Rhesus glycoprotein gene expression in the mangrove killifish Kryptolebias marmoratus exposed to elevated environmental ammonia levels and air

Most teleost fish excrete ammonia and urea as the primary nitrogenous waste, with ammonia contributing more than 80% of the total(Wood, 1993). The major site of ammonia elimination is the gills, but the kidney and skin may also be involved (for a review, see Wilkie,2002). Although ammonia is produced endogenously, it is a dangerous neurotoxin at elevated concentrations(Felipo and Butterworth, 2002; Rose, 2006). Normally ammonia is excreted predominantly by passive diffusion down the NH₃ partial pressure gradient (P_NH3) from blood to water across the gills (Wilson et al., 1994). In environments of high external ammonia or elevated pH, the P_NH3 gradient across the gills may be diminished or reversed, resulting in ammonia uptake from the environment and/or an inhibition of ammonia excretion (Wright and Wood, 1985; Wilson et al.,1994; Wilkie,2002).

Fish survive in environments of elevated ammonia by one or more of the following strategies: reducing endogenous ammonia production, converting ammonia to less toxic molecules (e.g. urea, glutamine), or continuing to excrete ammonia against the gradient (for a review, see Chew et al., 2006). Very little is known about the latter possibility because it is very difficult to define the true blood-to-water P_NH3 and electrochemical gradients(E_NH4₊)(Wilson et al., 1994). African sharptooth catfish Clarias gariepinus is a species that appears to excrete ammonia actively against high external ammonia, but the exact mechanism of the excretion is currently unknown(Ip et al., 2004). Indeed,only the mudskipper Periophthalmodon schlosseri has been reported to actively excrete ammonia against an indisputable gradient (8 and 30 mmol l^–1 NH₄Cl in the water). The mechanism likely involves basolateral Na⁺-K⁺(NH ⁺₄)-ATPase coupled to an apical Na⁺/H⁺ (NH ⁺₄) exchanger(Randall et al., 1999).

Another possible mechanism of NH₃/NH ⁺₄transport is via the Rhesus glycoproteins. The Rhesus glycoproteins(Rh genes) represent members from the Amt/MEP/Rh superfamily found in all three domains of living organisms: Bacteria, Archaea and Eukarya. Amt and Rh proteins are distantly linked (Huang and Liu, 2001). The first ammonia transporter genes were identified from yeast Saccharomyces cerevisiae (ScMep1) and plant Arabidopsis thaliana (AtAmt1)(Marini et al., 1994; Ninnemann et al., 1994). In mammals, three Rh genes are identified so far: RhAG, RhBG and RhCG. RhAG expression is restricted to the erythrocyte membrane,whereas RhBG and RhCG are expressed in various tissues in the mammalian systems, including liver, kidney and gastrointestinal tract(Handlogten et al., 2005; Liu et al., 2000; Liu et al., 2001; Weiner et al., 2003; Weiner and Verlander, 2003). Studies in mammalian and plant systems reveal that at least some of those Rh genes encode proteins that mediate NH₃/NH ⁺₄ movement (Khademi et al., 2004; Zheng et al.,2004; Mayer et al.,2006), but the form of ammonia (NH₃ gas or NH ⁺₄ ion) being transported, and whether the transport is active or passive, are still under much debate(Bakouh et al., 2004; Khademi et al., 2004; Nakhoul et al., 2005). A full-length cDNA of Rh-like protein (Rh-CM) has been identified recently from the gills of the aquatic crab Carcinus maenas, which has a similar predicted transmembrane structure as the mammalian Rh proteins (Weihrauch et al.,2004). However, the precise role of Rh-CM in crab ammonia excretion awaits elucidation.

To date, there have been only two reports of Rh sequences in fish(Huang and Peng, 2005; Nakada et al., 2007). RhAG, RhBG, RhCG1 and RhCG2 sequences have recently been identified in the pufferfish Takifugu rubripes, with unique spatial distribution within the fish gill, and these Rh genes mediate movement of the ammonia analogue methylammonia when expressed in Xenopus oocytes (Nakada et al.,2007). However, no physiological studies of Rh genes have been conducted. This family of Rh transporters may be an important missing link in our understanding of branchial ammonia excretion. The mangrove killifish Kryptolebias marmoratus is very tolerant of elevated external ammonia, surviving 48 h of 10 mmol l^–1NH₄Cl at pH 8.0 (Frick and Wright, 2002a). Remarkably, there was very little change in tissue ammonia levels in K. marmoratus exposed to external ammonia, nor were tissue urea or glutamine concentrations altered(Frick and Wright, 2002a). During prolonged air exposure (11 days), K. marmoratus volatilize NH₃ (Frick and Wright,2002b) by elevating both NH ⁺₄ concentration and pH on the cutaneous surface (Litwiller et al., 2006). Ammonia excretion during both elevated environmental ammonia and air exposure may require specialized transport mechanisms to move ammonia across cell membranes possibly against the gradient, similar to the mudskipper P. schlosseri(Randall et al., 1999).

The first objective of this study was to isolate Rh genes from K. marmoratus and determine their tissue distribution, with a particular focus on the gill and skin. The second objective was to quantify changes in Rh mRNA expression in K. marmoratus in response to elevated external ammonia concentrations and aerial exposure. We have cloned three Rh genes from K. marmoratus and investigated the tissue expression pattern of each Rh gene, in control and ammonia-exposed fish. Using quantitative real time-PCR, we have also measured the relative mRNA expression of all three Rh genes in response to (i)high external ammonia in the gills and skin and (ii) aerial exposure in the skin. In addition, relative RhBG expression levels in brain, liver and muscle of control and ammonia-exposed K. marmoratus were also quantified.

Fish were obtained from a breeding colony of Kryptolebias marmoratus Poey held in the Hagan Aqualab at the University of Guelph,Guelph, ON, Canada (Frick and Wright,2002a). Adult hermaphrodite fish, at least 1 year of age and weighing approximately 0.06–0.12 g, were used for experiments. Hermaphrodites were identified by the appearance of an overall mottled brown colouration, a characteristic caudal ocellus, and a whitish border on the anal fin. Fish were kept in individual containers under a constant photoperiod (12 h:12 h, L:D), in 15‰ artificial seawater (made with distilled water and Crystal Sea® Marinemix; Marine Enterprises International, Inc., Baltimore,MA, USA), 25°C, pH 8.1. Water changes were performed every 2 weeks and fish were fed Artemia five times per week. The experiments in this study were approved by the University of Guelph Animal Care Committee.

Two fish were used for Rh gene cloning. Fish were decapitated, the gills removed and immediately frozen in liquid nitrogen. Total RNA was extracted from killifish gill using Trizol (Invitrogen Canada Inc.,Burlington, ON, Canada). First-strand cDNA was synthesized using Superscript reverse transcriptase II (Invitrogen) with an adaptor oligodT primer (T17AP2:GACTCGAGTCGACATCGAT₁₇). A partial sequence (585 bp) was obtained using F primer (GCACACTGTTCCTGTGGATG) and R primer (CAGCAGGATCTCCCCAGA), with the thermal cycle (PTC-200, Peltier Thermal Cycler, MJ Research, Mississauga,ON, Canada): 94°C, 2 min, 45 cycles of 94°C, 30 s; 51°C, 30 s;68°C, 1 min and a final extension at 68°C, 5 min (Invitrogen Platinum Taq). Sequence of this PCR product was confirmed to be a Rhhomologue. Thereafter, 3′ RACE was conducted using the F primer and AP2 primer (GACTCGAGTCGACATCGA) with thermal cycles: 95°C, 5 min, 45 cycles of 94°C, 15 s; 56°C, 1 min; 68°C, 5 min and a final extension at 72°C, 10 min (QIAgen High Fidelity HotStar, Qiagen, Mississauga, ON,Canada). Three PCR products (1.3 kb, 1.6 kb and 1.8 kb) were obtained,gel-purified (QIAquick extraction kit), cloned into pGEM-T easy vectors(Promega, Madison, WI, USA) and transformed into E. coli XL-Blue strain. Plasmids from positive clones were extracted (QIAgen miniprep),sequenced (Mobix sequencing facility, McMaster University) and confirmed to be 3′ end sequences of RhBG, RhCG1 and RhCG2. 5′RACE was then carried out with cloning primers(Table 1) (Marathon cDNA Amplification Kit, Clontech, Moutainview, CA, USA).

Full-length sequences submitted to GenBank have the following accession number: DQ995211 (RhBG), DQ995210 (RhCG1) and DQ423779(RhCG2).

Fish were fasted for 48 h prior to the onset of experiments to eliminate the influence of diet on nitrogen metabolism and excretion. Three groups of animals were studied. Control fish were exposed to 15‰ seawater (pH 8.1) containing no added NH₄HCO₃. Experimental fish were exposed to either 1 or 2 mmol l^–1NH₄HCO₃ (15‰, pH 8.1). At the end of 5 days, fish were decapitated. Fish dissections were conducted under a dissecting microscope. Individual tissues (brain, eye, gill, gonad, gut, kidney, liver,skeletal muscle and skin) were excised and immediately frozen in liquid nitrogen and kept at –80°C for later RNA isolation within 2 weeks. Identical groups of fish were treated in the same way, except at the end of exposure, fish were decapitated and whole fish were frozen in liquid nitrogen,and stored at –80°C for whole-body ammonia analysis within 1 week.

We attempted to measure ammonia excretion to the water in fish exposed to 1 and 2 mmol l^–1 NH₄HCO_3, however, due to the small size of the fish and the high background ammonia, this proved to be difficult. In preliminary experiments we extended the flux period from 4 to 24 h to increase the signal:noise ratio, but this introduced the confounding problem of significant microbial metabolism, which removed ammonia from solution. Hence ammonia excretion data were unreliable and are not presented.

Fish exposed to terrestrial conditions were placed in individual plastic containers containing a moist paper of 15‰ seawater at 25°C, as described previously (Ong et al.,2007). After 24 h, fish were decapitated and skins were dissected,frozen immediately in liquid nitrogen and kept at –80°C until analysis within 1 week.

The frozen fish samples (including heads) were used for whole-body ammonia content measurement. Each sample was powdered using a mortar and pestle under liquid nitrogen, weighed and deproteinized in 10 vols of 10% trichloroacetic acid (TCA) (Fisher Scientific). The homogenate was centrifuged (Sorvall Legend RT, Mandel, ON, Canada) at 12 000 g at 4°C for 15 min. The deproteinized samples were neutralized to pH 6.5–7.0 with 2 mol l^–1 KHCO₃ (Sigma-Aldrich). Ammonia content was determined by the method of Kun and Kearney(Kun and Kearney, 1974). The change in absorbance at 340 nm (25°C) was monitored using a Spectra Max 90 spectrophotometer (Molecular Devices, Woodbridge, ON, Canada). Freshly prepared ammonium chloride (Sigma-Aldrich) was used as a standard.

Reverse-transcription PCR (RT-PCR) was used to determine the mRNA expression pattern of RhBG, RhCG1 and RhCG2 in control and ammonia-exposed tissues. Total RNA extraction and cDNA synthesis were done as described above. A DNase I (Invitrogen) digestion step was used (1 U per 1μg RNA, 15 min at room temperature) to ensure there was no genomic DNA contamination prior to cDNA synthesis. Three sets of gene-specific primers(Table 2) were used to examine the tissue-specific expression of RhBG, RhCG1 and RhCG2; and elongation factor 1alpha (EF1a) was used as the control gene to ensure that cDNA of individual samples were successfully synthesized.

The thermal cycle used was: 94°C, 3 min, 45 cycles of 94°C, 30 s;58°C, 30 s (RhBG and RhCG2) or 55°C, 30 s(RhCG1 and EF1a); 72°C, 1 min and a final extension at 72°C for 5 min (Invitrogen Taq polymerase).

Potential ammonia excretory organs, gill and skin, were used to test for all three Rh gene expressions in control and experimental fish by quantitative PCR (Q-PCR). Other internal organs, brain, liver and skeletal muscle, were selected to analyze for the expression of RhBG. In this study,two control genes, EF1a and 18S ribosomal RNA(rRNA), were used as normalization factors; and the choice of which normalization factor to use was dependent on which gene exhibited the most consistent expression in control and ammonia-exposed or air-exposed tissues. For each Q-PCR reaction, 5 μl of 5× diluted cDNA sample was used. Q-PCR primers used are listed in Table 3.

Quantification of gene expression was done using Platinum SYBR green qPCR Super-mix-UDG (Invitrogen) with PCR thermal cycle (Mx3000P QPCR System,Stratagene, Cedar Creek, Texas, USA): 50°C for 2 min, 95°C for 2 min,followed by 40 cycles of 95°C for 15 s and 60°C for 15 s. Gel electrophoresis was performed to ensure that only a single PCR product of the desired size was obtained for each reaction. The Q-PCR product was purified(QIAquick PCR purification kit) and sequenced to confirm its identity.

Mammalian Rh homologues are predicted to be membrane proteins with 12 membrane spanning domains and cytoplasmic N- and C-terminals(Huang and Liu, 2001), but nothing is known about fish Rh homologues. To predict the potential localization of killifish Rh proteins, we characterized the primary and secondary structure of the killifish Rh sequences by analyzing their hydropathy profiles (Kyte-Doolittle scale) via webserver SDSC Biology WorkBench 3.2 at http://seqtool.sdsc.edu/CGI/BW.cgi. To further confirm that these three Rh sequences encode for membrane proteins, which consist of transmembrane (TM), intra- and extracellular motifs, transmembrane domain organizations were predicted at TMHMM server,version 2.0(http://www.cbs.dtu.dk/services/TMHMM). Furthermore, mammalian Rh homologues have been shown to be glycoproteins(Quentin et al., 2003) and therefore, potential N-glycosylation sites on killifish Rh protein sequences were predicted via NetNGlyc 1.0 Server at http://www.cbs.dtu.dk/services/NetNGlyc.

Sequence alignments (Clustal W) were conducted using BioEdit software downloaded at http://www.mbio.ncsu.edu/BioEdit/bioedit.html. The same software was used for molecular mass, isoelectric point (pI), amino acid composition, and percentage sequence identity calculations.

Phylogenetic and molecular evolutionary analyses were conducted using MEGA version 3.1 downloaded at http://www.megasoftware.net(Kumar et al., 2004). Sequences were first aligned by ClustalW and subjected to a Neighbour-joining(NJ) matrix for tree reconstruction and evaluated by means of a Bootstrap of 1000 replicates.

All data are expressed as mean ± the standard error of the mean(s.e.m.) (N=6–8). Analysis of variance (ANOVA) was used to compare means of relative Rh mRNA expression in the gill with Statistix software (Analytical Software, Tallahassee, FL, USA), followed by the Least Significant Difference (LSD) test to determine where significant differences were present (P<0.05). For Rh mRNA expression level between control (immersion) and air-exposed (emersed) fish, an unpaired t-test was conducted to determine if expression was significantly different (P<0.05; control: N=6; air-exposed: N=4).

We have obtained full-length cDNA sequences of three Rhesus glycoproteins in the killifish K. marmoratus: RhBG, RhCG1 and RhCG2, and they are 1736, 1920 and 2021 bp long, respectively. RhCG1 and RhCG2 are highly homologous in both their nucleotide (72.5% identity) and amino acid (61.1% identity) sequences. RhCG1 and RhCG2 are similar to RhBG, with 50.6% and 49.5% identity, respectively, for nucleotide sequences and 54.4% and 52.3%identity at the amino acid level. All three Rh sequences appear to share high homology with other known Rh gene sequences.

K. marmoratus RhBG is 76.2–84.8% identical to other known fish RhBGs and 52.2–61.5% identical to known mammalian RhBGs. Sequence alignment revealed that K. marmoratus RhBG is highly homologous to other fish RhBGs, with no gap found between ⁵⁵H and ³⁸⁸E(TM2 to TM9 of K. marmoratus RhBG)(Fig. 1A). The open reading frame (ORF) of RhBG encodes a 462-amino acid polypeptide (50.5 kDa), with two potential N-glycosylation sites (⁴⁷NDSH⁵⁰,217NCSV²²⁰) (Fig. 1A). When K. marmoratus RhBG was aligned with all known RhBG sequences from fish and mammals, it was found that variable regions of fish RhBG sequences are located between amino acids 250–300 and the C-terminal sequences. However, these regions are rather conserved among the mammalian RhBG sequences (data not shown).

Many fragments of the RhBG sequences appear to be class-specific. For example, RhBG amino acid regions 200MVTRIL(Y/H)RPNLD²¹¹ and 312SMIVGF(L/M)AG(I/T)ISV³²⁴ are highly conserved among fish sequences. However, the mammalian RhBG sequences of the same regions are highly conserved among themselves and are encoded as 200(V/F)LS(R/W)VLYR(P/S)QLE²¹¹ and ³¹²ALAAGFLAGTVST³²⁴, respectively. Such class-specific regions are also located at the N- and C-terminal sequences (data not shown).

RhCG1 and RhCG2 ORFs, encode for polypeptides 490 (54.2 kDa) and 483 (53.3 kDa) amino acids in length, respectively. RhCG1 show 72–84.6% and 53.9–64.6% identity to known fish and mammalian RhCG sequences, respectively. Similarly, the identity between RhCG2 and other fish sequences is 71.2–81.0% and 51–61% with mammalian RhCGs (data not shown). RhCG1 has three potential N-glycosylation sites(⁶⁰NISS⁶³, 299NSTL³⁰²,479NFTV⁴⁸²), whereas RhCG2 contains two such sites(⁵⁷NLTD⁶⁰, 295NATL²⁹⁸)(Fig. 1B). The C-terminal sequences appear to be the least conserved among species and a class-specific conserved region is also found at the C-terminal of RhCG sequences (data not shown).

Hydropathy analysis revealed that all three Rh proteins are highly hydrophobic and negatively charged at physiological pH. They share similar topology that resembles bacterial Amt(Blakey et al., 2002) and mammalian Rh membrane channels (Liu et al., 2000; Liu et al.,2001), with cytoplasmic N- and C-terminals. K. marmoratusRhBG is predicted to have ten transmembrane domains, while both RhCG1 and RhCG2 have 12 predicted transmembrane domains(Fig. 2).

The phylogenetic tree reconstruction shows that the Rh proteins form two main clusters: Cluster I (RhCG) and II (RhBG), which are subdivided into smaller clusters (Fig. 3). The invertebrate Rh-like proteins (Ib) are divergent from the RhCG family (Ia). The vertebrate RhCG, on the other hand, diverge into a mammalian RhCG cluster and a cluster including other vertebrate RhCGs, which are then divided later into fish, amphibian and avian RhCG clusters. On the other hand, the RhBG cluster (cluster II), is also subdivided into two groups IIa and IIb. However,it is interesting to note that the fish RhBG (IIa) proteins apparently diverged earlier, forming one distinct cluster, whereas mammalian RhBGs, in this case, have closer relationships with the amphibian and avian homologues and diverged later than the fish RhBGs.

The whole-body ammonia level was not altered after 5 days of exposure to 1 mmol l^–1 NH₄HCO₃. However, after 5 days of 2 mmol l^–1 NH₄HCO₃ exposure,whole-body ammonia content increased significantly by 37%(Fig. 4).

Reverse-transcription PCR results show that RhBG expression in control K. marmoratus is consistently found in gill and skin, whereas RhCG1 and RhCG2 are predominantly expressed in gill. After 5 days of exposure to high external ammonia levels (2 mmol l^–1NH₄HCO₃), RhBG expression is induced in brain,eye, gonad, gut, kidney, liver and skeletal muscle and continues in gill and skin. RhCG1 expression is also induced in skin of ammonia-exposed fish, but RhCG2 expression remains restricted to the gill(Fig. 5).

Gill RhCG2 mRNA levels were not altered after 5 days of exposure to 1 mmol l^–1 NH₄HCO₃, but 2 mmol l^–1 NH₄HCO₃ exposure resulted in a 5.8-fold increase, relative to the control level. Gill RhBG and RhCG1 mRNA levels did not change in response to either ammonia level(Fig. 6A). In skin, RhCG1 mRNA levels increased significantly by 2.4-fold and 7.7-fold after 5 days of 1 and 2 mmol l^–1 of NH₄HCO₃ exposure, respectively, relative to control fish. As well, skin RhCG2 mRNA levels increased by 3.8-fold after exposure to 1 mmol l^–1 NH₄HCO₃, but there were no changes when fish were exposed to 2 mmol l^–1NH₄HCO₃ relative to control fish(Fig. 6B).

RhBG mRNA levels remained unchanged in the brain but were significantly higher in liver (3.4-fold) and muscle (7.2-fold) tissue in 2 mmol l^–1 NH₄HCO₃-exposed fish compared to control fish (Fig. 7).

Significant induction of RhCG1 (6.2-fold) and RhCG2(4.2-fold) mRNA levels were observed in killifish skin after 24 h of air exposure. The RhBG mRNA level was not altered(Fig. 8).

We have identified three Rh genes in killifish K. marmoratus: RhBG, RhCG1 and RhCG2, which encode transmembrane proteins with cytoplasmic N- and C-terminals. The amino acid sequences of these three Rh genes share high homology to known Rh sequences in the GenBank. Since all three Rh proteins contain potential N-glycosylation sites, they are probably expressed as glycoproteins, similar to their mammalian counterparts (Quentin et al., 2003). K. marmoratus RhBG shares higher sequence homology to three-spined stickleback (Gasterosteus aculeatus) (data not shown), which also has ten predicted transmembrane domains but is unlike zebrafish (Danio rerio) and pufferfish (T. rubripes), both of which have 12 predicted transmembrane domains (data not shown). K. marmoratus RhCG1 and RhCG2, on the other hand, have 12 transmembrane domains similar to their mammalian counterparts(Huang and Liu, 2001).

Our study represents the first report of the possible involvement of Rh gene/proteins in ammonia excretion in fish during adverse conditions such as high external ammonia and aerial exposure. All three Rh mRNAs (RhBG, RhCG1 and RhCG2) were expressed under control conditions in the gill, similar to the observation in pufferfish gills (Nakada et al., 2007),and RhBG is also expressed in the skin. After exposure to high environmental ammonia (2 mmol l^–1NH₄HCO₃, pH 8.1), RhBG was widely expressed in many tissues, and RhCG1 expression extended to the skin. Interestingly, RhBG induction is observed in almost all tissues screened (Figs 5 and 7) except gill and skin(Fig. 6). RhCG1 and RhCG2 appear to demonstrate tissue specificity during high ammonia exposure (Fig. 6), with RhCG1 mRNA levels dominating skin Rh expression and RhCG2 dominating gill Rh expression.

An earlier study (Frick and Wright,2002a) indicated that ammonia might be excreted against the gradients when K. marmoratus were exposed to high environmental ammonia (see Introduction). Indeed, in the present study, there was no elevation of body tissue ammonia levels after 5 days in 1 mmol l^–1 NH₄HCO₃ and only a small increase(37%) after the same period of time in 2 mmol l^–1NH₄HCO₃ (pH=8.1, NH₃≈100 μmol l^–1). This is surprising because in other fish, internal ammonia levels typically increase several-fold in response to external ammonia levels (Wilson and Taylor,1992; Wilson et al.,1994; Knoph and Thorud,1996; Kong et al.,1998; Steele et al.,2001; Anderson et al.,2002). For example, in the ammonia-tolerant marine toadfish Opsanus beta, muscle tissue ammonia concentrations were elevated by threefold after 4 days in water containing 3.5 mmol l^–1NH₄Cl (at pH 7.8, NH₃≈70 μmol l^–1) (Wang and Walsh,2000), a lower NH₃ level than in this study. In contrast, K. marmoratus is capable of preventing excessive accumulation of ammonia at high external ammonia and this appears to be similar to the mudskipper P. schlosseri(Randall et al., 1999), a species that is known to excrete ammonia actively during high environmental ammonia exposure.

Although the mode of ammonia excretion against high external ammonia in killifish has not been delineated, we hypothesize that Rh genes may be involved. All three Rh proteins are expressed in control pufferfish gills(Nakada et al., 2007). Constitutive expression of Rh proteins is probably required for routine removal of endogenous ammonia from nitrogen metabolism. There was no increase in any of the Rh mRNAs in the gills of K. marmoratus during 1 mmol l^–1 exposure, but the upregulation of RhCG1and RhCG2 in the skin (Fig. 6B) and constitutive expression of Rh proteins in the gill, may have been involved in facilitating ammonia excretion and thereby preventing an elevation of whole-body ammonia content(Fig. 4). Likewise, when external ammonia levels were increased to 2 mmol l^–1, the upregulation of RhCG2 mRNA levels in the gill and RhCG1 mRNA levels in the skin (Fig. 6B) is suggestive of a functional role of Rh genes in ammonia transport. It should be noted that transcriptional changes do not necessarily imply a corresponding upregulation of protein content, and verification of these ideas is necessary. Initial experiments using antibodies obtained from mammalian Rh proteins were unsuccessful and future work will require fish-specific antibodies.

In the mammalian kidney, polarized localization of the Rh proteins is observed in epithelial cells of specific regions, including the renal collecting segment and the collecting duct, with RhBG expression in the basolateral membrane and RhCG in the apical membrane(Quentin et al., 2003; Verlander et al., 2003). Such polarized localization is also present in the pufferfish. RhBG is localized basolaterally and RhCG2 proteins apically in the pavement cells of the gills,whereas RhCG1 proteins are found in the apical membrane of the mitochondrial-rich (MR) cells (Nakada et al., 2007). We predict a similar localization of the different Rh proteins in the killifish.

It has been suggested (Weiner,2006) that RhBG may be playing a role in ammonia-sensing or having roles other than ammonia-transport. This suggestion was based on the basolateral location of RhBG glycoproteins, together with the observation that expressions of RhBG mRNA and protein were not increased metabolic acidosis (which results in net excretion of acid and ammonia by the mammalian renal system) (Seshadri et al.,2006). Others have reported that genetic ablation of RhBGin the knockout mice did not affect ammonia transport in the kidney and tolerance to chronic acid-loading (Chambrey et al., 2005; Chambrey et al.,2006), although other genes may have been upregulated to compensate for the Rh knockdown. RhBG expressions are restricted to kidney, liver and gastrointestinal tract in the mammals (human and mouse) (Liu et al., 2001; Handlogten et al., 2005). In contrast, RhBG in K. marmoratus has a very broad tissue expression (Fig. 5). From the phylogenetic tree (Fig. 3), it is noted that RhBG is relatively more primitive and emerged earlier compared to RhCGs. Induction of RhBG was observed in a variety of tissues,whereas the expression of RhCGs was restricted to gill and skin during high ammonia exposure (Fig. 4). Fish normally experience much higher extracellular ammonia concentrations [0.1–1.3 mmol l^–1(Wood, 1993)] than mammals[0.03–0.1 mmol l^–1(Felipo and Butterworth,2002)]. Taken together, the data suggests that RhBG may have important functions in fish to facilitate ammonia transport in various internal organs.

The skin of K. marmarotus is an important site for NH₃volatilization during air exposure (Frick and Wright, 2002b; Litwiller et al., 2006). Following emersion, there was an 18-fold increase in the NH ⁺₄ concentration on the cutaneous surface,leading to a substantial elevation in the partial pressure of NH₃(Litwiller et al., 2006). The authors proposed that an active mechanism may be involved in moving ammonia across the skin surface when fish were emersed. Induction of ammonia/ammonium transporter genes (RhCG1 and RhCG2) in the skin during aerial exposure (Fig. 8),therefore, may be involved in the transport of ammonia across the epidermis for subsequent NH₃ volatilization, as well as in aquatic ammonia excretion. According to the model proposed by Khademi and co-workers(Khademi et al., 2004),ammonia approaches the Amt B channel as NH ⁺₄, then the proton is stripped away and ammonia is transported across the channel as NH₃ gas; it exits the channel as NH₃, then picks up a proton and forms NH ⁺₄. We propose that fish Rh proteins also conduct ammonia in the gaseous form. During air exposure, ammonia would leave the cutaneous surface of K. marmoratus in the form of NH₃, without being protonated and hence volatilized. While in water, RhBG and/or RhCG would facilitate NH₃ excretion across the gill and skin, which would be subsequently trapped by H⁺ to form NH ⁺₄. Acidification of the boundary water layer(Wright et al., 1989) in the cutaneous or gill surface of immersed K. marmoratus might be a possible mechanism to accelerate ammonia removal in fish, particularly in the presence of high external ammonia concentration.

The authors wish to thank Jess Vickruck for her assistance with experimental animals, as well as Meaghan Mitchell for her careful attention to fish husbandry. We also thank Allan Edelsparre for his help in fish dissections. This project was supported by NSERC Discovery Grants and C.F.I./O.I.T. awards to C.M.W. and P.A.W. C.M.W. is supported by the Canada Research Chair Programme.