Molecular characterization of sodium/proton exchanger 3 (NHE3) from the yellow fever vector, Aedes aegypti

Blood-feeding insects such as adult mosquitoes face an enormous challenge to void excess fluid imbibed from their vertebrate host. An exquisite interplay of hormonal and membrane transport processes facilitates rapid removal of excess ions and fluid from the insect hemolymph. Although the physiology of such diuretic mechanisms has been appreciated since the early studies (Ramsay, 1950; Wigglesworth, 1972), the molecular detail surrounding this efficient diuretic system remains largely elusive. In vertebrate models systems, a sodium/potassium ATPase(Na⁺/K⁺-ATPase) appears to play a critical role in facilitating membrane transport. This sodium motive paradigm is widely accepted. However, studies by Harvey and Wieczorek and their colleagues in insect epithelia indicate that a proton-motive transport process might be a key determinant in transport (Wieczorek et al., 2000). In lepidoteran midgut, the protons extruded from the cell into the lumen by vacuolar-type ATPase (V-ATPase) dictate ion transport through a functionally coupled potassium pump (K-pump or antiporter). The activities of V-ATPase together with the K-pump are crucial to maintain an alkaline midgut in lepidoptera (Dow,1999; Wieczorek et al.,1999). Pharmacological studies in various insect epithelia indicate the K-transport component to be reminiscent of sodium/proton exchangers (NHEs) that regulate intracellular pH as well as a variety of other functions in vertebrates (Burckhardt et al., 2002; Counillon and Pouyssegur, 2000; Grinstein and Wieczorek, 1994; Orlowski and Grinstein, 2004). However, no NHE `isoform' has been molecularly characterized in any insect system, although some studies have suggested their existence in some insect tissues involved in ion transport(Giannakou and Dow, 2001; Hart et al., 2002; Pannabecker, 1995; Petzel, 2000)

Given the importance of NHEs in regulating cellular and systemic pH, and ion transport and their potential to be functionally coupled to the activities of V-ATPases, molecular identification of members of this family is required in order to formulate a comprehensive model to understand how mosquitoes efficiently regulate ion and fluid balance following a blood meal. The cascades culminating in and initiated through these processes are crucial for the reproductive biology of mosquitoes and for disease transmission vectored by these insects.

Ion transport studies in insect epithelia show that the primary active transport of H⁺ by a V-ATPase drives secondary ion transport through Na⁺ or K⁺/H⁺ exchanger(s) that together function as primary generators of electrochemical gradients(Pannabecker, 1995; Wieczorek, 1992). Protons extruded by V-ATPase into the lumen are cycled by the exchangers allowing K⁺ (or Na⁺) and fluid secretion(Pannabecker, 1995; Wieczorek, 1992). Implicit in this model is that the K⁺ (or Na⁺)/H⁺exchanger virtually operates in reverse to pump H⁺ from the lumen in to the cell, which is a property exploited in vertebrate epithelial cells to clone NHE-deficient cell lines (Paris and Pouyssegur, 1983; Pouyssegur et al., 1984). The molecular identity of these Na⁺ or K⁺/H⁺exchangers has been suggested from pharmacological studies on isolated Malpighian tubules (Petzel,2000), or from the recently completed insect genome sequences(Adams et al., 2000; Holt et al., 2002; Pullikuth et al., 2003).

The Malpighian tubules of insects are robust ion- and fluid-transporting tissues (Maddrell and O'Donnell,1992; Maddrell,1991), and in the mosquito Aedes aegypti, Malpighian tubules secrete 0.4 nl min^-1 spontaneously. When stimulated with natriuretic peptides (MNP) or cAMP, tubules secrete at a rate of 2.8 nl min^-1, simulating secretion rates after a blood meal(Petzel et al., 1987; Petzel et al., 1986). Microfluorimetric and microelectrode measurements showed the rapid secretion,which is coupled to a bafilomycin-sensitive V-ATPase, is sensitive to amiloride or its analogs; these features are reminiscent of a Na⁺or K⁺/H⁺ exchanger(Beyenbach et al., 2000; Petzel, 2000). Further, NHE antagonists reduced recovery of pH_i following an acid load in isolated Malpighian tubules (Petzel et al., 1999). Thus, pH_i regulation by exchangers is directly linked to the fluid secreting efficiency of tubules.

The mosquito, Ae. aegypti, is a vector of human diseases including hemorrhagic dengue fever and yellow fever. Ingestion of a blood meal by the adult female results in an enormous Na⁺ load and increase in fluid volume, both of which have to be regulated rapidly. Although prevailing models suggest the involvement of NHE-like proteins in regulating ionic and fluid secretion in mosquitoes, their diversity, cellular expression, localization,functions and/or sensitivity to inhibitors remain incompletely understood in these insects. Here we present the molecular and functional characterization of Ae. aegypti NHE3 (GenBank accession numbers reported are AF187723). We show that AeNHE3 complements yeast NHE and heterologous expression in NHE-deficient epithelial cells results in the transport of ²²Na⁺ and in the recovery of pH_i after an acid load. Immunohistochemical evidence for its localization in the basolateral plasma membrane domain suggests that current models of ion transport in Malpighian tubules need to be re-evaluated.

Restriction and DNA modifying enzymes were from New England Biolabs. Cyanine3 (Cy3)- and Cy5-conjugated secondary antibodies were from Jackson ImmunoResearch Laboratories. A sodium/proton exchanger activity deficient cell line (PS120) derived from Chinese hamster lung fibroblast cell line CCL39 was obtained from Dr James Melvin (Attaphitaya et al., 1999) (University of Rochester, Rochester, NY, USA) with the kind permission of Dr Jacques Pouyssegur (Centre Antoine Lacassagne, Nice,France) (Pouyssegur et al.,1984). A dual-function expression vector, pXOON, was a gift from Dr Thomas Jespersen (University of Copenhagen, Denmark)(Jespersen et al., 2002). Nigericin, Alexa Fluor 488-phalloidin and 2′,7′-bis-(2-carboxyethyl-5-(and-6)-carboxyfluorescein,acetoxymetyl ester (BCECF-AM) were purchased from Molecular Probes. Geneticin(G418) and cell culture reagents were from Invitrogen (Carlsbad, CA, USA). Carrier free ²²Na⁺ was purchased from Amersham Biosciences (Piscataway, NJ, USA). Unless stated otherwise, all other reagents were purchased from Sigma-Aldrich.

Isolation of NHE3 cDNA and analysis of protein sequence Partial cDNAs were obtained using degenerate primers to conserved NHE amino acid sequences, LDAGYFMP and AVDPVAVFE, and mRNA isolated from the midgut and Malpighian tubules of Aedes aegypti L. Gene specific primers were then designed and used for screening an Ae. aegypti Malpighian tubule cDNA library as described (Ross and Gill,1996). The cDNA clone isolated was sequenced completely in both directions to obtain the nucleotide and deduced amino acid sequences.

Transmembrane predictions were inferred from SOSUI analysis at Sosui/proteome.bio.tuat.ac.jp(Hirokawa et al., 1998). Potential phosphorylation sites were identified through high stringency scans by NetPhos and ScanSite algorithms at www.cbs.dtu.dkand scansite.mit.edurespectively. Gene structure analysis was performed with Aedes aegypti NHE3 ORF against the recently released Ae. aegypti WGS scaffolds.

S. cerevisiae strains G19 (MATα, ade2, his3, leu2, trp1,ura3, Δena1::HIS3::ena4) and AXT3 (MATα, ade2, his3, leu2, trp1,ura3, Δena1::HIS3::ena4, nha1::LEU2, nhx1::TRP1) were gifts from Dr Jose M. Pardo (Consejo Superior de Investigaciones Cientificas, Sevilla, Spain) and have been previously characterized (Madrid et al., 1998; Quintero et al.,2000). Both strains are derivatives of W303-1B (MATα, ura3-1 leu2-3, 112his3-11, 15trp1-1, ade2-1, can1-100). pYES2.1 TOPO: Aedes NHE3 plasmid was transformed into yeast cells following the LiCl method. Growth in high Na⁺ was assayed in alkali cation-free arginine phosphate (AP)medium (Rodriguez-Navarro and Ramos,1984) containing known concentrations of NaCl. Tolerance to hygromycin B was assayed in minimal medium.

The Aedes NHE3 ORF was amplified using Expand Hi-Fidelity PCR system (Roche Biochemicals, Indianapolis, IN, USA) and cloned as two separate fragments into pXOON vector (Jespersen et al., 2002). First, a 2.2 kb 5′ fragment of ORF that codes for the N-terminal 731 amino acids was cloned in to a dual expression vector pXOON (designated NHE-731Δ). NHE-731Δ lacks the C-terminal cytoplasmic tail but contained the sequence (⁷³²TGDIGPAGHDRAAA stop) that was coded by the cloning site and the vector followed by several in-frame stop codons. Second, a 1.2 kb fragment coding for the C-terminal tail was seamlessly appended to NHE-731Δ to reconstitute the full-length Aedes NHE3 ORF. The expression vector pXOON also codes for EGFP from a different promoter site enabling isolation of transfected cells by GFP fluorescence. All constructs were fully sequenced to confirm that no PCR errors were introduced.

PS120 cells (Pouyssegur et al.,1984) in Dulbecco's modified Eagle medium (DMEM) supplemented with penicillin/streptomycin/fungizone (Invitrogen) and 10% fetal bovine serum(FBS) were transfected using Lipofectamine 2000 (Invitrogen) either with 5-12μg of PmeI linearized expression constructs or vector alone. After 2 days, selection medium containing G418 (1 mg ml^-1) (in DMEM, 10%fetal bovine serum and antibiotics) was added and growth maintained until individual foci were apparent. Transfected cells were also selected with HBS(20 mmol l^-1 Hepes-KOH, 5 mmol l^-1 glucose, 2 mmol l^-1 CaCl₂, 5 mmol l^-1 KCl, 1 mmol l^-1 MgCl₂) buffered HCO₃^--free DMEM, pH 6.9 to hasten the selection process. The parental PS120 cells are conditional for growth in bicarbonate-free medium(Pouyssegur et al., 1984) at acidic and neutral pH. Therefore, clones that survived the acid challenge in selection medium contained functionally expressed NHE. Fifteen independent clonal lines were selected for each NHE expression construct. Clones were initially assayed for recovery of intracellular pH (pH_i) with the cell-permeant fluorescent pH indicator, 2′,7′-bis -(2-carboxyethyl-5-(and -6) - carboxyfluorescein, acetoxymetyl ester (BCECF-AM:Molecular Probes) as described for measurement of pH_i below. Selected clones were expanded and frozen or maintained in selection medium. Alternatively, 2 weeks after transfection, plates were incubated in HCO₃^--free DMEM for 3-8 days, dead cells were removed and viable cells trypsinized and maintained as polyclonal populations in medium containing G418 (1 mg ml^-1).

PS120 cells were cultured in bicarbonate buffered DMEM with 10% fetal calf serum. At 90-95% confluency each 10 cm² plate was trypsinized. The cell pellet was resusupended in culture medium and equal volumes added to each well in a 24-well plate. The day after plating, cells were transfected with 0.8 μg plasmid DNA with Lipofectamine 2000 (Invitrogen) according to manufacturer's recommendation. Uptake assays were performed 2-3 days after transfection. Culture medium was aspirated and wells washed quickly twice with 1 ml of acid load buffer (50 mmol l^-1 NH₄Cl, 70 mmol l^-1 choline chloride, 1 mmol l^-1 MgCl₂, 2 mmol l^-1 CaCl₂, 5 mmol l^-1 glucose, 20 mmol l^-1 Hepes-Tris, pH 7.4) and incubated in 0.5 ml of the same buffer for 30 min at 37°C (in nominal CO₂)(Orlowski, 1993; Wakabayashi et al., 1992). Cells were washed twice and incubated for 5 min at room temperature in choline chloride buffer (135 mmol l^-1 choline chloride, 1 mmol l^-1 MgCl₂, 2 mmol l^-1 CaCl₂, 5 mmol l^-1 glucose, 20 mmol l^-1 Hepes-Tris, pH 7.4). To initiate ²²Na⁺ uptake, buffer solution was aspirated and 250 μl of uptake buffer was added and incubated for 20 min at room temperature. Uptake buffer contained 1 μCi ml^-1 (1 Ci=3.7×10¹⁰ Bq) of carrier-free ²²Na⁺in choline chloride buffer supplemented with 1 mmol l^-1 ouabain and 100 μmol l^-1 bumetanide from 0.5 mol l^-1 and 0.1 mol l^-1 stocks in 100% DMSO, respectively. Uptake was stopped by adding 1 ml of ice-cold stop solution (135 mmol l^-1 NaCl, 1 mmol l^-1 MgCl₂, 2 mmol l^-1 CaCl₂, 5 mmol l^-1 glucose, 4 mmol l^-1 KCl, 20 mmol l^-1Hepes-Tris, pH 7.4) and quickly rinsed four times with the same solution. Cells were solubilized in 0.5 ml of 0.5 mol l^-1 NaOH and added to 0.5 ml of 0.5 mol l^-1 HCl wash collected after rinsing the wells(Orlowski, 1993). Both were pooled, added to 2-4 ml of scintillation fluid and counted. A 50 μl sample of NaOH- and Hcl-extracted and pooled samples was saved for protein estimation with BCA reagent (Pierce, Rockford, IL, USA) with bovine serum albumin (BSA)as standard. The effect of inhibitors was assessed by adding appropriate amounts from stock solutions made in 100% DMSO. An equivalent volume of solvent was added to control wells to account for any carrier effect on uptake. Data were normalized to protein concentration in each well.

Monoclonal antibody to avian Na⁺/K⁺-ATPase(Takeyasu et al., 1988) was obtained from the Developmental Studies Hybridoma Bank, University of Iowa,IA, USA). Polyclonal antibody to B-subunit of vacuolar ATPase (V-ATPase) has been previously described (Filippova et al., 1998). A purified synthetic peptide corresponding to the C-terminal residues (E¹¹¹¹→G¹¹²⁸) of AedesNHE3 with an N-terminal cysteine was synthesized at Molecular Genetic Instrumentation Facility (University of Georgia, GA, USA). The synthetic peptide was conjugated to maleimide-activated KLH (Pierce) and used to immunize rabbits for antibody production. Production bleeds with highest titers, determined by ELISA, were used for immunohistochemical studies.

Fourth instar larvae and adult female Aedes aegypti were dissected in PBS and fixed overnight in 4% paraformaldehyde (PFA) at 4°C. Tissues were washed in PBS and dehydrated for 3-6 h each in 20, 40, 70 and 96% ethanol followed by three changes in 100% ethanol at room temperature. Ethanol was replaced with a xylene series (30% and 70%) and with two changes in 100% for 6-16 h at each step. Paraplast chips were added to tissues in xylene and incubated at 58°C, followed by complete infiltration with 100% paraplast. Microtome sections of 8 μm were cut and adhered to silane-prep slides(Sigma-Aldrich).

Tissue sections were deparaffinated with xylene, rehydrated in a descending ethanol series and washed with PBS-Triton X-100 (PBS-Tx; 0.1%Triton X-100,1×PBS, pH 7.4). Sections were blocked with 2%BSA for 2 h at room temperature. The appropriate dilution of antibodies (in PBS-Tx) were added after blocking and incubated at 4°C overnight. Unbound material was removed and tissues further blocked in 2% normal goat serum (NGS). Cy3- or Cy5-conjugated to either anti-rabbit or anti-mouse antibodies diluted in PBS-Tx were used as secondary antibodies. After 1-2 h of incubation with secondary antibody in the dark, tissues were washed with PBS-Tx and mounted for microscopic examination in 90%glycerol/4%N-propyl gallate.

Dual labeling experiments were done using whole rabbit serum against Aedes NHE3 (1:250 dilution) and mouse anti-avian Na⁺/K⁺-ATPase antibody (a5 culture supernatant, 1:10 dilution). In all experiments, preimmune serum, or antibody preabsorbed with antigen, were used as negative controls. Whole-mount immunohistochemistry was performed with similarly fixed and blocked tissues. Anti-NHE3 and anti-Na⁺/K⁺-ATPase antibodies were used at 1:500 and 1:20 dilution, respectively.

Sections and whole mounts were examined with a Zeiss Axioplan confocal microscopy (LSM510) at the Center for Advanced Microscopy and Microanalysis at University of California, Riverside, CA, USA. All images were imported to Adobe Photoshop (6.0) where final assembly and labeling were done.

Insect saline contained 156 mmol l^-1 NaCl, 6 mmol l^-1KCl, 5 mmol l^-1 glucose, 2 mmol l^-1 CaCl₂, 1 mmol l^-1 MgCl₂ and 20 mmol l^-1 Hepes-KOH, pH 7.3 (Petzel et al., 1999). Other solutions were prepared in Hepes-buffered saline (HBS: 20 mmol l^-1 Hepes-KOH, pH 7.4, 5 mmol l^-1 KCl, 1 mmol l^-1 MgCl₂, 2 mmol l^-1 CaCl₂, 5 mmol l^-1 glucose). Normal sodium medium was prepared in HBS containing 135 mmol l^-1 NaCl. For sodium-free medium, NaCl was replaced with 135 mmol l^-1N-methyl-d-glucamine(NMDG). Ouabain was prepared as 10 mmol l^-1 stock in HBS or choline chloride buffer. Bumetanide was prepared as 0.1 mol l^-1 stock in 100% dimethyl sulfoxide (DMSO).

Cells were plated at a density of 30-50×10⁴ per well in 48-well culture plates and grown in DMEM with 10% FBS,penicillin/streptomycin/fungizone (Invitrogen) with or without G418 (1 mg ml^-1) for transfected and untransfected cells, respectively. Two days after plating, the culture medium was aspirated, wells washed with normal sodium buffer twice and loaded with the fluorescent indicator dye BCECF (5μmol l^-1 in 135 mmol l^-1 NaCl, HBS, pH 7.4) for 30 min at 37°C. Cells were pulsed with ammonium chloride (60 mmol l^-1 NH₄Cl, 75 mmol l^-1 NaCl, in HBS, pH 7.4)for 10 min at 37°C. To increase intracellular acidity, cells were then washed with sodium-free medium (HBS, pH 7.3, 135 mmol l^-1 NMDG)twice and incubated further for 5 min at room temperature. Appropriate buffers were then added and intracellular change in BCECF fluorescence monitored as end-point measurements after 7-10 min. Plates were scanned in a Typhoon 9410 variable mode imager (Molecular Dynamics, Sunnyvale, CA, USA) equipped with blue laser with dual excitation of 457 and 488 nmol l^-1 and emission recorded through a 555 band pass 20 nmol l^-1 filter. Plates were scanned at a resolution of 200 μm and images acquired by ImageQuant 5.1 software, exported as Microsoft Excel spread sheet and analyzed in Microcal Origin 6.1. Calibration of pH_i was performed by high KCl/nigericin technique (Zhang et al.,1992). Briefly, cells were incubated with 10 μmol l^-1 nigiricin containing KCl buffers (HBS, 150 mmol l^-1KCl) adjusted for pH between 5.5 and 8.0. Duplicate wells in each plate were used for each pH value (total of 8 points in the range). Emissions at pH-insensitive excitation of 457 and pH-sensitive excitation of 488 were derived as ratios and converted to pH values by a mathematical fit. A calibration curve from each plate was used to determine the pH_i in experimental wells. The contributions of Na⁺/K⁺-ATPase and Na⁺/K⁺/Cl^- cotransporters to ion influx was determined by incubating cells in parallel with buffers containing 1 mmol l^-1 ouabain and 100 μmol l^-1 bumetanide,respectively.

Degenerate primers toward conserved vertebrate NHE sequences were used to isolate a PCR product from the midgut and Malpighian tubules mRNA of Aedes aegypti. This partial sequence was then used to isolate a full-length cDNA coding for the Aedes NHE homologue from a Malpighian tubule cDNA library. The cDNA was 3981 bp long with 3540 bp (from 165 to 3704 bp) open reading frame translating to 1179 amino acids. A perfect polyadenylation signal (AAUAAA) at 3951 bp is located immediately upstream of the poly(A)tail. Our preliminary analysis of Aedes aegypti genome at the NHE3 locus indicates that the open reading frame is coded by 21 exons covering a rather large genome segment (ca. 154 kb)(Fig. 1A, asterisks; data not shown). Exon 1 (5′UTR + amino acids 1-182) is the largest coding exon whereas exon 2 (amino acids 183-194) is the shortest. The size of introns varies from 63 bases (between exons 9 and 10) and 31.7 kb (between exons 3 and 4).

The Aedes NHE protein product is estimated to be 130.2 kDa(pI=6.78) (Fig. 1A). By sequence similarity to vertebrate NHEs and evolutionary relationship to cloned NHEs (not shown) (Pullikuth et al.,2003), we assign Aedes NHE to the NHE3 family (hereafter, AeNHE3). AeNHE3 is predicted to contain 12 transmembrane helices with both N and C termini oriented cytoplasmically(Hirokawa et al., 1998). The C terminus accounts for 56% of the protein and contains several putative regulatory sites (Fig. 1B). AeNHE3 apparently lacks potential N-glycosylation sites but contains potential sites for cAMP- and cGMP-dependent protein phosphorylation(T668, S820, S935, S1017, S1099), and casein kinase II phosphorylation(S1162). AeNHE3 is likely to be phosphorylated by ERK1&2 since a consensus site (PX(S/T)P) for ERK is found at T307 within the ERK substrate motif 305PLTP308. Further, two ERK D-domains necessary for ERK docking to its substrates are embedded around V808 and L1103. The preponderance of these sites in the C-terminal tail indicates that AeNHE3 function is likely regulated by phosphorylation by several kinases in metabolic, physiological and cell cycle contexts, as with many vertebrate NHEs(Counillon and Pouyssegur,2000; Orlowski and Grinstein,1997). AeNHE3 also contains a calcineurin B homologous protein (CHP) binding site in the cytoplasmic tail(Fig. 1D)(Pang et al., 2001).

The AeNHE3 protein is nearly identical to two PCR-derived Aedes NHEs (Hart et al.,2002) and is similar to Drosophila NHE2 (53%)(Giannakou and Dow, 2001) and vertebrate NHE3s (50%) (Biemesderfer et al., 1993; Brant et al.,1995; Tse et al.,1993) (Fig. 1A). Since Drosophila NHEs were presumably named by the order of discovery, we assign Drosophila NHE2, originally annotated by Giannakou and Dow to the NHE3 family(Giannakou and Dow, 2001),hence termed DmNHE3. Partially using the AeNHE3 sequence reported here, Hart et al. identified two PCR products that exhibited >95% identity(Hart et al., 2002). It is interesting to note that the shorter 2.8 kb PCR product (mNHE2.8 kb), which differs at 11 positions to the longer 3.7 kb fragment (mNHE3.7 kb), ends precisely after exon 10. This data, together with the presence of a distinct 3′ UTR in the mNHE2.8 kb PCR product(Hart et al., 2002), suggests that AeNHE3 is likely expressed as at least two splice variants.

Polyclonal antibodies to a C-terminal peptide(E¹¹¹¹→G¹¹²⁸) of AeNHE3 with an added N-terminal cysteine were generated and used to examine the distribution of AeNHE3 in larval and adult tissues of Ae. aegypti. In larvae, higher level of AeNHE3 expression was detected in the distal end of posterior midgut, Malpighian tubules and hindgut(Fig. 2A), while the anterior midgut showed reduced staining for NHE3. Preimmune antibody or immune antibody preabsorbed with the C-terminal synthetic peptide antigen did not show any specific labeling (Fig. 2B). In adult mosquitoes, both anterior midgut and the proximal section of posterior midgut expressed higher levels of NHE3, together with Malpighian tubule and hindgut. Remarkably, the distal section of posterior midgut exhibited only weaker labeling (Fig. 2C). In both anterior (Fig. 2D) and posterior (Fig. 2E) midgut,NHE3 was found to localize exclusively to the basolateral domain of the plasma membrane.

In gastric caeca, NHE3 expression predominates in the distal part of caeca known to actively transport ions (Fig. 3A,B) (Ramsay,1950; Ramsay,1951; Volkmann and Peters,1989a; Volkmann and Peters,1989b). Proximal segments of caeca contained weak, if any,staining for NHE3 (Fig. 3C). Immunohistochemistry with tissue sections revealed NHE3 to localize to the basal membrane; no apical staining was detected(Fig. 3B,C). Preimmune antibody(Fig. 3D) or immune antibody preabsorbed with the peptide antigen did not show any specific labeling. Double labeling with monoclonal antibody to Na⁺/K⁺-ATPase indicated that NHE3 expression is characteristically segregated from Na⁺/K⁺-ATPase expression (Fig. 3E-G). Na⁺/K⁺-ATPase is abundantly expressed in the proximal segment of the caeca (Fig. 3E)and localized to the basolateral membrane(Fig. 3F). In proximal and distal caecal cells, the V-ATPase localizes predominantly to the apical side(Fig. 3H,I), whereas in distal cells weaker staining was also detected in the basal side(Fig. 3H).

Ion and fluid transport in the Malpighian tubule is energized by a proton pump that combines a V-ATPase and a secondary ion transport system that returns luminal protons in exchange for cellular Na⁺ or K⁺ (Beyenbach, 1995; Pannabecker, 1995). Staining of Malpighian tubules with antibodies to B-subunit of the V-ATPase revealed that V-ATPase is primarily located in the apical membrane, consistent with its proposed role in proton secretion (Fig. 4A,B). Similar to gastric caeca and midgut, NHE3 localized to the basolateral membrane of Malpighian tubules in the proximal segment(Fig. 4D). Cytoplasmic staining as punctate structures were also evident in the Malpighian tubule, possibly representing an intracellular pool of NHE3(Fig. 4F). Basolateral expression of NHE3 reflects the localization pattern of Na⁺/K⁺-ATPase. In the principal ion-secreting cells in proximal tubules, Na⁺/K⁺-ATPase colocalized with NHE3 to the basal membrane (Fig. 4F). Interestingly, we also detected apical staining for NHE3 in median Malpighian tubule that appeared unique for this segment, as proximal and distal ends of the tubules showed only basal membrane staining with NHE3 antibody(Fig. 4E). Although hindgut expressed NHE3, the intensity of staining was much higher in the rectum where reabsorption of fluid and ions are known to occur in a variety of insects(Fig. 4G).

Yeast AXT3 cells that lack the Na⁺ efflux proteins ENA1-4, the plasma membrane Na⁺/H⁺ antiporter NHA1 and the vacuolar antiporter NHX1 are very sensitive to high sodium in growth media(Madrid et al., 1998; Quintero et al., 2000). Their salt-sensitive phenotype makes them a convenient tool for studying Na⁺ extrusion and/or sequestration ability of NHE-like proteins by heterologous protein expression. When AeNHE3 was expressed in this yeast strain under the control of the strong GAL 1 promoter, it restored salt tolerance at up to 70 mmol l^-1 sodium(Fig. 5A and data not shown),but failed to restore tolerance to hygromycin B(Fig. 5B). Thus, AeNHE3 rescues yeast plasma membrane NHE defects but does not complement vacuolar NHE (NHX) (see Discussion).

To further characterize AeNHE3 function, PS120 cells selected by H⁺-suicide that have been shown to lack a functional NHE(Pouyssegur et al., 1984) were used. These cells are conditional for growth in HCO₃^--free medium and lose viability in neutral and acidic pH (Pouyssegur et al.,1984). We selected AeNHE3-expressing cells by transfection and double selection in conditional medium. To avoid selecting spontaneous revertants, selected clones were screened for GFP fluorescence encoded by a non-contiguous gene under independent promoter control in the cloning vector, pXOON. All selected clones were positive for GFP and no clones survived acid load when transfected with the empty vector alone. We did not detect any spontaneous revertants in untransfected cells or in those transfected with pXOON alone.

Following an acid load, untransfected cells do not recover intracellular pH whereas unchallenged cells maintained in culture medium retain a near neutral pH (Fig. 6A). Polyclonal cells expressing AeNHE3 were capable of alkalinization following an acid load and subsequent change to Na⁺-buffer(Fig. 6B). We selected stable clones expressing AeNHE3 and assayed for intracellular pH recovery following an acid challenge (Fig. 6C). Our selection procedure included growth in bicarbonate-free medium buffered with 20 mmol l^-1 Hepes, pH 6.9. This growth condition was lethal to untransfected PS120 cells that are incapable of elevating intracellular pH in the absence of the activity of NHE1 and its coupled Cl^-/HCO₃^- transporter. The fact that we were able to select clonal lines under this conditional regime indicated that AeNHE3 was functional in PS120 cells, where it functions in a manner similar to vertebrate NHE1(Pouyssegur et al., 1984). Consistent with growth under non-permissive conditions in yeast AXT3 cells, AeNHE3-expressing PS120 cells recovered intracellular pH after an acid load (compare Fig. 6A,grey bar, with Fig. 6C).

In order to determine if the large cytoplasmic carboxy tail is required for AeNHE3 function we deleted the carboxy cytoplasmic tail of AeNHE (NHE-CΔ) and assayed for its capacity to recover intracellular pH following an acid load(Fig. 6D). In all cases,recovery was not significantly compromised, indicating that the cytoplasmic tail is not essential for NHE3 function, but could be required for functions that are distinct from ion and proton translocation or required for regulating its activity in response to cellular cues in mosquito tissues that are not recapitulated in this mammalian expression system.

Transiently transfected PS120 cells were assayed for ²²Na⁺ uptake in 12- or 24-well formats. Cells were either transfected with the vector alone (control) or with those expressing the full-length AeNHE3 ORF and assayed for sodium uptake following an acid load protocol. To determine the sensitivity of ²²Na⁺ uptake by AeNHE3 to amiloride and its analog we compared these agents at concentrations that were sufficient to completely inhibit mammalian NHE1. In the presence of 1 mmol l^-1amiloride or 100 μmol l^-1 EIPA, concentrations that completely abolish mammalian NHE1 mediated functions(Orlowski, 1993), ²²Na⁺ uptake was only reduced by 40%, showing that AeNHE3 is at best partially sensitive to amiloride and EIPA at very high concentrations (Fig. 7A). No further change in sensitivity to amiloride or EIPA was evident in cells lacking the carboxy tail of NHE3 (NHE-CΔ)(Fig. 7B). We conclude that the large cytoplasmic region of NHE3 is dispensable for its sodium transport function, similar to its proton translocation properties. However, this does not preclude the possibility that this region might possess subtle functional attributes by sensitizing the exchanger to small and transient changes in pH or other cellular aspects that are not reproduced in this vertebrate cell line. The data are consistent with the model that AeNHE3 is insensitive to traditional agents used to distinguish the contribution of several NHEs in Na⁺ and proton transport across membranes. Further,the transport properties require only the core domain of NHEs consisting of the transmembrane helices.

Here we report the cloning, tissue distribution, subcellular localization and functional characterization of an Aedes aegypti NHE3. High millimolar concentrations of sodium are toxic to the yeast, S. cerevisiae, and must be extruded through the plasma membrane(Rodriguez-Navarro et al.,1994), or sequestered in vacuoles(Darley et al., 2000). The four plasma membrane Na⁺-ATPases (ENA1-4)(Darley et al., 2000) and the plasma membrane Na⁺/H⁺ antiporter NHA1(Prior et al., 1996) extrude sodium from the cells; the vacuolar exchanger NHX1(Nass et al., 1997), on the other hand, sequesters it in vacuoles. AXT3 yeast cells have disruptions on ENA1-4, NHA1 and NHX1 genes, which render the strain highly sensitive to high sodium levels (Quintero et al.,2000). In addition, the functionality of vacuolar NHX1 can be determined by assaying the yeast cells' tolerance to hygromycin B; the precise mechanism by which NHX1 imparts this resistance is, however, not yet well understood (Gaxiola et al.,1999). AeNHE3 on expression in AXT3 cells conferred sodium-tolerance to a concentration as high as 70 mmol l^-1(Fig. 5 and data not shown),but offered no improvement in growth in media containing 100 μg ml^-1 hygromycin B. This indicates that the mosquito exchanger is not functional in vacuoles, and any salt-tolerance conferred on the mutant cells would be attributable to plasma membrane Na⁺/H⁺exchange. Wild-type yeast cells can grow in media with a sodium concentration as high as 200 mmol l^-1(Quintero et al., 2000). This genetic complementation assay adds to previous findings on Ae. aegypti NHE8 (W.K., A.K.P., K.A. and S.S.G., manuscript submitted for publication) showing that the genetic pliability of yeast eases the functional characterization of proteins, from organisms that are more difficult to manipulate such as the mosquitoes.

In insects, the Malpighian tubules, gastric caeca, midgut and hindgut are organs involved in fluid and ion homeostasis. Immunolocalization studies with an antibody to the carboxy tail epitope of NHE3 indicate that AeNHE3 is localized in almost all tissues examined, predominantly to the basal membrane. Interestingly, we also detected apically localized NHE3 in the median segment of Malpighian tubule (Fig. 4). Further, intracellular staining in Malpighian tubules is suggestive of NHE3 being sequestered in endomembrane compartments, likely representing the population being recycled through endocytosis. The different localization patterns of NHE3 might reflect its functional versatility. The basolateral localization of NHE3 was surprising since vertebrate NHE3 is expressed in apical membranes of renal tissues, apart from a distinct cytoplasmic pool (Biemesderfer et al.,1997; Chow et al.,1999; D'Souza et al.,1998; Janecki et al.,1998) that is recruited to the apical membrane in a phosphatidylinositol 3-kinase dependent manner(Kurashima et al., 1998). Thus, AeNHE3 seems to share a localization pattern similar to the vertebrate `housekeeping' NHE1 that resides in the basolateral membrane and also to vertebrate NHE3 that is present both in apical membrane and in endomembrane compartments. Future work should clarify the complex trafficking patterns that underlie NHE3 localization, dynamics and function in tubule physiology.

In his classical work in Aedes larvae, Ramsay showed that the midgut region corresponding to the 3-5 abdominal segment to be involved in ion and fluid transport (Ramsay,1951). Our results demonstrating the presence and enrichment of NHE3 precisely to that location in the midgut(Fig. 2)(Clements, 1992) support the possibility that NHE3 activity may play a role in ionic homeostasis in the midgut. This possibility is further strengthened by the expression pattern of NHE3 in gastric caeca (Fig. 3). Two distinct segments of the gastric caeca are known for ion and fluid transport. Reabsorbing/secreting cells (at proximal caeca) and ion transporting cells (at distal caeca) have been identified where secondary concentration and fluid secretion occur(Clements, 1992; Ramsay, 1950; Ramsay, 1951; Volkmann and Peters, 1989a; Volkmann and Peters, 1989b). Higher levels of NHE3 expression in distal gastric caeca thus might be reflective of NHE3's role in ion and proton exchange in the caeca. The proximal segment of gastric caeca expressed lower levels of NHE3, in contrast to Na⁺/K⁺-ATPase. Together with expression of V-ATPase in distal gastric caeca, our present results support the model for ion transport in insect epithelium where a V-ATPase and its coupled cation/H⁺ transporter constitute the major fluid and ion transport mechanism. However, both components of this pump are proposed to be located in the apical membrane, with little consideration given to a basolateral Na⁺/H⁺ exchanger. Because NHE3 is expressed in the basolateral membrane of gastric caeca and most of Malpighian tubules, we believe this isoform does not constitute the tightly coupled exchanger operating in parallel with V-ATPase in gastric caeca and most of the tubules. However, it is possible that homologue(s) of one of the other four NHEs identified by in silico methods(Pullikuth et al., 2003) or the recently characterized exchanger (W.K., A.K.P., K.A. and S.S.G.,manuscript submitted for publication) might constitute the exchanger coupled to the apical membrane proton pump. In addition, we cannot rule out an equally likely possibility that AeNHE3 might, by virtue of its basolateral localization, play a role in regulating intracellular pH that indirectly impacts V-ATPase and apical exchanger functions.

Drosophila NHE3 contains three splice variants, DmNHE3a, b and c. A shorter variant of NHE3 (here named AeNHE3b) was recently identified that is possibly expressed in Malpighian tubules(Hart et al., 2002). This splice variant results in a protein of 672 amino acids that lacks most of the carboxy cytoplasmic tail. The C-terminal deletion mutant of AeNHE3(NHE3ΔC) is analogous to the shorter version identified previously(Hart et al., 2002) that contains a highly conserved protein kinase A phosphorylation site(S⁶⁶⁸ and S⁶⁶⁴ in Aedes and AnophelesNHE3, respectively), analogous to vertebrate S⁶⁰⁵ that is the prime target for cAMP-mediated acute inhibition of NHE3 function(Kurashima et al., 1997). In contrast, trout red blood cell βNHE is activated by cAMP, which stimulates PKA phosphorylation of S⁶⁴¹ and S⁶⁴⁸(Malapert et al., 1997). Although similar adjacent PKA sites (S⁶⁵⁹ and S⁶⁶⁴) are present in Anopheles NHE3(Pullikuth et al., 2003)(Fig. 1A), it remains to be determined if these sites are indeed phosphorylated in a cAMP-dependent manner and whether such phosphorylation regulates the activity of mosquito NHE3s. AeNHE3 contains a CHP-binding motif in the cytoplasmic tail that is likely to regulate its transport function. The hydrophobic residues within this motif (asterisks in Fig. 1D) bind the ubiquitous CHP in vertebrate NHEs. Substitution of these hydrophobic residues with hydrophilic residues, while not affecting surface expression of NHE1-3, dramatically reduces transport of Na⁺through these exchangers (Pang et al.,2004). This raises an intriguing possibility that AeNHE3 and the shorter AeNHE3b (Hart et al., 2002) might be differentially regulated by CHP since the latter contains only a half-site for CHP binding.

Whole tubule electrophysiological and microfluorimetric studies have demonstrated that cAMP stimulates fluid secretion by Ae. aegyptiMalpighian tubules (reviewed in Beyenbach,1995; Pannabecker,1995). This stimulation could occur by the cAMP mediated responses on the basolateral bumetanide-sensitive Na⁺/K⁺/Cl^- cotransporter, or by increasing basolateral Na⁺ conductance(Beyenbach, 1995; Petzel, 2000; Petzel et al., 1999; Sawyer and Beyenbach, 1985; Williams and Beyenbach, 1984). The NHE antagonist amiloride inhibits basal secretion by isolated Malpighian tubules of Ae. aegypti (Hegarty et al., 1991) as well as serotonin-stimulated secretion by Rhodnius prolixus tubules(Maddrell and O'Donnell,1992). However, amiloride had little effect on basal membrane voltages or transepithelial resistance, suggesting this Na⁺conductive pathway is amiloride-resistant(Hegarty et al., 1991). Basolateral Na⁺ conductance is unlikely to occur through Na⁺ channels since a recent RT-PCR analysis indicated that no Na⁺ channels are expressed in Malpighian tubules of Drosophila (Giannakou and Dow,2001). The effect of amiloride thus could be on the apical membrane NHE, which acts as secondary cation transport coupled to the bafilomycin-sensitive active transport of H⁺ into the lumen by V-ATPase (Beyenbach et al.,2000; Pannabecker,1995).

Current models propose an active apical proton pump coupled to a cation/H⁺ exchanger as the primary generator of transepithelial gradients that facilitate fluid secretion in insect epithelia(Wieczorek et al., 1991). A basolateral NHE has not been implicated in contributing to this fluid secretion. However, our immunolocalization data clearly show that NHE3 is indeed localized to the basal membrane, where it can mediate Na⁺conductance that so far has been attributed to unidentified amiloride-resistant Na⁺ channels(Hegarty et al., 1991). As amiloride has been the only means of distinguishing the activity and requirement of NHEs in these studies, specific NHEs resistant or insensitive to these antagonists would mask their cellular relevance in fluid and ion secretion and in maintenance of intracellular pH. We showed the ²²Na⁺ uptake of AeNHE3 to be insensitive to amiloride and its anologue, EIPA (Fig. 7). These agents reduced uptake through AeNHE3 by only 40%, whereas at similar concentrations completely abolish uptake and pH recovery by amiloride-sensitive mammalian NHEs(Orlowski, 1993). Further,Malpighian tubules of Drosophila and Ae. aegypti exhibited profound sensitivities to these agents under an acid load protocol(Giannakou and Dow, 2001; Petzel, 2000). We conclude that the sensitivity to inhibitors in these tissues does not reflect the transport function mediated by NHE3, which could transport Na⁺ at the basolateral membrane and whose contribution cannot be unmasked with inhibitors that alter the properties of apical NHEs that are as yet undiscovered.

Accordingly, the very highly conserved leucine residue in the amiloride-binding pocket (FFLYLLPP) is substituted by phenyalanine(F313) in AeNHE3 (arrow in Fig. 1C). A single site mutation (L167F) in the amiloride-sensitive vertebrate NHE1 renders it 30-fold resistant to the amiloride analog,methylpropyl amiloride (Counillon et al.,1993). Similarly, a mutant form of mammalian NHE2 (L143F) is 5-and 20-fold resistant to amiloride and EIPA, respectively, compared to wild-type NHE2 (Yun et al.,1993). In AeNHE3, this residue corresponds to F313(Fig. 1A,C), which is also conserved in the amiloride-resistant vertebrate NHE3 (F116). Our direct evidence for amiloride insensitivity of AeNHE3 expressed in PS120 cells indicates that NHE3 function in basal membranes of Malpighian tubules is likely to be amiloride resistant. As a result, prevailing models of ion transport in Malpighian tubules discounted the relevance of a basolateral NHE. We suggest fluid secretion in the Malpighian tubule of Ae. aegyptimight involve ion and proton fluxes through the amiloride-resistant NHE3 located at the basolateral membrane.

The insect epithelial ion transport models suggest a K⁺/H⁺ or Na⁺/H⁺ exchanger situated in the apical membrane functions in reverse compared to physiological polarity of vertebrate NHEs (Beyenbach,1995; Maddrell and O'Donnell,1992; Pannabecker,1995; Wessing et al.,1993; Wieczorek,1992; Wieczorek et al.,2000). Protons extruded into the lumen by V-ATPase are taken into the cell in exchange for cellular Na⁺ or K⁺. Together with our results demonstrating NHE3 in the basolateral membranes, it is apparent that basal and apical NHEs need to operate in reverse orientation to each other to effect transepithelial ion transport. We demonstrated this directly by expressing AeNHE3 in NHE1-deficient epithelial cell line(Figs 6 and 7). After an acid load, the cell interior was estimated to be less than pH 5.5. Upon exchange with Na⁺- or K⁺-rich solutions, AeNHE3 expression resulted in efficient cellular alkalinization that showed no remarkable preference for Na⁺ over K⁺ or vice versa. Thus, AeNHE3 is capable of assuming a transport polarity similar to vertebrate NHEs in aiding the extrusion of cellular protons in exchange for extracellular cation(s). It is reasonable to suggest that AeNHE3 could function in a similar manner in the basolateral membrane of Malpighian tubule, gastric caeca and midgut of this mosquito.

Current models for ion and fluid secretion by Malpighian tubules are derived from studies on whole tubule physiology where regional specialization and functional correlation are not sufficiently delineated. We showed that distinct segments along the length of the tubule differ in their pattern of NHE3 expression and localization, which is likely to be the case for other transport proteins as well. These results suggest the need to re-evaluate unified models attempting to explain ion regulation in different parts of the Malpighian tubules. We hypothesize that fluid, proton and ion secretions by Malpighian tubules may require NHE3 activity in the basolateral membrane in the proximal and distal segments, whereas in the median segment of the tubule NHE3 can function additionally in the apical membrane where it might be regulated through recycling pathways.

In summary, we have characterized the AeNHE3 that is localized to the basolateral membrane of almost all tissues examined in Ae. aegypti. The presence of potential splice variants, apical staining in median Malpighian tubule, and intracellular pool of NHE3 immunoreactivity,support the possibility that distinct isoforms could act in concert with differential transport polarity in insect epithelia. Our results add an important, yet often neglected, aspect of basolateral NHE function in models describing insect epithelial ion transport. Further efforts in distinguishing pharmacological properties, and molecular identification of the remaining NHE isoforms and their specific localization and expression dynamics should clarify the roles of these integral membrane proteins in ionic homeostasis in insects.

List of abbreviations
 
• AP

  arginine phosphate

 
• BSA

  bovine serum albumin

 
• CHP

  calcineurin B homologous protein

 
• DMSO

  dimethyl sulfoxide

 
• FBS

  fetal bovine serum

 
• HBS

  Hepes-buffered saline

 
• K-pump

  potassium pump

 
• Na⁺/K⁺-ATPase

  sodium/potassium ATPase

 
• NGS

  normal goat serum

 
• NHE

  sodium-proton exchanger

 
• NMDG

  N-methyl-d-glucamine

 
• ORF

  open reading frame

 
• PFA

  paraformaldehyde

 
• V-ATPase

  vacuolar-type ATPase

We thank Dr Marjorie Patrick for suggesting the use of a5 Na⁺/K⁺-ATPase monoclonal antibody, Drs James Melvin,Jacques Pouysségur, Jose M. Pardo and Thomas Jespersen for sharing reagents with us. Research was funded through grants from the National Institutes of Health, AI 32572 and AI48049.