulated fluid transport and calcium signalling was demonstrated in the trp and trpl mutants.

Taken together, our data suggested that tubules required TRPL but not TRP for normal epithelial function – unlike photoreceptors, which are dependent on functional TRP. Furthermore, it is probable that in Malpighian tubule, TRPL-TPPL homomultimers or TRPL/TRPgamma multimers are formed, which allow normal tubule function in the trp null background. A novel mutant of trp in which the calcium pore selectivity was modified by a Asp621 mutation, thus eliminating calcium permeation in vivo and resulting in altered photoresponses and retinal degeneration(Liu et al., 2007), was without effect on tubule secretion rates (P. Cabrero, S.A.D. and R. C. Hardie,data not shown). Finally, a key scaffolding component of the TRPL/TRP complex,INAD (Li and Montell, 2000),is not expressed in tubules (MacPherson et al., 2005), further confirming that TRP/TRPL complexes do not occur in Drosophila tubules. Thus, TRPC channels, particularly TRPL,are important for epithelial function.

Analysis of gene expression levels of other TRP family members in tissues of the whole adult fly or larvae was undertaken using the Drosophilatranscriptome database, www.flyatlas.org(Chintapalli et al., 2007)(Table 1). The data suggests that other novel TRP family members may be important for epithelial function,either in Malpighian tubule or in the gut – in addition to the roles already suggested for these TRP channels.

Although IP₃R is clearly associated with the ER(Mikoshiba, 2007), studies in many tissues, including epithelia, demonstrate alternative and complex arrangements for IP₃R isoforms(Bush et al., 1994). In rat kidney, the three IP₃R isoforms are expressed in different cell types, with type 1 and type 2 IP₃R expressed in the cytoplasm. However, type 3, although also cytoplasmic, is localised to the basolateral side (Monkawa et al., 1998). In rat colonic epithelium, however, IP₃R2 is localised to the nucleus and IP₃R3 in the cytoplasm at the apical region.

Attempts to localise Drosophila IP₃R in tubule cells have had only limited success, even though PLC-mediated calcium signalling via IP₃R has been established for tubule principal and stellate cells. Use of a rabbit polyclonal antibody against Drosophila IP₃R has demonstrated perinuclear staining of IP₃R in principal cells(Pollock, 2005), suggesting either a potentially nuclear localisation of IP₃R or association of IP₃R with ER around the nucleus.

To add to the complexity of calcium signalling via intracellular release calcium channels, tubules also express ryanodine receptor (RyR), an intracellular calcium release channel normally associated with the sarcoplasmic reticulum of cardiac and striated muscle(Wehrens et al., 2005). In Drosophila, RyR is encoded by a single gene, RyR-44F, which is most highly expressed in brain, thoracicoabdominal ganglion, crop and hindgut in the adult fly(www.flyatlas.org). Although expression levels of the gene are not high in tubules,immunocytochemical localisation of RyR in tubules reveals staining only in principal cells, with stellate cells excluded; the staining pattern is also reticular throughout the cell (Fig. 5). Functionally, ryanodine itself, a plant alkaloid, has an inhibitory effect on neuropeptide-stimulated fluid transport by tubules:CAP_2b-stimulated fluid tranport is sensitive to ryanodine at all concentrations between 10^–8 and 10^–3 mol whereas drosokinin-stimulated fluid transport is only affected by very high concentrations of ryanodine, especially 10^–3 mol. The differential effect of ryanodine on capa- or drosokinin-stimulated fluid transport may reflect differences in the store mobilisation of calcium elicited by the different peptides (i.e. capa: Golgi and mitochondria, see below; and drosokinin: ER) or by the fact that ryanodine is transported into the principal cells but not into stellate cells; or by the simplest explanation that stellate cells do not have ryanodine receptors, so the inhibition of drosokinin-stimulated transport at 10^–3 mol ryanodine is non-specific. At 10^–5 mol, however, ryanodine inhibits the CAP_2b-induced rise in cytosolic [Ca²⁺] by∼50% and reduces the drosokinin-induced cytosolic [Ca²⁺] rise by ∼30% (Pollock,2005).

However, given the lack of RyR in stellate cells, this suggests that the inhibitory effects of ryanodine occur via the IP₃R. Furthermore, treatment of intact tubules from itpr mutants(Pollock et al., 2003) with ryanodine causes a reduction in the already reduced rate of CAP_2b-induced fluid transport associated with loss-of-function of the IP₃R. Thus, although tubules do express RyR, the inhibitory effects of ryanodine on calcium signalling and fluid transport may be due to the impact of ryanodine on IP₃R; however, this does not rule out the possibility of an interaction between RyR and IP₃R, in principal cells at least. Although RyR is the dominant calcium release channel in striated muscle (Missiaen et al.,2000), functional interactions between RyR and IP₃R have been shown Chinese Hamster Ovary cells(George et al., 2003) and in pancreatic acinar cells (Gerasimenko et al., 2006).

The Drosophila sarco/endoplasmic reticulum Ca²⁺ ATPase(SERCA) is encoded by CG3725 (Ca-P60A) – a complex gene containing nine exons. Studies of Drosophila SERCA in cell lines confirm the importance of SERCA in calcium homeostasis(Raymond-Delpech et al., 2004)but in vivo studies reveal physiological roles of SERCA. Generation of conditional mutants for the SERCA gene allowed analysis of the role of store-derived calcium in neuromuscular physiology; and demonstrated that muscle excitability is dependent on SERCA-regulated, voltage-gated calcium channels and calcium-activated potassium channels(Sanyal et al., 2005). SERCA mutants have also been used to model cardiac dysfunction in the larval heart and to show that heart beat frequency is reduced in SERCA mutants(Sanyal et al., 2006).

In vertebrates, SERCA exists as three main isoforms, localised to different cell types and sub-cellular locations. SERCA1 in fast-twitch skeletal muscle and SERCA2a in cardiac muscle are located in the sarcoplasmic reticulum (SR),however, SERCA 2 and 3 are located in the ER of other cell types, although mitochondrial localisations may not be excluded(Misquitta et al., 1999). In the Drosophila tubules, SERCA has been localised to the Golgi apparatus, an observation reached via immunocytochemistry to SERCA(Fig. 6A) (similar staining is observed with Golgi-localised secretory pathway ATPase, SpoCkA)(Fig. 7) and viaexpression of epitope-tagged SERCA in principal cells, which shows an intracellular, vesicle-like localisation(Fig. 6B). Furthermore, RNAi of the SERCA gene in vivo demonstrates an impact of Golgi-localised SERCA on the IP₃-derived calcium spike and associated calcium influx in principal cells (Fig. 6C,D). Thus, IP₃-induced calcium signalling in the principal cell is dependent on calcium storage in the Golgi and release by SERCA.

In addition to SERCA and plasma membrane calcium ATPase (PMCA) pumps, the secretory pathway calcium ATPase (SPCA) comprises a third class of calcium pumps (Wuytack et al., 2002). SPCA pumps have an equal selectivity for transporting Mn²⁺ and Ca²⁺, unlike the SERCA pumps, and are insensitive to potent SERCA inhibitors such as thapsigargin (Sorin et al., 1997). Work in yeast has shown that SPCAs may be involved in regulation of Mn²⁺-mediated removal of superoxide radicals(Lapinskas et al., 1995). Expression of C. elegans SPCA in vertebrate cell lines has shown that SPCA can establish baseline cytosolic Ca²⁺ oscillations, without involvement of ER (Missiaen et al.,2001). Thus, SPCA may be involved in non-ER-associated calcium homeostasis.

In Drosophila, SPCA is encoded by CG7651 (SPoCk), and is most highly expressed in brain and in thoracicoabdominal ganglion(www.Flyatlas.org). However, SPoCk encodes three transcripts, SPoCk A-C, with SPoCk A and C (but not SPoCk B) being expressed in tubules. Generation of transgenic lines for all three transcripts and targeted expression of these transcripts in tubule principal cells revealed that each SPoCk isoform was differentially localised to intracellular compartments in vivo(Fig. 7): SPoCk A to the Golgi,SPoCk B to ER (under conditions of ectopic expression) and SPoCk C to peroxisomes. These data were confirmed by expression of each SPoCk isoform in Drosophila S2 cells (Southall et al., 2006) and suggested regulation of tubule function via novel calcium stores and SPCA.

Using c42 aequorin flies, SPoCk isoform expression was driven in principal cells, and cytoslic calcium levels assessed(Fig. 8A,B,D). SPoCk A, in Golgi, mediates the rapid rise in intracellular calcium induced by capa-1 and impacts on re-calcium entry (Fig. 8A). Also, RNAi to the SPoCk gene results in reduction of the rapid rise in intracellular calcium induced by capa-1(Southall et al., 2006). Interestingly, only SPoCk B affects IP₃-induced calcium signalling via the ER in Drosophila S2 cells but does not modulate calcium signals in vivo (Fig. 8B) confirming the lack of transcript B function in tubules. Disruption of the Golgi apparatus with brefeldin A(Fig. 8C) reduces capa-1 induced calcium signalling (Fig. 8D); confirming that in tubules, the Golgi is a major site of calcium release and homeostasis. Targeted expression of SPoCk A to tubule principal cells results in markedly increased capa-1 stimulated fluid transport (Fig. 8E) whereas expression of SPoCk-RNAi in principal cells abolishes capa-1 stimulated fluid transport (Southall et al.,2006). Thus, calcium signalling by Golgi-associated SPCA regulates fluid transport by the tubule.

The idea that only the ER can act as a dynamic calcium store is out-dated. The Golgi, nucleus, mitochondria and even peroxisomes are now known to modulate calcium homeostasis in cells, and to be critically involved in cell function. Furthermore, experimental measurements of calcium in endosomes/lysosomes, the nucleoplasm and nuclear envelope, and secretory granules are possible (Gerasimenko and Tepikin, 2005).

In tubules, although IP₃-mediated calcium signalling via ER in the principal and stellate cells has been established,substantial involvement of other organelles in calcium homeostasis is now known.

Mitochondria are key calcium stores, and whose activation state is coupled to calcium dynamics, and to cellular events including cell death(Rizzuto et al., 2000; Romagnoli et al., 2007; Szabadkai and Duchen, 2008). Moreover, key mitochondrial enzymes are calcium regulated, and so mitochondrial function is a calcium–sensitive process.

In tubules, the measurement of mitochondrial calcium in vivo has been achieved using aequorin targeted to mitochondria via subunit VIII of human cytochrome c oxidase(Terhzaz et al., 2006). This allowed the measurement of a slow, 100 s rise in mitochondrial calcium–dependent on uptake via the mitochondrial calcium uniporter – which was triggered by activation of PLC-mediated IP₃ signalling by capa-1. Furthermore, development of a transgene encoding mitochondrial-targeted pericam (`Mitycam') allowed imaging of mitochondrial calcium in individual principal cells in vivo,confirming the slow calcium rise measured with aequorin, deriving from populations of principal cells in several tubules. Imaging with Mitycam provided novel data on the mechanism of calcium sensing upon capa-1 challenge,and showed that apical and basolateral mitochondria in tubule principal cells respond differentially to capa-1, with the apical mitochondria being responsive to the neuropeptide. Assessment of mitochondrial activity via a potential-sensitive dye indicated that only apical mitochondria`sensed' capa-1, which also increases ATP production by the tubule. The dependence of ATP generation upon calcium increases in mitochondria was demonstrated by the reduction of capa-1-induced tubule ATP production upon inhibition of the mitochondrial calcium uniporter. Thus, a novel mode of regulation exists for capa-1-stimulated fluid transport, where a concerted action of cytosolic calcium, mitochondrial calcium and alteration in the activity profile of mitochondrial proteins (as assessed by proteomics) occurs to re-model the mitochondrial matrix (Fig. 9).

Transcript A of the Drosophila SPCA Ca²⁺/Mg²⁺ pump, SPoCk A, is expressed in the Golgi of tubule principal cells and significantly impacts on capa-induced calcium signalling and fluid transport (Fig. 8) (Southall et al.,2006). Thus, in tubule principal cells, the Golgi, and not ER, is the primary IP₃-inducible calcium store.

The novel localisations of SPoCk isoforms in principal cells also provide new insights into the role of novel calcium stores and pools in epithelial function. Analysis of SPocK C gene expression in regions of the tubule demonstrates that transcript C is expressed more highly in tubules from females compared with males. This is associated with the higher expression of SPocK C protein in female tubules compared with males(Fig. 10A). It is possible that SPoCk C is more abundant in female tubules due to the extra demands of oogenesis for calcium, as the chorion is calcium rich (35%)(Keramaris et al., 1991). The greater abundance of SPoCk C vesicles in females may be associated with the storage excretion of calcium in spherites throughout life providing an osmotically inactive bulk pool of calcium that could be mobilised during oogenesis.

Drosophila contain a pair of anterior tubules and a pair of posterior tubules in the abdomen, connected to the hindgut but free floating in the hemocoel. Whilst the entire tubule secretes calcium, the initial and transitional segments of the anterior tubules are capable of transporting the entire animal's Ca²⁺ content in less than 2 h(Dube et al., 2000b). Although some calcium is secreted in soluble form, most is sequestered into vesicles called spherites (Wessing and Eichelberg,1978) commonly observed in insect epithelia(Turbeck, 1974). Antibodies to SPoCk C labelled the periphery of such vesicles in the lumen of the initial segment of the tubule (Fig. 10B–E). Thus, spherites are formed as concretions within specialized peroxisomes with high levels of SPoCk C in their membranes, so facilitating calcium uptake and storage(Fig. 10F,G) and then discharged into the tubule lumen. Thus, whereas SPoCk A modulates fluid transport via Golgi calcium, SPoCk C modulates another important property of the tubule: bulk calcium transport and sequestration. This was demonstrated by targeted over-expression of SPoCk C in principal cells, which resulted in increased calcium storage and excretion by the tubules, as determined by transport experiments using ⁴⁵Ca²⁺(Fig. 10F,G).

In vertebrates, peroxisomes are abundant in kidney and liver and are associated with disease and stress conditions(Wanders and Waterham, 2006). Thus, deciphering the mechanisms of peroxisomal calcium storage, transport and their impact on cellular calcium homeostasis in a genetically tractable epithelial model like the tubule may be a useful avenue of research. This may be achieved by measuring peroxisomal [Ca²⁺] using targeted aequorin or GFP-aequorin and monitoring links between the change of [Ca²⁺]in the cytoplasm and peroxisome. Furthermore, defective peroximsomal biogenesis has been shown to impact on mitochondrial ultrastructure and activity (Baumgart et al.,2001), so potential interactions between peroxisomal and mitochondrial calcium pools should be investigated.

The Golgi is a critically important organelle for calcium homeostasis(Dolman and Tepikin, 2006) and is clearly an essential calcium store in the tubule, thus modulating tubule physiology. Work in acinar cells has also shown that Golgi and mitochondria associate, and influence calcium gradients at particular zones in the cell(Dolman et al., 2005). Given the prominence of mitochondrial calcium stores in tubule principal cells, such interactions between mitochondrial or other stores (e.g. ER), and Golgi may well occur. Thus, it will be important to make measurements of Golgi[Ca²⁺] in the intact tubule. The existence of multiple protein probes in which to survey intracellular calcium pools in vivo,coupled with transgenic technology, should open up several key avenues of research into cellular calcium homeostasis in defined cellular sub-types and epithelial tissue function.

We have described calcium signalling mechanisms in organelles of the tubule principal cell, and also the complexity of capa action that mobilises calcium signals from different intracellular organelles. The action of capa is in marked contrast to that of Drosokinin, which acts on tubule stellate cells. Although capa peptides have been shown to act via IP₃ and IP₃R (Pollock et al.,2003), the multiple organellar targets of capa-1 suggest that either IP₃ is generated at these different sites in the cell or,more likely, that capa-1 utilises additional routes for calcium mobilisation via calcium entry channels like CNG(Broderick et al., 2003; MacPherson et al., 2001) or TRPL (MacPherson et al.,2005), which promote calcium release and/or uptake into the Golgi,mitochondria or peroxisomes. Drosokinin, however, seems to act only via IP₃ (Pollock et al., 2003; Radford et al.,2002). Drosokinin stimulates a calcium rise in cells, which are transfected with the ER-localised SPoCk-B (which is not expressed in principal cells) but does not do so in either SPoCk A- or SPoCk C-transfected cells(Southall et al., 2006). This confirms that Drosokinin modulates an ER-derived IP₃ signal, thus increasing [Ca²⁺]cyt. Therefore, although both neuropeptides can act via IP₃, in the tubule, Drosokinin is diagnostic of ER-derived IP₃ signalling whereas capa-1 is not, suggesting that calcium signalling mechanisms in the principal cell are necessarily more complex.

Although calcium signalling has been shown to modulate fluid transport by the Malpighian tubule, it is clear that the other important roles of the tubule in immune sensing and detoxification, may also be calcium-regulated processes. A recent study on larval fat body has demonstrated involvement of calcineurin in innate immunity (Dijkers and O'Farrell, 2007); and our unpublished work (S.T. and S.A.D.)shows involvement of principal cell calcium signalling in the innate immune response by the tubule.

The impact on calcium signalling and calcium homeostasis on tubule function is extremely significant, and affords challenges in the understanding of novel regulation of epithelial cell and tissue function. Given the importance of the tubule in detoxification, immunity and survival of the organism, there are many novel calcium-mediated events to be discovered. Targeted protein-based probes for such studies will be important tools for such studies; and will reveal dynamics of calcium signalling between distinct cellular pools.

List of abbreviations

 
• BRET

  bioluminescence resonance energy transfer

 
• CFP

  cyan fluorescent protein

 
• CNG

  cyclic nucleotide-gated channel

 
• DAG

  diacylglycerol

 
• EGFP

  enhanced green fluorescent protein

 
• ER

  endoplasmic reticulum

 
• FRET

  fluorescence resonance energy transfer

 
• GFP

  green fluorescent protein

 
• IP₃

  inositol 1,4,5-triphosphate

 
• IP₃R

  inositol 1,4,5-triphosphate receptor

 
• PLCβ

  phospholipase Cβ

 
• PMCA

  plasma membrane calcium ATPase

 
• PTS1

  peroxisomal targeting sequence 1

 
• PUFAs

  polyunsaturated fatty acids

 
• RyR

  ryanodine receptor

 
• SERCA

  sarco/endoplasmic reticulum Ca²⁺ ATPase

 
• SPCA

  secretory pathway calcium ATPase

 
• TRP

  transient receptor protein

 
• YFP

  yellow fluorescent protein