Allatostatin A-like immunoreactivity in the nervous system and gut of the larval midge Chironomus riparius: modulation of hindgut motility, rectal K⁺ transport and implications for exposure to salinity

Juvenile hormones are synthesized by the corpora allata and control the juvenile characteristics of insects during growth and development. In adult insect females, juvenile hormones can also regulate reproductive activities such as the maturation of oocytes. Synthesis of juvenile hormones is inhibited by allatostatins (ASTs), members of a large and diverse family of peptides that are structurally characterized into three distinct groups (Gäde and Goldsworthy, 2003).

The C-type allatostatins with an N-terminal pGlu residue and a free C terminus have been discovered in the moth Manduca sexta, the fly Drosophila melanogaster and the mosquito Aedes aegypti (Kramer et al., 1991; Williamson et al., 2001; Li et al., 2006). The B-type allatostatins all have a common Trp-(X6)-Trp-amide C terminus and were first discovered in the cricket (Wang et al., 2004). The A-type allatostatins (ASTs) were the first to be discovered in the cockroach and share a Phe-Gly-Leu-amide C terminus (Donly et al., 1993; Siju et al., 2014).

All allatostatins inhibit juvenile hormone synthesis in at least one group or species of insect; however, many do not in other insects, and instead have known inhibitory activity on contractile properties of muscle (Gäde and Goldsworthy, 2003). In particular, the ASTs have been shown to inhibit contractions of visceral muscle including the foregut and hindgut (Lange et al., 1993; Sarkar et al., 2003; Predel et al., 2001; Matthews et al., 2008), and have been localized to endocrine cells of the gut (Veenstra et al., 1995; Veenstra, 2009; Reichwald et al., 1994; Robertson and Lange, 2010), suggesting that they may have a role in feeding and digestion. In the cockroach, ASTs have been shown to decrease food consumption (Aguilar et al., 2003). There is increasing evidence that ASTs modulate ion transport of various regions of the gut (Onken et al., 2004; Vanderveken and O'Donnell, 2014). An important group of insects in which ASTs have not been studied are the midges.

Midge (Chironomus riparius Meigen 1804) larvae are ubiquitous benthic inhabitants of freshwater lakes, ponds, marshes and rivers in temperate regions of the Northern Hemisphere and have become important bioindicators of environmental pollution resulting from industrial effluents, sewage effluents and other anthropogenic sources (Moreau et al., 2008; Faria et al., 2006). Chironomus riparius have also been observed in naturally occurring and anthropogenically induced salinated water (Colbo, 1996; Williams and Williams, 1998). Endocrine disruptors affect midge larvae by altering the accumulation of yolk protein in eggs, inducing expression of Hsp70, activating the enzymes catalase and glutathione-S-transferase, and reducing the rate of successful adult emergence (Lee and Choi, 2007; Morales et al., 2011; Hahn et al., 2002). Despite their importance, very little is known about the physiological mechanisms that underlie the observed responses to xenobiotics, heavy metals, endocrine disruptors and salinity in midge larvae. In particular, to the best of our knowledge, there is only a single study that has examined neuroendocrine factors in midge larvae, revealing the presence of serotonin and the peptides proctolin and bombesin in neurons of Chironomus tentans (Johansson et al., 1986).

Because allatostatins play a role in development and control aspects of feeding and digestion, factors that are affected by pollutants and endocrine disruptors in midges, it was hypothesized that allatostatins are present in midge larvae, where they could affect gut motility and ion transport. This study examined the presence of

AST-like immunoreactivity (AST-IR) in the nervous system and the gut of the larval midge C. riparius. A novel midge larval hindgut motility assay was developed to assess potential effects of ASTs on the hindgut musculature. Effects of ASTs on rectal ion transport and rectal Na⁺/K⁺-ATPase (NKA) and vacuolar-type H⁺-ATPase (VA) activities were also determined. The results led to an unexpected link between the effects of ASTs in midge larvae and larval habitat salinity.

AST-IR is present throughout the ventral nerve cord and gut of larval C. riparius (Fig. 1). Diagrammatical representations of AST-IR in neurons, neural processes and neuroendocrine cells in the gut and nervous system are shown in Fig. 2. The midgut contains three regions with AST-IR endocrine-like cells (Fig. 1A,B, Fig. 2A). One region is located at the middle midgut (Fig. 1A, Fig. 2A), another region is located at the posterior midgut near the attachment of the Malpighian tubules (Fig. 1B, Fig. 2A), and some endocrine-like cells were found in the vicinity of the gastric caecae, albeit in fewer numbers with respect to the other two regions (Fig. 2A). The endocrine-like cells possessed a single short process that extended from the soma towards the midgut lumen (Fig. 1A, inset).

A number of AST-IR cells are located in the brain and give rise to neuropile and processes that run posteriorly into the nerve cord (Fig. 1C, Fig. 2B). These processes extend the length of the abdominal nerve cord through the connectives between ganglia and within the ganglia (Fig. 2B). As the processes pass through the ganglia, they have varicosities along their length (Fig. 1D). No AST-IR cell bodies are found in the abdominal ganglia; however, the terminal ganglion contains four AST-IR cells (Fig. 1E, Fig. 2B) and at least two of these give rise to processes in the proctodeal nerves, which innervate the rectum, ileum and posterior midgut (Fig. 1B,F–H, Fig. 2). The AST-IR in these is not continuous and instead appear as distinct punctate varicosities along the process (see Fig. 1B,G,H). Controls where the primary AST antiserum was omitted or pre-absorbed with 10⁻⁵ mol l⁻¹ Diploptera puncata allatostatin 7 (Dippu-AST7) resulted in no staining throughout the gut and nervous system (data not shown).

The semi-isolated larval hindgut of C. riparius bathed in a physiological saline contracted in a non-rhythmic manner and these contractions can be detected with an impedance monitor and viewed with data acquisition software. Each single contraction produced a vertical deflection of the recording voltage trace that was easily discernible above background noise (Fig. 3). It was not possible to localize contractions to specific regions of the hindgut because contractions of the ileum or rectum were detected by the impedance electrodes. Contractions that were visibly seen to occur at the posterior midgut were not detected by the impedance electrodes because the minutin pins inserted at the posterior midgut limited movements of this region.

In the presence of Dippu-AST7, the frequency of hindgut contractions decreased with respect to the frequency of contractions recorded in physiological saline without the peptide (Fig. 3, compare saline with Dippu-AST7 doses). This effect was dependent on the dose of Dippu-AST7, where increasing doses resulted in fewer hindgut contractions. Dippu-AST7 inhibited hindgut contractions with a threshold dose of ~0.5 nmol l⁻¹ and an EC₅₀ of ~10 nmol l⁻¹ (Fig. 4). Application of 1 μmol l⁻¹ Dippu-AST7 did not abolish contractions but inhibited the contractions by ~95%.

The rectum of larval C. riparius is a site of K⁺ absorption (see Jonusaite et al., 2013). The mean baseline K⁺ flux (absorption, lumen to bath) measured at the recta in this study was 104.3±13.7 pmol cm⁻² s⁻¹ (n=38). Addition of various doses of Dippu-AST7 to the recta resulted in a dose-dependent decrease in the K⁺ flux (Fig. 5). Addition of saline (control) or a low dose of Dippu-AST7 (1 pmol l⁻¹) did not affect the K⁺ fluxes. The highest dose of Dippu-AST7 tested (1 μmol l⁻¹) resulted in K⁺ fluxes that were roughly half of the initial baseline K⁺ flux.

NKA and VA activities in the rectum of larval C. riparius were ~15 to 20 times higher than in other regions of the gut including the Malpighian tubules, and evidence was provided that NKA and VA drive K⁺ absorption at the rectum (see Jonusaite et al., 2013). Incubation with 10 nmol l⁻¹ Dippu-AST7 for 10 min decreased the NKA and VA activities of the rectum by approximately one-third and two-thirds, respectively (Fig. 6).

A previous study demonstrated that recta of brackish-water-reared larvae have lower activities of NKA and VA compared with recta of freshwater-reared larvae (Jonusaite et al., 2013). Therefore, salinity and AST7 have similar effects on rectal NKA and VA activities. The actions of AST7 occur rapidly and hence cannot be mediated by changes in NKA and VA expression; however, it was suggested that larvae reared in brackish water may have lower expression of rectal NKA and VA than those reared in freshwater (Jonusaite et al., 2013). In an effort to determine whether the mechanisms of AST7- and salinity-induced changes in NKA and VA rectal activities are similar or different, NKA and VA protein expression studies on freshwater- and brackish-water-reared larvae were carried out.

The NKA α subunit was detected by western blot as a single band at ~95 kDa in recta collected from C. riparius larvae (Fig. 7A). Recta from larvae reared in brackish water, however, exhibited approximately fourfold less NKA α subunit protein expression than recta collected from freshwater (control)-reared larvae (Fig. 7B). In all samples examined, the VA B subunit was detected as a single band at ~56 kDa by western blot; however, a second band at ~58 kDa was also occasionally detected, but this did not appear to be correlated with rearing salinity (Fig. 7A). Densitometric analysis of the ~56 kDa band revealed relatively variable VA B subunit protein expression within recta collected from the freshwater-reared group (Fig. 7B). VA B subunit protein expression within recta collected from freshwater-reared larvae was not significantly different from protein levels in recta from the brackish-water-reared group (Fig. 7B). Actin expression, which was used as a loading control to normalize NKA and VA protein abundance, was not significantly different (Student's t-test, P=0.4232; data not shown) between recta collected from control and brackish-water -reared larvae.

Allatostatin immunoreactivity has been well documented in the nervous systems of insects with AST-IR cell bodies routinely found in the brain and abdominal ganglia (Skiebe et al., 2006; Žitňan et al., 1993; Kreissl et al., 1999; Lange et al., 1993; Hernández-Martínez et al., 2005). In C. riparius larvae, the brain contains at least five paired AST-IR cell bodies at the anterior edge of the frontal lobes (Fig. 1C, Fig. 2B). Processes originating from these cells appear to enter the central complex, which contains extensive AST-IR. Processes cross the center of the brain to the respective contralateral side and are seen running posteriorly into the nerve cord. In C. riparius larvae, only the terminal ganglion contained AST-IR cell bodies while only processes with many varicosities were present in the remaining ganglia (Fig. 1D,E, Fig. 2B). The processes in the ganglia are likely to originate from the neurons in the brain, suggesting that allatostatins act as neurotransmitters relaying information from the brain to other neurons in the ganglia of the ventral nerve cord in larval C. riparius. Similar processes have been found in other insects (Skiebe et al., 2006; Žitňan et al., 1993). At least one of the two pairs of lateral neurons in the terminal ganglion give rise to processes that innervate the hindgut (see Fig. 1F–H, Fig. 2B) and these processes extend the length of the rectum, ileum and onto the posterior midgut (Fig. 1B, Fig. 2A). Nerves containing AST-IR cell bodies on the hindgut that originate from neurons in the terminal ganglion/neuromere have been found in the cockroach, locust, moth, fruit fly and mosquito (Maestro et al., 1998; Yoon and Stay, 1995; Duve et al., 1997; Robertson and Lange, 2010; Hernández-Martínez et al., 2005). In the cockroach, branches of at least one pair of AST-like IR neurons project to the hindgut and were also seen to extend onto the midgut (Maestro et al., 1998). In the larval moth Cydia pomonella, larval Drosophila and adult mosquitoes, the processes from the terminal ganglion neurons are confined to the hindgut (Duve et al., 1997; Yoon and Stay, 1995; Hernández-Martínez et al., 2005).

In larval C. riparius, AST-IR endocrine cells are found in the posterior midgut near the attachment of the Malpighian tubules to the gut (Fig. 1B, Fig. 2A). Another group of endocrine cells is located in the middle midgut region (Fig. 1A, Fig. 2A) and a third group is found in the anterior midgut associated with the gastric caecae (Fig. 2A). Other dipteran species have regional groups of AST-IR endocrine cells in the midgut. The adult female mosquito contains AST-IR endocrine cells only in the posterior region of the midgut (Hernández-Martínez et al., 2005), as do larval and adult Drosophila (Yoon and Stay, 1995). The larval lepidopteran C. pomonella also has AST-IR endocrine cells confined to the posterior midgut (Duve et al., 1997), while other insects have an apparently even distribution of cells along the entire midgut (Robertson and Lange, 2010; Sarkar et al., 2003; Maestro et al., 1998). The regionalization of AST-IR endocrine cells in the midgut of dipterans and lepidopterans may be a consequence of digestive processes requiring high luminal alkalinity because these larvae have regional differences in luminal pH in the midgut (Boudko et al., 2001; Shanbhag and Tripathi, 2005; Coddington and Chamberlin, 1999). The function of the AST-IR endocrine cells in the midge is unclear. These cells may sense nutritional content and AST may affect midgut muscle, enzyme production and/or ion transport. Further study in these areas can determine the function of AST in the endocrine cells.

AST inhibits the contractions of the larval midge hindgut in a dose-dependent manner (Figs 3, 4). This suggests that the hindgut muscle expresses receptors for AST. Many previous studies have demonstrated that AST is a myoinhibitory peptide in insects (Sarkar et al., 2003; Lange et al., 1993; Onken et al., 2004; Duve et al., 2000). A dose of 1 pmol l⁻¹ non-native AST on the foregut and hindgut of Lacanobia oleracea and Rhodnius prolixus, respectively (Sarkar et al., 2003; Duve et al., 2000), is effective at inhibiting muscle contractions. The larval midge hindgut responds to a threshold dose of 5 nmol l⁻¹ non-native AST from the cockroach. This result, together with the presence of AST-IR in processes on the hindgut that originate from cells in the terminal ganglion, suggest that AST-like neuropeptides are acting as inhibitory neurotransmitters and/or neuromodulators of hindgut muscle in larval C. riparius. The inhibition of hindgut muscle contraction by AST should prolong the time luminal contents remain in the hindgut prior to excretion. In a previous study, our laboratory has shown that the rectum of larval C. riparius is a site of K⁺ reabsorption and that this is dependent on the activities of NKA and VA, which are expressed on the basal and apical membranes of epithelial cells, respectively (Jonusaite et al., 2013). In that study, it was shown that the recta of larvae reared under dilute freshwater or in ion-poor conditions had relatively high NKA and VA activities and exhibited high rates of K⁺ reabsorption. Conversely, the recta of larvae reared in brackish water had low NKA and VA activities and exhibited low rates of K⁺ reabsorption (see Jonusaite et al., 2013). Application of AST to the larval recta decreased K⁺ reabsorption in a dose-dependent manner (Fig. 5). A dose of 10 nmol l⁻¹ AST also decreased the NKA and VA activities in larval recta (Fig. 6). Together, these results indicate that AST inhibits K⁺ reabsorption by lowering the activities of NKA and VA in the larval recta. The actions of exogenously applied AST on hindgut motility and ion reabsorption suggest that AST functions as a general depressant of hindgut functions; however, because the rectum receives direct innervation from neurons of the terminal ganglion, it is conceivable that the modulation of muscle activity and epithelial transport can be independently controlled. Studies have shown that AST modulates gut muscle activity (Sarkar et al., 2003; Duve et al., 2000). Studies have also provided evidence that AST directly modulates ion transport activities in gut epithelia (Onken et al., 2004; Vanderveken and O'Donnell, 2014). By simultaneously and perhaps independently modulating transit time of hindgut luminal contents and the magnitude of epithelial ion reabsorption through direct neural control, the midge would greatly increase its ability to fine-tune the reabsorptive and excretory functions of the hindgut.

The larvae of C. riparius are found in waters of widely varying ionic content. In dilute freshwater and ion-poor conditions, the anal papillae take up ions while the rectum reabsorbs ions, thereby countering the loss of ions and osmotic uptake of water from the dilute habitat (Nguyen and Donini, 2010; Jonusaite et al., 2013). Larvae survive abrupt transfer to, and can be reared in, brackish water (20% seawater) (Jonusaite et al., 2011; Jonusaite et al., 2013). Abrupt transfer to brackish water is accompanied by a sharp reduction in whole-body VA activity and a less pronounced decrease in whole-body NKA activity (Jonusaite et al., 2011). The greatest amount of NKA and VA activity can be attributed to the rectum, and the recta of larvae reared in brackish water have greatly reduced NKA and VA activities, which coincide with reduced ion reabsorption (Jonusaite et al., 2013).

Although the larval response to abrupt and long-term (rearing) exposure to brackish water is the same, the mechanisms with which this is achieved are likely to be different. Abrupt exposure to elevated salinity would require a rapid response that could be achieved with localized release of AST at the rectal epithelium, resulting in a rapid reduction in NKA and VA activity (Fig. 6) and subsequent rapid reduction in ion reabsorption (Fig. 5). The observed rapid response to application of AST to the rectal epithelium is likely to result from activation of AST receptors on epithelial cells and subsequent second messenger pathways that impinge on the activities of NKA and VA. This scenario would be similar to the serotonin modulation of VA activity in the blowfly salivary glands (Baumann and Walz, 2012). Larvae of C. riparius that inhabit regions of southern Canada and the northern United States encounter abrupt salination of their freshwater habitats in the spring when snow melt adjacent to roadways that were treated with salt in the winter enters surrounding water bodies that are inhabited by C. riparius. Often, these events produce substantial, acute increases in the ionic content of these bodies of water (Allert et al., 2012; Fay and Shi, 2012), and as such C. riparius would require a mechanism to regulate internal ion levels over the short term. Our results suggest that localized modulation of rectal reabsorption by AST could be such a mechanism.

When C. riparius larvae encounter sustained, long-term, elevated ionic conditions, continued synthesis and release of AST may be metabolically detrimental; therefore, reducing NKA and VA expression in the rectal epithelium would be a more feasible solution. Indeed, protein expression of the catalytic NKA α-subunit was significantly downregulated in brackish-water-reared C. riparius larvae when compared with freshwater-reared animals (Fig. 7). However, VA B subunit protein expression was unaltered by rearing salinity (Fig. 7). Regulation of VA enzyme activity in insect and vertebrate models has been shown to occur via many mechanisms, notably reversible dissociation of V₁ (cytoplasmic sector) and V_o (membrane sector) domains, oxidation of disulfide bonds, and translocation of VA molecules to the plasma membrane via endocytosis and exocytosis from an intracellular vesicular pool (Baumann and Walz, 2012). Our results suggest that VA activity in the rectum of C. riparius is regulated in such ways by AST and also by long-term exposure to elevated salinity.

In summary, the larvae of C. riparius possess neurons and midgut endocrine cells containing AST A-like factors. The extensive AST-IR processes in the brain and within the ganglia of the central nervous system (CNS) suggest that AST A-like factors may coordinate various targets to carry out important physiological roles. The presence of AST A-like factors in midgut endocrine cells suggests that one of these roles may be related to feeding and digestion. AST-IR neurons in the terminal ganglion send processes through the proctodeal nerves to directly innervate the hindgut, where AST A inhibits muscle contraction and ion reabsorption. The inhibition of ion reabsorption at the rectum is achieved by decreasing the activities of NKA and VA and could serve as an effective mechanism of ion regulation in the face of abrupt, acute increases in habitat salinity.

A laboratory colony of Chironomus riparius was maintained in the Aquatic Facility in the Department of Biology at York University at 24°C under a 12 h:12 h light:dark cycle. The larvae were reared in aerated 6 l aquaria with a 2.54 cm deep layer of sand on the bottom and filled with 3 l of dechlorinated tap water. Larvae were fed every second day with TetraFin Goldfish flakes (Tetra Holding Inc., VA, USA) and the water was renewed weekly. Studies were conducted on third and fourth instar larvae.

The CNS and digestive system of larvae were processed to reveal AST-like immunoreactivity utilizing a rabbit polyclonal antiserum against Diploptera puncata allatostatin 7 (Dippu-AST7) (R-AST; kind gift from Dr Stephen Tobe, Toronto, Canada). Cross-reactivity of this antiserum to other ASTs has been previously determined (Yu et al., 1995). Larvae were pinned through the cuticle to the bottom of a Sylgard® (Dow Corning, Mississauga, ON, Canada)-lined dish filled with saline containing (in mmol l⁻¹): 5 KCl, 74 NaCl, 1 CaCl₂, 8.5 MgCl₂, 10.2 NaHCO₃, 8.6 HEPES, 20 glucose and 10 glutamine; pH 7.0 (BioShop, Burlington, ON, Canada). The CNS and gut were removed under saline and fixed with a solution of 2% paraformaldehyde in Millonig's buffer, at 4°C overnight or room temperature for 2 h, with the following composition (in mmol l⁻¹): 136 NaH₂PO₄H₂O, 120 NaOH, 67 glucose and 0.78 CaCl₂2H₂O; pH 7.0. All subsequent incubations were performed at 4°C unless otherwise indicated. Tissues were rinsed with phosphate-buffered saline (PBS) four times at 5 min intervals and incubated in a solution containing 2% normal sheep serum (NSS), 4% Triton-X-100 in PBS for 1 h at room temperature and subsequently in a solution of 3% skim milk powder (Nestlé, North York, ON, Canada) in PBS for 24 h. The R-AST was pre-incubated at a dilution of 1:1000 in 0.4% Triton-X-100, 2% NSS in PBS for 24 h. Tissues were incubated in the 1:1000 R-AST solution for 48 h. Tissues were rinsed with PBS five times at 10 min intervals and subsequently incubated in a solution containing 1:200 affinity purified Cy3-labeled sheep anti-rabbit antibody (Sigma-Aldrich, Oakville, ON, Canada) and 10% NSS in PBS for 24 h. Tissues were rinsed with PBS 10 times over 24 h. Controls were processed in the same manner except that some were incubated in the absence of R-AST and others were incubated in a 1:1000 R-AST solution pre-absorbed with 10 μmol l⁻¹ of Dippu-AST7 (APSGAQRLYGFGLamide, AnaSpec, San Jose, CA, USA). Tissues were mounted in glycerol on glass slides and viewed with a Fluoview 300 laser scanning confocal microscope (Olympus, Markham, ON, Canada) using a helium–neon (543 nm) laser, and digital images were captured with the accompanying Fluoview software package. Composite image figures were constructed using Adobe Photoshop CS2 (Adobe, Etobicoke, ON, Canada).

The frequency of larval hindgut contractions was monitored with an impedance monitor (UFI Model 2991, Morro Bay, CA, USA) connected to a Powerlab 4/30 and desktop computer running LabChart 6.0 software (ADInstruments, Colorado Springs, CO, USA). A semi-isolated preparation contained in a Sylgard®-lined dish and bathed in 2 ml of saline was developed. The preparation consisted of a portion of the posterior midgut with surrounding cuticle, exposed ileum and rectum, and the intact posterior section of the larva complete with anal papillae and prolegs. Minuten pins inserted into the cuticle at the base of the prolegs and the posterior midgut secured the preparation to the dish. Electrodes were fashioned from minuten pins soldered to wires connected to the impedance monitor inputs. With the aid of a stereomicroscope (Model K400 Series, Motic, Richmond, BC, Canada), the electrodes were positioned on either side of the preparation at the anterior rectum. Preparations were rinsed several times with fresh saline and then allowed to equilibrate in 1.98 ml of saline for 10 min. Contractions were monitored for the subsequent 10 min intervals prior to and after the addition of Dippu-AST7. The procedure was repeated to test several doses of Dippu-AST7 on individual preparations. Dippu-AST7 was administered by adding 20 μl of 100 times the desired dose of the peptide to the saline bathing the preparation. The addition of 20 μl of saline to the bath served as a control. The Peak Analysis feature in LabChart Pro 6.0 software was utilized to determine the number of contractions that occurred over the 10 min intervals and yield the frequency of contractions (contractions min⁻¹). Effects of individual doses of Dippu-AST7 on the frequency of contractions were determined by expressing the frequency of contractions in the presence of Dippu-AST7 as a percentage of the frequency of baseline contractions recorded during the 10 min interval prior to the addition of the peptide.

In aquaria identical to those outlined above, eggs were hatched and larvae were reared to third and fourth instar (~30 days) in ion-poor water (composition in μmol l⁻¹: [Na⁺] 20; [Cl⁻] 40; [Ca²⁺] 2; [K⁺] 0.4; pH 6.5). Larval recta were then isolated under saline and individually incubated with either saline (control) or saline containing 10⁻⁸ mol l⁻¹ Dippu-AST7 for 10 min at room temperature. After incubation, the recta were collected in 1.5 ml microcentrifuge tubes (30–35 recta per tube), flash-frozen and stored at −85°C until further analysis. For NKA and VA activity assays, recta were thawed on ice and processed according to previously described methods (Jonusaite et al., 2011). NKA and VA activity assays were subsequently conducted using the methodology of McCormick (McCormick, 1993) with modifications as described by Jonusaite et al. (Jonusaite et al., 2011).

In the following experiment, eggs were hatched and larvae were reared as described above in brackish water composed of 7 g l⁻¹ Instant Ocean Sea Salt (i.e. 20% seawater, approximately equivalent to 96 mmol l⁻¹ NaCl; United Pet Group, Blacksburg, VA, USA). Freshwater (control) animals also were reared in dechlorinated tap water as described above. Recta from larvae reared in freshwater and brackish water were then isolated under saline and collected into 1.5 ml microcentrifuge tubes (25–30 recta per tube), flash-frozen and stored at −85°C until further processing. For western blot analysis, recta were thawed and homogenized on ice in chilled homogenization buffer (50 mmol l⁻¹ Tris-HCl pH 7.4, 1 mmol l⁻¹ phenylmethylsulfonyl fluoride) containing 1:200 protease inhibitor cocktail using glass Kontes handheld tissue grinders. Homogenates were then centrifuged at 8000 g for 10 min at 4°C, and protein content of collected supernatant was determined using the Bradford assay (Sigma-Aldrich) according to the manufacturer's guidelines. Samples were prepared for SDS-PAGE by either incubating for 10 min at 65°C for NKA samples or by boiling for 5 min at 100°C for VA samples with 5× loading buffer containing 225 mmol l⁻¹ Tris-HCl (pH 6.8), 3.5% (w/v) SDS, 35% glycerol, 12.5% (v/v) β-mercaptoethanol and 0.01% (w/v) Bromophenol Blue. Equal amounts of sample (i.e. 5 μg each) were electrophoretically separated by SDS-PAGE and western blot analysis of NKA and VA was conducted according to procedures outlined by Chasiotis and Kelly (Chasiotis and Kelly, 2008) using a 1:100 dilution of mouse monoclonal anti-α5 antibody (Developmental Studies Hybridoma Bank, Iowa City, IA, USA) for NKA α-subunit and a 1:1000 dilution of rabbit polyclonal anti-ATP6V1B1 antibody (Abnova, Neihu District, Taipei City, Taiwan) for VA B subunit. After examination of NKA and VA protein expression, membranes were stripped and re-probed with a 1:100 dilution of mouse monoclonal anti-JLA20 antibody (Developmental Studies Hybridoma Bank) for actin as a loading control. Densitometric analysis of NKA, VA and actin protein was conducted using Quantity One 1D analysis software (Bio-Rad Laboratories Canada, Mississauga, ON, Canada). NKA and VA protein abundance were expressed as normalized values relative to actin abundance.

The authors thank Dr Stephen Tobe (Toronto) for providing the allatostatin antisera as a kind gift, Hang Nguyen for conducting preliminary immunohistochemical studies and Dr Marjorie Patrick (San Diego) for advising on western blot protocols for small tissues. Mouse monoclonal anti-α5 and anti-JLA20 antibodies, developed by D. M. Fambrough and J. J.-C. Lin, respectively, were obtained from the Developmental Studies Hybridoma Bank, created by the National Institute of Child Health and Human Development of the National Institutes of Health and maintained at The University of Iowa, Department of Biology, Iowa City, USA.