Occludin immunolocalization and protein expression in goldfish

Hydromineral balance in freshwater (FW) fishes is regulated by strategies of ion retention and ion acquisition across ionoregulatory epithelia. While ion retention mechanisms limit outwardly directed solute movement across epithelia, ion acquisition compensates for obligatory losses to the environment (e.g. ion loss to FW due to epithelial permeability). Studies examining the role of ionoregulatory epithelia in the maintenance of hydromineral status in fishes have generally focused on transcellular mechanisms/routes of ion movement, through which actively driven ion transport takes place. In contrast, far less emphasis has been placed on mechanisms that enhance ion retention, i.e. those that limit paracellular ion loss. As a result, while transcellular pathways are well characterized and are known to incorporate a multifaceted suite of pumps, exchangers and channels that associate with either apical or basolateral cell membranes(Loretz, 1995; Marshall, 2002; Evans et al., 2005), the paracellular pathway, which is controlled by the tight junction (TJ) complex,remains poorly understood in aquatic vertebrates.

TJs are composed of transmembrane and cytosolic protein complexes that form strands around the apical domain of an epithelial cell. TJ proteins of adjacent epithelial cells associate with one another to form a semi-permeable paracellular seal that restricts solute movement between cells(Cereijido and Anderson, 2001). In addition to regulating paracellular permeability and limiting solute movement, TJs also demarcate apicobasal polarity and establish cell–cell contacts, which aid in the regulation of cellular processes such as transcription and proliferation (reviewed by Schneeberger and Lynch,2004). To date, over 40 TJ and TJ-associated proteins have been identified at the epithelial TJ complex(González-Mariscal et al.,2003). Isolated from chick liver, occludin was the first transmembrane TJ protein identified(Furuse et al., 1993). The presence of this tetraspan protein within TJ fibrils(Fujimoto, 1995) and its capacity to form TJ-like strands when transfected into cells lacking TJs(Furuse et al., 1998) quickly underscored a structural role for occludin within TJ complexes. Several other lines of evidence, however, also suggested a vital functional role for occludin in TJ sealing and the regulation of solute movement through the paracellular pathway. For example, the over-expression of chick occludin in Madin–Darby canine kidney (MDCK) epithelial cells led to an increase in transepithelial resistance (TER) (Balda et al., 1996; McCarthy et al.,1996). In contrast, treatment of Xenopus A6 epithelial cells with a synthetic peptide designed to disrupt occludin associations between adjacent cells led to a significant decrease in TER and an increase in permeability to paracellular markers (e.g. [³H]mannitol,[¹⁴C]inulin and dextrans) (Wong and Gumbiner, 1997). Moreover, microinjection of mRNA encoding C-terminally truncated occludin into Xenopus embryos resulted in TJs exhibiting a `leaky' phenotype (Chen et al., 1997). This `leaky' phenotype could be rescued by co-injection with full-length occludin mRNA(Chen et al., 1997).

While various TJ forms (e.g. `leaky' versus `tight') are widely accepted to play critical roles in the way fish epithelia function (for reviews, see Loretz, 1995; Marshall, 2002; Evans et al., 2005), little is known about TJ proteins in fishes. Given its role in regulating TJ barrier function and its use as an indicator of changes in paracellular permeability,we conducted studies on the integral TJ protein occludin in goldfish to investigate a potential role for occludin in the maintenance of the hydromineral balance of fishes. Using Western blot analysis and immunohistochemistry, we first examined occludin protein expression and localization in select goldfish epithelial tissues (i.e. gill, intestine and kidney). We then hypothesized that occludin protein expression would alter in response to hydromineral imbalance. To test this hypothesis we subjected goldfish to varying periods of food deprivation, working on the assumption that the restricted nutritional state would disrupt normal, energy-dependent mechanisms of ion acquisition in a time-dependent manner. Using endpoint measurements of hydromineral status and Na⁺,K⁺-ATPase activity, we characterized the response of goldfish to negative energy status in conjunction with measurements of occludin protein expression.

Goldfish Carassius auratus L. were obtained from a local supplier and held at 18–19°C under a simulated natural photoperiod (12 h light:12 h dark) in aerated 200l opaque polyethylene tanks at a density of 10–12 fish per tank. Each tank was supplied with flow through dechlorinated FW (approximate composition in mmol l^–1:[Na⁺] 0.59, [Cl^–] 0.92, [Ca²⁺] 0.76 and[K⁺] 0.43) at a rate of ∼250–300 ml min^–1. Fish were held for at least 4 weeks prior to experimentation and during this period were fed ad libitum once daily with commercial koi and goldfish pellets (Martin Profishent, Elmira, ON,Canada).

Fish were randomly selected and anaesthetized using 1 gl^–1MS-222 (Syndel Laboratories Ltd, Qualicum Beach, BC, Canada). For immunohistochemistry, gill and kidney tissues were carefully isolated and fixed in Bouin's solution for 3–4 h. A standardized region of the intestine (i.e. a section approximately one-third from the anterior-most region, relative to full gastrointestinal tract length), was also isolated and gently flushed of any gut contents with Bouin's solution. The tissue was then immersed in Bouin's solution and fixed for 3–4 h. Following fixation,all tissues were rinsed twice with 70% ethanol and stored in 70% ethanol at 4°C until further processing. For Western blot analysis, samples of goldfish blood cells, gill, kidney and intestine were collected. Blood was sampled from anesthetized fish via caudal puncture using a 25 gauge needle, following which fish were killed by spinal transection. Blood was allowed to clot at 4°C for 1 h and was then centrifuged at 10 600 g for 5 min at 4°C. The resulting serum was discarded and the pellet of packed blood cells was quick-frozen in liquid nitrogen and stored at –85°C until further processing. The intestinal segment collected for Western blot analysis was the same as that described above;however, any residual gut contents were gently flushed out with 0.7% NaCl(i.e. not flushed with Bouin's solution). Rat kidney tissue was donated by G. Sweeney (York University, Toronto, ON, Canada). After collection, all tissues were quick-frozen in liquid nitrogen and stored at –85°C until further analysis.

Fixed tissues were dehydrated through an ascending series of ethanol rinses(70–100%), cleared with xylene and infiltrated and embedded in Paraplast Plus Tissue Embedding Medium (Oxford Worldwide, LLC, Memphis, TN, USA). Sections (3 μm thick) were cut using a Leica RM 2125RT manual rotary microtome (Leica Microsystems Inc., Richmond Hill, ON, Canada), collected on 2% bovine serum albumin (BSA; BioShop Canada Inc., Burlington, ON,Canada)-coated glass slides and incubated overnight at 45°C. Sections were deparaffinized with xylene, rehydrated to water via a descending series of ethanol rinses (100–50%), and subjected to heat-induced epitope retrieval (HIER). HIER was accomplished by immersing slides in a sodium citrate buffer (10 mmol l^–1, pH 6.0) and heating both solution and slides in a microwave oven for 4 min. The solution was allowed to cool for 20 min, reheated for 2 min and cooled for a further 15 min. Slides were then washed three times with phosphate-buffered saline (PBS, pH 7.4) and quenched for 30 min in 3% H₂O₂ in PBS. Following quenching, slides were then successively washed with 0.4% Kodak Photo-Flo 200 in PBS (PF/PBS, 10 min), 0.05% Triton X-100 in PBS (TX/PBS, 10 min), and 10%antibody dilution buffer (ADB; 10% goat serum, 3% BSA and 0.05% Triton X-100 in PBS) in PBS (ADB/PBS, 10min). Slides were incubated overnight at room temperature with rabbit polyclonal anti-occludin antibody (1:100 dilution in ADB; Zymed Laboratories, Inc., South San Francisco, CA, USA) and mouse monoclonal anti-Na⁺,K⁺-ATPase α-subunit antibody(α5, 1:10 in ADB; Developmental Studies Hybridoma Bank, Iowa City, IA,USA). The anti-occludin antibody is epitope-affinity purified and directed against the C-terminal region of the human occludin protein. As negative controls, two sets of slides were also incubated overnight with ADB alone or with normal rabbit serum. Normal rabbit serum was donated by P. Moens (York University, Toronto, ON, Canada). Following overnight incubation, sections were successively washed with PF/PBS, TX/PBS and ADB/PBS (10 min each) as described above, and incubated with tetramethyl rhodamine isothiocyanate(TRITC)-labelled goat anti-rabbit antibody (1:500 in ADB; Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) and fluorescein-isothiocyanate (FITC)-labelled goat anti-mouse antibody (1:500 in ADB; Jackson ImmunoResearch Laboratories, Inc.) for 1 h at 37°C. Slides were then successively washed with PF/PBS, TX/PBS and PF/PBS (10 min each) and rinsed 3 times with 0.4% PF in distilled water (PF/dH₂O, 1 min each). Slides were air dried for 1 h and mounted with Molecular Probes ProLong Antifade (Invitrogen Canada Inc., Burlington, ON, Canada) containing 5μgml^–1 4′,6-diamidino-2-phenylindole (DAPI;Sigma-Aldrich Canada Ltd, Oakville, ON, Canada). Fluorescence images were captured using a Reichert Polyvar microscope (Reichert Microscope Services,Depew, NY, USA) and Olympus DP70 camera (Olympus Canada Inc., Markham, ON,Canada), and merged using Adobe Photoshop CS2 software (Adobe Systems Canada,Toronto, ON, Canada).

Goldfish tissues (blood pellet, gills, intestine, kidney) and rat kidney were thawed and homogenized on ice in chilled homogenization buffer (200 mmol l^–1 sucrose, 1 mmol l^–1 EDTA, 1 mmol l^–1 PMSF, 1 mmol l^–1 DTT in 0.7% NaCl)containing 1:200 protease inhibitor cocktail (Sigma-Aldrich Canada Ltd). Tissues were homogenized at a 1:3 w:v tissue to homogenization buffer ratio using a PRO250 homogenizer (PRO Scientific Inc., Oxford, CT, USA). Homogenates were centrifuged at 3200 g for 20 min at 4°C and supernatants were collected after centrifugation. Protein content was quantified using the Bradford assay (Sigma-Aldrich Canada Ltd) according to the manufacturer's guidelines with BSA as a standard. Samples were prepared for SDS-PAGE by boiling at 100°C with 6× sample buffer(360mmoll^–1 Tris-HCl, 30% glycerol, 12% SDS,600mmoll^–1 DTT, 0.03% Bromophenol Blue); 20μg of rat kidney and 75μg of goldfish blood pellet, gill, intestine and kidney were electrophoretically separated by SDS-PAGE in 12% acrylamide gels at 150 V. After electrophoresis, protein was transferred to a Hybond-P polyvinylidene difluoride (PVDF) membrane (GE Healthcare Bio-Sciences Inc., Baie d'Urfé, QC, Canada) over a 2 h period using a TE 70 Semi-Dry Transfer unit (GE Healthcare Bio-Sciences Inc.). Following transfer, the membrane was washed in Tris-buffered saline with Tween-20 [TBS-T; TBS (10 mmol l^–1 Tris, 150 mmol l^–1 NaCl, pH 7.4) with 0.05% Tween-20], and blocked for 1 h in 5% non-fat dried skimmed milk powder in TBS-T (5% skimmed milk TBS-T). The membrane was then incubated for ∼16 h at 4°C with rabbit polyclonal anti-occludin antibody (1:1000 dilution in 5% skimmed milk TBS-T; Zymed Laboratories, Inc.). Following incubation with primary antibody, the membrane was washed with TBS-T and incubated at room temperature with horseradish peroxidase (HRP)-conjugated goat anti-rabbit antibody (1:5000 in 5% skimmed milk TBS-T; Bio-Rad Laboratories, Inc.,Mississauga, ON, Canada) for 1 h, and then washed with TBS-T and TBS,respectively. Protein bands were visualized using Enhanced Chemiluminescence Plus Western blotting detection system (GE Healthcare Bio-Sciences Inc.).

Goldfish (mean mass 19.2±0.7 g, N=60), were acclimated to conditions as outlined above and randomly assigned to one of six experimental groups. Fish were food deprived for 1, 2 or 4 weeks, and for each unfed group a corresponding fed control group was run. Fed control fish were provided with pellets at a ration of 1.5% their body mass once daily. Control fish were not fed 24 h prior to sampling. Goldfish were weighed immediately prior to commencing experiments and at the end of each experimental period (i.e. at the time of tissue sampling), enabling calculation of body mass change. Individual fish were identified by unique markings. At 1, 2 and 4 weeks, fed and unfed fish were anaesthetized using 1 g l^–1 MS-222 and blood was rapidly sampled (within ∼2 min) via caudal puncture using a 25 gauge needle. Blood was processed as described above and serum was collected,quick-frozen in liquid nitrogen and stored at –85°C until further analysis. For analysis of muscle moisture content, a standardized region of epaxial white muscle was removed. Gill, kidney and intestinal tissues for enzyme and Western blot analyses were removed, frozen in liquid nitrogen and stored at –85°C until further processing.

Pre-weighed muscle tissue was placed in an oven and dried to a constant mass at 60°C. Muscle moisture content was subsequently determined gravimetrically. Serum osmolality was measured using a Model 5500 vapor pressure osmometer (Wescor, Inc., Logan, UT, USA). Serum Na⁺concentration was determined by atomic absorption spectroscopy using an AAnalyst 200 spectrometer (PerkinElmer Life and Analytical Sciences,Woodbridge, ON, Canada). Serum Cl^– concentration was determined using a colorimetric assay as previously described(Zall et al., 1956) and measured using a Multiskan Spectrum microplate reader (Thermo Fisher Scientific, Nepean, ON, Canada).

Na⁺,K⁺-ATPase activity was examined using methods previously outlined (McCormick,1993), with some minor modifications. Briefly, gill, kidney or intestinal tissues were homogenized at 4°C in a 1:10 w:v pre-chilled SEI(150 mmol l^–1 sucrose, 10 mmol l^–1 EDTA, 50 mmol l^–1 imidazole, pH 7.3):SEID (0.5 g sodium deoxycholate/100ml SEI) buffer mixture (4:1 mixture of SEI:SEID) using a PRO250 homogenizer. Homogenates were centrifuged at 3200 g for 10 min at 4°C and supernatants were collected, quick-frozen in liquid nitrogen and stored at –85°C until enzyme analysis. For analysis,supernatants were thawed on ice and assayed for Na⁺,K⁺-ATPase activity using solutions that couple ATP hydrolysis to ADP with the oxidation of NADH. The sensitivity of goldfish Na⁺,K⁺-ATPase activity to ouabain inhibition varies from tissue to tissue (see Busacker and Chavin,1981). Therefore, to distinguish Na⁺,K⁺-ATPase activity from total ATPase activity,samples were run in assay solutions either with or without K⁺present, under the assumption that K⁺-dependent ATPase activity is almost exclusively Na⁺,K⁺-ATPase activity. The use of K⁺-free assay solutions yielded equivalent results to using 0.5 mmol l^–1 and 10 mmol l^–1 ouabain for kidney and gill/intestine tissues, respectively. This corresponds with the observations in a previous report (Busacker and Chavin, 1981) that maximal inhibition of gill and kidney Na⁺,K⁺-ATPase activity occurred at an ouabain concentration of 10 mmol l^–1, but that gill activity was less sensitive to higher ouabain concentrations than kidney tissue. Na⁺,K⁺-ATPase activity was standardized to ADP release and was expressed as μmol ADP mg protein^–1h^–1. The protein content of supernatants used for analysis was measured using a Bradford assay with BSA as a standard, as described above.

Western blots for occludin were carried out as outlined above using equal amounts of protein from each tissue sampled. As a loading control, membranes were subsequently stripped with stripping buffer (100 mmol l^–1 glycine, 30 mmol l^–1 KCl, 20 mmol l^–1 sodium acetate, pH 2.2), washed with TBS-T, blocked with 5% skimmed milk TBS-T and incubated for ∼16 h at 4°C with mouse monoclonal anti-α-tubulin antibody (12G10; 1:10000 in 5% skimmed milk TBS-T; Developmental Studies Hybridoma Bank). Membranes were then washed with TBS-T and incubated at room temperature with HRP-conjugated goat anti-mouse antibody (1:5000 in 5% skimmed milk TBS-T; Bio-Rad Laboratories, Inc.) for 1 h, washed with TBS-T then TBS, and visualized as described above. Occludin andα-tubulin protein expression were quantified using Labworks image acquisition and analysis software (UVP BioImaging Systems and Analysis Systems, Upland, CA, USA), and α-tubulin was used for normalization of occludin expression.

All data are presented as mean values ± s.e.m. A one-way ANOVA was used to examine for significant differences between control groups at 1, 2 or 4 weeks. In all cases no significant differences were found. Therefore, data were subsequently analysed using Student's t-test to examine for significant differences (P<0.05) between control (fed) and experimental (unfed) groups within a time point. A one-way ANOVA was also used to examine for significant differences between experimental (unfed) groups at 1, 2 or 4 weeks. All statistical analyses were run using Graphpad Instat software version 3.00 (GraphPad Software, Inc., San Diego, CA, USA).

Na⁺,K⁺-ATPase immunolocalized to select cells within the interlamellar (IL) region of primary filaments and at the base of the secondary gill lamellae of goldfish gills(Fig. 1A). Immunofluorescence microscopy revealed pronounced and discontinuous occludin immunostaining along the edges of secondary lamellae and medial parts of lamellae that are embedded within the body of the primary filament(Fig. 1B,C,E). In addition,occludin immunostaining appeared to be associated with cells of the secondary lamellae that also express Na⁺,K⁺-ATPase(Fig. 1C,D). Furthermore,occludin immunofluorescence was also prominent in parts of the IL region that line the lateral walls of the central venous sinus (CVS; Fig. 1B). While there appeared to be no difference in staining for occludin between the afferent and efferent edges of gill filaments, Na⁺,K⁺-ATPase immunostaining was generally prominent along the trailing edge of primary gill filaments(cross-sections not shown). No co-localization of occludin and Na⁺,K⁺-ATPasewas found, and no TRITC or FITC fluorescence was observed in control sections that had been probed with secondary antibody only (Fig. 1F) or with normal rabbit serum (not shown).

Immunohistochemical analysis of goldfish intestine revealed prominent basolateral Na⁺,K⁺-ATPase immunostaining and distinct apical occludin immunostaining in columnar epithelial cells lining the intestinal lumen (Fig. 2A,B,C). While basolateral Na⁺,K⁺-ATPase immunostaining also extended to epithelial cells lining the base of intestinal villi, apical occludin immunostaining appeared less prominent in this region(Fig. 2A). Observation of intestinal villi and occludin immunostaining at higher magnification revealed a honeycomb-like TJ protein apical distribution pattern facing the intestinal lumen (Fig. 2C). No TRITC or FITC fluorescence was observed in control sections probed with secondary antibody only (Fig. 2E) or with normal rabbit serum (not shown).

Immunofluorescence microscopy of the goldfish kidney revealed differential immunostaining patterns in discrete regions of the nephron for both occludin and Na⁺,K⁺-ATPase(Fig. 3B). No TRITC or FITC fluorescence was observed in control sections probed with secondary antibody only (Fig. 3C) or with normal rabbit serum (not shown). Within the proximal region of the nephron,immunostaining for Na⁺,K⁺-ATPase appeared to be restricted primarily to basal regions of the plasma membrane of renal epithelial cells (Fig. 3B, Fig. 4A). In contrast,Na⁺,K⁺-ATPase immunostaining in renal epithelial cells of the distal tubule and collecting duct exhibited patterns consistent with localization in basolateral regions of the plasma membrane(Fig. 3B, Fig. 4B,C). An observable difference in Na⁺,K⁺-ATPase immunostaining between the distal tubule and collecting duct was the intensity of immunoreactivity;basolateral Na⁺,K⁺-ATPase immunostaining was consistently weaker in the collecting duct(Fig. 3B, Fig. 4C) than in the distal tubule (Fig. 3B, Fig. 4B). While no occludin immunostaining could be observed in proximal segments of the goldfish nephron(Fig. 3B, Fig. 4D), strong and moderate occludin immunostaining occurred at the apical membrane of renal epithelial cells lining the lumen of the distal tubule and collecting duct, respectively(Fig. 3B, Fig. 4E,F). There appeared to be no occludin or Na⁺,K⁺-ATPase immunostaining at the goldfish renal corpuscle.

Western blot analysis of protein isolated from goldfish gill, intestine and kidney revealed single occludin immunoreactive bands at ∼68 kDa; however,no immunoreactive bands were detected for protein isolated from goldfish blood cells (Fig. 5). A single occludin immunoreactive band for protein isolated from rat kidney resolved at∼65 kDa (Fig. 5).

The effects of 1, 2 and 4 weeks of food deprivation on goldfish body mass are shown in Fig. 6. Control fish, fed 1.5% their initial body mass, gained an average (±s.e.m.) of 4.5±0.6%, 13.5±1.7% and 26.6±1.6% of their initial body mass during the 1, 2 and 4 week experimental periods, respectively. In contrast, food-deprived fish lost an average of 8.6±0.6% (1 week),12±1% (2 weeks) and 17.5±1.1% (4 weeks) their initial body mass.

After 1 week of food deprivation, no significant (P>0.05)alterations in serum osmolality and Na⁺ or Cl^–levels were observed (Fig. 7A,B,C). Similarly, 1 week of food deprivation had no significant(P>0.05) effect on muscle moisture content(Fig. 7D). In contrast,following 2 and 4 weeks of food deprivation, serum osmolality, and Na⁺ and Cl^– levels significantly(P<0.05) decreased, while muscle water content significantly(P<0.05) increased (Fig. 7).

When compared with fed fish groups, goldfish gill Na⁺,K⁺-ATPase activity significantly(P<0.05) decreased (∼16%) following 1 week of food deprivation, increased (∼11%, P>0.05) following 2 weeks of food deprivation and significantly (P<0.05) increased (∼13%)following 4 weeks of food deprivation (Fig. 8A). Gill occludin protein expression in food-deprived goldfish was significantly (P<0.05) lower than in control groups by∼41%, ∼58% and ∼31% following 1, 2 and 4 weeks of food deprivation, respectively (Fig. 8B). Intestinal Na⁺,K⁺-ATPase activity in goldfish following 1, 2 and 4 weeks of food deprivation significantly(P<0.05) decreased by ∼46%, ∼66% and ∼44%,respectively, when compared with fed fish(Fig. 9A). Intestinal occludin protein expression did not significantly (P>0.05) alter in goldfish following 1 and 2 weeks of food deprivation when compared with control groups; however, intestinal occludin protein expression significantly(P<0.05) decreased by ∼34% following 4 weeks of food deprivation when compared with fed fish(Fig. 9B). Food deprivation did not significantly (P>0.05) alter goldfish kidney Na⁺,K⁺-ATPase activity at any point during the experiment (Fig. 10A);however, goldfish kidney occludin protein expression significantly(P<0.05) increased by ∼640% and ∼160% following 1 and 2 weeks of food deprivation, respectively, and significantly(P<0.05) decreased by ∼60% following 4 weeks of food deprivation when compared with fed groups(Fig. 10B). The loading control, α-tubulin, was detected at ∼52 kDa and its expression did not change following 1, 2 and 4 weeks of food deprivation (not shown).

Despite the universal inclusion of TJs in illustrative models of ion transport across fish epithelia, studies specifically investigating TJs in fishes have largely been limited to morphological analyses by electron microscopy (Sardet et al.,1979; Bartels and Potter,1991; Freda et al.,1991; McDonald et al.,1991). To the best of our knowledge, no studies have examined the immunohistochemical localization and protein expression of any integral TJ protein in fish gill, intestine or renal epithelial tissue, although the localization of zonula occludens-1 (ZO-1, a cytosolic TJ-related protein) was recently described in puffer fish gills(Kato et al., 2007). Furthermore, no studies have examined whether or how TJ protein expression may adjust in response to altered hydromineral status in aquatic vertebrates. We have examined the localization and expression of the integral transmembrane TJ protein occludin in fish tissues for the first time. Our observations suggest a role for occludin in the maintenance of hydromineral balance and that its regulation of epithelial `tightness' may be tissue specific. Furthermore,within tissue composed of heterogeneous regions of physiological function,such as the nephron, the role of occludin is likely to vary between discrete zones. Overall, alterations in occludin protein expression in response to hydromineral imbalance suggest that occludin may play a dynamic role in the regulation of epithelial permeability in fish.

Na⁺,K⁺-ATPase immunolocalization to cells within the IL region of goldfish primary filaments and at the base of the secondary gill filaments (particularly along the trailing edge of primary gill filaments; Fig. 1A) corresponds with the location of mitochondria-rich cells (MRCs) in goldfish gills(Kikuchi, 1977) and FW fish gills in general (Perry, 1997; Wilson et al., 2000). Pronounced and discontinuous occludin immunostaining along the edges of goldfish gill secondary lamellae (Fig. 1B,E), and along the edges of gill epithelial cells that immunostain for Na⁺,K⁺-ATPase (and are thus presumed to be MRCs; Fig. 1D), suggests that occludin may be associated with TJs between cells of the lamellar epithelium (e.g. pavement cells or MRCs) and/or with TJs between pillar cells that surround and form the lamellar blood spaces. This generally agrees with previous reports in which freeze fracture and electron microscopy observations of fish gill epithelia have shown that pavement cells (PVCs), MRCs and pillar cells all form TJ complexes with adjacent cells – i.e. PVCs with adjacent PVCs, PVCs with MRCs, or pillar cells with adjacent pillar cells(Hughes and Grimstone, 1965; Sardet et al., 1979; Bartels and Potter, 1991; Kudo et al., 2007). Furthermore, a recent study has immunolocalized ZO-1, which is believed to associate with the C-terminal region of occludin(Furuse et al., 1994), to pillar cells within the gills of marine puffer fish(Kato et al., 2007). When taken together, the results of our study combined with the immunolocalization of ZO-1 in puffer fish gills (Kato et al.,2007) indicate that the close association between ZO-1 and occludin that has been observed in mammals is likely to exist in fishes as well. While our data suggest a potential role for occludin in gill permeability, future studies using higher resolution microscopy techniques will be beneficial to ascertain the exact nature of occludin expression and interaction between gill cells (e.g. PVCs, MRCs and/or pillar cells) within branchial lamellae.

Similar to immunohistochemical studies in other fish species(Giffard-Mena et al., 2006),Na⁺,K⁺-ATPase immunostaining in the goldfish intestine was concentrated along the basolateral membrane of columnar epithelial cells lining the intestinal lumen (Fig. 2A,B). In contrast, occludin immunostaining was most prominent in apical regions of intestinal epithelial cells(Fig. 2A,C). When observed more closely, occludin immunostaining along the apical membrane of intestinal epithelial cells appeared to be distributed in a honeycomb-like arrangement(Fig. 2C), a typical TJ protein distribution pattern that has been observed along the gastrointestinal tract of other vertebrates (Inoue et al.,2006; Ridyard et al.,2007).

The goldfish kidney revealed differential immunostaining patterns in discrete regions of the nephron for both Na⁺,K⁺-ATPase and occludin (Fig. 3B). Similar differential Na⁺,K⁺-ATPase staining patterns have been observed in other fish species and basolateral localization of Na⁺,K⁺-ATPase concurs with models illustrating region-specific ion transport mechanisms in fish renal epithelia(Nebel et al., 2005; Beyenbach, 2004). Furthermore,specific patterns of Na⁺,K⁺-ATPase distribution along the nephron have also been reported for several other vertebrate groups(Piepenhagen et al., 1995; Kwon et al., 1998; Sabolić et al., 1999; Sturla et al., 2003). The differential occludin expression patterns observed in the goldfish nephron were similar to those observed in the mammalian kidney(Kwon et al., 1998; González-Mariscal et al.,2000). In mammals, differential occludin immunostaining patterns correlate with renal epithelial `tightness', such that `tighter' nephron segments (as determined by TER measurements) express higher levels of occludin protein than `leakier' nephron regions. In human and rabbit renal tubules,occludin immunostaining was weakest in `leaky' proximal tubules and strongest in `tight' distal tubules (Kwon et al.,1998; González-Mariscal et al., 2000). Furthermore, Western blot analysis of microdissected rabbit renal tubules revealed low occludin protein expression in `leaky' proximal tubules and significantly higher occludin protein expression in `tight' distal and collecting segments(González-Mariscal et al.,2000). The proximal tubule of the FW fish nephron is characterized as a relatively water-permeable and `leaky' epithelium that reabsorbs a small percentage of Na⁺ and Cl^– ions, as well as glucose and other organic solutes, from glomerular filtrate(Logan et al., 1980; Dantzler, 2003). The distal tubule and collecting duct of the FW teleost nephron, on the other hand,reabsorb the majority of salts from glomerular filtrate and are characterized as `tight' epithelia (Nishimura et al.,1983; Dantzler,2003). In the goldfish nephron, no occludin immunostaining was observed in `leaky' proximal regions (Fig. 4D); however, strong apical occludin immunoreactivity was detected in the `tighter' distal regions and moderate expression was observed in the collecting segments (Fig. 4E,F)These observations suggest that occludin may regulate goldfish renal epithelial `tightness' in a manner similar to mammals, and may thus influence the re-absorptive capacity of the different segments of the nephron.

Western blot analysis of occludin protein expression in homogenized rat kidney revealed a single immunoreactive band at ∼65 kDa, while single immunoreactive bands at ∼68 kDa were detected for homogenized goldfish gills, intestine and kidney (Fig. 5). No occludin immunoreactivity was detected for goldfish blood cells, a non-epithelial tissue (Fig. 5). Although predominantly detected as a 65 kDa protein in mammals, many reports have identified several occludin immunoreactive bands between ∼62 and 82kDa (Sakakibara et al., 1997; Wong,1997), therefore the occludin immunoreactive band found for goldfish epithelial tissue is consistent with the molecular mass range found in other vertebrates.

Food deprivation in goldfish resulted in a negative energy balance(negative changes in fish mass) at all time periods examined in the current study (Fig. 6). However, only fish that were food deprived for 2 weeks or longer exhibited alterations in the endpoints associated with salt and water balance. The observed reductions in serum osmolality, and Na⁺ and Cl^– levels and a concomitant increase in muscle hydration(Fig. 7) indicate that food deprivation can elicit changes in the hydromineral status of goldfish. These observations are in line with other studies that have described the reorganization of ionoregulatory machinery in response to restricted dietary regimes or food deprivation (Kültz and Jürss, 1991; Vijayan et al., 1996; Kelly et al.,1999).

Short-term food deprivation (1 week) in goldfish resulted in a significant reduction in gill Na⁺,K⁺-ATPase activity while a longer period without feeding (e.g. 4 weeks) resulted in a significant increase in gill Na⁺,K⁺-ATPase activity(Fig. 8A), suggesting an initial (temporary) down-regulation of active ion transport across the gills followed by a significant up-regulation. An up-regulation of gill Na⁺,K⁺-ATPase activity after 4 weeks of food deprivation was unexpected. A previous study(Kültz and Jürss,1991) reported that food deprivation (albeit 6 weeks) in FW tilapia caused a reduction in gill Na⁺,K⁺-ATPase activity. However, results may be time or species specific as another study(Vijayan et al., 1996)reported no change in gill Na⁺,K⁺-ATPase activity after 2 weeks of food deprivation in the same species. In contrast to Na⁺,K⁺-ATPase activity, occludin protein expression decreased in response to food deprivation after 1 week and remained consistently low after 2 and 4 weeks (Fig. 8B). Although these results allow us to accept our original hypothesis, that occludin protein expression would alter in response to hydromineral imbalance, it is difficult to rationalize a reduction in occludin protein expression, as opposed to an increase that might be expected to occur in association with gill epithelial tightening and reduced passive ion loss. There are several possible explanations: (1) gill epithelia may become`leakier' in food-deprived goldfish; however, this would seem maladaptive[furthermore, a previous study (Nance et al., 1987) reported reductions in gill epithelial permeability in response to food deprivation in a FW fish]; (2) the role of occludin in regulating gill epithelial permeability during periods of food deprivation may be overshadowed by other TJ proteins such as claudins, a family of transmembrane TJ proteins that also significantly contribute to the TJ barrier function (for a review, see Koval,2006); or (3) the use of whole-gill homogenates and the heterogeneous nature of the gill epithelium may mask specific changes in occludin expression between specific gill cell types. For example, an earlier study reported (Kültz and Jürss,1991) a significant reduction in MRC number in response to food deprivation in a FW fish. A reduction in MRC number, and thus a reduction in MRC–PVC TJ interactions, could potentially result in an overall reduction in gill occludin expression with little to no change in gill permeability, since PVC–PVC TJs would presumably remain intact. Regardless, the exact reason(s) for a reduction in gill occludin expression in food-deprived goldfish requires further study.

Dietary Na⁺ and Cl^– as well as nutrient absorption by intestinal epithelial cells of FW fishes is dependent upon an electrochemical gradient generated by Na⁺,K⁺-ATPase(Loretz, 1995). Food deprivation at all time points resulted in a significant reduction in goldfish intestinal Na⁺,K⁺-ATPase activity(Fig. 9A), suggesting a diminished capacity for active dietary salt and nutrient absorption by starved fish. Reduced intestinal Na⁺,K⁺-ATPase activity as a result of food deprivation has also been reported in FW tilapia(Kültz and Jürss,1991) and may be indicative of a depletion of the intestinal absorptive mucosa, a starvation-associated condition observed in other fish species (Bogé et al.,1981; Avella et al.,1992). Intestinal occludin expression significantly decreased following 4 weeks of food deprivation only(Fig. 9B), suggesting that over longer periods of dietary restriction, occludin may become involved in modifications of the barrier function of the goldfish intestine. Significant reductions in intestine epithelial TER and increased Na⁺-independent intestinal influx of proline in FW-adapted coho salmon following 2weeks of food deprivation have previously been reported(Collie, 1985), indicating impairment of barrier function in response to starvation in a FW fish. In other vertebrates, occludin down-regulation occurs in association with decreased intestine epithelial resistance, TJ protein re-distribution and intestinal barrier dysfunction (Zeissig et al., 2007; Musch et al.,2006). These areas require further attention.

The FW fish kidney actively reabsorbs salts from glomerular filtrate producing dilute urine. Solute reabsorption across proximal and distal tubules of the nephron is driven by an electrochemical gradient of Na⁺generated by Na⁺,K⁺-ATPase(Dantzler, 2003). Although, in the current study, negative energy balance appeared to have no overall effect on kidney Na⁺,K⁺-ATPase activity(Fig. 10A), it is possible that food deprivation may have resulted in nephron-specific alterations in Na⁺,K⁺-ATPase activity such that there was no observable alteration in `total' activity. For example, in humans, dietary restriction and food deprivation are associated with reduced Na⁺ reabsorption across the proximal tubule of the nephron and a concomitant increase in Na⁺ reabsorption by distal segments to counterbalance natriuresis,presumably both of which are associated with opposing alterations in ionomotive enzyme activity (Satta et al.,1984). Occludin protein expression in the goldfish kidney,however, exhibited an apparent biphasic pattern in food-deprived fish,markedly increasing after 1 week and significantly decreasing after 4 weeks(Fig. 10B), suggesting that food deprivation provokes a biphasic effect on renal function in the goldfish. A starvation-induced biphasic response in renal function has previously been documented in both humans and rats (Boulter et al., 1973; Boim et al.,1992; Wilke et al.,2005), where short-term starvation can result in natriuresis and polyuria that are eventually corrected and compensated for over longer experimental periods (Boulter et al.,1973; Wilke et al.,2005). Assuming the observed biphasic alterations in kidney occludin expression in food-deprived goldfish lead to adaptive function, one can rationalize that resulting regional changes in nephron permeability would enhance ion reabsorption and augment water elimination. In this regard, it is noteworthy that the highest and lowest renal expression of occludin in food-deprived goldfish occur in association with reduced and elevated gill Na⁺,K⁺-ATPase activity, respectively, indicating an interplay of strategies worthy of further investigation.

To summarize, we have immunolocalized occludin in goldfish ionoregulatory epithelia and demonstrated that occludin protein expression levels alter in response to hydromineral imbalance. The changes that occur in occludin protein abundance in response to starvation-induced hydromineral imbalance are tissue specific and, based on morphological evidence, are likely to be regionally different within specific tissues. The current study suggests that occludin should be expected to play an important role in the regulation of paracellular solute movement in aquatic vertebrates. While the response of occludin to hydromineral imbalance in goldfish often fits with its currently accepted role as an integral transmembrane TJ protein involved in regulating epithelial permeability, alterations in gill tissue are less easily explained. This underscores the paucity of information in the area of TJ physiology and the role these proteins play in the homeostatic control of hydromineral balance in aquatic vertebrates. This alone is an impetus for further study.

This work was supported by an NSERC Discovery Grant and a CFI New Opportunities Fund to S.P.K. H.C. was supported by an Ontario Graduate Scholarship from the Government of Ontario. All procedures conformed to the guidelines of the Canadian Council of Animal Care. The monoclonal antibodies developed by D. M. Fambrough (α5) and J. Frankel and E. M. Nelsen(12G10) were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa,Department of Biological Sciences, Iowa City, IA, 52242, USA.