Renal responses to salinity change in snakes with and without salt glands

The reptilian kidney has long been known to be incapable of eliciting urine hyperosmotic to the blood plasma (Dantzler, 1976) (but see Yokota et al., 1985); purely renal regulation of NaCl, therefore, is thought to be insufficient for maintaining ion balance in reptiles inhabiting desiccating (e.g. marine and desert) environments. Despite this, both marine and desert environments are rich in reptile diversity, suggesting that reptiles integrate renal and various extra-renal osmoregulatory systems effectively. Notably, many marine and desert species possess salt glands for enhanced excretion of excess ions. Interestingly, only a few studies of reptiles have attempted to analyze renal function in species with and without salt glands, and clear correlations between possession of salt glands and kidney structure and function in reptiles remain to be examined. Though reptile kidney physiology has been studied for many decades, a recent review of reptile renal function (Dantzler and Bradshaw, 2009) reveals just how much remains to be discovered in the field of renal ion regulation.

Although reptilian kidneys are incapable of excreting urine significantly hypertonic to the blood plasma, many species are capable of modifying urine composition in response to environmental pressures (e.g. water diuresis). The ability of the reptilian kidney to modify urine composition likely derives from at least three important features of this tissue: (1) the rate of filtration at the glomerulus, (2) the heterogeneity of cell types populating the various segments of the nephron and (3) variation in the number of functioning nephrons (Dantzler and Bradshaw, 2009). Much like the nephron of mammalian kidneys, the reptilian kidney is comprised of several segments: the proximal tubule is connected to the distal tubule through a short intermediate segment, and the distal tubule connects to the collecting duct through a connecting segment. Among snakes (and other squamates), the connecting segment is a sexually dimorphic structure (sometimes called the renal sex segment) thought to be involved in the seasonal production and/or modification of the seminal fluid (Cuellar et al., 1972). Results from studies of garter snakes (Thamnophis sirtalis) suggest that Na⁺ and Cl^– are reabsorbed in both the proximal and distal segments of the snake nephron, a process that is thought to require active Na⁺ and passive Cl^– transepithelial transport (Dantzler and Bradshaw, 2009). Among mammals, apical Na⁺ uptake is modulated by the apical Na⁺/H exchanger (NHE) in the proximal tubule, a combination of NHE and the absorptive isoform of the Na⁺/K⁺/2Cl^– cotransporter (NKCC2) in the loop of Henle, the Na⁺/Cl^– cotransporter (NCC) in the distal tubule, and the epithelial Na⁺ channel (ENaC) in the connecting tubules and collecting ducts. By contrast, the basal extrusion of Na⁺ is facilitated by Na⁺/K⁺-ATPase (NKA) in all segments of the mammalian kidney (Kinne and Zeidel, 2009). Although early studies of renal function in snakes provide indirect evidence that some of the same ion transporters may regulate NaCl balance in reptiles as well (Beyenbach and Dantzler, 1978; Dantzler et al., 1991), to our knowledge, no studies have directly examined the distribution and/or abundance of these ion transporters in the kidneys of any reptile species. Further, the relationship between the distribution and/or abundance of these transporters and environmental salinity has yet to be determined in any reptile species.

In concert with reabsorption of NaCl, modification of urine can be achieved through reabsorption of water from the filtrate. In many vertebrates, this process is stimulated primarily through the action of aquaporin (AQP) 1, in the proximal tubules, and AQP2, -3 and -4 in the distal tubules, connecting segments and collecting ducts (Borgnia et al., 1999). Both AQP2 and AQP3 are known to be hormonally regulated (via vasopressin) in mammals (Kinne and Zeidel, 2009; Terris et al., 1996), and a similar mechanism of regulation has been proposed for avian AQP2 (Lau et al., 2009) and amphibian AQP2 (Ogushi et al., 2007). Upon stimulation by vasopressin [or arginine vasotocin (AVT) in birds and amphibians], AQP2 is mobilized from the cytoplasmic vesicles, where it is stored, to the apical membrane of the collecting duct cells, facilitating luminal passage of water into the cell (Nielsen et al., 1995). Water then exits the cell via the basolaterally located AQP3 (Kinne and Zeidel, 2009; Sugiura et al., 2008). Although renal expression of AQP3 is restricted to the basolateral membranes of collecting duct cells in mammals and birds, its distribution among amphibians appears to extend into the distal tubules (Akabane et al., 2007; Mochida et al., 2008) and among fishes the localization of AQP3 remains equivocal (Cutler and Cramb, 2002). As reviewed by Dantzler (Dantzler, 1976), the permeability of the distal tubule to water is quite variable among reptiles (and can also vary considerably with hydration status within a given species). Though some evidence suggests that the basolateral membrane of the distal tubule in snakes may in fact be quite permeable to water (Beyenbach, 1984), the potential role of AQP3 in regulating basolateral renal water transport in any reptile has yet to be studied.

Animals with an extra-renal means for secreting a concentrated NaCl solution might be expected to drink saltwater and absorb NaCl across the gut, cloaca and nephron epithelia (even when experiencing high environmental salinity) because they can effectively excrete the salt and retain the water. Those animals without such means, however, might be expected to minimize drinking and salt reabsorption across these epithelia because they are unable to excrete the excess salt. Furthermore, within a species, regulation of the renal mechanisms for NaCl absorption may be expected to vary with the salinity of the environment; the ways in which renal water economy is affected by aquaporins is entirely unknown among reptiles. Thus, the objectives of this study were to examine changes in the structure and function of snake kidneys, including changes in the localization and abundance of NKA, NKCC(2) and AQP3, after acclimation to 0, 50 and 100% seawater (SW; 32 ppt). To determine whether the structural and/or functional responses of the kidneys were related to the presence of an extra-renal site for salt excretion, we compared one marine species with a salt gland (Laticauda semifasciata Reinwardt 1837) with one marine species without a salt gland (Nerodia clarkii clarkii Conant and Collins 1991). To determine whether kidney structure and/or function was related to habitat use, we compared two congeneric species lacking salt glands: one which inhabits marine environments (N. c. clarkii) and one which inhabits freshwater environments (Nerodia fasciata Linnaeus 1766).

Adult banded sea kraits (L. semifasciata; 497.4±121.2 g initial mass) were collected by hand from Orchid Island, Taiwan, and housed individually in plastic aquaria in 100% SW (32 ppt) prior to the beginning of the experiment. Aquarium water was mixed fresh daily using Instant Ocean (Spectrum Brands, Inc., Madison, WI, USA) and tap water from National Taiwan Normal University (Taipei, Taiwan) and changed daily. For the first 5 days in the lab, all animals were acclimated in 100% SW at room temperature (29.67±0.62°C). At the end of this 5 day period, control animals (N=6) were selected randomly and killed. The remaining animals were assigned to one of three treatments: 0, 50 or 100% SW (N=6, per treatment). To assess the response of kidney structure and/or function to these defined salinity treatments and to avoid the response to salinity shock (from direct transfer), we reduced the salinity of the cage water in small increments over a period of 7 days until animals reached their final treatment salinity. Animals from all three treatments were then held in their final salinities for 1 week before being killed by rapid decapitation, as outlined in the AVMA Guidelines on Euthanasia. Throughout the experiment, animals were fasted and maintained in enough water such that they could rest, submersed, on the bottom of the cage while still being able to reach the surface easily for respiration. In accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC), animals were blotted dry and weighed (±0.1 g) daily throughout the duration of the experiment, and cage water salinity was checked daily using either an Atago S/Mill refractometer (Tokyo, Japan) or a YSI 85 salinity meter (Yellow Springs, OH, USA). As in previous studies (Dunson, 1980; Pettus, 1963; Winne et al., 2001), rate of ‘dehydration’ was determined as the percent loss of initial body mass per day for each individual.

Adult salt marsh snakes (N. c. clarkii; 118.8±79.7 g) were collected from Seahorse Key, FL (Levy Co.; permit #05-012) and adult banded watersnakes (N. fasciata; 136.3±95.2 g) were collected from public roadways near Paynes Prairie, FL (Alachua Co.). Because N. fasciata was expected to be highly intolerant of 100% SW, we modified the design outlined above such that the control animals for both species of Nerodia were held in 0% SW (Gainesville, FL, tap water) for the laboratory acclimation period (room temperature: 23.23±0.65°C). As above, control animals (N=5, per species) were killed after the laboratory acclimation period, and the remaining animals were assigned to the indicated treatments (N=5 per treatment, per species).

Whole trunk blood was collected in unheparinized tubes within 1 min of decapitation and centrifuged immediately to separate serum. Hematocrit was estimated as the percent of the total volume in the tube composed of red blood cells. Serum was then removed to a clean unheparinized tube, snap frozen in liquid nitrogen and stored at –80°C prior to analysis. Total osmolality was measured on 10 μl triplicates of thawed serum using a Vapro 5520 vapor pressure osmometer (Wescor, Logan, UT, USA). Individual electrolytes were measured on 125 μl samples using a Stat Profile pHOx Plus C machine (Nova Biomedical, Waltham, MA, USA). The right kidney was also removed from each animal following decapitation and immediately fixed in 4% paraformaldehyde for 24 h at 4°C. Tissues were then washed three times (15 min each) in 10 mmol l^–1 phosphate buffered saline (PBS). Fixed and washed tissues were stored at room temperature in 75% ethanol before being embedded in paraffin wax, sectioned and mounted on charged slides as previously described (Babonis et al., 2009).

To examine the morphology of the kidneys (organization of tubules, distribution of blood vessels, etc.) we used the Lillie modification of Masson's trichrome technique (Humason, 1972). Additionally, because the secretion of acidic mucosubstances from the distal segments of the snake nephron is thought to protect the nephron epithelium from damage caused by the passage of colloids and/or organic osmolytes (More, 1977), we examined changes in the secretion of mucins and/or their precursors (glycogen) by pairing the Alcian Blue (AB) technique with a modified periodic acid-Schiff (PAS) technique. Sections stained with AB were counterstained by incubation for 30 s in Nuclear Fast Red (Humason, 1972) and those stained with PAS were counterstained in hematoxylin, as previously described (Babonis et al., 2009).

To localize NKA, NKCC and AQP3, we used the immunohistochemical techniques as described by Babonis et al. (Babonis et al., 2009). Briefly, after blocking endogenous peroxidases and non-specific proteins, we incubated tissue sections with anti-NKA (1/100), anti-NKCC (1/1000) or anti-AQP3 (1/500) overnight at 4°C. All antibodies were diluted in Protein Block (BioGenex, San Ramon, CA, USA). Sections were rinsed of primary antibody and prepared for visualization using BioGenex's Supersensitive Link-and-label universal secondary antibody kit with a DAB (3,3′-diaminobenzidine tetrahydrochloride) chromagen. Negative controls were produced by incubating sections in BioGenex Protein Block rather than primary antibody, and positive controls were produced via western blotting (for NKA and NKCC; see below) or by peptide preabsorption (AQP3). To preabsorb anti-AQP3, primary antibody was incubated in approximately 200-fold molar excess of peptide while shaking at 4°C overnight (Pandey et al., 2010). Enough BioGenex Protein Block was then added to bring anti-AQP3 to a final concentration of 1/500 before use. A minimum of three individuals per treatment were examined for each species.

Monoclonal anti-NKA (α5), developed by Dr Douglas Fambrough, and monoclonal anti-NKCC (T4), developed by Drs Christian Lytle and Bliss Forbush III, were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by The University of Iowa, Department of Biological Sciences (Iowa City, IA, USA). Although anti-NKA is directed against the α1 subunit of the NKA heterodimer (Takeyasu et al., 1988), anti-NKCC is directed against a conserved epitope in the carboxyl tail of NKCC1, NKCC2 and NCC (Lytle et al., 1995). Anti-AQP3 (Hc-3) and its blocking peptide were generous gifts from Dr David Goldstein at Wright State University [see Pandey et al. (Pandey et al., 2010) for epitope].

Frozen kidneys were homogenized on ice and prepared for electrophoresis using the methods previously described (Babonis et al., 2009). We then electrophoresed 25 μg of total protein in 10% Tris-HCl polyacrylamide Redi-gels (Bio-Rad, Hercules, CA, USA) before transferring proteins to polyvinylidine fluoride membranes for blot analysis (Bio-Rad). Dry blots were rehydrated in 100% methanol and rinsed in de-ionized (DI) water before blocking in a solution of 5% non-fat dry milk in Tris-buffered saline (TBS; 25 mmol l^–1 Tris, 150 mmol l^–1 NaCl; pH 7.4) for 2 h at room temperature while shaking. Following the blocking step, blots were incubated in anti-NKA or anti-NKCC overnight at room temperature while shaking. Primary antibody was removed in three washes with TTBS (TBS with 0.1% Tween-20; pH 7.4) before blots were incubated in alkaline-phosphatase-conjugated goat anti-mouse IgG (1/3000 diluted in blocking solution) while shaking for 1 h at room temperature. Secondary antibody was removed from the blots with three washes in TTBS before Immun-Star alkaline-phosphatase-conjugated chemiluminescent signal (Bio-Rad) was applied, following the manufacturer's protocol. Following analysis, each blot was stained with 0.02% Coomassie Blue (diluted in 50% methanol, 40% water and 10% acetic acid) to visualize total protein. All blot images were scanned and digitized for analysis and brightened using Photoshop CS3 (Adobe, San Jose, CA, USA).

We followed a previously published protocol (Choe et al., 2005) for our molecular techniques. In short, total RNA was extracted from tissues fixed in RNAlater® (Ambion, Austin, TX, USA) using Tri-Reagent (Sigma, St Louis, MO, USA), quantified and checked for purity using a micro-volume spectrophotometer (Nanodrop ND-1000, Thermo Scientific, Wilmington, DE, USA), and cDNA was synthesized from mRNA using Oligo-dT primers and the Superscript III reverse-transcription kit (Invitrogen, Carlsbad, CA, USA). Degenerate primers were designed to amplify NKA, NKCC2 and AQP3 in all three taxa using the CODEHOP online primer design software (Rose et al., 2003). Amplification was accomplished using standard PCR cycles for 0.5 μl Oligo-dT cDNA and Ex Taq™ Hot-Start DNA polymerase (Takara Bio, Madison, WI, USA) in an Express thermocycler (ThermoHybaid, Franklin, MA, USA). Amplicons were then transfected into PCR®-4 TOPO vectors and transformed into TOP10 chemically competent cells using the TOPO-TA cloning kit for sequencing (Invitrogen). Plasmids were sequenced in both directions by the marine DNA sequencing facility at the Mount Desert Island Biological Laboratory (Salisbury Cove, ME, USA) and the resulting species-specific sequences were used to design primers for all other applications (see Table 1 below). Quantitative real-time PCR (qRT-PCR) primers were designed to amplify an amplicon of ∼100–150 bp and tissue distribution primers amplified an amplicon of ∼450 bp using the Primer-3 Plus online primer design software (Untergasser et al., 2007). All specific sequences were deposited in GenBank (see Table 2 for accession numbers).

To examine changes in the abundance of NKA, NKCC2 and AQP3 mRNA across treatments, we performed qRT-PCR, as has previously been described (Choe et al., 2005). In brief, we loaded 24 μl triplicates of reaction mixture (1 μl of 1/10 diluted cDNA, 7.4 pmol specific primers and SYBR® Green Mastermix; Applied Biosystems, Foster City, CA, USA) into 96 well optical plates (BioRad) and PCR-amplified using an I-cycler IQ thermocycler (Bio-Rad) and the following cycling protocol: step 1, 95°C for 10 min (initial denaturing step); step 2, 95°C for 35 s, 60°C for 30 s and 72°C for 30 s (repeat for a total of 40 cycles); and step 3, melting curve analysis (to ensure amplification of only a single product). Each plate also contained 24 μl triplicates of a five-point dilution series, which was mixed fresh for each use from a mixed sample of species-specific undiluted cDNA. No-template control reactions, lacking cDNA, and negative control reactions, made with pre-reverse transcription RNA rather than cDNA, were amplified using the preceding procedure to ensure amplification was either absent or occurred >10 cycles later than the latest cycle of amplification for target DNA. To ensure specificity of amplified products, a random selection of samples from each plate were extracted, sequenced and identity-searched using BLAST (NCBI, Bethesda, MD, USA).

To determine the sequence of a full-length mRNA for AQP3, we amplified both the 5′ and 3′ ends of the AQP3 transcript, from L. semifasciata (lsAQP3), following the manufacturer's protocol for the GeneRacer kit (Invitrogen). Specific RACE primers were designed using Primer-3 Plus.

Nucleotide comparisons were made using the coding sequence only for lsAQP3 and the blastn algorithm. Predicted amino acid sequences were compared using the tblastn algorithm. For comparisons of Laticauda and Anolis, we used NCBI's bl2seq function with the blastn (nucleotide) or blastp (amino acid) algorithm. Accession numbers: Anolis carolinensis (ENSACAT00000012739), Gallus gallus (XM_424500.2), Homo sapiens (NM_004925.3) and Hyla chrysoscelis (DQ364245.1).

To examine the distribution and relative abundance of AQP3 across snake tissues, we extracted RNA as described above from the brain, duodenum, esophagus, Harderian gland, kidney, liver, lung, muscle (skeletal), pancreas, salt gland, stomach and testis of L. semifasciata. cDNA was then reverse transcribed from total RNA using random hexamer primers (SSIII kit, Invitrogen). Specific primers were designed to amplify a 450 bp amplicon of AQP3 and duplexing PCR was then performed by amplifying cDNA in the presence of both AQP3-specific primers and control primers (Quantum RNA™ 18S internal standard primer kit; Ambion). To ensure accurate representations of relative cDNA abundance, reactions were terminated in the exponential phase of the PCR protocol. Consistency in 18S amplification across tissues indicates low variability in cDNA quality and quantity across tissues. To visualize amplicons, PCR products were electrophoresed at 60 V in a 2% agarose gel, stained with ethidium bromide and digitized using the Gel Doc™ XR system (Bio-Rad). Negative control reactions were prepared with RNA rather than cDNA for each tissue.

Mean rates of daily mass loss, serum electrolytes (including total osmolality), hematocrit and mRNA expression values were compared among species and treatments using ANOVA with Tukey's honestly significant difference (HSD) post hoc test. For qRT-PCR analysis, cycle threshold values were compared at the arbitrary threshold position of 100 using the MyIQ Optical System software version 1.0 (BioRad). Expression values for samples loaded onto each plate were adjusted to the standard curve run on the same plate and log-transformed to homogenize variance. Transformed expression values were then normalized to the expression value for the reference gene, ribosomal protein L8, chosen because it was invariant across treatments (L.S.B., unpublished). These normalized gene expression values were then standardized to the control treatment for each species (thus, mRNA expression values for the control treatment will always appear as 1.0). Error estimates were calculated from the log-transformed data and rescaled to the standardized mean. All analyses were performed in the R statistical environment (R Development Core Team, 2008).

There was no effect of treatment on rate of mass loss in any of the three species examined (Table 3). Furthermore, mass loss in N. c. clarkii (the marine watersnake) did not differ from mass loss in N. fasciata (the freshwater watersnake) in any treatment, though total mass loss (calculated as a percentage of initial body mass) was considerably more variable (see standard deviations in Table 3) in these two species in all treatments than in L. semifasciata in any treatment. Both N. c. clarkii and N. fasciata lost more mass per day in 100% SW than did L. semifasciata (the sea snake) in 0% SW, but perhaps more interestingly, N. fasciata lost mass at a greater rate in 0% SW than did L. semifasciata in all treatments. Because rates of mass loss were not different in the freshwater and saltwater treatments for any species, we consider these rates to reflect merely the effect of fasting rather than dehydration.

Survival differed among treatments for N. fasciata only. Although L. semifasciata and N. c. clarkii were found to have 100% survival in all three treatments, among N. fasciata survival decreased from 100% in 0% SW to 80% in 50% SW and 60% in 100% SW.

For each species, treatment means were compared with the mean value for the species-specific control group to determine whether there was an effect of salinity on serum. Total osmolality was not affected by treatment in L. semifasciata. For N. c. clarkii, total osmolality decreased in the 0% SW group (P=0.026, relative to control) and in N. fasciata it increased in the 50% SW group (P=0.014, relative to control) (Fig. 1A). Conversely, L. semifasciata experienced a decrease in both Na⁺ (P=0.028) and K⁺ (P=0.019) in the 0% SW treatment, whereas Na⁺ and K⁺ levels in N. c. clarkii and N. fasciata did not differ significantly among treatments (Fig. 1B,C). Cl^– levels were not affected by treatment in L. semifasciata or N. c. clarkii but exhibited an increase in both the 50% SW (P=0.003) and the 100% SW (P=0.041) treatments in N. fasciata (Fig. 1D). Though hematocrit levels (not measured in L. semifasciata) did not differ among treatments for either N. c. clarkii or N. fasciata (Fig. 2), the values we obtained for both N. c. clarkii and N. fasciata through this experiment were similar to published values for these species and their marine and freshwater congeners (Dunson, 1980; Pough, 1979).

Using histology, we examined the cell types populating each segment of the snake nephron (Fig. 3). Originating at the glomerulus, the neck segment is characterized by low cuboidal cells with very little cytoplasm. Filtrate passes through the neck segment to the proximal tubule, which is characterized by relatively large cells with abundant cytoplasm and basally or centrally positioned nuclei surrounding a lumen that can vary in size from essentially collapsed (Pc in Fig. 3A) to quite open (Po in Fig. 3B). Following the proximal tubule is the relatively short intermediate segment, typified by low cuboidal cells with little cytoplasm and basally or centrally positioned nuclei organized around a relatively small lumen (Fig. 3A). The distal tubule follows the intermediate segment and is comprised of two sub-segments. The early distal tubule (De) is comprised of cells that are intermediate in size between those of the intermediate segment and those of the proximal tubule, and has nuclei that are positioned basally and often appear flattened against the basal membrane (Fig. 3A). Cells in this sub-segment appear to have more cytoplasm than those of the intermediate segment, which, combined with the basal nuclei, makes cells in this sub-segment appear empty. Additionally, cells in the early distal tubule are more rounded at the apical margin than cells from other segments. By contrast, the cells of the late distal tubule (Dl) are much more regular in shape and have basally or centrally positioned round nuclei and less cytoplasm than cells of the proximal tubule (Fig. 3B). The diameter of the distal tubule (both sub-segments) is always smaller than that of the proximal tubules and slightly larger than that of the intermediate segments.

The connecting tubule (hereafter referred to as the renal sex segment) is sexually dimorphic, appearing engorged with secretory granules in the males (so much so that the lumen is often not visible; Fig. 3B,C). In females, this segment often resembles that of the proximal tubule but with flatter and more basally positioned nuclei. This segment is also relatively short in females, being confined only to the outer margins of the kidney (near the collecting ducts). Though many hypotheses about the function of the renal sex segment have been proposed [Cuellar et al. (Cuellar et al., 1972) and references therein], this segment does not appear to contribute to water or ion balance and will not be considered further in this study. At the distal end of the renal sex segment is a short (often <20 cells in length) segment connecting the renal sex segment to the primary collecting duct (Fig. 3C). This connecting segment is similar in appearance to the primary collecting duct (columnar cells with basal flattened nuclei) and often can be distinguished from primary collecting ducts only by their smaller diameter (Fig. 3D). Primary collecting ducts from each nephron eventually join with other primary collecting ducts to become secondary collecting ducts (Fig. 3D), which merge to form the ureter (Fig. 3E,F). The primary and secondary collecting ducts and the ureter are distinguishable only by the diameter of their lumena.

There was no effect of salinity on the secretion of mucus or glycogen in the kidneys of any of the three species examined (Fig. 4). In L. semifasciata, AB⁺ material was detected only at the apical margin of the cells comprising the distal tubules (early and late), the connecting segments and the collecting ducts (Fig. 4A–E). This pattern was consistent in both N. c. clarkii (Fig. 4F–J) and N. fasciata (Fig. 4K–O); however, unlike L. semifasciata, the distribution of AB⁺ material in the early distal tubules of the Nerodia was diffuse, extending all the way through the basal cytoplasm of the cell. Similarly, we detected PAS⁺ material in the apical margins of the distal tubules, connecting segments and collecting ducts of all three species yet no effect of treatment in any of them (Fig. 5). The basement membranes of the nephrons (especially around the renal sex segment) and the apical membrane of the proximal tubule were also PAS⁺ in all three species.

NKA localized to the basolateral membranes of the distal tubules (early and late), the connecting segments and the collecting ducts of all three species (Fig. 6). Early distal tubules often exhibited strong staining in the basal membrane and only faint staining of the lateral membranes whereas late distal tubules exhibited prominent staining in both basal and lateral membranes (compare De and Dl in Fig. 6J). There was no effect of treatment on the localization of NKA in any of the three species examined (Fig. 6C–E,H–J,M–O). NKCC was undetectable in the kidney from any of the three species studied (Fig. 7). Putative AQP3 localized to the basolateral membranes of the connecting segments and collecting ducts of all three species (Fig. 8B,G,L); interestingly, this protein was also detected in the apical membrane and subapical cytoplasm of the late distal tubules in L. semifasciata (Fig. 8A) but was absent from these tubules in N. c. clarkii (Fig. 8F) and N. fasciata (Fig. 8K). The localization of putative AQP3 did not vary with treatment in any of the three species examined (Fig. 8C–E,H–J,M–O).

The specificity of anti-NKA (α5) and anti-NKCC (T4) for their target proteins has already been verified via western blotting for L. semifasciata (Babonis et al., 2009). Here, we further demonstrate specificity of α5 for a protein of approximately 110 kDa in both N. c. clarkii and N. fasciata (Fig. 9). Interestingly, when kidney homogenates were probed with T4, no proteins were detected in any of the three species (data not shown). Whether this suggests an affinity of T4 for NKCC1 (which is far less abundant than the absorptive isoform, NKCC2, in the vertebrate kidney) (Russell, 2000) or extremely low abundance of NKCC1, NKCC2 and NCC in the kidney of these three species cannot be evaluated at this time. Though we were unable to confirm specificity of the AQP3 antibody (Hc-3) through western blot analysis, peptide pre-absorption of Hc-3 completely abolished staining from the distal tubules of L. semifasciata (Fig. 10A) and the connecting segments and collecting ducts of all three species (Fig. 10B–D). Because Hc-3 is a heterologous antibody, we cannot exclude the possibility that the positive immunoreaction we observed is a result of non-specific interaction with another protein (perhaps another AQP). Further studies are required to confirm the localization of putative AQP3 in the distal nephron of snakes.

We found only minor effects of treatment on mRNA expression (Fig. 11). mRNA expression for NKA was variable but not statistically different among treatments in any of the three species examined. In L. semifasciata, AQP3 abundance was approximately twice as high in the 0% SW treatment as in the 50 or 100% SW treatments (P=0.013; Fig. 11A). There was no effect of salinity on AQP3 in either N. c. clarkii or N. fasciata. NKCC2 expression was higher in the 50% SW treatment than in the 0% SW treatment for N. c. clarkii only (P=0.048; Fig. 11B). There was no effect of salinity on mRNA expression for NKA, NKCC2 or AQP3 in N. fasciata (Fig. 11C).

We sequenced the full-length mRNA transcript from L. semifasciata and compared both the predicted amino acid and nucleotide sequences of lsAQP3 with that of H. chrysoscelis (the species against which the Hc-3 antibody was made) as well as A. carolinensis, G. gallus and H. sapiens (Table 4). When comparing amino acid sequences, lsAQP3 shared the highest percent identity with AQP3 from A. carolinensis (85% identical); percent identities between lsAQP3 and AQP3 from G. gallus, H. chrysoscelis and H. sapiens were all very similar (81–82%). Nucleotide sequences showed slightly lower congruence but lsAQP3 still shared the highest similarity with AQP3 from A. carolinensis. Importantly, lsAQP3 was determined to have both asparagine–proline–alanine (NPA) motifs (a defining characteristic of aquaporins) as well as the conserved lysine (D) residue ∼50 amino acids upstream of the second NPA motif (Fig. 12), confirming that it is a member of the glycerol transporter subgroup of aquaporins (Zardoya and Villalba, 2001).

lsAQP3 was detected in all tissues examined, though expression was notably low in brain, liver and pancreas (Fig. 13A). Intermediate expression was detected in the kidney, skeletal muscle and testis, and relatively high expression was detected in the duodenum, esophagus, Harderian gland, lung, salt gland and stomach. No amplification occurred in the negative control (Fig. 13B).

The ability of some species of snake to inhabit marine environments without the aid of a specialized salt gland has been well documented; however, the mechanisms by which these animals carry out water and ion regulation are unknown. Both of the marine species studied herein (L. semifasciata, the sea snake, and N. c. clarkii, the marine watersnake) appear to have been largely unaffected by changes in environmental salinity. Survival was 100% in all treatments for each of these species and three metrics of plasma ion homeostasis (total osmolality, K⁺ and Cl^–) were robust to increases in salinity in these species as well (Fig. 1). Together, these observations suggests that either the acclimation period was not long enough to induce salinity stress and elicit an osmoregulatory response in the two marine species, or these animals utilized some mechanism to keep plasma ion levels low while experiencing increases in environmental salinity. By contrast, survival among N. fasciata (the freshwater watersnake) decreased with an increase in salinity and both total osmolality and Cl^– ion concentration increased with salinity. Though total osmolality in 100% SW was not significantly different from the 0% SW treatment, this likely reflects the reduction in sample size due to death of two of the animals in this treatment. Because Cl^– ion concentration was also significantly elevated in both the 50 and 100% SW groups and there was a trend toward increased Na⁺ in these groups as well (though these groups were not statistically different from the 0% treatment), the increase in total osmolality in N. fasciata seems to be a result of inadequate NaCl regulation. These results suggest that further investigation into the specific mechanisms by which N. c. clarkii maintains low serum Na⁺ and Cl^– concentrations may reveal the functional differences underlying variation in salinity tolerance among marine and freshwater watersnakes.

To determine whether elevated Na⁺ and Cl^– levels in N. fasciata from the 50 and 100% SW groups were a result of increased ion intake or a reduction in plasma water (indicating dehydration), we examined both rates of mass loss and changes in hematocrit across treatments. Mean rates of mass loss in L. semifasciata and N. fasciata were similar to those previously reported for these species (Dunson, 1978; Lillywhite et al., 2009), whereas estimates in N. c. clarkii were slightly higher than previously reported (Dunson, 1980; Pettus, 1963). Among-individual variation in mean daily mass loss was very high in both species of Nerodia, likely reflecting the large overall range in body mass in these two species, and mass loss did not differ among treatments in any of the species examined. Furthermore, hematocrit neither differed across treatments in either species nor differed between N. c. clarkii and N. fasciata in any treatment (Fig. 2). Because rate of mass loss did not differ with the salinity of the environment, these results are likely a result of fasting during the experiment rather than dehydration. It is important to note that a previous study of salinity acclimation in marine snakes has suggested that these animals cannot maintain water balance while fasting (Dunson and Robinson, 1976). Because some amount of both water and salt are likely taken up orally (indeed, in all species examined – marine, estuarine and freshwater – the majority of Na⁺ influx occurs orally) (Dunson and Robinson, 1976), the osmoregulatory stress experienced by wild animals may not be easily extrapolated from these results. Future tests of the effects of salinity on water and ion balance in these and other species should make explicit comparisons of fasted and fed animals to determine, for example, the effect of access to prey on survival times of N. fasciata in saltwater.

Early studies of water and ion balance in reptiles suggest that the transport properties of the integument may vary with habitat type (Dunson and Robinson, 1976; Stokes and Dunson, 1982). Specifically, N. fasciata and other freshwater species are known to experience greater Na⁺ influx than efflux and greater water efflux than influx but this pattern is reversed in N. c. clarkii and other marine species (Dunson, 1978). Thus, the increases in serum osmolality among N. fasciata may also be explained by the relatively greater influx of ions across the skin. It was surprising, however, to find that rates of mass loss observed in this study were not higher in N. fasciata than either N. c. clarkii or L. semifasciata. One possible explanation for the lack of differences among rates of mass loss is that the freshwater watersnakes used in this study in fact did undergo relatively higher rates of mass loss [similar to those observed by Dunson (Dunson, 1978)] but that this loss was balanced by water intake via drinking. If true, this scenario could also explain the relative increases in the concentration of Na⁺ and Cl^– ions in the blood of N. fasciata without dramatic reductions in body mass.

Among fishes, kidney morphology has been shown to vary with habitat use, extreme examples of which include nephrons that completely lack glomeruli, proximal tubules or distal tubules [see references in Evans and Claiborne (Evans and Claiborne, 2009)]. Although limited examples of aglomerular nephrons have been reported among squamates as well (Dantzler and Bradshaw, 2009), we found no evidence for a correlation between nephron morphology and habitat use among the species examined in this study. In fact, the only morphological variation evident among individuals in these experiments was the previously described sexual dimorphism in the renal sex segment. These observations suggest that differences in the osmoregulatory ability of the kidney across species are likely reflected in the physiology (and perhaps the microanatomy) of the kidney, rather than in the gross differences seen among the Osteichthyes.

As a first indication of the effect of salinity on kidney function, we examined the kidneys of all three species for changes in the production or secretion of mucosubstances. We found no evidence for increased secretion of mucus (no change in the expression domain of AB⁺ material across treatments in any of the species; Fig. 4) and no evidence for the upregulation of the muco-synthesis pathway (no change in the domain of expression of PAS⁺ material across treatments; Fig. 5). These results suggest that either these animals do not secrete increased amounts of mucus in concert with increased osmolyte excretion or that these animals did not undergo increases in osmolyte excretion coincident with these treatments. It is interesting to note, however, the general observation that AB⁺ material appeared to be more abundant in the distal tubules of N. fasciata from all treatments than in L. semifasciata from any treatment. Functional studies of changes in the organic components of urine composition as well as studies aimed at understanding the integration of renal and post-renal mechanisms of urine modification in species from marine and freshwater environments would be very interesting in this context.

Using a combination of histology and immunohistochemistry, we demonstrate a basolateral localization of NKA in the distal tubules, connecting segments and collecting ducts of all three species (Fig. 6). A previous study demonstrating ouabain-inhibited Na⁺ reabsorption in the distal tubules of snake kidneys (Beyenbach and Dantzler, 1978) supports this finding. Though we were unable to detect NKA in the proximal tubule using immunohistochemistry, previous studies of proximal tubule membrane transport in snakes suggest that NKA may be present in the basolateral membrane of this tubule as well (Benyajati and Dantzler, 1988; Dantzler, 1972). Given that Na⁺ reabsorption from filtrate is likely facilitated in the proximal tubule by the presence of an apical Na⁺/H⁺ exchanger (Dantzler et al., 1991), and that K⁺-activated p-nitrophenylphosphatase (the enzyme responsible for the dephosphorylation of NKA) has been localized to the basolateral membrane of this tubule (Benyajati and Dantzler, 1988), our results likely reflect low abundance of NKA in the proximal tubules of these species (relative to other renal tubules in these species) rather than absence. Interestingly, and in contrast with what is known about Na⁺ transport in snake distal tubules, ouabain has been shown to be ineffective at inhibiting fluid reabsorption in the proximal tubule (Dantzler and Bentley, 1978), suggesting that Na⁺-independent fluid absorption may be occurring in this tubule as well. Finally, the localization of NKA did not change with treatment in any of the three species examined but this was not surprising given that a previous study of the effects of diuresis on renal function in watersnakes has suggested that changes in total solute secretion are a result of changes in the number of functioning nephrons rather than changes in nephron membrane physiology (Lebrie and Sutherland, 1962).

Early studies of AVT-induced anti-diuresis in watersnakes demonstrated increased tubular reabsorption of Na⁺ and K⁺ upon stimulation by AVT (Dantzler, 1967a), a process that, in mammals, is mediated by apical NKCC2 in the proximal tubules (Giménez and Forbush, 2003). Further research (Beyenbach and Dantzler, 1978) identified a transepithelial K⁺ flux in the distal tubule of Thamnophis (the garter snake) that was inhibited by ethacrynic acid (a known NKCC2 inhibitor). Considering that NKCC2 is also expressed in the apical membranes of the proximal tubules in birds (Nishimura and Fan, 2002) and in the apical membranes of the distal tubules in both fishes (Katoh et al., 2008) and amphibians (Guggino et al., 1988), it was surprising that NKCC was not detectable in the kidneys of any individuals in this experiment (Fig. 7). Because anti-NKCC (antibody T4) has been demonstrated previously to react with the NKCC1 isoform in the salt gland of L. semifasciata (Babonis et al., 2009), it is unlikely that these negative results represent a methodological anomaly. Thus, studies aimed at assessing the effects of furosemide (another known inhibitor of NKCC2) on ion reabsorption as well as those aimed at understanding the distribution of putative AVT receptors in reptilian kidneys would be very informative in unraveling the mechanisms by which K⁺ ion reabsorption is regulated in reptiles.

Using immunohistochemistry, we demonstrate positive immunoreaction for putative AQP3 in the basolateral membranes of the cells comprising the connecting segments and collecting ducts of all three species examined herein (Fig. 8). This localization is consistent with the localization of this protein in other vertebrate taxa and supports earlier observations of transepithelial water flux in the distal portion the snake nephron (Beyenbach, 1984; Dantzler, 1967b). Importantly, we were unable to confirm antibody specificity via western blot for anti-AQP3; thus, future studies aimed at expressing putative snake AQP3 in oocytes to assess functionality and confirm our immunolocalization results are required to positively identify the protein we call putative AQP3. Further, we demonstrate an apparent apical or subapical localization of putative AQP3 in the cells comprising the late distal tubule epithelium in L. semifasciata (Fig. 8A). This apical localization of putative AQP3 in the distal tubule of sea snakes was surprising but is also reminiscent of the localization of AQP3 to the apical membranes of chloride cells in the gills of silver eels (Lignot et al., 2002) and the localization of AQP2 in the apical membranes and subapical vesicles of the collecting duct in mammals (Kinne and Zeidel, 2009). Interestingly, peptide preabsorption of Hc-3 completely abolishes all staining in the kidney of all three species examined (Fig. 10), suggesting that the immunoreactivity was a specific result of the interaction of antibody Hc-3 with its antigen. Additionally, a BLAST search for the Hc-3 antibody sequence returns an overwhelming number of records for other vertebrate AQP3 homologs and only two other vertebrate proteins: a recombination activating gene in the bull shark (Carcharhinus leucas, AAB17267.1) and a protein of unknown function in the zebrafish (Danio rerio, XP_002667244.1). Lastly, no significant similarity is found when the Hc-3 antibody sequence is BLASTed against AQP2 from A. carolinensis (ENSACAP00000008275), G. gallus (ENSGALP00000016674) or H. sapiens (NP_000477.1). Despite all of these controls, it is possible that the heterologous antibody Hc-3 cross-reacted with another protein (e.g. putative snake AQP2) in the kidney of these three species of snakes; thus, additional studies aimed at characterizing all of the kidney AQPs in snakes combined with expression studies to confirm their function and localization are necessary to resolve the identity of both the basolateral and apical forms of putative AQP3 in snakes. Finally, because there was no effect of treatment on the localization of putative AQP3 in any of the three species examined, we suggest that modification to the function of this protein, when it occurs, likely involves post-translational modification rather than changes to the position of the protein within the cell.

Animals without an extra-renal means to excrete excess salts (i.e. those species lacking salt glands) may be expected to alter the composition of the urine (inasmuch as reptiles have this capacity) to excrete a maximally concentrated urine when experiencing dehydration. Thus, the kidneys of N. c. clarkii and N. fasciata (the two species used in this study that lack salt glands) were expected to exhibit decreases in NKA and NKCC2 abundance (to minimize reabsorption of NaCl) and increases in AQP3 abundance (to facilitate reabsorption of water) under these conditions. Despite these predictions, but in support of our findings that the localization of NKA, NKCC and putative AQP3 did not differ across treatments, we found only minor differences in the mRNA expression across treatments. NKA expression values were variable but not statistically different among treatments for all three species. Because changes in the function of NKA are often associated with post-translational modification (e.g. phosphorylation and dephosphorylation) (Bertorello et al., 1991), our finding that transcription of this molecule did not decrease significantly does not rule out the possibility that the activity of this enzyme changed with treatment. Furthermore, because NKA is likely found in the basolateral membranes of all parts of the nephron (which vary in function), the relationship between the abundance of this ion transporter and environmental salinity may vary along the length of the nephron. Further investigations into the activity and abundance of NKA should examine isolated segments of the nephron to resolve these issues.

NKCC2 mRNA, though also highly variable, was found to be significantly higher for N. c. clarkii in 50% SW than in 0% SW. At the present time, we cannot determine whether NKCC2 is transcribed in the kidney but not translated or whether the abundance of the protein is simply too low to detect (via immunohistochemistry or western blot) in this tissue. Furthermore, the high variability in NKCC2 expression within a treatment for these species suggests that more data are necessary to interpret the potential role of NKCC2 in regulating renal ion balance among snakes.

In L. semifasciata, we found significantly higher expression of AQP3 in the freshwater treatment; whether this unexpected pattern is coincident with the apical localization of putativeAQP3 protein or whether it reflects a role in facilitating the production of dilute urine in low salinity environments cannot be determined at this time. A previous study of AQP3 expression in the kidneys of chicken demonstrated increases in AQP3 only with water deprivation, not with salt loading (Sugiura et al., 2008). Because we demonstrated no difference in the dehydration rate across species and, further, no difference in hematocrit between N. c. clarkii and N. fasciata, it is not surprising that we detected no difference in the expression of AQP3 across treatments in N. c. clarkii or N. fasciata; however, the large overall variation in mRNA expression seen among individuals in the same treatment suggests that the response of snake kidneys to salinity acclimation may be very complex.

We found that the predicted amino acid sequence for lsAQP3 shares a high percent identity with AQP3 from A. carolinensis, G. gallus, H. chrysoscelis and H. sapiens (Table 4), lending support to the hypothesis that lsAQP3 is, in fact, an ortholog of AQP3 from these other taxa. Importantly, lsAQP3 exhibits the two NPA motifs characteristic of all aquaporins as well as two aspartic acid (D) residues at positions 163 and 219 (Fig. 12), characteristics of the aquaglyceroporin subgroup (Borgnia et al., 1999). Although we cannot confirm that the protein detected by the Hc-3 antibody was the same protein encoded by the lsAQP3 mRNA that was extracted from these kidneys, we think these results warrant further investigations into the localization and potential functions of the putative distal tubule form of AQP3 identified from a marine snake.

Similar to the results of AQP3 distribution studies in fishes, amphibians and mammals (Mobasheri et al., 2005; Pandey et al., 2010; Tipsmark et al., 2010), lsAQP3 mRNA was detected in a wide range of tissues (Fig. 13). Interestingly, those tissues with the highest relative expression are those that are composed of mucous epithelia, supporting the hypothesis that AQP3 may play a role in the water transport associated with epithelial mucus secretion (Lignot et al., 2002). Interestingly, lsAQP3 was also expressed in both the Harderian gland (a cephalic seromucous gland) and the salt gland (a specialized serous gland). Given its potential role in facilitating production of mucus, the expression of lsAQP3 in the Harderian gland is not surprising. Although mucus secretion is an unlikely role for AQP3 in the salt gland, our evidence of AQP3 in this tissue combined with recent evidence of AQP3 from the rectal gland of the dogfish (Cutler, 2007) and evidence of AQP1 and AQP5 from the salt glands of marine birds (Muller et al., 2006) suggests that much remains to be learned about the concerted roles of the various AQP isoforms in vertebrate salt gland physiology.

In summary, though we largely found no renal effects of salinity acclimation in any of the three species examined in this study, we were able to contribute to a general understanding of the regulation of water and ion balance at various locations along the snake nephron. Our results regarding the distribution of NKA and putative AQP3 in the snake nephron combined with previous studies of Thamnophis (the garter snake) are summarized in Fig. 14. Because neither the localization nor the abundance of NKA, NKCC2 or AQP3 changed with treatment, our results are consistent with the hypothesis that changes in kidney function are not the result of changes in the physiology of the functioning nephrons but are, potentially, a result of changes in the number of functioning nephrons [as suggested by Lebrie and Sutherland (Lebrie and Sutherland, 1962)]. Further studies of long-term salinity acclimation and variation in the number of functioning nephrons are required to fully evaluate this hypothesis. Interestingly, we found no differences in the structure or function of the kidneys when comparing N. c. clarkii with N. fasciata (sister species that use very different habitats). The mechanism by which N. c. clarkii is able to regulate osmotic and ionic balance in the marine environment, therefore, remains elusive.

 
• AB⁺

  Alcian Blue positive

 
• AQP

  aquaporin

 
• AVT

  arginine vasotocin

 
• cDNA

  complimentary DNA

 
• Cl^–

  chloride ion

 
• ENaC

  epithelial Na⁺ channel

 
• K⁺

  potassium ion

 
• lsAQP3

  Laticauda semifasciata ortholog of AQP3

 
• mRNA

  messenger RNA

 
• Na⁺

  sodium ion

 
• NaCl

  sodium chloride

 
• NCC

  Na⁺/Cl^– cotransporter

 
• NHE

  Na⁺/H exchanger

 
• NKA

  Na⁺/K⁺-ATPase

 
• NKCC2

  Na⁺/K⁺/2Cl^– cotransporter

 
• NPA

  asparagine–proline–alanine motif

 
• PAS⁺

  periodic acid Schiff positive

 
• qRT-PCR

  quantitative real-time PCR

 
• RACE PCR

  rapid amplification of cDNA ends PCR

 
• SW

  seawater

We thank David Butcher, Chelsey Campbell, David Hall, Kelly Hyndman, Ryan McCleary, Brandon Moore, Ming-Chung Tu and Molly Womack for their help with animal collections and/or sample preparation. We also thank Charles A. Wingo and his laboratory for the use of their vapor pressure osmometer and pHOx machine.