An animal homolog of plant Mep/Amt transporters promotes ammonia excretion by the anal papillae of the disease vector mosquito Aedes aegypti

Ammonia (NH₃/NH₄⁺) is a compound of major importance to life on earth, serving as an important nutrient for microorganisms and plants while presenting challenges to animals because of its toxic nature. While plants and microorganisms typically scavenge for ammonia in order to incorporate the nitrogen into proteins, animals excrete it as a toxic by-product of metabolic protein breakdown. Members of the methylammonium permease/ammonium transporter (Mep/Amt) family of transporters play a critical role in ammonia uptake by plants and microorganisms (Thornton et al., 2006; Nygaard et al., 2006; Willmann et al., 2007; Gu et al., 2013; Wu et al., 2015). The annotated genomes of vertebrate animals lack Mep/Amt homologs and excretion of ammonia in animals has been linked to members of a related family of glycoproteins, the Rhesus glycoproteins (RhGPs) (Nawata et al., 2007; Braun et al., 2009; Shih et al., 2013). Interestingly, homologs of the Mep/Amt transporters are present in the invertebrates along with homologs of RhGPs (Weihrauch et al., 2012b; Fig. 1).

In the mosquito Aedes aegypti, three putative ammonia transporter genes were identified in silico and named AeRh50-1, AeRh50-2 (sharing sequence similarity with RhGPs) and AeAmt1 (similar to Mep/Amt) (Weihrauch et al., 2012b). In a gene analysis, the AeAmt1 gene aligns with other invertebrate Mep/Amt homologs and these are most similar to the Mep/Amt genes of plants (see Fig. 1). The AeRh50-1 and AeRh50-2 genes align with other RhGP homologs of insects and other invertebrates (Fig. 1). The Aedes RhGPs possess the conserved ammonia conducting amino acid residues that have been identified in the vertebrate RhGPs, and AeAmt1 shares 22% sequence identity with the bacterial AmtB and also contains amino acid residues in the pore that are critical to ammonia transport (Weihrauch et al., 2012b). There is ample evidence from other animals to suggest that RhGPs transport NH₃ preferentially over NH₄⁺ (see Gruswitz et al., 2010; Nawata et al., 2010; Nakhoul and Hamm, 2014). Although the exact mechanism of transport by RhGPs is still up for debate, a model for NH₃ transport by human RhCG relies on comparisons with a model proposed for transport by AmtB (see below) and suggests that RhCG may recruit NH₄⁺ (Gruswitz et al., 2010). The NH₄⁺ is then deprotonated and NH₃ is transported and released at the other side of the membrane (Gruswitz et al., 2010). The RhGPs have also been implicated in conducting CO₂ across cellular membranes (Soupene et al., 2004; Kustu and Inwood, 2006; Peng and Huang, 2006; Perry et al., 2010). In contrast, evidence suggests that the plant and microorganism Mep/Amts transport NH₄⁺ (Willmann et al., 2007; Yuan et al., 2009; Gu et al., 2013; Wu et al., 2015). The exact mechanism of transport is the subject of debate; however, one study proposes that AmtB recruits NH₄⁺ and as the NH₄⁺ travels further into the channel it is deprotonated yielding NH₃ (Khademi et al., 2004). The NH₃ then passes through the central hydrophobic region of the AmtB, and as it emerges it receives a proton, yielding NH₄⁺, which is released to the other side of the membrane (Khademi et al., 2004). Hence, AmtB mediates net NH₄⁺ transport. The Amt transporters of Arabidopsis have also been shown to carry out net NH₄⁺ transport (see Yuan et al., 2009). To date, the only animal Mep/Amt homolog that has been studied in any detail is the mosquito AgAmt of Anopheles gambiae (Pitts et al., 2014). It was suggested that AgAmt transports NH₄⁺ as heterologous expression of AgAmt in Xenopus oocytes bathed in NH₄Cl (200 nmol l⁻¹ to 500 mmol l⁻¹) resulted in a positive inward current; however, the ionic nature of the current was not determined and a yeast complementation study for AgAmt failed to demonstrate ammonia transport (Pitts et al., 2014). Transcript abundance of AgAmt was enriched in the antennae of adult mosquitoes, leading the authors to suggest a role in sensing volatile ammonia during host-seeking behavior (Pitts et al., 2014).

Whereas A. gambiae is a major vector for malaria, Aedes aegypti is the vector for re-emerging tropical diseases dengue and chikungunya virus. Larval control measures traditionally focus on freshwater habitats; however, it was shown that A. aegypti naturally breeds in brackish water and ammonia-rich septic tanks in Asia and the Caribbean (Burke et al., 2010; Ramasamy and Surendran, 2012). In the case of the latter, it was suggested that this discovery explains the persistence of dengue during drier seasons as the mosquitoes breeding in septic tanks act as a reservoir for the disease (Burke et al., 2010). As ammonia is toxic to all animals, investigating how larval A. aegypti excrete ammonia is particularly important in light of their ability to breed in septic tanks. Other aquatic freshwater animals exploit the habitat and excrete ammonia directly into the water without the costly conversion to less-toxic forms (García Romeu and Salibián, 1968; Weihrauch et al., 2004; Nawata et al., 2007; Weihrauch et al., 2012a,b; Cruz et al., 2013). Through work on frog skin, planarians, and crustacean and fish gills, a number of cellular mechanisms and transporters for ammonia excretion in freshwater animals have been discovered. In addition to the RhGPs and the Mep/Amts, these include members of the solute carrier family 9 (SLC9, NHEs) (Orlowski and Grinstein, 2004; Zachos et al., 2005); the Na⁺/K⁺-ATPase (NKA) (Masui et al., 2002; Furriel et al., 2004; Weihrauch et al., 2012b; Ip et al., 2012); and the V-type H⁺-ATPase (VA) (Wilson et al., 1994; Shih et al., 2008; Braun et al., 2009; Weihrauch et al., 2009; Wright and Wood, 2009).

In general, the excretory organs of insects include the Malpighian tubules and hindgut; however, larvae of A. aegypti also possess four anal papillae that are finger-like structures surrounding the anus and composed of a simple, syncytial epithelium covered by a thin cuticle with a luminal space that is continuous with the hemocoel of the body (Edwards and Harrison, 1983). A previous study detected ammonia effluxes (excretion) by the anal papillae using the scanning ion-selective electrode technique (SIET) (Donini and O'Donnell, 2005). It was determined that the anal papillae of larval mosquitoes excrete ammonia at a rate of 360 nmol cm⁻² h⁻¹, which is consistent with rates of ammonia excretion by the locust hindgut and by the cockroach Periplaneta americana (Mullins, 1974; Thomson et al., 1988; Donini and O'Donnell, 2005). Therefore, the anal papillae appear to be major sites of ammonia excretion in larval mosquitoes.

The aim of this study was to elucidate the epithelial transport mechanisms of ammonia excretion by the anal papillae of A. aegypti. We hypothesized that the anal papillae express the putative ammonia transporters and that these transporters are involved in the observed ammonia excretion from the anal papillae. To test this hypothesis, the study examined the transcript abundance of the three putative Aedes ammonia transporters in the anal papillae and pursued further studies on the putative transporter that showed enriched transcript abundance (AeAmt1).

The amino acid sequences were aligned by MUSCLE alignment in MEGA 5. Gene analysis of the aligned sequences was also performed with MEGA 5 using the maximum likelihood method with the Jones–Taylor–Thornton+four categories of gamma substitution rates+invariable sites model and nearest-neighbor interchange (NNI) heuristic method. Bootstrap values were determined from 1000 bootstrap replicates.

Larvae of A. aegypti (Liverpool) were obtained from a colony reared in the Department of Biology, York University (Toronto, ON, Canada). Larvae were reared in reverse-osmosis (RO) water at room temperature and on a 12 h:12 h light:dark cycle. Larvae were fed daily with a solution of liver powder and yeast in water. Rearing water was refreshed every other day. Fourth instar larvae were used 24 h post-feeding for all pharmacological studies. For molecular experiments, the larvae were fasted for 72 h prior to use.

Seven biological samples, each consisting of a pool of 200 anal papillae from 50 larvae, were isolated in cold A. aegypti physiological saline (Clark and Bradley, 1996) prepared in diethylpyrocarbonate (DEPC; Sigma-Aldrich, Oakville, ON, Canada) treated water. Harvested tissues were transferred to tubes containing TRIzol^® RNA isolation reagent (Invitrogen, Burlington, ON, Canada) and sonicated for 5 s at 5 W using an XL 2000 Ultrasonic Processor (Qsonica, LL, CT, USA). RNA was extracted according to TRIzol^® specifications and treated with the TURBO DNA-free™ Kit (Applied Biosystems, Streetsville, ON, Canada) to remove genomic DNA. RNA samples were used to synthesize cDNA using the iScript™ cDNA Synthesis Kit (Bio-Rad, Mississauga, ON, Canada) according to the manufacturer's instructions. The cDNA was stored at −20°C until subsequent use. Genes encoding putative A. aegypti ammonia transporters (AeRh50-1, AeRh50-2 and AeAmt1) were found in the A. aegypti genome in the National Center for Biotechnology Information (NCBI) database (http://www.ncbi.nlm.nih.gov) and the sequences used to design primers. Primer sequences, amplicon sizes and related accession numbers are listed in Table 1. Reverse transcription polymerase chain reaction (RT-PCR) was used to check for the presence of transcripts of putative ammonia transporters in A. aegypti anal papillae with 18S rRNA serving as an internal control. Amplicons were resolved by gel electrophoresis and sequenced at the York University sequencing facility (York University Core Molecular Facility, Toronto, ON, Canada). Sequence identities were confirmed by a BLAST search (NCBI database; http://www.ncbi.nlm.nih.gov).

To determine the relative mRNA abundance of AeRh50-1, AeRh50-2 and AeAmt1 in larval A. aegypti anal papillae, quantitative PCR (qPCR) was performed using the primers listed in Table 1 and SsoFast™ Evagreen^® Supermix (Bio-Rad) according to the manufacturer's protocol. Reactions were carried out using the CFX96™ real-time PCR detection system (Bio-Rad). To confirm the presence of a single product after each reaction, a melting curve analysis was performed. For each gene of interest, a standard curve was generated to optimize reaction efficiency. Quantification of transcripts was determined according to the Pfaffl method (Pfaffl, 2004). Samples were run in duplicate and 18S rRNA was used as an internal control because it showed consistent levels of expression under experimental treatments. A no-template negative control was also used.

Western blotting and immunohistochemistry for AeAmt1 and NHE3 were carried out at York University and the University of San Diego (CA, USA), respectively. Biological samples consisting of pools of 10–15 A. aegypti larval whole bodies or pools of anal papillae that were isolated from 25–50 larvae under ice-cold saline were collected and stored at −80°C until later processing. For examination of AeAmt1 expression, samples were thawed on ice and larval whole bodies were sonicated in a homogenization buffer containing 50 mmol l⁻¹ Tris-HCl pH 7.4, 1 mmol l⁻¹ PMSF and 1:200 protease inhibitor cocktail (Sigma-Aldrich), while anal papillae were sonicated in a homogenization buffer containing 50 mmol 1⁻¹ Tris-HCl pH 7.5, 150 mmol 1⁻¹ NaCl, 1% sodium deoxycholate, 1% Triton X-100, 0.1% SDS, 1 mmol 1⁻¹ PMSF and 1:200 protease inhibitor cocktail (Sigma-Aldrich). All samples were sonicated for 2×10 s at 3.5 W using an XL 2000 Ultrasonic Processor (Qsonica). For NHE3 expression, 80 µl of homogenization buffer (as described above) was added to a 1.5 ml centrifuge tube containing anal papillae tissue and then homogenized on ice for 30 s using a hand-held tissue grinder (Kontes Pellet Pestle). Homogenates were then centrifuged at 8000 or 10,000 g for 8 or 10 min at 4°C, and the protein content of the collected supernatants was determined using the Bradford assay (Sigma-Aldrich or Bio-Rad) according to the manufacturer's guidelines. Whole-body samples and anal papillae samples for NHE3 expression were prepared for SDS-PAGE by heating for 10 min at 95°C in 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, while anal papillae samples for AeAmt1 expression were prepared by heating for 5 min at 100°C in 6× loading buffer containing 360 mmol l⁻¹ Tris-HCl pH 6.8, 12% (w/v) SDS, 30% glycerol, 600 mmol l⁻¹ DTT and 0.03% (w/v) Bromophenol Blue. Samples (5–7 µg protein) were then electrophoretically separated by SDS-PAGE and western blot analysis of AeAmt1 was conducted according to a previously described protocol (Chasiotis and Kelly, 2008), while analysis of NHE3 was carried out by overnight transfer of protein to Immobilon-P polyvinylidene-difluoride (PVDF) membrane (EMD Millipore, USA) at 4°C. A custom-made polyclonal antibody that was produced in rabbit against a custom-made synthetic peptide (CPLWEREVELSDGLM) corresponding to a 14-amino acid region of AeAmt1 (GenScript USA Inc., Piscataway, NJ, USA) was used at 1:500 dilution. To confirm the specificity of the custom-made AeAmt1 antibody, a comparison blot was also run with the AeAmt1 antibody pre-absorbed with 10× molar excess of the immunogenic peptide for 1 h at room temperature prior to application to blots. After examination of AeAmt1 expression, blots were stripped and re-probed with a 1:100 dilution of mouse monoclonal anti-JLA20 antibody (J. J.-C. Lin, Developmental Studies Hybridoma Bank, Iowa City, IA, USA) for actin or a 1:1000 dilution of rabbit monoclonal anti-GAPDH (14C10) antibody (New England BioLabs, Whitby, ON, Canada) as loading controls. Densitometric analysis of AeAmt1, actin and GAPDH was conducted using Quantity One 1D analysis software (Bio-Rad). AeAmt1 abundance was expressed as a normalized value relative to the abundance of the loading control. Detection of NHE3 was performed using Life Technologies Western Breeze anti-rabbit chemiluminescent kit and an antibody generated against the C-terminal residues (E¹¹¹¹-G¹¹²⁸) of A. aegypti NHE3 (Pullikuth et al., 2006) diluted to 1:5000. Detection of protein bands were performed using Bio-Rad ChemiDoc XRS system.

Immunolocalization of AeAmt1, NKA, VA and NHE3 in paraffin-embedded sections of anal papillae was conducted according to previously described protocols (Pullikuth et al., 2006; Chasiotis and Kelly, 2008) using a 1:40 dilution of the custom-made anti-AeAmt1 antibody described above, a 1:10 dilution of a mouse monoclonal anti-α5 antibody for NKA (Douglas Fambrough, Developmental Studies Hybridoma Bank, IA, USA), a 1:100 dilution of a mouse polyclonal anti-ATP6V0A1 antibody for VA (Abnova, Taipei, Taiwan) or a 1:250 dilution of the NHE3 antibody described above. A sheep anti-mouse antibody conjugated to Cy2 or a goat anti-mouse antibody conjugated to Cy5 (Jackson Immunoresearch) was applied at 1:500 and 1:1000, respectively, to visualize NKA. The same sheep anti-mouse antisera applied at 1:500 was used to visualize VA. A goat anti-rabbit antibody conjugated to Alexa Fluor 594 (Jackson Immunoresearch) was applied at 1:500 dilution to visualize AeAmt1, while a goat anti-rabbit antibody conjugated to Alexa Fluor 647 (Jackson Immunoresearch) was applied at 1:1000 dilution to visualize NHE3. Comparison negative control slides were also processed as described above with either all primary antibodies omitted or the AeAmt1 antibody pre-absorbed with 10× molar excess of the immunogenic peptide for 30 min at room temperature prior to application to sections. Slides were mounted using ProLong Gold antifade reagent with or without DAPI (Life Technologies, Burlington, ON, Canada), and fluorescence images were captured using either an Olympus IX71 inverted fluorescence microscope (Olympus Canada, Richmond Hill, ON, Canada) and CellSense^® 1.12 Digital Imaging software (Olympus Canada) or a Nikon A1R laser scanning confocal microscope. Images were merged using Adobe Photoshop CS6 software (Adobe Systems Inc., Ottawa, ON, Canada).

Total RNA was extracted from whole bodies of fourth instar A. aegypti larvae and cDNA was generated as described above. Using this cDNA template, a fragment of the AeAmt1 gene (849 bp) was amplified by RT-PCR using primers (forward 5′ TTTCGTCATCTTCCACAACC 3′; reverse 5′ TCAATACCCACAGCAAGGTC 3′) designed based on GenBank Accession no. DQ011229. A fragment of β-lactamase (βLac; 799 bp) was also amplified by RT-PCR from a pGEM-T-Easy vector (kind gift from J. P. Paluzzi, York University) using the following primers: forward 5′ ATTTCCGTGTCGCCCTTATTC 3′; reverse 5′ CGTTCATCCATAGTTGCCTGAC 3′. PCR products were concentrated and purified using a QIAquick PCR Purification kit (Qiagen Inc., Toronto, ON, Canada) and used to generate double stranded (ds)RNAs by in vitro transcription using the Promega T7 RiboMAX Express RNAi Kit (Promega, WI, USA). AeAmt1 or βLac dsRNA were delivered to larvae according to a previously described protocol (Singh et al., 2013). Briefly, groups of 50 larvae (1st and 2nd instar), which were initially reared as described above, were soaked for 2 h in 75 µl PCR-grade water containing 0.5 µg µl⁻¹ dsRNA and then transferred into 20 ml RO water. Ingestion of the dsRNA solution by larvae was confirmed in preliminary studies by soaking larvae in a solution containing 0.5% Bromophenol Blue and then assessing accumulation of the dye in the larval gut as described previously (Singh et al., 2013). Rearing water was refreshed every 2 days following dsRNA treatment. To confirm reductions in AeAmt1 transcript as a result of dsRNA treatment, total RNA was extracted and cDNA was generated (as described above) from pools of 10–15 larval whole bodies at days 3 and 6 post-dsRNA treatment. cDNA templates were then used in RT-PCR with the above primers and PCR products were resolved electrophoretically in a 1% agarose gel stained with ethidium bromide. Reductions in AeAmt1 in larval whole bodies and anal papillae as a result of dsRNA treatment were examined by western blotting as described above. Larval mortality between day 3 and day 6 post-dsRNA treatment was determined by subtracting the number of surviving larvae on day 6 from the number of larvae counted on day 3. NH₄⁺ flux across anal papillae of A. aegypti larvae was determined at 6 days post-dsRNA treatment by SIET (see below). The [NH₄⁺] in hemolymph was measured using ion-selective microelectrodes as previously described (Jonusaite et al., 2011). Microelectrodes were backfilled with 100 mmol l⁻¹ NH₄Cl and front filled with NH₄⁺ ionophore I, cocktail A. Microelectrodes were calibrated in solutions of 0.1, 1 and 10 mmol l⁻¹ NH₄Cl.

The SIET system used in this study has been described previously (Donini and O'Donnell, 2005; Nguyen and Donini, 2010). Microelectrodes were backfilled with 100 mmol l⁻¹ NH₄Cl and front filled with NH₄⁺ ionophore I, cocktail A (Sigma-Aldrich). Microelectrodes were calibrated in solutions of 0.1, 1 and 10 mmol l⁻¹ NH₄Cl. Larvae were blotted dry on a piece of filter paper and then immobilized in a Petri dish using beeswax, such that one side of an anal papilla was exposed for measurements. Voltage gradients adjacent to the anal papilla were recorded under 4 ml of 0.5 mmol l⁻¹ NH₄Cl.

The recording protocol was as follows: the ion-selective microelectrode was moved to a target site ∼5 µm away from the surface of the anal papilla and a voltage was recorded. A second voltage recording was then obtained at ∼105 µm from the surface of the anal papilla. The protocol was repeated four times at each target site and consisted of 4 s wait time and 1 s sampling time. Readings were taken at the middle one-third portion of the anal papilla at four target sites separated by 25–30 µm. A voltage gradient was calculated by Automated Scanning Electrode Technique software (ASET; Science Wares, East Falmouth, MA, USA), using the differences between voltage readings at the two points of each site, and is reported as the average of the four readings at each site. Background voltage readings were taken by moving the microelectrodes 2500 µm away from the preparation and employing the same sampling protocols. Background voltage gradients (noise) were then subtracted from the gradients reported at each target site.

The NH₄⁺ fluxes were recorded from larvae treated with AeAmt1 or βLac dsRNA and larvae treated with pharmacological transport inhibitors in a bathing solution of 0.5 mmol l⁻¹ NH₄Cl. For the latter, larvae reared in RO water were pre-incubated for 30 min under control or experimental conditions. Pre-incubated experimental larvae were exposed to the following concentration of inhibitor (transporter/mechanism targeted is given in parentheses): 100 µmol l⁻¹ bafilomycin (VA); 100 µmol l⁻¹ ouabain (NKA); 100 µmol l⁻¹ 5-(N,N)-hexamethylene amiloride, HMA (NHE); 5 µmol l⁻¹ S3226 (NHE3); 100 µmol l⁻¹ methazolamide (carbonic anhydrase, CA). Selection of the concentration for each inhibitor was based on previous invertebrate or vertebrate tissue studies (MacVicker et al., 1993; Marcaggi et al., 1999; Vallon et al., 2000; Weihrauch, 2006; Tsui et al., 2009; Nguyen and Donini, 2010; Del Duca et al., 2011). All the inhibitors contained 0.1% DMSO in 0.5 mmol l⁻¹ NH₄Cl, therefore control larvae were pre-incubated in 0.5 mmol l⁻¹ NH₄Cl containing 0.1% DMSO. The Nernst slopes of NH₄⁺ ion-selective microelectrodes were (mean±s.e.m.) 51.6±0.5, N=6, for the dsRNA experiments and 54.3±0.5, N=54, for the pharmacology experiments.

Data were analyzed using Prism^® 5.03 (GraphPad Software Inc., La Jolla, CA, USA) and expressed as means±s.e.m. Two-tailed t-tests were used to determine significance between control and experimental groups. Control groups as well as ion-selective microelectrode calibration data were compared using a one-way ANOVA to determine statistical similarity. For SIET data, a single biological replicate is defined as the average flux of 6 sites along a single papilla from a single larva.

Quantitative analysis of transcript abundance of three putative ammonia transporters in anal papillae revealed that AeRh50-1 transcript levels were 10-fold greater than AeRh50-2 transcript levels (Fig. 2A). Furthermore, AeAmt1 transcript abundance was 12- and 120-times greater than that of AeRh50-1 and AeRh50-2, respectively. Western blot analysis of anal papillae from larvae revealed an AeAmt1 pre-protein, based on its putative amino acid sequence, at ∼55 kDa, and potential AeAmt1 monomer and trimer forms, which were all blocked by antibody pre-absorption with the immunogenic peptide (Fig. 2B). While the AeAmt1 monomer was always detected in western blots, the pre-protein and trimer forms were not consistently detected between samples; therefore, only the monomer form could be quantified for dsRNA-treatment experiments (see below).

AeAmt1 co-immunolocalized with basally expressed NKA in the anal papillae epithelium (Fig. 3A,D,G). However, AeAmt1 immunoreactivity was also observed in a few regions of the epithelium where NKA was not present (Fig. 3B,E,H). Although there was overlap in immunoreactivity between AeAmt1 and the V₀ subunit of VA at the interface between the apical and basal domains of the anal papillae epithelium, AeAmt1 did not co-immunolocalize with the apically expressed V₀ subunit (Fig. 3C,F,I). Immunostaining was absent in control sections that were probed with secondary antibodies only or with AeAmt1 antibody that was pre-absorbed with the immunogenic peptide (not shown).

Larvae treated with AeAmt1 dsRNA showed a qualitative reduction in AeAmt1 transcript at day 3 post-AeAmt1 dsRNA treatment (Fig. 4A), and while there was an apparent recovery of transcript levels at day 6 post-treatment, the protein monomer was significantly reduced (Fig. 4B,C). Also, at day 6 post-treatment with AeAmt1 dsRNA, larvae had elevated NH₄⁺ levels in hemolymph (Fig. 4D); however, although there was an apparent increase in larval mortality, this was not statistically significant (Fig. 4E). Anal papillae of larvae at day 6 post-AeAmt1 dsRNA treatment showed a ∼37% reduction in AeAmt1 monomer levels when compared with βLac dsRNA-treated groups (Fig. 5A,B). NH₄⁺ flux measurements across the anal papillae of larvae by SIET at day 6 post-AeAmt1 dsRNA treatment also showed a ∼3.3-fold decrease relative to control animals (Fig. 5C).

Incubation of larvae in the VA inhibitor bafilomycin A1 resulted in a reduction of net ammonia efflux (measured as [NH₄⁺] gradient) at the anal papillae from 47.6 to 13.3 pmol cm⁻² s⁻¹ (Fig. 6A). A similar result was obtained with ouabain, an inhibitor of NKA, which resulted in net NH₄⁺ fluxes that were less than half of flux rates measured at the papillae of control larvae (Fig. 6A). In contrast, the NHE inhibitor HMA did not alter the NH₄⁺ fluxes at the anal papillae (Fig. 6A); however, the specific NHE3 inhibitor S3226 decreased NH₄⁺ efflux (Fig. 6A). NHE3 was localized to the apical side of the anal papilla epithelium (Fig. 6B) and western blots of anal papillae homogenate revealed a 120 kDa band corresponding to the predicted size of the high molecular mass variant of A. aegypti NHE3 (Fig. 6C). The CA inhibitor methazolamide resulted in NH₄⁺ fluxes that were only 15% of flux rates recorded at papillae of control larvae (Fig. 6A).

The transcripts of three putative ammonia transporters AeRh50-1, AeRh50-2 and AeAmt1 were detected in the anal papillae of A. aegypti larvae in agreement with our hypothesis (Weihrauch et al., 2012b; Fig. 2A). The transcript abundance of AeAmt1 was significantly greater than that of the RhGP homologs (Fig. 2A). In light of this and the previously recorded ammonia efflux at the anal papillae, we focused our work on the Aedes Mep/Amt homolog. NKA and VA have been localized to the basal and apical membranes of the anal papillae epithelium, respectively (Patrick et al., 2006), and were therefore used to delineate the localization of AeAmt1 in larval anal papillae. Co-immunolocalization of AeAmt1 with NKA (Fig. 3G) indicates expression of AeAmt1 on the basal side. In freshwater-reared A. aegypti larvae, the basal membrane of the anal papillae epithelium forms extensive and deep folds into the cytoplasm that can extend more than halfway into each epithelial cell (Sohal and Copeland, 1966). Given that AeAmt1 did not co-immunolocalize with VA, but some overlap at the interface between AeAmt1 and VA was observed (Fig. 3I), it is likely that AeAmt1 expression extends into the deep infoldings of the basal membrane of the anal papillae epithelium where NKA is not present (Fig. 3H).

dsRNA-mediated reductions of AeAmt1 in larval A. aegypti (Figs 4 and 5) were associated with significantly decreased NH₄⁺ efflux from anal papillae (Fig. 5C) and elevated levels of NH₄⁺ in larval hemolymph (Fig. 4D). In larvae of freshwater zebrafish, an Rh50 protein, Rhcg1, is required along with VA to actively excrete ammonia against a higher external concentration (Shih et al., 2013). The transcript abundance of Rhcg1 and VA increases with exposure to high environmental ammonia (HEA) and the capacity for ammonia excretion by ammonia-excreting cells increases (Braun et al., 2009). HEA exposure of the fully aquatic, freshwater-inhabiting African clawed frog caused a decrease in the transcript levels of Rhbg, VA and NKA in the dorsal and ventral skin and reduced the capacity of the skin to excrete ammonia (Cruz et al., 2013). Therefore, observations from ammonia-excreting tissues of freshwater animals exhibit a positive correlation between ammonia transporter transcript levels and ammonia excretion rates. The comparatively high levels of AeAmt1 transcript along with the elevated ammonia levels of the hemolymph, with decreased ammonia efflux at the anal papillae in AeAmt1 dsRNA-treated larvae suggest that AeAmt1 is an important ammonia transporter for ammonia excretion at the anal papillae of larval A. aegypti. Therefore, the results are in agreement with our hypothesis that the anal papillae express the putative ammonia transporters that are involved in ammonia excretion.

Exposing A. aegypti larvae to ouabain (an NKA inhibitor) resulted in NH₄⁺ fluxes that were less than half of those measured at the anal papillae of control larvae (Fig. 6A). There are at least two possible explanations for the effect of ouabain on ammonia excretion. First, NKA may transport ammonia as NH₄⁺, which competes with K⁺ or where there are separate binding sites for these two ions (Skou, 1960; Masui et al., 2002; Cruz et al., 2013). There is considerable evidence for this ammonia transport mechanism in a number of animal epithelia, such as crab gills and the antennal gland (see Weihrauch et al., 2004, for review), frog skin (Cruz et al., 2013), fish gills (Mallery, 1983), the epidermis of planaria (Weihrauch et al., 2012b) and potentially larval zebrafish skin (see Shih et al., 2013). Second, as NKA is co-localized with AeAmt1 on the basal side, it can provide a strong voltage gradient to drive NH₄⁺ through AeAmt1 from the hemolymph to the cytosol. Either of these two mechanisms could result in decreased ammonia excretion if NKA is inhibited by ouabain.

Application of bafilomycin A (a VA inhibitor) to larval A. aegypti resulted in significantly lower NH₄⁺ fluxes measured at the anal papillae relative to control larvae (Fig. 6A). VA in the apical membrane of the anal papillae epithelium excretes H⁺, which could then be used in ammonia trapping, with NH₃ being supplied via a putative NH₃ transporter, e.g. RhGP. Evidence for this mechanism of ammonia trapping at the apical membrane of epithelia in freshwater exists in a number of animals, such as zebrafish larval skin (Shih et al., 2013), the planarian epidermis (Weihrauch et al., 2012a), frog skin (Cruz et al., 2013) and fish gills (Wilson et al., 1994). The apical localization of VA in the anal papillae suggests that VA contributes to an ammonia-trapping transport mechanism; however, the specific putative ammonia transporter involved remains to be identified. As AeAmt1 localizes to the basal side, an apical transporter participating in an ammonia-trapping mechanism with VA is likely to be one of the AeRh50 proteins. In addition, like NKA, VA likely contributes to a strong, cytosol negative voltage gradient and inhibition of VA may then also affect NH₄⁺ entry across the basal membrane through AeAmt1.

Incubation of mosquito larvae with methazolamide, a CA inhibitor, resulted in NH₄⁺ efflux at the anal papillae that was significantly reduced compared with controls (Fig. 6A). This result lends further support for an ammonia-trapping mechanism at the anal papillae, whereby cytoplasmic CA, at least in part, supplies VA with H⁺ or where an apical membrane-bound CA directly acidifies the boundary layer. CA has been implicated in contributing to acidification of the fish gill boundary layer through the hydration of CO₂, which results in the production of H⁺ which can then serve for ammonia trapping (see Wright and Wood, 2009). Apical membrane-bound CA may do this directly in freshwater; however, cytoplasmic CA may also contribute by providing H⁺ for VA (Georgalis et al., 2006). Additionally, one may speculate that CO₂, the substrate for CA, enters the papilla syncytium, via one of AeRh50 proteins on the basal side, as RhGPs have also been strongly implied to function as CO₂ channels (Kustu and Inwood, 2006; Perry et al., 2010).

The amiloride-based inhibitor of NHEs HMA did not affect NH₄⁺ fluxes at the anal papillae (Fig. 5); hence, we conclude that amiloride-sensitive NHEs are not involved in ammonia transport by anal papillae. A leucine amino acid residue substitution in the NHE3 of A. aegypti renders this exchanger amiloride insensitive (Pullikuth et al., 2006). Application of S3226, a specific inhibitor of NHE3 in mammals (Schwark et al., 1998), to larval mosquitoes resulted in decreased NH₄⁺ efflux at the anal papillae (Fig. 6A), suggesting that an NHE, likely NHE3, plays a role in ammonia excretion. The high molecular mass variant of Aedes NHE3 is present in the anal papillae (Fig. 6C) and is located on the apical side of the anal papilla epithelium (Fig. 6B). These results suggest that NHE3 may directly transport ammonia out of the cytosol in exchange for a cation (e.g. Na⁺), or by exchanging H⁺ with cations that may participate in ammonia trapping.

The anal papillae of mosquito larvae are important sites for ammonia excretion, similar to the gills of fish and crustaceans, and the skin of planarians, frogs and larval zebrafish. Based on the results of this study, we propose a working model (see Fig. 7) whereby NH₄⁺ enters the cytosol of the syncytial epithelium from the hemolymph through the basal membrane-localized AeAmt1 transporter. This is driven by the cytosol negative voltage potential established by the basal membrane-localized NKA, aided by the apical VA. In the cytosol, NH₄⁺ could be directly extruded by an apical NHE3 in exchange for a cation (e.g. Na⁺) or may dissociate into NH₃ and H⁺. The AeRhGPs have not been localized in the anal papillae; however, if located in the basal membrane, an AeRhGP may mediate entry of NH₃ and CO₂ from the hemolymph. NH₃ could be extruded via an apical AeRhGP in conjunction with H⁺ transport by an apical VA (NH₃ trapping). The acidification of the boundary layer by VA, and perhaps also by NHE3, is supported by the hydrolysis of CO₂ to yield H⁺ through the activity of CA.

Although a number of the mechanisms in our working model are supported solely by pharmacological studies and remain to be confirmed, the identification and localization of AeAmt1 expression, coupled to the reduction in ammonia excretion when AeAmt1 expression is repressed, strongly implicate AeAmt1 in ammonia excretion by anal papillae of mosquito larvae. The direction of NH₄⁺ transport by AeAmt1 is consistent with the Mep/Amt transporters of plants, bacteria and yeast, in that NH₄⁺ is conveyed into the cytosol. This is the first animal Mep/Amt protein family member to be implicated in ammonia excretion, a vital physiological process for all animals.

The authors thank Jean-Paul Paluzzi for advice on dsRNA experiments and Alex R. Quijada-Rodriguez for aiding with gene analysis.