Molecular Characterization of Branchial aquaporin 1aa and Effects of Seawater Acclimation, Emersion or Ammonia Exposure on Its mRNA Expression in the Gills, Gut, Kidney and Skin of the Freshwater Climbing Perch, Anabas testudineus

We obtained a full cDNA coding sequence of aquaporin 1aa (aqp1aa) from the gills of the freshwater climbing perch, Anabas testudineus, which had the highest expression in the gills and skin, suggesting an important role of Aqp1aa in these organs. Since seawater acclimation had no significant effects on the branchial and intestinal aqp1aa mRNA expression, and since the mRNA expression of aqp1aa in the gut was extremely low, it can be deduced that Aqp1aa, despite being a water channel, did not play a significant osmoregulatory role in A. testudineus. However, terrestrial exposure led to significant increases in the mRNA expression of aqp1aa in the gills and skin of A. testudineus. Since terrestrial exposure would lead to evaporative water loss, these results further support the proposition that Aqp1aa did not function predominantly for the permeation of water through the gills and skin. Rather, increased aqp1aa mRNA expression might be necessary to facilitate increased ammonia excretion during emersion, because A. testudineus is known to utilize amino acids as energy sources for locomotor activity with increased ammonia production on land. Furthermore, ammonia exposure resulted in significant decreases in mRNA expression of aqp1aa in the gills and skin of A. testudineus, presumably to reduce ammonia influx during ammonia loading. This corroborates previous reports on AQP1 being able to facilitate ammonia permeation. However, a molecular characterization of Aqp1aa from A. testudineus revealed that its intrinsic aquapore might not facilitate NH3 transport. Hence, ammonia probably permeated the central fifth pore of the Aqp1aa tetramer as suggested previously. Taken together, our results indicate that Aqp1aa might have a greater physiological role in ammonia excretion than in osmoregulation in A. testudineus.


Introduction
Aquaporins (AQPs) are an extended family of integral membrane proteins that mediate facilitated transmembrane water transport [1]. They exist as tetramers [2] with each monomer possessing its own functional channel [3,4]. At the center of the four monomers lies a fifth pore composed mainly of hydrophobic amino acids. In mammals, 13 homologs of AQPs (0-12) are known, with some also involved in the exchange of glycerol and other low molecular weight solutes such as urea, CO 2 , or NH 3 [5,6].
The most ubiquitous and extensively studied AQP is aquaporin 1 (AQP1), which was the first AQP discovered while Preston and Agre [7] were identifying Rh blood group polypeptides in the erythrocyte plasma membrane. Human AQP1 has 269 amino acids of 28 kDa, and contains 6 transmembrane regions with 5 connecting loops, of which 3 (A, C, and E) are located outside the cell and 2 (B and D) within the cytoplasm [8]. Two identical asparagine-proline-alanine motifs at residues 76-78 (in cytoplasmic loop B) and 192-194 (in extracellular loop E) are connected to each other within the membrane, forming a single narrow aqueous pathway (aquapore) of 2.8 Å in diameter at the narrowest point (the constriction region) as calculated by electron crystallography [9,10,11]. The outer constriction region contains an aromatic/ arginine motif, which acts as a selective filter [12]. In some AQPs, the polarity and diameter of this constriction region is tuned to facilitate the transport of polar solutes other than water [13,14].
Since H 2 O and NH 3 have similar molecular sizes and charge distribution, several studies examined the role of aquaporins, in particular AQP1, AQP3, AQP8, and AQP9, in transmembrane NH 3 transport. Nakhoul et al. [15] expressed human AQP1 in Xenopus oocytes, which have low NH 3 permeability, and concluded that NH 3 permeability was enhanced by AQP1. However, not all studies have confirmed that AQP1 can transport NH 3 [16,17]. Holm et al. [16] expressed human aquaporins AQP8, AQP9, AQP3, and AQP1 in Xenopus oocytes to study the transport of NH 3 and NH 4 + under open-circuit and voltage-clamped conditions, and concluded that apart from being water channels, AQP3, AQP8 and AQP9 also supported significant fluxes of NH 3 and NH 4 + . Yet, based on a similar technique, Musa-Aziz et al. [18] reported recently that human AQP1 enhanced NH 3 influx significantly more than AQP4 and AQP5 in Xenopus oocytes, pointing to facilitated transport of NH 3 by AQP1 and contradicting the report of Holm et al. [16] that AQP1 did not significantly affect NH 3 transport.
The climbing perch, Anabas testudineus (Bloch), is a freshwater teleost belonging to Order Perciformes and Family Anabantidae. It can be found in canals, lakes, ponds, swamps and estuaries in tropical Asia, and can tolerate extremely unfavorable water conditions [36]. It possesses accessory breathing organs (or labyrinth organs) in the upper part of the gill-chambers, which facilitate the utilization of atmospheric air [37,38,39]. Periodically, it approaches the water surface to gulp air, which is channeled to the accessory breathing organs for gaseous exchange. During drought, A. testudineus stays in pools associated with submerged woods and shrubs [40], or buries under the mud [41]. To search for a new habitat, it can travel long distances on land between pools of water, covering several hundred metres per trip when the air is sufficiently humid [42]. During emersion, A. testudineus can maintain relatively low plasma urea and ammonia concentrations due to its ability to actively excrete ammonia through the gills and skin [43]. Since it is capable of active ammonia excretion, it also exhibits extraordinarily high tolerance of environmental ammonia (,100 mmol l 21 NH 4 Cl at pH 7.0). In addition, A. testudineus can acclimate from freshwater to seawater through a progressive increase in salinity [44]. Recently, it has been demonstrated that active extrusion of Na + during seawater acclimation and active excretion of NH 4 + during exposure to environmental ammonia in freshwater in the gills of A. testudineus involve similar transport mechanisms, Na + /K + -ATPase, Na + :K + :2Cl 2 cotransporter and cystic fibrosis transmembrane conductance regulator, but different types of mitochondrion-rich cells [45,46,47]. NH 4 + can be transported, in substitution of K + , from plasma into mitochondrion-rich cells through the basolateral Na + :K + :2Cl 2 cotransporter [45], and exit the apical membrane through an unknown NH 4 + transporter down a favorable electrochemical potential generated by the excretion of Cl 2 and/or HCO 3 2 through the apical cystic fibrosis transmembrane conductance regulator [47]. The main function of Na + /K + -ATPase in active NH 4 + excretion is to maintain intracellular Na + and K + homeostasis, instead of transporting NH 4 + directly into mitochondrion-rich cells [46]. Since Aqp1aa is exclusively localized in the branchial epithelium of gilthead seabream [48], and Aqp1-like water channels are found in mitochondrion-rich cells in the gills of rainbow wrasse [33], the first objective of this study was to obtain the full cDNA sequence of aqp1aa from the gills of A. testudineus. The second objective was to examine the tissue expression of aqp1aa in A. testudineus. The third objective was to determine the mRNA expression of aqp1aa in the gills, anterior gut, posterior gut, kidney and skin of A. testudineus kept in freshwater (control) or exposed to seawater (salinity 30; 1 or 6 days), terrestrial conditions (1 day), or environmental ammonia (100 mmol l 21 NH 4 Cl; 1 day) using quantitative real-time PCR (qPCR). The hypothesis tested was that aqp1aa/Aqp1aa might have a more prominent role in ammonia excretion than in osmoregulation in A. testudineus which, despite being regarded commonly as a freshwater teleost, could acclimate to seawater, survive terrestrial exposure and tolerate high concentrations of environmental ammonia.

Animals
Specimens of A. testudineus (25-45 g body mass) were purchased from a local fish distributor. Fish were kept in dechlorinated tap water (freshwater; pH 6.8-7.0) at 25uC in fiberglass tanks with a continuous flow through system for at least 2 weeks under a 12 h light: 12 h dark regime before experiments. No aeration was provided because A. testudineus is an obligatory air-breather. They were fed frozen blood worms once every two days. Procedures adopted in this study were approved by the Institutional Animal Care and Use Committee of the National University of Singapore (IACUC 021/10 and 098/10).

Experimental conditions and collection of samples
Control fish (N = 6) were immersed in 25 volumes (v/w) of freshwater. For fish exposed progressively to seawater, they (N = 12) were randomly selected and transferred to fiberglass tanks containing freshwater (pH 7.0) on day 0 and subsequently, to salinity 10 (pH 7.4) on day 1, salinity 15 (pH 7.6) on day 2, salinity 20 (pH 7.8) on day 3, salinity 25 (pH 8.1) on day 4, and salinity 30 (seawater, pH 8.3) on day 5. Some fish were kept in seawater for an additional 6 days. Natural seawater was collected from the sea at least 1 km away from the coast of the Singapore main island. Waters of different salinities were prepared by mixing seawater with an appropriate quantity of freshwater. Salinity was monitored using a YSI Model 30/10 FT salinometer (Yellow Springs Instrument Co. Inc, Ohio, USA). Fish (N = 6) were killed for sample collection after 1 or 6 days in seawater. During salinity acclimation, fish were fed frozen blood worms on alternate days but fasted 2 days before sample collection. For fish exposed to terrestrial conditions, they (N = 6) were randomly selected, fasted for 2 days and transferred to fiberglass tanks containing a thin film of freshwater (10 ml) for one day. Another batch of fish (N = 6) were randomly selected and exposed to 100 mmol l 21 NH 4 Cl at pH 7.0 for one day. Fish were anaesthetized with 0.05% neutralized MS-222 and killed with a strong blow to the head. Gills, gut, kidney, skin, brain and accessory breathing organs were quickly excised, cooled in liquid N 2 and stored at 280uC.

Total RNA extraction and cDNA synthesis
The total RNA of the gill sample was extracted using the chaotropic extraction protocol of Whitehead and Crawford [49], and further purified using the Qiagen RNeasy Mini Kit (Qiagen GmbH, Hilden, Germany). Following isolation, RNA was quantified spectrophotometrically using a Hellma traycell (Hellma GmbH & Co. KG, Müllheim, Germany). The RNA quality was checked electrophoretically to verify RNA integrity and RNA was stored at 280uC. First strand cDNA was synthesized from 1 mg of total RNA using oligo(dT) 18 primer and the RevertAid TM first strand cDNA synthesis kit (Fermentas International Inc., Burlington, ON, Canada).

Polymerase Chain Reaction (PCR)
The partial aqp1aa sequence was obtained using primers (Forward: 59-ASATMAGYGGHKCCCA-39; Reverse: 59-CCAG-TAHACCCARTG-39) designed from the highly conserved regions from multiple alignments of the aqp1 sequences from various fish species available in Genbank (http://www.ncbi.nlm. nih.gov/Genbank/). Polymerase chain reaction (PCR) was performed in Biorad Peltier thermal cycler (Biorad, Hercules, CA, USA) using Dreamtaq polymerase (Fermentas International Inc.). The cycling conditions were 95uC for 3 min, followed by 35 cycles of 95uC for 30 s, 55uC for 30 s, 72uC for 2 min and a final extension of 72uC for 10 min. PCR products were separated by electrophoresis in 1% agarose gel. Bands of the estimated aqp1aa sizes were excised and purified from the gel using QIAquickH Gel Extraction Kit (Qiagen GmbH) according to manufacturer's protocol. Purified PCR products were subjected to cycle sequencing using BigDyeH Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) and sequenced using the 3130XL Genetic Analyzer (Applied Biosystems).

Rapid amplification of cDNA ends (RACE)-PCR
Total RNA (1 mg) isolated from the gills of A. testudineus in freshwater was reverse transcribed into 59-RACE-Ready cDNA and 39RACE-Ready cDNA using SMARTer TM RACE cDNA Amplification kit (Clontech Laboratories, Mountain View, CA, USA). RACE-PCR was performed using the AdvantageH 2 PCR kit (Clontech Laboratories) to generate the 59 and 39 cDNA fragments, with 59-GGCTTAACGCTCTCAGTGGTGT-TACCC-39 and 59-GTAACACCACTGAGAGCGTTAAGC-39, respectively. RACE-PCR cycling conditions were 25 cycles of 94uC for 30 s, 65uC for 30 s and 72uC for 4 min. RACE-PCR products were separated using gel electrophoresis, purified and sequenced.
The partial fragments of aqp1aa obtained from the gills of A. testudineus were aligned using BioEdit [50] to obtain the full-length nucleotide coding sequence, which were then translated into amino acid sequence. The deduced amino acid sequence was aligned and compared with selected Aqp from various animal species using BioEdit. The sequence identity generated was used to confirm the identity of the Aqp1aa from A. testudineus. Transmembrane domains were identified using the MEMSATS & MEM-SAT-SVA provided by PSIPRED protein structure prediction server (http://bioinf.cs.ucl.ac.uk/psipred/) [51].

Tissue expression
Total RNA (1 mg) isolated from gills, anterior gut, posterior gut, kidney, skin, brain and accessory breathing organs of A. testudineus kept in freshwater were reverse transcribed into cDNA using oligo(dT) 18 primer and the RevertAid TM first strand cDNA synthesis kit (Fermentas International Inc.). PCR was performed on the cDNAs of these tissues using forward primer 59-AATTCAAGAGCAAGAACTTCTG-39 and reverse primer 59-GAGCGACACCTTCACCTC-39 to detect the mRNA expression of each gene in various tissues. Each PCR was carried out in 10 ml reaction volumes using Dreamtaq polymerase (Fermentas International Inc.) with thermal cycling conditions: 95uC for 3 min, followed by 30 cycles of 95uC for 30 s, 55uC for 30 s, 72uC for 30 s and a final extension of 72uC for 10 min. PCR products were then separated by electrophoresis in 2% agarose gel. qPCR RNA from gill samples were treated with Deoxyribonuclease I (Sigma-Aldrich Co., St. Louis, MO, USA), to remove any contaminating genomic DNA. First strand cDNA was then synthesized from 1 mg of total RNA using random hexamer primer and the RevertAid TM first strand cDNA synthesis kit (Fermentas International Inc.). qPCR was performed in triplicates using a StepOnePlus TM Real-Time PCR System (Applied Biosystems). The standard cDNA (template) was serially diluted in 1X TE buffer (1 mmol 21 Tris, 0.1 mmol l 21 EDTA, pH 8.0) (from 10 6 to 10 2 specific copies/2 ml). The qPCR reactions contained 5 ml of 2X Fast SYBRH Green Master Mix (Applied Biosystems), 0.3 mmol l 21 of forward (59-AATTCAAGAGCAAGAACTTCTG-39) or reverse primers (59-GAGCGACACCTTCACCTC-39), and cDNA (equivalent to 1 ng of RNA) or standard (2 ml) in a total volume of 10 ml. Cycling conditions were 95uC for 20 s (1 cycle), followed by 45 cycles of 95uC for 3 s and 60uC for 30 s. Data (threshold cycle as C T values) were collected at each elongation step. Runs were followed by melt curve analysis by increasing from 60uC to 95uC in 0.3uC increments to confirm the presence of only a single product. The PCR products were separated in a 2% agarose gel to verify the presence of a single band.
In order to determine the absolute quantity of aqp1aa transcripts in a qPCR reaction, efforts were made to produce a pure amplicon (standard) of a defined region of aqp1aa cDNA from the gills of A. testudineus following the methods of Gerwick et al. [53]. PCR was performed with aqp1aa qPCR primers and cDNA as a template in a final volume of 25 ml with the following cycling conditions: initial denaturation of 95uC for 3 min, followed by 35 cycles of 95uC for 30 s, 60uC for 30 s and 72uC for 30 s and 1 cycle of final extension of 72uC for 10 min. The PCR product was separated in a 2% agarose gel. The product was excised and purified using QIAquick gel extraction kit (Qiagen GmbH). The aqp1aa fragment in the purified product was cloned using pGEMH-T Easy vector (Promega Corporation, Madison, WI, USA). The presence of the insert in the recombinant clones was confirmed by sequencing. The cloned circular plasmid was quantified using a spectrophotometer. A standard curve was obtained from plotting threshold cycle (C T ) on the Y axis and the natural log of concentration (copies/ml) on the X axis. The C T , slope, PCR efficiency, Y intercept and correlation coefficient (R 2 ) were calculated using the default setting of StepOne TM Software v2.1 (Applied Biosystems). Diluted standards were stored at 220uC. The PCR efficiency for aqp1aa was 96.9%. The quantity of transcript in an unknown sample was determined from the linear regression line derived from the standard curve and expressed as copies of transcripts per ng cDNA.

Statistical analysis
Results were presented as means 6 standard errors of the mean (S.E.M.). Independent two-tailed t test or one-way analysis of variance (ANOVA), followed by multiple comparisons of means by the Tukey test, were used in the evaluation of the differences between means where applicable. Differences were regarded as statistically significant at P,0.05.

Nucleotide sequence, translated amino acid sequence and phylogenetic analysis
The complete cDNA coding sequence of aqp1aa obtained from the gills of A. testudineus consisted of 786 nucleotides (Genbank accession number JX645188), coding for 261 amino acids with an estimated molecular mass of 27.4 kDa (Fig. S1). An alignment of the deduced amino acid sequence of Aqp1aa from A. testudineus with those from human, frog and three other fishes (lungfish, pufferfish and seabream) revealed six transmembrane regions, six potential phosphorylation sites and one N-glycosylation site (Fig. 1). The substrate discrimination sites at the aromatic/arginine constriction and the asparagine-proline-alanine motifs were conserved. A comparison of A. testudineus Aqp1aa with other teleost Aqp sequences reveals that it shares the highest amino acid sequence identity with Aqp1/Aqp1a (67.7-92.3%), followed by Aqp1b (57.5-64.3%; Table 1). This is highly indicative of its identity as Aqp1aa. A phylogenetic analysis further confirms that the Aqp1aa of A. testudineus is grouped together with teleost Aqp1/ Aqp1a, separated from teleost Aqp1b or lungfish and tetrapod Aqp1 (Fig. 2).

Tissue expression
Expression of aqp1aa were detected strongly in the gills, brain, liver, kidney and skin, but weakly in the anterior gut, accessory breathing organs and posterior gut (Fig. 3).

mRNA expression
Based on qPCR results, the highest expression of aqp1aa mRNA (copies of transcripts per ng cDNA) was detected in gills (,1000 copies; Fig. 4A), followed by skin (,800 copies; Fig. 4E) and kidney (,200 copies; Fig. 4D) of A. testudineus in freshwater. In comparison, the mRNA expression of aqp1aa in the anterior (,27 copies; Fig. 4B) and posterior (,17 copies; Fig. 4C) gut of these fish were very low.
The mRNA expression of aqp1aa in the gills (Fig. 4A), anterior gut (Fig. 4B) and posterior gut (Fig. 4C) of A. testudineus exposed to seawater for 1 or 6 days after a progressive increase in salinity were comparable to that of the freshwater control. By contrast, 1 day of seawater exposure led to a significant increase and a significant decrease in the mRNA expression of aqp1aa in the kidney and skin, respectively; however, such changes were transient and were not observed in fish after 6 days of exposure to seawater.
As for exposure of A. testudineus to 100 mmol 21 NH 4 Cl in freshwater, results obtained were different from those of terrestrial exposure, with significant decreases in mRNA expression of aqp1aa in gills after 6 days (Fig. 6A), kidney after 1 day (Fig. 6D), and skin after 1 or 6 days (Fig. 6E). Exposure to 100 mmol 21 NH 4 Cl in freshwater had no significant effects on the mRNA expression of aqp1aa in the anterior (Fig. 6B) and posterior (Fig. 6C) gut.

Discussion
Despite being regarded commonly as a freshwater teleost, A. testudineus can acclimate to seawater, survive terrestrial exposure and tolerate high concentrations of environmental ammonia. Since the gills and skin of A. testudineus had the highest expression of aqp1aa, Aqp1aa could have an important physiological function in these organs. However, the major function of Aqp1aa in A. testudineus might not be in osmoregulatory acclimation because of two reasons: (1) seawater acclimation had no significant effects on the mRNA expression of aqp1aa in the gills and gut, and (2) the mRNA expression of aqp1aa in the gut was extremely low. Terrestrial exposure led to significant increases in the mRNA expression of aqp1aa in the gills and skin of A. testudineus, but Aqp1aa could not have functioned predominantly in water permeation which would result in deleterious water loss through evaporation. Since it has been established previously that A. testudineus utilizes amino acids as energy sources for locomotor activity leading to increased ammonia production while on land [43], it is logical to deduce that increased aqp1aa mRNA expression might be necessary to facilitate increased ammonia excretion during emersion. The proposition that Aqp1aa could facilitate ammonia permeation is further supported by the observation that exposure to environmental ammonia led to significant decreases in mRNA expression of aqp1aa in the gills and skin, probably to reduce the influx of ammonia during ammonia loading. Hence, our results indicate that Aqp1aa could have a greater physiological role in ammonia excretion than in osmoregulation in A. testudineus.
Molecular characterization of Aqp1aa from the gills of A. testudineus: the intrinsic aquapore is permeable to water but not NH 3 An alignment of the deduced Aqp1aa sequence of A. testudineus with those from other species shows highly conserved segments, which include the pore-lining residues of the aquapore, the asparagine-proline-alanine motifs, the AQP1-inhibitor (HgCl 2 ) binding site and the outer aromatic/arginine constriction in the aquapore. The substrate discrimination sites of the aromatic/ arginine constriction consist of Phe63, His187, Cys189 and Arg202 in A. testudineus Aqp1 (corresponding to Phe56, His180, Cys189 and Arg195 in human AQP1). His187 and Arg202 provide a hydrophilic edge with Phe56 [54]. The sulfhydryl group of Cys189 extends into the pore and is the binding site for the AQP1-inhibitor HgCl 2 [55,56]. The remaining part of the aquapore contains hydrophobic residues, exposing the main-chain carboxyl oxygens to the pore surface [54]. They act as hydrogen bond acceptor sites to channel small hydrogen bond donor molecules, such as water, through the aquapore.
Beitz et al. [17] analyzed the function of three residues in the aromatic/arginine constriction (Phe56, His180, and Arg195) in rat AQP1. Individual or joint replacement of His180 and Arg195 by alanine and valine, respectively (AQP1-H180A, AQP1-R195V, and AQP1-H180A/R195V), did not affect water permeability, but the double mutant AQP1-H180A/R195V allowed urea to pass through. In line with the predicted solute discrimination by size, replacement of both Phe56 and His180 (AQP1-F56A/H180A) enlarged the maximal diameter of the aromatic/arginine constriction by 3-fold and enabled the passage of glycerol or urea. Beitz et al. [17] showed that NH 3 could not permeate through the aromatic/arginine constriction of rat AQP1, but it passed through all four AQP1 mutants. Since A. testudineus Aqp1aa possesses equivalents of Phe56, His180, and Arg195 in its aromatic/arginine constriction, its intrinsic aquapore probably facilitates water but not NH 3 movement. However, the possibility of NH 3 permeation through the central pore of the tetramer cannot be ignored (see below).

A. testudineus during seawater acclimation
To compensate for passive water loss, marine teleosts drink seawater and actively secrete salt via the gills and kidneys. In contrast, freshwater teleosts do not drink (or drink very little) water, but actively absorb salt from the environment through the gills and produce copious hypoosmotic urine to remove excess water via the kidney [57,58,59]. Euryhaline teleosts, such as A. testudineus, can survive in both freshwater and seawater environments due to their ability to alter osmoregulatory mechanisms upon exposure to media of different salinity.
Fish gills are in direct contact with the surrounding aquatic medium and have a potential risk of large transepithelial water fluxes due to the osmotic gradient [60]. It has been reported that salinity changes would lead to changes in mRNA expression of aqp1aa/aqp1ab in the gills of several fish species [29,30,31,32,60,61], indicating the involvement of Aqp1aa/ Aqp1ab in branchial osmoregulatory acclimation. However, there is no significant difference in the branchial aqp1aa mRNA expression between A. testudineus exposed to seawater for 1 or 6 days and the freshwater control. From an osmoregulatory perspective, transepithelial water permeability via the branchial epithelium should be kept to a minimum by reducing passive loss of water through a down-regulation of water channels during exposure to hyperosmotic environments. Hence, it can be concluded that, despite having an aquapore that would facilitate water permeation, Aqp1aa does not play a major role in osmoregulation in the gills of A. testudineus during seawater acclimation. However, the branchial mRNA expression of aqp1aa is the highest among all tissues/organs studied. Therefore, Aqp1aa probably has an important physiological function unrelated to seawater acclimation in the gills of A. testudineus. Marine or seawater acclimatized fish counteract the osmotic loss of water by drinking; and their guts play a crucial osmoregulatory role in water absorption. Several studies have demonstrated the mRNA and/or protein expression of aqp1aa/aqp1ab in guts of teleost fish [25,26]. Transcript and protein abundance of aqp1aa/ aqp1ab typically increase towards the distal portions of the gut with  the highest level in the posterior region and rectum [22,23,31,60,62]. The increased Aqp1aa/Aqp1ab expression in the gut would probably contribute to increased water permeability and hence higher rates of intestinal water absorption during seawater acclimation [22,60,62]. Indeed, seawater acclimation leads to significant increases in aqp1aa/aqp1ab mRNA expression in the guts of several fish species [22,23,25,30,31,60,63]. Furthermore, an injection of cortisol into freshwater eels increases water flux in isolated guts [64] and up-regulates the expression of Aqp1aa throughout the intestines, thereby enhancing intestinal permeability and water absorption [35]. However, the mRNA expression of aqp1aa in the anterior and posterior gut of freshwater A. testudineus was extremely low and unaffected by seawater acclimation. Thus, it can be concluded that unlike other species of fish, Aqp1aa does not function as a key water channel in water absorption in the gut of A. testudineus acclimated to seawater.
In mammalian kidneys, AQP1 is essential for concentrating urine in the thin descending limb of loop of Henle [65]. Kidneys of teleost fish are unable to produce hypertonic urine, but isotonic conditions may be reached in the urine of euryhaline fish during acclimation to hyperosmotic environments. This is largely explained by increased re-absorption of water from the glomerular filtrate in the distal segments of nephrons [66]. In the European eel, Aqp1aa is present within the apical brush border of epithelial cells of a subset of renal tubules, and it seems to be more prominent in the proximal rather than the distal tubules [63,67]. However, the functional role of Aqp1aa in fish kidney is controversial at present [30,32,60,63], although there are indications that the osmoregulatory role of renal Aqp1aa may be different from that of renal Aqp1ab [26]. For A. testudineus, there is a significant increase in the mRNA expression of aqp1aa in the kidney after 1 day, but not 6 days, of acclimation to seawater, indicating that Aqp1aa may have at most a transient role in water re-absorption during seawater acclimation.
Today, there is a dearth of knowledge on the response of Aqp1aa to salinity stress in the skin of fish. We report for the first time a decrease in the mRNA expression of aqp1aa in the skin of A. testudineus after 1 day, but not 6 days, of seawater exposure. The transient nature of this decrease in aqp1aa expression in the skin of A. testudineus indicates that Aqp1aa might not have a significant role in regulating water loss in a hyperosmotic environment, because the experimental fish were confronted with osmotic water loss not just for 1 day, but throughout the 6 days of acclimation period. Of note, Chang et al. [44] reported that A. testudineus reduced ammonia excretion and simultaneously accumulated certain amino acids, presumably for cell volume regulation, during a progressive increase in salinity. However, after long term acclimation to seawater, tissue amino acid concentrations returned to normal, and there is a significant increase in ammonia excretion instead [44]. Since it has been reported previously that AQP1 can also act as an ammonia transporter [15,18], our results, when taken together with those reported by Chang et al. [44], suggest that Aqp1aa might be involved in increased ammonia excretion in A. testudineus during seawater acclimation.
Up-regulation of mRNA expression of aqp1aa and the possible role of Aqp1aa in ammonia excretion in A. testudineus during terrestrial exposure Terrestrial exposure poses a number of challenges to teleosts; the two major problems are (1) desiccation due to water loss, and (2) ammonia intoxication due to inefficient ammonia excretion resulting from a lack of water to flush the branchial epithelium. To the best of our knowledge, there is no information on the effects of terrestrial exposure on the expression of any aqp isoform in airbreathing fishes in the literature. Results obtained from this study indicate for the first time that 1 day of terrestrial exposure leads to significant increases in the mRNA expression of aqp1aa in several organs, including gills and skin, of A. testudineus. To deal with desiccation during terrestrial exposure, it would be essential for A. testudineus to reduce water loss through the gills and skin, which have large surface areas. Hence, it is highly unlikely that the increase in expression of aqp1aa represents a provision for increased evaporative water loss through the branchial and cutaneous surfaces. This further supports the proposition that Aqp1aa may not function predominantly as a water channel in the gills and skin of A. testudineus during osmoregulatory acclimation. AQP1 is known to facilitate CO 2 permeation [5], but the increased expression of aqp1aa in the gills and skin of A. testudineus could be unrelated to CO 2 excretion during emersion. Since A. testudineus is an obligatory air-breather and possesses accessory breathing organs for air-breathing, it is unlikely that it would be confronted with problems related to CO 2 excretion while on land. Rather, our results indicate a possible relationship between increased aqp1aa expression and increased ammonia excretion in A. testudineus during terrestrial exposure.
Although some aquaporins, such as AQP8 [68], are known to facilitate NH 3 permeation, whether mammalian AQP1 can enhance ammonia conductance is controversial [15,16,18,54,69]. The first study on the possible role of AQP1 as an ammonia transporter was performed by Nakhoul et al. [15] who expressed human AQP1 in Xenopus oocytes and concluded that it facilitated NH 3 transport. Subsequently, Holm et al. [16] used Xenopus oocytes under open-circuit and voltage-clamped conditions (to exclude NH 4 + and H + transport) to study the effect of several human AQPs on NH 3 transport by monitoring the rate of acidification of a weakly buffered external medium. They reported that, except for AQP1, expression of AQP3, AQP8, and AQP9 increased acidification, confirming their functional roles in enhancing NH 3 influx across the cell membrane [16]. Based on a technique similar to that of Holm et al. [16], Musa-Aziz et al. [18] reported recently that AQP1 enhanced NH 3 influx significantly more than AQP4 and AQP5 in Xenopus oocytes, pointing to facilitated transport of NH 3 by AQP1. While the discrepancies between results of Holm et al. [16] and those of Musa-Aziz et al. [18] could be due to differences in sensitivities of the methods employed in their studies [8], they also point to the possibility that the capacity (or lack thereof) of an AQP channel to conduct ammonia cannot be determined solely by amino acid residues in the aromatic/arginine constriction of the AQP monomer as described by Beitz et al. [17].
Indeed, AQP homologs in yeasts and plants can facilitate ammonia transport, despite having completely dissimilar amino acid residues in the aromatic/arginine constriction [54,70]. For plant AQP homologs, the tonoplast intrinsic proteins (TIPs) from wheat (TaTIP2;1) and Arabidopsis (AtTIP2;1 and AtTIP2;3) can also facilitate NH 3 transport in addition to being water channels [69,71]. However, the conduction of water and ammonia through TaTIP2;2 from wheat is differentially affected by inhibitors [72], indicating that NH 3 permeation may not occur through the monomeric channel pores. Indeed, in the wheat TaTIP2;2, NH 3 is not transported in file with water, but through a separate pathway, which could be supplied by the fifth central pore in the TaTIP2;2 tetramer conformation [72]. Dynowski et al. [70] conducted molecular simulations on Arabidopsis thaliana AtTIP1;2 and AtPIP2;1 (plasma membrane intrinsic protein 2-1) to test the relevance of different selectivity filters, and their results did not support the proposition that an Arg/His pair in the aromatic/ arginine selectivity filter region would promote ammonia conductance [54]. They showed that ammonia could cross the membrane via the central pore instead of the aquapores [70]. The fifth central pore can function as a gated channel moderated by cGMP interaction with the cytoplasmic loop D [73]. It consists of mainly hydrophobic amino acids, which provides a path for non-polar molecules [8]. For human AQP1, which is known to facilitate ammonia transport according to Nakhoul et al. [15] and Musa-Aziz et al. [18], amino acid residues with side chains lining the cytoplasmic and periplasmic constriction regions of the central pore include Val50, Leu54, Leu170 and Leu174 (Fig. 1). Since A. testudineus Aqp1aa possesses equivalent amino acid residues of Val44, Leu48, Leu162 and Leu166, it is reasonable to suggest that the central pore formed in its tetrameric conformation has the physicochemical potential for ammonia permeation.
Despite the absence of supportive evidence, it has been proposed previously that Aqp would facilitate transepithelial ammonia fluxes in gills of fish [33,74,75]. Our results offer for the first time indirect support to the proposition that Aqp1aa could be involved in ammonia excretion through the gills and skin of A. testudineus. Since A. testudineus utilizes amino acids as energy sources for locomotor activity on land [43], which leads to an increase in ammonia production, it is highly probable that aqp1aa expression was up-regulated in gills and skin to facilitate passive ammonia excretion at the initial phase of terrestrial exposure before the buildup of an ammonia electrochemical gradient that requires the participation of active transport mechanisms [45,46,47]. A similar functional role in passive NH 3 permeation has been suggested for AQP8 in the inner mitochondrial membrane of liver cells [76].
In general, the gills are the major site of ammonia excretion in fish, although smaller quantities of ammonia may also be eliminated by the kidney [77]. However, unlike gills and skin, the kidney is not directly exposed to the external environment. More importantly, ammonia excretion through the kidney requires a constant supply of water for urine production. Therefore, the kidney probably plays a minimal role in ammonia excretion terrestrial exposure during which desiccation is a major issue. This could account for the lack of change in aqp1aa expression in the kidney of A. testudineus after 1 day of exposure to terrestrial conditions.
Of note, while a predisposition based on the type of amino acid residues along the fifth central pore of the Aqp1aa may satisfy certain requirements towards NH 3 transport, other factors such as the size of, and the orientation of the amino acid residues in, the central pore need to be considered. Additionally, the Aqp1aa tetramer may need to interact with certain protein partners in order for the central pore to act as a NH 3 channel. Indeed, mammalian AQP1 has been co-immunoprecipitated with transporters such as Na + /H + and Cl 2 /HCO 3 2 exchangers, and with heterotrimeric complexes of PDZ domain (PDZ is derived from the first three proteins in which these domains were found: PSD-95 which is a 95 kDa protein involved in signaling in the postsynaptic density, Dlg which is the Drosophila discs large protein, and ZO1 which is the zonula occludens 1 protein involved in maintaining epithelial cell polarity) proteins [78]. Interactions with other signaling molecules have also been reported, reinforcing the idea that AQP1 may be regulated by multiple levels of signaling cascades, the onset of which depends on the local cytosolic environment and physiological needs [79]. All these could contribute to the differences in physiological functions of Aqp1aa/Aqp1ab in various fish species, and our results indicate that Aqp1aa could have a more predominant physiological role in ammonia excretion than in osmoregulation in A. testudineus.
Down-regulation of mRNA expression of aqp1aa, presumably to reduce the influx of exogenous ammonia, during ammonia exposure Elevated ammonia concentrations in the environment can lead to impaired ammonia excretion and/or a net influx of ammonia from the environment. The end result is an elevation in body ammonia levels, leading to convulsions and death. Most fish species cannot tolerate high environmental ammonia concentrations, but some species, including A. testudineus, have high environmental ammonia tolerance [75,80,81]. Anabas testudineus has the extraordinary ability to actively excrete ammonia (as NH 4 + ) against an unfavorable blood-to-water P NH3 gradient during exposure to environmental ammonia. However, active excretion of ammonia prescribes that the influx or backflux of NH 3 into the fish must be prevented. The gills are the primary excretory organ where active ammonia excretion occurs in A. testudineus. Moreover, its large surface area implies that it would be a major site prone to the influx of external ammonia. One would therefore expect the gills to reduce its permeability through the reduction of Aqp1aa to prevent the entry of ammonia into the fish body, and the same argument can be applied to the skin. Indeed, there are significant decreases in the mRNA expression of aqp1 in the gills of A. testudineus after 6 days of exposure to ammonia, and in the skin of fish exposed to ammonia for 1 or 6 days. Therefore, it is logical to deduce that down-regulation of aqp1aa could be an essential mechanism to reduce the net influx of NH 3 in A. testudineus during ammonia exposure. These results also corroborate the proposition that increased aqp1aa in the gills and skin of A. testudineus during terrestrial exposure served to facilitate NH 3 and not CO 2 excretion. Besides transcriptional regulation of aqp1aa expression, there could be a decrease in the translation of the existing aqp1aa transcripts and/or an increase in the removal of Aqp1aa from the cell membranes to prevent NH 3 influx, the confirmation of which awaits future studies.

Conclusion
In conclusion, our results indicate that Aqp1aa could have a more prominent physiological role in ammonia excretion than in osmoregulation in A. testudineus, which indirectly support previous propositions that AQP1 can act as an ammonia transporter. A molecular characterisation of A. testudineus Aqp1aa indicates that its aquapore may not be able to facilitate NH 3 movement, but highlights the physicochemical potential for ammonia permeation through the central pore of the Aqp1 tetramer. Therefore, efforts should be made in the future to elucidate the functional role of the central pore of the Aqp1aa tetramer in NH 3 homeostasis in A. testudineus and other fish species. Furthermore, future studies should also focus on the functional roles of Aqp3 and Aqp8 in ammonia excretion versus osmoregulation in fish in general. Figure S1 Nucleotide sequence (GenBank accession number JX645188) and translated amino acid sequence of the full coding region of Aqp1aa from the gills of Anabas testudineus. The start codon is indicated by the first ATG, while the stop codon is indicated by an asterisk. (TIF)