Functional characterization of a glutamate/aspartate transporter from the mosquito Aedes aegypti

Glutamate is a well-established neurotransmitter in both vertebrates and invertebrates. In mammals it is the principal excitatory transmitter in the central nervous system, although excess extracellular glutamate concentrations are associated with various neurological states including ischemia, amyotropic lateral sclerosis and Alzheimer's disease, and must therefore be tightly regulated (Maragakis and Rothstein,2001). In contrast to vertebrates, glutamate primarily acts as an excitatory transmitter at the neuromuscular junction in invertebrates,analogous to the role played by acetylcholine in vertebrate neuromuscular junctions (Kerkut et al.,1965; Usherwood,1967; Usherwood and Machili,1966). The action of glutamate in insects is further complicated by its additional ability to elicit inhibitory responses at insect muscles, as well as mixed excitatory and inhibitory responses in the central nervous system (Burrows, 1996; Cull-Candy, 1976; Darlison, 1992; Homberg, 1994; Sattelle, 1992).

In mammals, it is believed that Na⁺- and K⁺-dependent high-affinity transport systems are responsible for regulating extracellular glutamate levels as well as terminating glutamatergic signals in the central nervous system (for a review, see Danbolt,2001). Five transporters with the ability to transport l-glutamate and l-and d-aspartate have been identified in humans. They have been designated EAAT (Excitatory Amino Acid Transporter) 1–5, having 50–60% amino acid identity with each other (Seal and Amara, 1999). These proteins drive the uphill transport of glutamate by dissipating the electrochemical Na⁺ gradient generated by Na⁺/K⁺-ATPase. It is currently accepted in mammals that glutamate transport occurs with the cotransport of 3Na⁺ and 1H⁺, and a countertransport of 1K⁺ with every molecule of glutamate (Levy et al.,1998; Zerangue and Kavanaugh,1996b). Additionally, glutamate evokes an anion channel activity intrinsic to the EAATs: a property which, however, is not directly coupled to substrate translocation (Fairman et al.,1995; Wadiche et al.,1995a).

In insects, early biochemical studies have shown the existence of Na⁺-dependent l-glutamate transport systems in both muscle and central nervous systems. Examples include Na⁺-dependent glutamate transport on isolated abdominal nerve cords from adult Periplaneta americana, as well as autoradiographic transport studies in the excitatory neuromuscular junctions of locust localizing Na⁺-dependent glutamate transport to nerve endings and surrounding glia (Evans, 1975; van Marle et al., 1983).

The availability of cDNA sequences encoding mammalian EAATs, coupled with the advent of molecular cloning, has allowed for the cloning and isolation of EAAT homologues from a number of insect species within the last 5 years. In particular, insect EAAT homologues have been cloned by independent groups from the brain and embryo of Drosophila melanogaster (dEAAT1 and dEAAT2)as well as the brains of caterpillar Trichoplusia ni (TrnEAAT),cockroach Diploptera punctata (DipEAAT) and honeybee Apis mellifera (AmEAAT) (Besson et al.,1999; Donly et al., 1997, 2000; Kawano et al., 1999; Kucharski et al., 2000; Seal et al., 1998). With the exception of AmEAAT, all insect homologues cloned thus far have been heterologously expressed and functionally characterized as Na⁺-dependent high-affinity glutamate and aspartate transporters with affinity in the micromolar range (Besson et al., 1999, 2000; Donly et al., 1997, 2000; Kawano et al., 1999; Kucharski et al., 2000; Seal et al., 1998). Electrophysiological studies on dEAAT1 have illustrated a substrate-elicited anion conductance that is stoichiometrically uncoupled from substrate translocation, much like the mammalian EAATs(Fairman et al., 1995; Seal et al., 1998; Wadiche et al., 1995b). Amongst such similarities, however, certain differences have been noted, such as a preference for l- and d-aspartate as substrates over l-glutamate for dEAAT2; a preference for l-asparatate over d-aspartate by DipEAAT; and abolished substrate transport by dEAAT1 in the absence of both intracellular and extracellular chloride (Besson et al.,2000; Donly et al.,2000; Seal et al.,1998).

Attempts at elucidating the spatial localization of the insect EAATs at the mRNA level for dEAAT1, dEAAT2 and AmEAAT, have demonstrated almost exclusive localization to parts of the brain insect brain(Besson et al., 1999; Seal et al., 1998; Soustelle et al., 2002). A recent immunohistochemical study on TrnEAAT, however, demonstrated expression in larval skeletal muscle fibers (Gardiner et al., 2002).

In spite of the fact that glutamate performs a plethora of roles in insects, and that a number of insect transporters for glutamate have been characterized, information available for the termination of insect glutamatergic signals is still relatively sparse. In light of this, we present a comprehensive study on a glutamate and aspartate transporter from the mosquito Aedes aegypti (AeaEAAT), which encompasses cloning and characterization from pharmacological, electrophysiological, and spatial localization standpoints. Ae. aegypti is the major vector for the dengue virus; the work presented here therefore represents the first such kind from a blood-feeding insect. Given the recent completion of the Anopheles gambiae genome project, together with the historic interest in the biology of the mosquito as a vector for various human diseases, the work presented here is a timely contribution to the understanding of this disease vector.

Aedes aegypti L. were maintained in a light:dark photoperiod of 16 h:8 h, at 26°C and 77% relative humidity. Larvae were fed with a mixture of yeast and dry dog food (1:3 w/w) dissolved in water, adults were reared on sugar water and females were given weekly mouse blood meals.

Poly(A)⁺ RNA was isolated from isolated midgut and Malpighian tubules of adult female Aedes aegypti. The mRNA was then used for the construction of cDNA using a cDNA synthesis kit from (GIBCO BRL Life Technologies, Gaithersburg, MD, USA). The cDNA (approximately 1.5 μg) was separated by size-exclusion chromatography on a Superose 6 column using a SMART system (Pharmacia LKB Biotechnology, Piscataway, NJ, USA) as previously described (Mbungu et al.,1995). The >2kb cDNA was ligated to NotI/SalI-cut, dephosphorylated pSport1 vector and electroporated into DH10B cells. The size of the unamplified library was 130 000 clones, 95% of which were recombinants. Isolation and analysis of individual cDNA clones demonstrated the presence of large inserts in this library.

The gene encoding AeaEAAT, initially named BC10, was identified during an EST project that randomly sequenced approximately 2000 different genes expressed in the cDNA library above. The open reading frame (ORF) was subcloned into the expression vector pBSAMV containing a T7 promoter, through the NcoI and FseI restriction enzyme sites(Chiu et al., 2000). These restriction sites were engineered into the 5′ and 3′ ends of the ORF, respectively, by polymerase chain reaction (PCR) using primers containing the sites. The PCR product generated by Expand (Roche, Palo Alto, CA, USA) was gel purified, and initially ligated to pCR2.1 (Invitrogen, Carlsbad, CA, USA)according to the manufacturer's specifications. Clones putatively containing the desired product were sequenced to confirm the lack of any PCR-generated errors using ABI Prism cycle sequencing. The clone with the correct sequence was subsequently subcloned into the pBSAMV vector(Chiu et al., 2000), and will herein be referred to as pBSAMVBC10.

The plasmid pBSAMVBC10 containing the AeaEAAT ORF was in vitrotranscribed and translated using the TNT T7-Coupled Reticulocyte Lysate System(Promega, Madison, WI, USA). A sample (4 μl) of the total reaction was separated by discontinuous 8% SDS-PAGE as described previously(Laemmli, 1970). The gel was stained with Coomassie Blue, destained, treated in Entensify enhancing solutions (Dupont-NEN, Boston, MA, USA), and dried before exposure to X-ray film for 3 days.

HeLa cells were maintained in Minimum Essential Medium (Invitrogen,Carlsbad, CA, USA) containing 1.5 g NaHCO₃, sodium pyruvate(Invitrogen), non-essential amino acids (Invitrogen), penicillin/streptomycin(Invitrogen) and 10% fetal bovine serum (Summit Biotechnologies, Greeley, CO,USA) at 37°C and 5% CO₂. Transient transfections utilizing the T7/vaccinia virus method were generally carried out in 48-well plates using serum-free medium for all steps (Blakely et al., 1991). Briefly, 5–7×10⁴ cells per well were plated the night before transfection in the above medium lacking antibiotics, of which 2 wells were counted on the day of the transfection to obtain an average number of cells per well. A virus multiplicity of infection of 5, 0.3 μl of lipofectin (Invitrogen), 1 μl PLUS reagent (Invitrogen),and 100 ng of DNA was used per well in all cases. Lipofectin was diluted in medium 30–35 min before mixing with DNA, which itself was preincubated with the PLUS reagent for 15 min beforehand. Cells were infected with virus(30 m, 37°C/5% CO₂) with shaking every 10 min, and the lipofectin/DNA/PLUS mixture was overlayed.

Transport assays were carried out between 18 and 24 h after infection/transfection. Cells were washed once with PBS, and incubated for 15 min with uptake buffer (120 mmol l^-1 NaCl, 4.7 mmol l^-1KCl, 10 mmol l^-1 Hepes, pH 7.5, 5 mmol l^-1 Tris-Cl, pH 7.4, 1.2 mmol l^-1 KH₂PO₄, 2 mmol l^-1 CaCl₂, 1.2 mmol l^-1 MgSO₄)containing glucose before performing the transport assay in duplicate. In order to determine substrate specificity, cells were incubated with a mixture of cold substrate (100 μmol l^-1) and ³H-labeled substrate (30 nmol l^-1; Amersham, Piscataway, NJ, USA) used as a tracer, initially for 10–20 min. The substrates that were transported(l-glutamate, l-aspartate and d-aspartate)were further analysed for ion dependence, time course and dose–response studies.

A time course study was performed on all three substrates, where the duration of transport was varied. Upon determining the time during which transport followed a linear time course, this time point (5 min) was used in all subsequent transport assays.

The dependence of substrate transport on the major extracellular ions was determined by substituting equimolar choline chloride for NaCl in the assay buffer, while two types of anions, gluconate and acetate, were used in independent experiments to replace chloride ions.

Dose–response studies were performed to obtain the affinity and kinetic parameters (K_m and V_max), by varying the cold substrate concentration between 0.1 μmol l^-1and 500 μmol l^-1 on a logarithmic scale. Experiments were performed in triplicate to obtain mean K_m and V_max values, calculated by a least-squares fit to the Michaelis–Menton equation and Eadie–Hofstee transformation using Origin 6.0 (Microcal, Northampton, MA, USA).

Inhibition studies were carried out with 1 μmol l^-1 cold substrate and 30 nmol l^-1 3H-labeled substrate, so as to render the concentration of substrate negligible when mathematically determining the K_i, such that the IC₅₀ approximates the K_i as described (Cheng and Prusoff, 1973a). A three-point screen of 3, 100 and 3000μmol l^-1 inhibitor concentration was used initially in order to eliminate compounds that essentially had no effect on transport. Others were consequently subjected to a further inhibition assay, where inhibitor concentrations were varied on a logarithmic scale, typically between 1 nmol l^-1 and 3 mmol l^-1.

Heterologous expression in Xenopus laevis oocytes was utilized to study the electrophysiological properties of AeaEAAT viatwo-electrode voltage clamp (TEV-200, Dagan Instruments, Minneapolis, MN, USA)interfaced to an IBM-compatible PC using a Digidata 1200 A/D controlled using the pCLAMP 6.0 program suite (Axon Instruments, Foster City, CA, USA). The expression construct pBSAMVBC10 was cut with NotI for runoff in vitro transcription using the Ambion T7 mMessage mMachine kit (Ambion,Austin TX, USA). Stages V and VI Xenopus oocytes were prepared using standard protocols, injected with 35 ng of cRNA, and incubated for 3–6 days in ND96 (96 mmol l^-1 NaCl, 2 mmol l^-1 KCl, 1 mmol l^-1 CaCl₂, 1 mmol l^-1 MgCl₂, 5 mmol l^-1 Hepes, pH 7.5) supplemented with gentamycin at 18°C before recordings. Current responses were acquired at room temperature and analyzed with either Clampfit 8.0 (Axon Instruments, Foster City, CA, USA) or ORIGIN 6.0 (MicroCal Software, Northampton, MA, USA). Signals were filtered with a 2 kHz low-pass filter and a 3 mol l^-1 KCl agar bridge was used to reduce junction potentials. Microelectrodes were filled with filtered 3 mol l^-1 KCl solution, with resistances of less than 1 MΩfor both current passing and recording electrodes. For the superfusing of saline solution, a Perkin–Elmer Model 410 liquid chromatography pump was used, supplying a continuous flow of ND-96 buffer at a flow rate of 1.0 ml min^-1. A Rheodyne injection valve was used to apply substrates for dose–response analysis in 170 μl volumes, providing a 5 s exposure of substrate to the oocyte. Concentration–response relationships to determine EC₅₀ values were obtained from oocytes clamped at -60 mV by measuring the peak current amplitude as a function of substrate concentration. Results were normalized to current evoked by the highest substrate concentration used, and subsequently fit to the equation I=I_maxS/(EC₅₀+S),where I is the peak current amplitude, I_max is the peak current amplitude evoked by the highest substrate concentration, and S is the concentration of substrate.

In reversed-transport studies, oocytes clamped at -60 mV were exposed to a high potassium buffer containing 98 mmol l^-1 KCl, 1 mmol l^-1 MgCl₂, 1.8 mmol l^-1 CaCl₂ and 5 mmol l^-1 Hepes (pH 7.5).

Current–voltage relationships were determined either by measurement of steady-state currents in response to bath application of substrates or by off-line subtraction of control current records obtained during 1 ms voltage pulses to potentials between -60 and +50 mV from corresponding current records in the presence of substrate. Sodium substitution studies were carried out in saline containing equimolar choline chloride instead of NaCl. External chloride was replaced by either gluconate, nitrate or thiocyanate salts as indicated, at concentrations equivalent to the chloride salts used in ND96. Intracellular chloride depletion experiments were performed as described in Wadiche et al. (1995a), by placing oocytes in chloride-free buffer substituted with gluconate for at least 20 h prior to use.

Substrate-transport assays in Xenopus oocytes were performed by exposing oocytes to 3H-radiolabeled substrate (30 nmol l^-1l-glutamate, l-aspartate and d-aspartate;Amersham) for 5 min at room temperature. Oocytes were then rapidly washed three times in ice-cold buffer in order to prevent any possibility of reversed transport, upon which they were lysed in 10% SDS and subjected to scintillation counting.

For cases in which transport assays were performed under two-electrode voltage clamp, oocytes were clamped at -60 mV and exposed to radioactive ligand for 5 s, then washed for 10 s with the respective saline, and subsequently lysed and counted for accumulated radioactivity as described above.

A synthetic peptide of 15 amino acids, corresponding to the C-terminal 14 amino acids of AeaEAAT, plus a cysteine at the N terminus of these 14 residues(CPSSEINGKTQRNSL) was synthesized (MGIF, University of Georgia). The peptide was conjugated to a maleimide-activated form of keyhole limpet hemocyanin(KLH; Pierce, Rockford, IL, USA) according to the manufacturer's instructions,immunized in rabbits and affinity-purified essentially as described(Umesh and Gill, 2002).

Membranes were isolated from adult Ae. aegypti using a differential centrifugation protocol. Head, thorax and abdomen were separated from male and female Ae. aegypti anesthetized in CO₂. Briefly, tissues were suspended in ice-cold Tris-HCl buffer (50 mmol l^-1, pH 7.4) containing phenylmethylsulfonyl fluoride (PMSF; 1 mmol l^-1) and protease cocktail inhibitor (Roche Diagnostics,Indianapolis, IN, USA), homogenized with 30 strokes using a Dounce homogenizer, and centrifuged (3000 g, 4°C, 10 min). The post-nuclear supernatant was centrifuged (10 000 g, 4°C,10 min) to remove mitochondria, and the resulting supernatant was further centrifuged to pellet the membranes (100 000 g, 1 h, 4°C)using a Beckman Ti 42.2 rotor. Final pellets were resuspended in the homogenization buffer, upon which protein concentrations were determined with the BCA Protein Assay Kit (Pierce).

Membranes were prepared from Xenopus oocytes injected with cRNA of pBSAMVBC10 and that of a D. melanogaster sodium channel, of which the latter served as a negative control (Buller and White, 1992). Briefly, oocytes were homogenized in Buffer A(150 mmol l^-1 NaCl, 10 mmol l^-1 magnesium acetate, 20 mmol l^-1 Tris-Cl, 0.1 mmol l^-1 PMSF, pH 7.6) containing sucrose (10% w/v) at a concentration of 25 oocytes ml^-1 of buffer. Homogenates were overlayed on a discontinuous sucrose gradient consisting of 20% and 50% sucrose, and subsequently centrifuged (15 000 g,30 min, 4°C). The layer visible at the interface of 20% and 50% sucrose,representing the membrane portion, was carefully removed, diluted in 5 volumes of buffer A and recovered by centrifugation (100 000 g, 1.5 h,4°C).

Membranes and cytosol of Ae. aegypti, as well as X. laevis oocyte membranes, were analyzed by SDS-PAGE and western blotting. Protein samples were treated with Laemmli sample buffer (37°C, 30 min). Subsequently, samples were separated by an 8% polyacrylamide gel at a concentration of 25 and 9 μg per lane for the tissue samples and oocyte membranes, respectively. For western analysis, samples separated by SDS-PAGE were electrotransferred onto Immobilon P membranes (0.45 mm; Millipore,Billerica, MA, USA) using a tank transfer system (80 V, 3 h, 4°C). Upon blocking (1 h, room temperature) with 3% BSA in PBS containing 0.05% Tween-20(PBST), membranes were incubated with affinity-purified AeaEAAT antibody (1μg, overnight, 4°C) diluted in PBST containing 1% BSA. As a negative control, membranes with identical samples were incubated with primary antibody that had been preadsorbed with peptide (10 μmol l^-1). Upon washing in PBST, membranes were incubated with horseradish peroxidase(HRP)-coupled donkey anti-rabbit IgG (Amersham) diluted 1:3000 in 1% BSA/PBST(1 h, room temperature), washed extensively, and proteins were ultimately detected using enhanced chemiluminescence (Amersham).

Immunohistochemical studies were performed on adult thorax. Upon separation of the thorax and fixation in alcoholic Bouin's solution (overnight, 4°C),tissues were sectioned after paraffin embedding as described before(Umesh and Gill, 2002). Sections were blocked in PBS-Triton X-100 (0.2%) (PBS-Tx, pH 7.4) containing normal goat serum (NGS, 1%) and BSA (0.1%) for 30 min, and then incubated with the α-AeaEAAT (1/200 dilution in blocker, 4°C, overnight). After extensive washes with PBS-Tx, sections were incubated with HRP-labeled goat anti rabbit IgG (American Qualex, San Clemente, CA, USA) diluted 1/1000 in blocker, washed thoroughly with PBS-Tx and stained using 3-amino 9-ethylcarbazol as a substrate (Umesh and Gill, 2002). Slides were visualized with a Zeiss Axiovert microscope (Carl Zeiss, Thornwood, NY, USA). All images were ultimately imported into Adobe Photoshop (version 6.0, Adobe, Sunnyvale, CA, USA) in which they were assembled and labeled.

Clones were isolated from a cDNA library prepared from the midgut and Malpighian tubule tissues of Ae. aegypti, and sequenced by ABI cycle sequencing. Of the clones isolated and sequenced one, designated BC10, showed sequence homology to the known EAATs. Its open reading frame of 1443 base pairs (bp) encoded a protein of 481 amino acids (aa), designated AeaEAAT(Ae. aegypti excitatory aa transporter; GenBank Accession number AY249244). At the aa level, AeaEAAT was 40–50% identical to other known EAATs, with highest identity to known insect EAATs, having 55.3, 51.5 and 48.2% identity to DipEAAT, dEAAT1 and TrnEAAT, respectively. Fig. 1A shows an alignment generated by ClustalW, representing AeaEAAT sequence similarity with insect EAATs that have been functionally characterized to date. As for known EAATs,hydropathy analysis of AeaEAAT gave six strongly hydrophobic domains in the N-terminal half of the protein, corresponding to distinct regions of transmembrane domains (overlined in Fig. 1A). The C-terminal half, having the highest degree of sequence conservation, gave one large domain of hydrophobicity. The deduced amino acid sequence of AeaEAAT additionally gave one putative N-linked glycosylation site between transmembrane domains 2 and 3 (asterisk in Fig. 1A).

The dendrogram analysis (Fig. 1B) includes members of the human EAAT superfamily consisting of hEAAT1-5 and related system ASC transporters, ASCT1 and ASCT2, which transport alanine, serine, threonine and cysteine. Also included are cloned insect EAATs and homologous sequences from the completed genome of Anopheles gambiae. Amongst the five subtypes of mammalian EAATs identified thus far, the above mentioned insect transporters collectively are most identical to hEAAT3, with 47.7% identity to AeaEAAT. The completed genomic sequence of An. gambiae gave three sequences representing putative members of the EAAT family: AAB01008807, AAB01008797 and AAB01008964, with the deduced amino acid sequences yielding proteins with 75.1, 52.0 and 33.5% identity to AeaEAAT, respectively.

The AeaEAAT open reading frame was subcloned into the expression vector pBSAMV, whose expression is driven by the T7 promoter. In vitrotranscription and translation of this subclone gave a protein migrating at 42 kDa when separated by SDS-PAGE (Fig. 2, lane 1).

Heterologous expression of pBSAMVBC10 in both HeLa(Fig. 3) and CV-1 (data not shown) cells showed transport of substrates typical for this family of proteins, namely, l-glutamate, and l- and d-aspartic acid. Negative controls were initially conducted using the same vector containing a different transporter, MasIne(Chiu et al., 2000), but showed no difference from transfections conducted without DNA (data not shown).

Transport was Na⁺-dependent, as replacement of Na⁺ions in the assay buffer with choline reduced uptake to background levels(Fig. 3A). On the other hand,transport was independent of CL^-, since replacement with each of two types of anions, acetate and gluconate, made no significant difference(Fig. 3B).

AeaEAAT is a high-affinity transporter of its substrates, having K_m values of approximately 30 and 10 μmol l^-1 for l-glutamate and the two types of aspartate,respectively (Table 1). In all three cases transport was saturable, with the dose–response relationship following Michaelis–Menten kinetics. Similar affinity parameters were observed when AeaEAAT was independently expressed in both HeLa and CV-1 cells(data not shown). The V_max values were determined to be 6.0, 3.0 and 6.9 fmol cell^-1 h^-1 for l-glutamate, l-aspartate and d-aspartate,respectively. From these values, the efficacy(V_max/K_m) of transport by AeaEAAT was highest for d-aspartate (0.6), being double that of l-aspartate (0.3), and triple that of l-glutamate (0.2). Substrate transport was not inhibited by d-glutamate(Table 2), indicating that for glutamate, AeaEAAT demonstrates stereoselective transport of the l-congener.

We tested the ability of the 20 amino acids to compete with transport of d-aspartate (Fig. 4), the non-metabolizable substrate, at a competitor concentration of 5 mmol l^-1. As expected, the earlier determined high-affinity substrates (i.e. l-glutamate and l-aspartate) most strongly competed d-aspartate transport. In addition, we found that l-cysteine inhibited d-aspartate uptake to 12% of control levels, and that l-asparagine and l-glutamine also inhibited the substrate transport by 27.5% and 70% of control levels,respectively. At the concentrations used, inhibition by these amino acids was statistically significant at the 99% confidence interval.

In an initial screen, various compounds in amounts of 3, 100 and 3000μmol l^-1 were tested for their ability to independently block the AeaEAAT-mediated transport of l-glutamate and d-aspartate. 30 nmol l^-1 3H-labeled and 100 μmol l^-1 unlabeled substrate were used in the initial screen. Of the compounds tested, N-methyl-d-aspartate and kainic acid had no effect on transport, even at 3 mmol l^-1(Table 2). Thereafter, the amount of substrate used was reduced to a total of 1 μmol l^-1,so as to approximate the IC₅₀ to K_i(Cheng and Prusoff, 1973b). The IC₅₀ (equivalent to the calculated K_i)values of 11 compounds, dl-threo-benzoxyaspartic acid(dl-TBOA), trans-pyrrolidine-2,4-dicarboxylate(t-PDC), l-aspartate-β-hydroxymate, dl-threo- β-hydroxyaspartic acid (TBHA), l-cysteine, l-cysteic acid, serine-O-sulfate, l-cysteine sulfinic acid, α-amino adipic acid,β-glutamate and d-glutamate were determined by using logarithmically increasing concentrations of the various inhibitors. Overall,the inhibitors were equally effective at inhibiting the transport of both d-aspartate (Table 2) and l-glutamate (data not shown). Of the compounds tested, those involved in the cysteine oxidation pathway such as l-cysteic acid and l-cysteine sulfinic acid, as well as TBHA, were the most effective inhibitors, with calculated K_i values of approximately 6.6±3.3, 7.1±3 and 9.2±2.9 μmol l^-1, respectively. The most notable difference in AeaEAAT pharmacology was inhibition by serine-O-sulfate, for which the K_i was 29 μmol l^-1. This compound was twice as potent at inhibiting transport by AeaEAAT than TrnEAAT, and three times as potent than hEAAT3. A universal inhibitor of all EAAT subtypes, t-PDC, blocked d-aspartate transport with a K_i of 91±19 μmol l^-1, and dl-TBOA, a non-substrate EAAT inhibitor, gave a K_i value of 23.7±2.1 μmol l^-1.

In an attempt to understand the ion transport characteristics of AeaEAAT,we turned to heterologous expression of AeaEAAT in Xenopus laevisoocytes where two-electrode voltage clamp analysis was performed. Xenopus oocytes injected with cRNA in vitro-transcribed from pBSAMVBC10 displayed an inward current that is characteristic of this family of transporters upon application of substrate(Fig. 5A). This is consistent with a net inward flow of cations during the transport cycle(Zerangue and Kavanaugh,1996b). Typically, the current elicited by the application of 100μmol l^-1 substrate was as large as 200–400 nA. Similar currents were produced regardless of the substrate (i.e. l-glutamate, l-and d-aspartate, l-cysteine).

Reversal of extracellular ionic composition by raising K⁺ ion concentrations caused outward currents to be elicited by oocytes expressing AeaEAAT, reflecting the reversed transport of endogenous excitatory amino acids (Fig. 5B). No outward current was observed in water-injected oocytes upon exposure to high extracellular K⁺ (data not shown). Endogenous excitatory amino acid concentration in X. laevis oocytes has been reported to be as high as 12 mmol l^-1 (Wadiche et al.,1995a). Reversed transport is a physiologically relevant phenomenon in mammals that occurs in disease states such as ischemia, when extracellular concentrations of K⁺ rise to 60 mmol l^-1,thereby depolarizing cells and allowing the release of glutamate by reversed functionality of the glutamate transporters(Attwell et al., 1993).

Difference I–V curves obtained by off-line subtraction of voltage-jump protocols without substrate from those with substrate, revealed a substrate-activated current reversing at 25–30 mV for all substrates tested (Fig. 5C). The substrate-activated current was abolished upon substitution of extracellular Na⁺ with choline (Fig. 5D).

It is now well established that excitatory amino acid transporters carry a substrate-gated anion conductance that is not coupled to transport (reviewed in Seal and Amara, 1999). The ability of AeaEAAT to carry this type of current was determined by investigating the difference I–V curves upon replacement of extracellular CL^- with other anions. Outward current was abolished when external chloride was replaced by gluconate(Fig. 6A,B), an impermeable anion, but enhanced when replaced with more permeant anions such as nitrate and thiocyanate (Fig. 6A,B). We then compared the substrate-activated currents of oocytes depleted of intracellular chloride by dialysis, in both the presence and absence of extracellular chloride (Fig. 6C). Chloride-dialysed oocytes gave identical inward currents regardless of the presence of extracellular chloride. However, outward currents were only seen in dialysed oocytes in the presence of extracellular CL^-.

We addressed the question of whether substrate transport was dependent on the presence of either intracellular or extracellular CL^-, by comparing radiolabeled substrate transport of dialysed oocytes to undialysed oocytes while holding the membrane potential at -60 mV(Fig. 6D). Under these conditions, substrate transport was not compromised, in spite of a reduction in the transport associated current.

We addressed the question of AeaEAAT localization by generating a peptide antibody to a variable region of the protein. Affinity purification produced a specific antibody with high affinity for AeaEAAT, as characterized by ELISA(data not shown). When used in immunoblots against adult Ae. aegyptihead membrane fractions (25 μg per lane), as well as membrane preparations of AeaEAAT expressing Xenopus oocytes (9 μg per lane), the antibody recognized a band at 52.5 kDa(Fig. 2, lanes 2 and 4). When membrane fractions isolated from the adult head, thorax, and abdomen regions were subjected to western analysis (25 μg per lane), a specific signal of 52.5 kDa was evident in all three regions(Fig. 7, lanes 1, 3, 5), with the most intense signal in the thorax (Fig. 7, lane 3). These bands could not be detected when peptide preadsorbed antibody was used to probe the blots(Fig. 7, lanes 7, 9, 11). Immunoreactive bands were absent in the cytosolic fractions(Fig. 7, lanes 2, 4, 6). The results shown in Fig. 7 are from adult males: identical results were obtained from adult females (data not shown).

Analogous experiments were conducted with larval and pupal membrane fractions to gather information on the developmental expression profile. Interestingly, only low levels of AeaEAAT immunoreactivity could be detected in larval head, thorax and abdomen membranes fractions even when 40 μg of protein was loaded per lane (data not shown). In the pupal stage, AeaEAAT immunoreactivity was negligible when 40 μg of membranes were probed (data not shown).

Based on the results from western analysis, we focused on the adult thorax for immunohistochemical studies. Paraffin sections of the adult thorax probed with the affinity-purified antibody revealed staining of neuropile regions of the thoracic ganglia (Fig. 8B,C). Immunoreactivity in this region was abolished when the antibody had been preadsorbed with the antigenic peptide(Fig. 8A). Staining of fibers in the neuropile region, as well as cell bodies peripheral to this region,could be observed (Fig. 8C,arrowheads). The muscles within the thorax, however, did not show immunoreactivity to this antibody.

Here we have presented a comprehensive study on the cloning and functional characterization of a glutamate transporter isolated from a tissue-specific cDNA library from the midgut and Malpighian tubules of the adult female mosquito, Aedes aegypti, prior to blood feeding. We presented pharmacological and biophysical data on the heterologously expressed protein,and explored its tissue distribution using western analysis and immunohistochemistry.

At the amino acid level, AeaEAAT has greatest identity to its cloned insect counterparts, especially to DipEAAT, dEAAT1, and TrnEAAT. When compared to the five human homologues, AeaEAAT shares highest homology with the neuronal hEAAT3, whose rabbit homologue was originally isolated from the kidney(Fig. 1B). For all cloned EAATs thus far, including AeaEAAT, sequence conservation is greatest at the C terminus, where 50–60% identity is observed. Hydropathy analysis gives proteins with intracellular N and C termini, having six clear transmembrane segments in the N terminus. The C-terminal transmembrane topology for this family of proteins has been a point of much debate, as despite the high degree of sequence conservation in this region, predictions based on hydrophobicity gave 0–4 transmembrane segments for different EAATs(Slotboom et al., 1999). More recent experimental studies of some mammalian counterparts suggest that the C-terminal half comprises a re-entrant loop structure with an outward-facing hydrophobic linker between the loop and final transmembrane domain(Grunewald et al., 1998; Seal and Amara, 1998; Seal et al., 2000).

In humans, apart from the high-affinity glutamate transporters (EAAT1-5),there are two Na⁺-dependent neutral amino acid transporters with approximately 30% identity to EAAT1–5, possessing the 6-amino-acid signature sequence AAI/VFIAQ. The two proteins, ASCT1 and ASCT2, have the transport properties of system ASC, which transport alanine, serine, threonine and cysteine (Arriza et al.,1993; Shafqat et al.,1993; Utsunomiya-Tate et al.,1996). The five EAAT proteins and two ASCT proteins thus form a carrier superfamily. As with other EAATs, AeaEAAT shares lower identity(30–35%) with ASCT1 and ASCT2.

Analysis of the two completely sequenced insect genomes, namely D. melanogaster and Anopheles gambiae, show the presence of two and three EAAT homologues, respectively. The two in D. melanogaster,dEAAT1 and dEAAT2, have been cloned and shown to functionally transport glutamate and aspartate, while the three in An. gambiae (GenBank accession numbers: AAB01008807, AAB01008797 and AAB01008964) have yet to be cloned (Besson et al., 1999, 2000; Kosakai and Yoshino, 2001; Seal et al., 1998). Based on this information, it is likely that there are at least two genes in mosquitoes encoding glutamate/aspartate transporters. However, the lower identity of AeaEAAT (33.5%) with the third homologous EAAT sequence in An. gambiae (AAB01008964) resembles the level of identity observed with the mammalian System ASC transporters (approximately 30–35%), suggesting that it is a functional homologue of System ASC transporters rather than EAATs.

Based on the initial amino acid sequence analysis, as well as the fact that AeaEAAT was originally isolated from a midgut/Malpighian tubule cDNA library,we heterologously expressed AeaEAAT in mammalian cells and Xenopusoocytes to test the hypothesis that it is the functional homologue of hEAAT3. Heterologously expressed AeaEAAT indeed transports glutamate and aspartate with high affinity (Table 1),sharing many of the properties seen with the mammalian EAATs, such as a strict dependence on extracellular Na⁺ for function, and ability to perform reversed transport in the presence of high concentrations of extracellular K⁺. A few lines of evidence prove our hypothesis that AeaEAAT is the functional homologue of hEAAT3. For example, analysis of AeaEAAT pharmacology excludes it from being the functional homologue of hEAAT2, as AeaEAAT is insensitive to DHK and KA, a property distinct to hEAAT2(Table 2). With the exception of a unique sensitivity to serine-o-sulfate, pharmacological analysis of AeaEAAT using known blockers of glutamate transport shows its similarity to hEAAT1 and hEAAT3. Examination of substrate-elicited currents shows AeaEAAT to be responsive to l-cysteine(Fig. 5C), an amino acid which causes neuronal damage in mammals (Olney and Ho, 1970). Application of l-cysteine produces transporter-mediated currents specific for mammalian EAAT3, and thus additionally classifies AeaEAAT as an hEAAT3 orthologue(Zerangue and Kavanaugh,1996c).

Profiles of difference I–V curves representing the substrate-activated transporter currents demonstrate AeaEAAT to be most similar to hEAAT3 as both reverse between +35 and +40 mV(Fig. 5C)(Wadiche et al., 1995a). Difference I–V curves for mammalian EAATs show clear reversal potentials ranging from +9 to +40 mV for hEAAT1–3, to -20 mV in the case of hEAAT4 and 5 (Arriza et al.,1997; Fairman et al.,1995; Wadiche et al.,1995a). Sodium-dependent solute transporters acting as secondary active carriers are not expected to reverse in polarity when exposed to extracellular substrate, regardless of the membrane potential, as seen for the rat GABA transporter (Mager et al.,1993). While reversals are not expected of classical secondary active carriers, they have been observed with transporters mediating substrate-evoked conductances that are not coupled to transport(Chaudhry et al., 2001; Fairman et al., 1995; Wadiche et al., 1995a). In particular, for mammalian EAATs the reversal has been attributed to a stoichiometrically uncoupled substrate-activated anion conductance, which is a property intrinsic to EAATs comprising (1) an anion leak in the absence of substrate, and (2) an anion channel-like activity evoked by the substrates of glutamate transport, but is not required for substrate transport(Arriza et al., 1997; Bergles and Jahr, 1997; Fairman et al., 1995; Otis and Jahr, 1998; Seal et al., 1998; Wadiche et al., 1995a; Wadiche and Kavanaugh, 1998). Although termed `stoichiometrically uncoupled', this channel activity requires the presence of Na⁺ and substrate, and is thought to be activated early in the transport cycle (Otis and Kavanaugh, 2000). The necessity for Na⁺ and substrate is also apparent in our studies, since absence of both Na⁺ and substrate evokes no transporter-mediated current(Fig. 5D).

While all five human EAATs exhibit a substrate-activated anion conductance,its contribution to the overall substrate-activated current differs. For example, more than 95% of the substrate-induced current is carried by CL^- ions for hEAAT4 and hEAAT5 and 50–73% for hEAAT1-3(Arriza et al., 1997). dEAAT1,the only other insect glutamate transporter studied by two-electrode voltage clamp, possesses a substrate-activated anion conductance that is responsible for most of the substrate-induced current. Furthermore, in the absence of intracellular and extracellular CL^-, both substrate transport and substrate-induced currents by dEAAT1 are completely abolished(Seal et al., 1998).

Difference I–V curves upon anion substitution show AeaEAAT to elicit an analogous substrate-activated anion conductance, although the effect of CL^- on AeaEAAT is not as dramatic as for dEAAT1 or hEAAT4 and 5. Replacement of extracellular CL^- with gluconate, a more bulky anion, reduces the amplitude of both inward and outward currents,as for dEAAT1 (Fig. 6A,B),whereas substitution with more permeant anions such as NO₃^- and SCN^- increases the outward currents and shifts the reversal potential to values more negative than E_Cl, as seen for mammalian EAATs(Wadiche et al., 1995a; Zerangue and Kavanaugh,1996a). The gluconate-mediated effects can be explained by employing explanations similar to those used for dEAAT1, where (1) the reduced outward current results from a reduction in inward anion permeation, and (2)the reduced inward current results from a reduced outward flow of CL^- due to a change in the chemical gradient(Seal et al., 1998). On the other hand, effects of NO₃^- and SCN^- are due to increased permeation of these anions through AeaEAAT. Removal of both extracellular and intracellular CL^- by dialysis with gluconate abolishes the reversal of the substrate-elicited current, while retaining an inward current (Fig. 6C),reflecting the coupled transport of substrate, Na⁺, H⁺and countertransport of K⁺. We conclude that the anion channel activity is not stoichiometrically coupled to substrate transport, since radiolabeled substrate transport under voltage clamp is not compromised regardless of the presence of CL^-(Fig. 6D).

For the mammalian EAATs, the anion channel activity was recently separated from transport activity on a molecular level, where sulfhydryl modification of mutated residues V449C or V452C in hEAAT1 abolished substrate transport but not the anion conductance (Ryan and Vandenberg, 2002; Seal et al.,2001). These amino acids are conserved in AeaEAAT (V398 and V402)and could thus provide the basis for its ability to also exhibit an anion conductance. In summary, studies on the anion channel properties of AeaEAAT demonstrate its similarity to hEAAT1, 2 and 3, rather than hEAAT4 or hEAAT5. Coupled with the sequence comparisons and pharmacological studies discussed previously, these data strengthen the hypothesis that this protein is a functional homologue of hEAAT3.

Competition studies for d-aspartate transport by the 20 amino acids revealed statistically significant inhibition by l-glutamine and l-asparagine, amidated forms of the excitatory amino acid substrates (Fig. 4). While preliminary studies in our laboratory implicate the ability of AeaEAAT to transport these amidated amino acids (data not shown), these results, as well as the l-glutamine and l-asparagine inhibition noted here, may be an artifact of trace contamination of l-glutamate and l-aspartate in the commercially available amino acids.

The final question addressed in this study is the localization of AeaEAAT,as it is still unclear whether insect EAATs participate in signal termination at invertebrate neuromuscular junctions. Only two recent reports have addressed this issue, where Soustelle et al.(2002) examined dEAAT1 and dEAAT2 expression via in situ hybridization, and Gardiner et al.(2002) studied TrnEAAT localization via immunohistochemistry. Results from both groups demonstrate the respective EAATs to be glial rather than neuronal. Gardiner et al. (2002) provided additional evidence for TrnEAAT localization in glia at both neuromuscular junctions and in neuropile regions of larval T. ni. The glial localization of these insect transporters illuminates the importance of glia in regulation of extracellular glutamate concentrations. Indeed, in humans it is believed that the glial transporters (hEAAT1 and hEAAT2) principally contribute to glutamate clearance in the brain, thereby preventing glutamate induced excitotoxicity. This has been demonstrated by EAAT knockout mice, in which mice lacking glial EAATs develop neurodegeneration and paralysis that are symptomatic of glutamate excitotoxicity (Rothstein et al., 1996; Tanaka et al.,1997).

Initial experiments to localize AeaEAAT by western analysis showed the greatest immunoreactivity in membrane fractions of the adult thorax(Fig. 8). Given the intensity of the immunoreactive band in thorax membranes compared with the head and abdomen membrane fractions, we initially hypothesized AeaEAAT to be present in adult thorax muscles, since the adult thorax is rich in muscles involved in flight and leg movement. However, immunohistochemistry results from the adult thorax showed specific staining in the neuropile regions of the thoracic ganglia but not in muscles (Fig. 8). The present data do not discriminate between the neuronal or glial localization of AeaEAAT, an issue which is currently being pursued. Nonetheless the results do suggest that AeaEAAT is not involved in termination of glutamatergic signals at the adult mosquito neuromuscular junction. Yet,based on the findings of Gardiner et al.(2002) that insect EAATs are indeed localized in muscles, as well as our hypothesis that a second functional EAAT exists in the mosquito genome (as discussed earlier), we postulate that the second mosquito EAAT would be responsible for termination of glutamatergic signals at the mosquito neuromuscular junction.

It is interesting to note that western analysis of analogous membrane fractions from fourth-instar larvae and day-2 pupae gave virtually negligible immunoreactivity, implying an increase of the AeaEAAT signal with development. Indeed, this is the case for dEAAT1, where the authors have noted that the onset of expression occurs in larval stages rather than in embryonic stages(Soustelle et al., 2002). Tissue-specific RT–PCR performed to localize AeaEAAT gave inconclusive results, as PCR products from genomic and cDNA gave identical sizes,indicating a lack of intronic sequences (data not shown).

In spite of the fact that this clone was originally isolated from an EST library of the adult female midgut/Malpighian tubule library, whole-mount immunolocalization studies using these tissues did not yield any positive results. We believe that this can be attributed to one of the following: (1)the present antibody is not capable of detecting AeaEAAT in the whole-mount preparations used, (2) the transcript, and therefore the protein is expressed below the limits of detection of immunocytochemical procedures, or (3) the clone was isolated from contaminating nerve or muscle tissue surrounding the midgut and Malpighian tubule tissues used to create the cDNA library.

The results presented in this study provide an insight to the role played by excitatory amino acid transporters in the mosquito, Aedes aegypti.Mosquitoes have historically been and still are the cause of many human diseases worldwide, including malaria, dengue, lymphtic filariasis, yellow fever, Japanese encephalitis and West Nile virus(Roberts, 2002). Together,there are over 500 million cases a year of mosquito-borne illnesses. Amidst the heightened interest in mosquito biology, the work presented is a timely contribution to the further understanding of this disease vector.

We thank Drs V. Filippov, Y. Park, C. S. Chiu, and M. E. Adams for valuable discussions throughout the course of this work; Dr Y. Zhao for expert technical guidance; Drs V. Filippov, M. L. Patrick and Ms H. R. Sanders for critically perusing the manuscript; and Ms A. Soderini-Coviella for maintaining the Ae. aegypti colony and help with dissections. This research was funded by grants AI 48049 to S.S.G. and a University of California Dissertation Research Grant to A.U.