claudin-10 isoform expression and cation selectivity change with salinity in salt-secreting epithelia of Fundulus heteroclitus

Teleost fish gill and opercular epithelia, containing high densities of ionocytes, are used as models to study the plasticity of ion transporters and transport-related proteins in vertebrates. The mummichog (Fundulus heteroclitus L. 1766) is an important genomic and physiological model from which to reveal the molecular and cellular mechanisms underlying gill function (Burnett et al., 2007; Whitehead et al., 2011; Cozzi et al., 2015). Mummichogs readily inhabit salinities ranging from freshwater (FW) to strongly hypersaline conditions for extended periods of time (Griffith, 1974). For several euryhaline teleosts, transfer from FW to seawater (SW) increases the number of ionocytes and accessory cells and enhances expression of ion secretion transporters such as Na⁺/K⁺/2Cl⁻ cotransporter (NKCC) and cystic fibrosis transmembrane conductance regulator (CFTR) (Marshall et al., 1999; Wilson et al., 2004; Marshall and Grosell, 2006). In euryhaline Mozambique tilapia (Oreochromis mossambicus), transfer from SW to hypersaline SW (2SW) increases the number and size of ionocytes (Kültz et al., 1995). In mummichogs, exposure to 2SW produces ultrastructural changes to the apical crypt of ionocytes, wherein the accessory cells hypertrophy and contact the ionocyte at multiple punctate loci, thus interrupting the immunofluorescence of CFTR to produce a bead-like distribution (Cozzi et al., 2015). Thus increases in environmental salinity are associated with intercellular junctions (i.e. tight junctions, TJs) becoming specialized for cation secretion, providing an opportunity to identify the composition of these porous junctions.

The paradigm of salt secretion by vertebrate epithelia is the two-step transcellular transport of Cl⁻ involving NKCC1 at the basolateral side of the ionocyte that facilitates the accumulation of Cl⁻ intracellularly and the anion channel CFTR at the apical membrane where the ionocyte typically develops a cup-shaped apical crypt (Evans et al., 2005; Marshall and Grosell, 2006; Edwards and Marshall, 2013). Sodium ions are instead translocated via a paracellular pathway between cells through single-stranded, cation-permeable intercellular TJs known to form exclusively between accessory cells and ionocytes (Sardet et al., 1979; Karnaky, 1991). Because of the elaboration of mummichog accessory cell–ionocyte TJs in 2SW conditions (Cozzi et al., 2015), it can be hypothesized that acclimation to 2SW will require an increased contribution of TJ proteins that create cation permeability of the paracellular pathway between ionocytes and accessory cells.

Claudin (Cldn) proteins are incorporated into intercellular TJs of vertebrate epithelia where they play a dominant role in the control of tight junction permselectivity; this is true in teleost fish epithelia (Chasiotis et al., 2012) and in tetrapods (Krug et al., 2012, 2014). Some of these epithelial intercellular junctions are non-selective ‘leaky’ junctions, whereas others are selective to certain molecular species (e.g. cations or anions) and constitute cation- or anion-permeable ‘pores’ (Shen et al., 2011). Because the gill epithelium changes its transport characteristics from ion uptake in FW to ion secretion in SW, with concomitant changes in paracellular permeability, euryhaline species have been used to study the functional dynamics of cldn/Cldn responses to these changing transport demands (for recent reviews, see Chasiotis et al., 2012; Kolosov et al., 2013). Emerging from studies conducted thus far is the idea that select Cldn-10 TJ protein isoforms play a central role in facilitating the switch of branchial epithelia to salt secretion in a SW environment. There are several lines of evidence to support this notion: (1) Cldn-10 isoforms are often found to exhibit a limited distribution pattern in fish organs, being expressed primarily in osmoregulatory organs, and in some cases, principally in the gill and skin (i.e. organs that interface directly with the surrounding environment and that become salt secreting in SW) (Loh et al., 2004; Tipsmark et al., 2008; Bui et al., 2010; Bui and Kelly, 2014; Kolosov et al., 2014); (2) within the gill epithelium of euryhaline fishes, select cldn-10/Cldn-10 isoforms exhibit cell-specific distribution patterns, being present either exclusively or enriched in gill epithelium cells (i.e. ionocytes) that are involved in salt secretion in SW (Bui et al., 2010; Bui and Kelly, 2014; Kolosov et al., 2014); (3) select Cldn-10 isoforms in gill ionocytes exhibit alterations in subcellular distribution when a euryhaline fish is acclimated to SW versus FW (Bui and Kelly, 2014); (4) select Cldn-10 isoforms in the gill exhibit increased mRNA and/or protein abundance in euryhaline fishes under conditions where the gill epithelium becomes salt secreting (i.e. following transfer from FW to SW) (Tipsmark et al., 2008; Bui et al., 2010; Bui and Kelly, 2014, 2015; Bossus et al., 2015); and (5) salmonid cldn-10e responded to endocrine factors that direct the functional re-organization of the gill epithelium following salinity change (Tipsmark et al., 2009; Trubitt et al., 2015). Despite this growing evidence that select Cldn-10 isoforms underlie branchial salt secretion, no study has exclusively focused on elucidating a role for Cldn-10 isoforms in the gill of a euryhaline fish, combining the molecular physiology of Cldn-10 salinity regulation with measured cation selectivity across branchial tissue in SW and hypersaline SW. Taking into consideration the well-established response of mummichog branchial and opercular epithelia to SW (Karnaky, 1991; Zadunaisky, 1996; Hoffmann et al., 2002) and recent morphological investigation into mummichog branchial ion transport in hypersaline SW (Cozzi et al., 2015), the objective of this study was to use the mummichog model as a means to provide new insight into the role of claudins in cation-selective ion movement across vertebrate epithelia.

Adult mummichogs of both sexes were collected from an estuary in Avery Point, Connecticut, USA by seine net and transported to the Skidmore College Animal Care Facility. Fish were maintained in FW (5.31 mmol l⁻¹ Na⁺, 5.25 mmol l⁻¹ Cl⁻, 0.10 mmol l⁻¹ Ca²⁺) in recirculating stock tanks with particle and charcoal filtration and continuous aeration at 24–27°C under 12 h light:12 h dark. Fish were fed Omega One mini pellets (Omega Sea, Painesville, OH, USA) twice daily. At the time of transfer (0 h), fish were quickly netted and distributed into six recirculating 38-liter tanks (N=8 fish/tank); three tanks contained FW and three tanks contained artificial SW (35‰ Instant Ocean, Blacksburg, VA, USA). Time 0 fish were sampled directly from the stock tanks. Fish were fasted for the duration of the transfer experiment. At the time of sampling, fish (N=8) from one FW and one SW tank were netted and anaesthetized with a lethal dose of 2-phenoxyethanol (2 ml l⁻¹). Fish were sampled at 0, 1, 3 and 7 days after transfer. Fish were rapidly decapitated and filaments from the branchial arches were stored in TRIzol Reagent (Ambion, Carlsbad, CA, USA) at −80°C until tissue homogenization and RNA isolation. White muscle was sampled from the caudal musculature and the water content was measured gravimetrically after drying overnight at 90°C. The Institutional Animal Care and Use Committee (IACUC) of Skidmore College approved all housing and experimental protocols (IACUC protocol number 130).

Adult mummichogs of both sexes were trapped in Jimtown estuary, Antigonish, Nova Scotia, and transported to the St Francis Xavier University Animal Care Facility. The fish were placed in full strength SW (35‰) in 450-liter recirculating tanks at room temperature (20±1°C), under 12 h light:12 h dark photoperiod under artificial lighting and were held for several weeks prior to experimentation. Salinity was monitored daily using a YSI Pro2030 conductivity meter (YSI Inc., Yellow Springs, OH, USA); part changes were performed at the rate of one-third of aquarium volume (same salinity and temperature) per 48 h period. Thirty fish were acclimated to 2SW by flow-through salinity change to 45‰ water for 48 h, then to 60‰ for 10 days; SW was made hypersaline by addition of artificial sea salt (Instant Ocean). A second group of 30 control animals was transferred from SW to SW in similar aquaria. Fish were fed Nutrafin flakes (R.C. Hagen, Montreal, Quebec, Canada) twice daily to provide each fish with 1.0 g of food daily per 100 g of body weight. Fish were also fed meal-worms (Tenebrio molitor) 3 days a week. Ten fish were sampled on days 1, 3 and 10 for 2SW and on day 10 for the SW controls. Fish were anaesthetized in 1.0 g l⁻¹ tricaine methane sulfonate (buffered to neutral pH) and were killed by pithing. The tail region was blotted dry and blood was collected from razor-severed caudal vessels into heparinized capillary tubes, centrifuged immediately at 600 g for 2 min and plasma frozen in microcentrifuge tubes for osmolality in duplicate by vapor pressure osmometry (Vapro model 5520; Wescor, Logan, UT, USA). All animal handling and procedures were approved by Canadian Council on Animal Care guidelines, approval protocol 16-003-R1 from St Francis Xavier Animal Care Committee.

A sample (0.5–1.0 g wet mass) of body skeletal muscle with skin removed was dissected from each fish, weighed, then dried to constant mass and expressed as wet mass minus dry mass divided by wet mass. This procedure provides a measure of body hydration level, which we predict should decrease in high osmolality environments if the animal is not fully acclimated to the new environment.

Gill tissue (approximately 25 mg) was placed frozen in chilled (0°C) TRIzol reagent, homogenized (5×3 min; Bullet Blender 24, NextAdvance, Averill Park, NY, USA), and total RNA isolated according to the manufacturer’s instructions. RNA was resuspended in 50 μl RNase-free water (Sigma, St Louis, MO, USA) and stored at −80°C until use. RNA samples were screened for quality (electrophoresis at 120 V for 45 min) in 1.5% agarose in TAE (Tris base, acetic acid and EDTA) buffer with 10 μl SybrSafe stain in 100 ml agarose; and quantified in duplicate spectrophotometrically (A260/A280 and A260/A230, Nanodrop 2000 spectrophotometer, Thermo Fisher Scientific, Waltham, MA, USA). Genomic DNA was removed by on-column DNase treatment (RNeasy minispin column, Qiagen, Toronto, ON, Canada) according to the manufacturer's instructions.

A high-capacity cDNA reverse transcriptase kit (Applied Biosystems, Warrington, UK) was used in duplicate, according to the manufacturer's instructions to convert mRNA to cDNA using random primers, reverse transcriptase (5 U μl⁻¹ final concentration) and with added RNase inhibitor (0.25 μl per reaction well of 20 μl: 10 μl master mix with 2 μg mRNA sample in 10 μl, total reaction volume 20 μl); thermocycler (model PTC-20, MJResearch, Mississauga, ON, Canada) programme: 25°C for 10 min, 37°C for 120 min, 85°C for 5 min, and 4°C hold.

Primers were selected (Table 1) from output of primer design software (Primer Express 3.0, Thermofisher, Burlington, ON, Canada), tested with mummichog gill cDNA standard (a combined cDNA sample composite) to ensure the amplification of a single product of appropriate size prior to qPCR analysis. qPCR used 1.0 μg cDNA of unknowns and a combined cDNA sample composite as standard curve (2.0 μg and fivefold dilutions from 1.0 μg to 0.008 μg cDNA) in duplicate in a 96 clear well qPCR plate, total volume 20 μl [SYBR Green Master Mix, 10 μl; forward and reverse primers (100 µmol l⁻¹), 0.5 μl each; Sigma water, 4 μl; and sample, 5 μl], using a model CFX-96 RealTime System (Bio-Rad, Hercules, CA, USA) and the thermocycler programme: 95°C for 10 min, 40 cycles of 95°C for 30 s, 60°C for 60 s, read every cycle, 65°C for 5 s to 95°C for 5 s melt curve, read every 0.5°C. The quantification (Cq) and back-calculated starting quantity (Sq) were calculated using CFX Manager 3.1 software (Bio-Rad, Hercules, CA, USA). For each unknown sample, Sq was divided by the 18S RNA Sq, averaged with the duplicate and statistics were performed on the raw Sq ratios. 18S RNA was used as a reference because it was unchanged by acclimation to SW or to 2SW. For figure presentation, the Sq ratios were presented relative to the control group (FW–FW transfer control in the FW–SW test, and SW–SW transfer control in the SW–2SW test), yielding the fold change in mRNA abundance relative to control.

Dissected opercular epithelia were pinned flat to a wax square and immersion fixed for 1 h in a solution of 2% formaldehyde and 2% glutaraldehyde in 0.1 mol l⁻¹ phosphate buffer at pH 7.4. The tissue was transferred to a glass vial, containing the same fixative, and stored overnight at 4°C. The fixative was rinsed by two changes of phosphate buffer followed by one change of 0.1 mol l⁻¹ sodium cacodylate (10 min per change). This was followed by immersion in a solution of 1% osmium tetroxide and 0.0015% Ruthenium Red in distilled water for 1 h. Ruthenium Red (molecular mass, 740 Da) bears six positive charges and stains the cell surfaces, but does not cross the plasma membrane to the cell interior (Chambers, 1973). The tissue was then rinsed with distilled water (2×10 min per change) and kept at 4°C for 4 days, after which it was stained en bloc in saturated aqueous uranyl acetate for 1 h then dehydrated through an ascending series of ethanol. The ethanol was replaced by two changes of propylene oxide (10 min each) and the tissue was then slowly infiltrated with epoxy resin (EMBed-812, EMS, Hatfield, PA, USA), placed in flat rectangular wells, and cured at 60°C for 2 days. Thick sections (1 μm) were cut on the transverse plane (i.e. parallel to the apical face of the opercular epithelium), and stained with Toluidine Blue until tissue was encountered. The blocks were then thin-sectioned (100 nm) on a Reichert microtome and the sections stained with uranyl acetate (1 h) and lead citrate (4 min) for viewing on a Philips 410 transmission electron microscope. Digital images were captured using a SIA L12C digital camera and entire images were adjusted for contrast and brightness using Adobe Photoshop.

The FW to SW transfer experiment was analysed by two-way ANOVA. Significant effects of treatment, time, or an interaction (P<0.05) are indicated in the figures: *P<0.05, **P<0.01, ***P<0.001 and ****P<0.0001. When significant main or interaction effects were detected, Student's t-tests were employed at each time point. Significant differences between groups at a given time point are also indicated in figures: ^†P<0.05, ^††P<0.01, ^†††P<0.001 and ^††††P<0.0001. Analysis of plasma, muscle, relative permeability and qPCR data used one-way or two-way ANOVA (as appropriate) followed by Tukey’s a posteriori tests. Transmission electron microscopy incidence of accessory cell contact with apical crypt formed ordinal data that were analysed using chi-squared two sample (control SW versus 2SW, 10 days) with control scores as the expected values. Differences were considered significant when P<0.05 in a two-tailed test. All statistical analyses were performed using GraphPad Prism 6 (GraphPad Software, La Jolla, CA, USA).

There was a significant interaction of treatment and time on muscle moisture content; a modest increase in muscle moisture content from controls occurred at 1 day after transfer to SW (Fig. 1).

Whereas there was no change in mRNA abundance of cldn-10c and cldn-10f (Fig. 2A,D), transfer to SW produced significant elevation in mRNA abundance of cldn-10d at 3 and 7 days (P<0.05 and 0.01, respectively) after transfer (Fig. 2B) and of cldn-10e at 1, 3 and 7 days after transfer (P<0.01, 0.05 and 0.05, respectively) (Fig. 2C). As expected, there were parallel significant increases in cftr mRNA abundance at 3 and 7 days post-transfer (P<0.0001 and 0.001, respectively, Fig. 2E).

Muscle moisture content (Fig. 3A) was significantly decreased 3 days after transfer to 2SW conditions, but was restored to control levels by 10 days post-transfer. Hematocrit was significantly elevated at 1 day by 31% (P<0.01) and at 3 days by 27% (P<0.05) by 2SW treatment, compared with SW controls (Fig. 3B). Plasma osmolality was significantly increased by 21.3 mosmol kg⁻¹ (5.9%, P<0.01) at 3 days but not at other times (Fig. 3C).

Transfer to 2SW produced significant elevation in mRNA abundance of cldn-10c and cldn-10f at 3 and 10 days (P<0.01 for both) after transfer (Fig. 4A,D) and of cldn-10e at 1 and 3 days post-transfer (P<0.05 and 0.01, respectively, Fig. 4C), whilst there was no change in mRNA abundance of cldn-10d (Fig. 4B). As expected, there were parallel significant increases in cftr mRNA abundance at 3 and 7 days post-transfer (P<0.01 and 0.001, respectively, Fig. 4E). The fold increases for cldn-10c, cldn-10f and cftr mRNA abundance were similar to each other.

Transmission electron microscopy (Fig. 5) of sections through apical crypts of SW- and 2SW-acclimated fish (N=5 SW fish, 66 crypt sections; N=5 2SW fish, 60 crypt sections scored), using sections normal to the plane of the epithelial surface, revealed an increase in the number of apical crypts with multiple (two or more) accessory cell contacts from 17/66=26% to 28/60=47% (chi-square, 13.016; d.f.=1; P<0.001), demonstrating approximately twice as many multiple contact apical crypts. The representative figure has two single-stranded junctions (arrowheads) in SW (Fig. 5A), and in 2SW (Fig. 5B) the accessory cell has apparently added a second cellular process and two more single-stranded junctions. The average number of accessory cell processes increased from 0.994 in SW to 1.256 per apical crypt section in 2SW; the large number of zeros in the data preclude parametric analysis of these data.

Estimates of relative permeability for the pore in SW- and 2SW-acclimated fish were made using two techniques: indirectly by measuring voltage and directly by measuring conductance changes (Fig. 6). Based on voltage responses to mucosal SW mixtures (Fig. 6A), there was strong selectivity against Li⁺ and lesser permeability in order of smaller to larger unhydrated ion size (Na⁺>K⁺>Rb⁺>Cs⁺). Two-way ANOVA revealed strong row (ion type) significance (P<0.0001), no significant column (salinity acclimation) result (P>0.05) and a significant interaction term (P<0.001). Multiple comparisons were significant at P<0.0001 for Li⁺ versus K⁺, K⁺ versus Rb⁺ and K⁺ versus Cs⁺; P<0.05 for Li⁺ versus Cs⁺; and P>0.05 (not significant) for Li⁺ versus Rb⁺ and Rb⁺ versus Cs⁺. Overall, the sequence approximates Eisenman VII: Na⁺>K⁺>Rb⁺∼Cs⁺>Li⁺. Also from voltage responses, the selectivity in 2SW epithelia was less than that of SW-acclimated epithelia for K⁺ (P<0.05), Rb⁺ (P<0.01) and Cs⁺ (P<0.01). Finally, cation:anion selectivity was measured with Na⁺ SW on the mucosal side: there was a selectivity of Na⁺:Cl⁻ of 4.65±0.604, N=28, for SW-acclimated fish and 3.49±0.703, N=20, for 2SW fish (P=0.221, unpaired t-test), demonstrating a strong cationic selectivity of the native tissue, even with the Cl⁻ secretory transcellular pathway operating.

From conductance estimates (Fig. 6B), the relative conductances were similar to the voltage results in terms of the sequence of selectivity, but this method did not detect the stronger selectivity against Li⁺, neither did the conductance method detect any clear differences between pore selectivity of opercular epithelia from SW- and 2SW-acclimated fish.

In the current set of experiments, mummichogs were able to acclimate from FW to SW, or SW to 2SW, and exhibit either unperturbed or modest (transient) alterations in systemic end-points of salt and water balance (i.e. muscle moisture content, plasma osmolality and hematocrit), indicating that experimental animals had acclimated to SW and 2SW conditions as expected. Also consistent with the idea that fish acclimated ‘normally’ to SW and 2SW conditions, cftr mRNA abundance increased in SW versus FW, and 2SW versus SW. Taken together, these observations are in line with previous studies using mummichogs to examine mechanisms of transcellular ion transport in SW conditions (Degnan and Zadunaisky, 1980; Marshall et al., 1999; Wood and Marshall, 1994) and verify that any observation made of TJ complex molecular physiology in the present work occurred under ‘typical’ systemic conditions.

A first observation from the current research is confirmation that increased cldn-10 transcript abundance is associated with the development of NaCl secretion by ionocytes in the gills and opercular epithelium of euryhaline teleost fishes acclimating from FW to SW. This idea can be extended to the mummichog because following transfer from FW to SW, (1) increased cldn-10e mRNA abundance occurred in the first 24 h and remained significantly elevated for up to 7 days, and (2) cldn-10d mRNA abundance increased progressively over the 7-day experimental period. Increased transcript abundance of these specific cldn-10 isoforms in the gill of the SW-acclimated mummichog is consistent with the response of the euryhaline puffer fish (Tetraodon nigroviridis) (Bui et al., 2010; Bui and Kelly, 2014) and Japanese medaka (Oryzias latipes) (Bossus et al., 2015). More specifically, both T. nigroviridis and O. latipes exhibited elevated cldn-10d and cldn-10e mRNA abundance in the gills of SW- versus FW-acclimated animals (Bui et al., 2010; Bui and Kelly, 2014; Bossus et al., 2015). In addition, transcript abundance of cldn-10e has also been reported to increase in the gill of SW-acclimated Atlantic salmon (Tipsmark et al., 2008). Therefore, a conserved role for cldn-10d and cldn-10e in SW acclimation is emerging.

In contrast, no alterations in the mRNA abundance of cldn-10c and cldn-10f were observed in mummichogs transferred from FW to SW. This result underscores two possibilities worthy of future consideration. The first is the idea of isoform-specific roles for Cldn-10s in FW to SW acclimation of mummichogs (i.e. a role for Cldn-10d and Cldn-10e but not Cldn-10c and Cldn-10f). The second possibility is that the roles for specific Cldn-10 isoforms in FW to SW acclimation may differ between teleost species. This is because the absence of any response from gill cldn-10c and cldn-10f mRNA in the current work is in contrast to previous observations that both cldn-10c and cldn-10f mRNA abundance increased in the gill of SW- versus FW-acclimated O. latipes (Bossus et al., 2015, 2017). In a cell line from rainbow trout (Oncorhynchus mykiss) gills, expression of cldn-10e was stimulated by prolactin, growth hormone (GH) and cortisol, but most strongly by GH (Trubitt et al., 2015); such a response is appropriate because GH is strongly associated with SW acclimation in salmonid fish (McCormick, 2001).

In addition to extending knowledge on putative roles for cldn-10 isoforms in euryhaline fishes, molecular analysis of branchial tissue taken from mummichogs transferred from SW to 2SW resulted in some truly novel findings. The first of these is that transcript abundance of cldn-10d was not significantly elevated in the gills of fish transferred from SW to 2SW, suggesting that any role for this cldn-10 isoform in salt secretion had been maximized following transfer from FW to SW. This was also somewhat apparent for cldn-10e which, by day 10 in 2SW, was not significantly elevated in 2SW versus SW gills. In contrast, both cldn-10c and cldn-10f mRNA abundance elevated substantially following SW to 2SW transfer. Therefore it would appear that recruitment of particular cldn-10 isoforms for the purposes of salt secretion across branchial epithelia of the mummichog might be determined by the degree of environmental salinity beyond that of SW. To our knowledge, this phenomenon has not been reported before, but it would be an important adaptation for an organism that can experience hypersaline conditions under natural circumstances. Future studies will provide insight into this unique finding, and should incorporate observations of isoform localization and protein abundance.

The increase in the number of single-stranded junctions observed between ionocytes and accessory cells in 2SW confirms previous findings (Cozzi et al., 2015) and suggests that accessory cells may extend cellular processes to increase the number of contact points by the accessory cell in the apical crypt. The increase in single-stranded junctions supports the notion that more of the appropriate claudin protein would be needed, especially in hypersaline conditions.

Most previous studies have used site-directed mutagenesis and expression of claudin proteins in model epithelial systems to compare pore characteristics; here, we leveraged natural variation of the isoforms, presumably cldn-10f in 2SW versus cldn-10c in SW, and compared pore selectivity in native opercular epithelia from SW- versus 2SW-acclimated fish. The observed series for the mummichog opercular epithelium is Eisenman series VII: Na⁺>K⁺>Rb⁺>Cs⁺>Li⁺, using the voltage-based response and as estimated by comparing conductance. When comparing SW- and 2SW-acclimated fish, there was a tendency for the 2SW-acclimated fish to exhibit a stronger selectivity against K⁺ (P<0.05, t-test, two-tailed), a lesser selectivity against Rb⁺ (P<0.01, t-test, two-tailed) and Cs⁺ (P<0.005, t-test, two-tailed), and no difference in the strong selectivity against Li⁺ (P=0.8027, t-test, two-tailed), indicating clear differences in the pore selectivity in the two conditions, as one would expect if different claudins comprised the cation-selective pores in the two salinities. The interaction term from the two-way ANOVA (P<0.001) emphasizes that whereas the SW and 2SW pores were similar with respect to Li⁺, they were different with respect to K⁺, Rb⁺ and Cs⁺. An interpretation of the interaction is that whereas the large hydrated Li⁺ ion is equally excluded from both pore types (i.e. the outer vestibule restriction is equal), the central part of the pore is shaped differently (i.e. more selective for Na⁺ against K⁺ in 2SW conditions). Also of interest is the protracted time to steady state for the Li⁺ SW exposure, which was between 1 and 2 h, an interval sufficient for the morphology of the TJ to change, as has been observed previously (Karnaky, 1991). Thus the high selectivity against Li⁺ could have been in part a tissue-level reaction. It is unclear at present whether the addition of phenylalanine residues in Cldn-10e and Cldn-10f in 2SW (Figs 7, 8) based on mRNA abundance (Fig. 4) are responsible for the apparent shift in cation selectivity in fish acclimated to 2SW (Fig. 6).

In Madin-Derby canine kidney (MDCK) epithelial cells overexpressing Cldn-10b, the ion selectivity of the epithelium changes from Eisenman sequence IV to sequence X (Na⁺>Li⁺>K⁺>Rb⁺>Cs⁺), suggesting a high field strength in the pore that removes most of the hydration shell of permeant ions (Günzel et al., 2009) and sequence X has a smaller predicted pore radius than sequence VII (Eisenman, 1962). Our results suggest that Li⁺ is excluded because of its large hydrated ion radius (3.4 versus 2.76 Å for sodium), consistent with the estimate of a pore radius of 3.25 Å for Cldn-10 pores (Laghaei et al., 2016). The rest of the series approximates the order of the dehydrated ion size (Na⁺>K⁺>Rb⁺>Cs⁺), suggesting that some or all of the water of hydration, the solvation shells, are stripped away as the ion passes through the pore. In a detailed comparison between Cldn-2 and Cldn-10b pores expressed in MDCK cells, Cldn-10b wild type had Eisenman sequence VII and high Na⁺ selectivity (Li et al., 2013). These authors discovered that hydrophobic residues near the pore help confer cation selectivity.

The dehydration of permeant ions (above) predicts that water is excluded from passing easily through the pore. This low water permeability is consistent with the observation that there is low osmotic permeability across the gills of SW-acclimated fish (Evans, 1969; Isaia, 1984), with the observation that the ions and water pass through the same pathway in the Cldn-10 pore (Laghaei et al., 2016) and with the observation that the mouse Cldn-10b pore (similar to Fundulus Cldn-10c pore-forming region) is impermeant to water (Milatz and Breiderhoff, 2017).

The transepithelial potential at the ionocytes in SW is approximately +37 to +40 mV (Guggino, 1980; Cozzi et al., 2015), sufficient to drive K⁺ down its electrochemical gradient and out across cation-permeable pores in the gills. Marine teleost fish actively secrete K⁺ (Marshall, 1981; Sanders and Kirschner, 1983). Recently, renal outer medullary potassium (ROMK) type channels in fish (Romk) have been identified in the apical membrane of ionocytes of a SW-acclimated euryhaline fish (Mozambique tilapia), and Romk is up-regulated in tilapia by high environmental K⁺ (Furukawa et al., 2012). Although a transcellular K⁺ secretion route exists, it is not known if some K⁺ secretion could also occur via the paracellular pathway. The lower K⁺ permeability of the putative Cldn-10 pore in 2SW fish (see above) could be an adaptation to possible high external K⁺ in hypersaline conditions that could reverse the flow of K⁺ to the detriment of the fish.

Many claudins have symmetrical motifs in an area of the first extracellular domain that are stabilized by a disulfide bridge between cysteine residues 53 and 63, e.g. CLDN-2 (Li et al., 2013). To gain some insight into the mummichog Cldn-10 pore structure, hypothetical anti-parallel alignments in this pore-forming zone are presented (Fig. 7). If the core D65 residues of CLDN-10 are mutated (D65N), Na⁺ permeation decreases and Na⁺ occupancy in turn blocks water permeation on a mole for mole basis, suggesting that water and Na⁺ permeate via the same pathway (Laghaei et al., 2016). The anion-selective pore-forming CLDN-17 has a critical positively charged residue in the same locus, K65, which when mutated, destroys the anion selectivity of the pore (Conrad et al., 2016). Modelling of CLDN-15 cation-selective pore formation predicts that an anti-parallel arrangement of beta pleated sheet extracellular loops is the most stable conformation (Suzuki et al., 2015) and the outer portion (the 12 residues distal to the disulfide linkage) is most closely juxtaposed in this conformation. In comparing the Fundulus cldn-10 isoforms, there is considerable agreement with this anti-parallel prediction. Considering only the zone of extracellular loop one near the disulfide linkage, parallel alignment of residues (Fig. 8A,B) naturally aligns similar charges, which would produce repulsion and no pore formation, whereas anti-parallel alignment of Cldn-10c, Cldn-10d, Cldn-10e and Cldn-10f produces two, three or four ionic bonds near the half cysteine residues (Fig. 8C–F), thus with the potential to form ionic attractions that would stabilize the pore. The central portion of the putative pore formed by all combinations has two unpaired acidic residues at D56 with (Cldn-10e and Cldn-10f) or without (Cldn-10c and Cldn-10d) nearby hydrophobic phenylalanine residues. The simplicity of the putative pore, involving only one set of negatively charged residues, is consistent with voltage-clamp studies performed previously with ionocyte-containing jaw skin of the long-jaw goby, Gillichthys mirabilis (Pequeux et al., 1988), where Na⁺ unidirectional fluxes behaved linearly as a single barrier to ion diffusion (as opposed to single file diffusion through serial binding sites, which produces complex flux dynamics). The extra phenylalanine residues in Fundulus Cldn-10e is an interesting addition, as increasing salinity from FW to SW, or again from SW to 2SW, transiently increased mRNA abundance of this isoform and Cldn-10e is the only pore combination to predict two hydrophobic bonds between the two peptides. The presence of the single phenylalanine in the pore region of Cldn-10f could be important in maintaining cation selectivity in hypersaline conditions. Meanwhile Cldn-10c, which lacks aromatic residues in the pore region, follows the expression pattern of cldn-10f in acclimation to 2SW, but the Cldn-10c–Cldn-10f hybrid pore has three stabilizing bonds and one unpaired phenylalanine. It is logical that the ionic strength of increased salt content of 2SW could weaken ionic bonds in tight junctions, and thus the increased numbers of ionic bonds (in Cldn-10f) and the addition of hydrophobic interactions (as in Cldn-10e) could help maintain pore structural integrity at high salinity.

Many thanks to StFX animal care facility for expert animal care, and to Alex Young and Dave Metzger for technical support.