SUMMARY
Juvenile gilthead sea bream were exposed to diluted seawater (2.5‰salinity; DSW) for 3 h or, in a second experiment, acclimated to DSW and fed a control or calcium-deficient diet for 30 days. Branchial Ca2+influx, drinking rate and plasma calcium levels were assessed. Sea bream plasma parathyroid hormone related protein (sPTHrP) was measured, and mRNAs of pthrp, its main receptor, pth1r, and the calcium-sensing receptor (casr) were quantified in osmoregulatory tissues and the pituitary gland. When calcium is limited in water or diet, sea bream maintain calcium balance; however, both plasma Ca2+ and plasma sPTHrP concentrations were lower when calcium was restricted in both water and diet. Positive correlations between plasma sPTHrP and plasma Ca2+(R2=0.30, N=39, P<0.05), and plasma sPTHrP and body mass of the fish (R2=0.37, N=148, P<0.001) were found. Immunoreactive sPTHrP was demonstrated in pituitary gland pars intermedia cells that border the pars nervosa and co-localises with somatolactin. In the pituitary gland, pthrp, pth1r and casr mRNAs were downregulated after both short-and long-term exposure to DSW. A correlation between pituitary gland pthrp mRNA expression and plasma Ca2+(R2=0.71, N=7, P<0.01) was observed. In gill tissue, pthrp and pth1r mRNAs were significantly upregulated after 30 days exposure to DSW, whereas no effect was found for casr mRNA expression. We conclude that in water of low salinity,declining pituitary gland pthrp mRNA expression accompanied by constant plasma sPTHrP levels points to a reduced sPTHrP turnover and that sPTHrP, through paracrine interaction, is involved in the regulation of branchial calcium handling, independently of endocrine pituitary gland sPTHrP.
Introduction
Parathyroid hormone related protein (PTHrP) is a hypercalcemic factor in fish (Guerreiro et al., 2001). Phylogenetically it is the ancestor of parathyroid hormone (PTH), which is the major hypercalcemic hormone in terrestrial vertebrates. A sea bream(Sparus auratus L.) pthrp cDNA has been cloned(Flanagan et al., 2000) and the genomic structure of pthrp was clarified in fugu [Fugu rubripes Temminck and Schlegel, 1850(Power et al., 2000)]. Recently, pth genes were discovered in the zebrafish (Danio rerio Hamilton 1822) genome (Gensure et al., 2004). The two hormones share a high N-terminal amino acid sequence homology and both peptides bind and activate shared G-coupled PTH/PTHrP receptors (Gardella and Jüppner, 2001). Three different PTHrP receptors were identified in fish (PTH1R, PTH2R and PTH3R), of which PTH1R is the most common and shares homology with the mammalian PTH1R(Rubin and Jüppner,1999). PTHrP has a key function in several physiological and biochemical processes in fish, including tissue differentiation and proliferation, vitellogenesis (Guerreiro et al., 2002; Bevelander et al., 2006), cortisol production(Rotllant et al., 2005a),calcium regulation (Guerreiro et al.,2001; Abbink et al.,2004) and calcium resorption from bone and scales(Rotllant et al., 2005b),which strongly indicates that PTHrP is involved in (skeletal) calcium physiology. The presence of PTHrP in a large number of tissues suggests PTHrP to be an auto-/intra- or paracrine factor. However, the immunohistochemical detection of PTHrP in the pituitary gland could also suggest a classical endocrine function for PTHrP in fish, as suggested by Danks et al.(Danks et al., 1993).
Fish have access to infinite sources of readily available calcium in the water. Calcium from water and diet can be taken up via gills and intestine, and calcium balance is achieved by branchial efflux and intestinal excretion. About 99% of the total calcium in fish is incorporated in the skeleton and dermal scales (Flik et al.,1986); the latter have a protective function, but also serve as an internal calcium buffer. In fish blood, the plasma total calcium concentration is about 2-3 mmol l-1, of which the ionic fraction accounts for about half (Hanssen et al.,1991). This ionic fraction is important for numerous physiological and biochemical processes and is therefore tightly regulated within narrow limits by calcemic endocrines (Flik et al., 1995). As the calcium availability in water and diet vary, as does the need for calcium, the calcemic endocrine system should react swiftly to changes in calcium availability or need(Björnsson et al.,1999).
This study focused on the regulation of the hypercalcemic sPTHrP and the calcium balance in response to a short and long-term calcium constraint in water and/or diet. Juvenile sea bream were rapidly transferred from full-strength seawater (SW; 34‰ salinity; 10.5 mmol l-1Ca2+) to diluted seawater (DSW; 2.5‰ salinity; 0.7 mmol l-1 Ca2+) and sampled 3 h later (the short-term experiment). In a second experiment, juvenile sea bream were exposed to SW or DSW and were fed a calcium-sufficient (Ca+) or calcium-deficient (Ca-)diet for 30 days (the long-term experiment). The experiments were carried out under controlled laboratory conditions. Gill Ca2+influx (FinCa2+), drinking rate(DR), plasma sPTHrP, as well as plasma total and ionic calcium concentrations were assessed. pthrp, pth1r and casr mRNA expression levels were quantified in gill, intestine, kidney and the pituitary gland and immunostaining was used to examine pituitary glands for sPTHrP immunoreactivity.
Materials and methods
Fish
Juvenile gilthead sea bream (Sparus auratus L.) were obtained from a commercial fish farm (Viveiro Vilanova, Lda., V.N. Milfontes, Portugal) and flown to the Netherlands without mortality. The fish were kept in round 600 l tanks with aerated flow through, a salinity of 34‰ and a temperature of 23±1°C. The fish were fed commercial pellets (Trouvit, Trouw,Putten, The Netherlands) at a ration of 2% of the total body mass daily.
Experimental set-up
Short-term experiment
Fourteen fish were placed in a tank with identical water conditions as in the stock. After 1 week of acclimatisation, the fish were rapidly transferred to a second tank with identical conditions (SW; control transfer; N=7) or to a tank containing diluted seawater of 2.5‰ salinity(DSW; N=7). After 3 h, the fish were killed with 2-phenoxyethanol(1:200; Sigma-Aldrich, St Louis, MO, USA) and blood was taken from the caudal vessels using a 1 ml tuberculin syringe, rinsed with 5× diluted sodium heparin (Leo Pharma, Weesp, The Netherlands; 1000 units ml-1). The collected blood was centrifuged at 13 600 g for 10 min and the plasma so obtained stored at -20°C. Fish were not fed for 24 h before sampling.
Long-term experiment
Fish (N=160) were randomly selected from stock, placed in four round tanks with 40 fish per tank and left to acclimatise. After 1 week, the salinity was gradually lowered by continuous flow-through with demineralised water until the test salinity of 2.5‰ (0.7 mmol l-1Ca2+) was reached, after 48 h. The diet was changed from control pellets to the test pellets (Hope Farms, Woerden, The Netherlands). The experimental animals were fed first and the controls received an equivalent amount of food as taken up by the experimental fish. After 3 days, the fish fully accepted the new diet and ate all the food provided (2% of the total body mass per day). The four groups of fish included a control (group a:34‰ salinity; Ca+ diet) and three test groups: group b (34‰salinity; Cadiet), group c (2.5‰ salinity; Ca+ diet) and group d(2.5‰ salinity; Cadiet). After 30 days, the fish were sampled as described for the short-term experiment; on the day before sampling, feeding was discontinued.
Calcium influx and drinking
After 30 days into the experiment, 20 fish from each group were randomly selected and placed in two identical vessels. After 24 h of acclimatisation, 51Cr-EDTA (1.9 kBq ml-1) or 45CaCl2 (2.5 kBq ml-1) was added to the tanks to assess drinking rate (DR) and gill Ca2+ influx(FinCa2+), respectively(Flik et al., 1985). Fish were sampled 2 h (DR) or 4 h(FinCa2+) after addition of the isotopes. Water samples were collected and the fish were killed by adding 2-phenoxyethanol (1:200; Sigma-Aldrich) to the water. The fish were rinsed with demineralised water, quick-frozen in solid CO2 and the frozen intestinal track was removed. Samples were weighed and rapidly digested in H2O2 (35%; 2 ml g-1; Lamers & Pleuger, `s Hertogenbosch, The Netherlands). Water calcium content was measured with a calcium kit (Roche, Mannheim, Germany; cat. no. 1489216) and radioactivity in the water and digested fish samples was counted with a liquid scintillation counter (Wallac 1410; Wallac, Turku, Finland). An OptiPhase HiSafe 3 liquid scintillation cocktail (Perkin-Elmer, Boston, MA, USA) was added before counting.
DR was calculated as: DR=Ai/(Awtm), where Ai is the total activity of 51Cr-EDTA in the intestinal track (c.p.m.), Aw is the total activity in the water (c.p.m. nl-1), t is the exposure time to 51Cr-EDTA (h) and m is the mass (mg) of the fish(Flik et al., 2002).
FinCa2+was calculated as: FinCa2+=(AfCw)/(Awt),where Af is the total activity of 45Ca2+ in the fish (d.p.m.), Cw is the calcium concentration in the water (pmol l-1), Aw is the total activity in the water (d.p.m. l-1), and t is the duration of exposure to 45Ca2+ (h). There were no differences in total mass between groups, therefore the results are expressed as nl h-1(DR) and nmol h-1(FinCa2+) and were not normalised for body mass (Guerreiro et al.,2004).
Plasma analyses
Plasma Ca2+ (mmol l-1) was measured using a Stat Profile pHOx plus analyser (Nova Biomedical, Waltham, MA, USA). Plasma total calcium (mmol l-1) was assessed using a calcium kit (Roche,Mannheim, Germany) and plasma PTHrP (nmol l-1) was measured with a homologous radioimmunoassay according to the method of Rotllant et al.(Rotllant et al., 2003).
Immunohistochemistry
Juvenile sea bream pituitary glands were fixed in Bouin's fixative for 90 min, dehydrated and embedded in paraffin wax. Sections were cut at 5 μm and dewaxed using xylene and degraded alcohols. The immunostaining procedure followed the protocols described earlier for PTHrP(Danks et al., 1993) and somatolactin (SL) (Kaneko et al.,1993). Rabbit anti-sea bream (1-34)sPTHrP (1:100) and rabbit anti-rainbow trout SL (1:3000; a generous gift from Dr Sho Kakizawa, Ocean Research Institute, Tokyo, Japan) were used as primary antibodies. For sPTHrP immunostaining, the sensitive immunoperoxidase method with the Vectastain avidin-biotinylated enzyme complex (Vectastain ABC; Vector Laboratories Inc.,Burlingame, CA, USA) was used to increase the staining intensity. Periodic-acid Schiff (PAS) staining was used to distinguish the two cell populations that are found in the pituitary gland, the PAS-positive somatolactin cells in the pars intermedia (pi), and the PAS-negative melanocyte stimulating hormone (MSH) cells.
Expression of pthrp, pth1r and casr mRNA
Tissue samples from the gill, kidney, anterior part of the intestine and pituitary gland were collected. Real-time, quantitative PCR (qPCR) was used to quantify the mRNA expression levels for pthrp, pth1r and the casr according to Hang et al.(Hang et al., 2005), with the housekeeping gene β-actin as endogenous control.
Statistics
All data are expressed as means ± standard deviation (s.d.);differences among groups were assessed by ANOVA. Significance of differences was assessed by parametric (Student's t-test) or non-parametric(Mann-Whitney U-test) when appropriate and P<0.05 was taken as fiducial limit.
Results
Gill calcium influx and drinking rate
Fig. 1A shows that branchial FinCa2+ declined significantly from 105±47 and 77±17 nmol h-1 in the SW groups a and b to 21.4±3.7 and 18.0±4.8 nmol h-1 in the DSW groups c and d. This same pattern was found for drinking(Fig. 1B), with a significant decrease from 14.8±6.5 and 10.7±4.7 nl h-1 in SW to 5.2±1.93 and 4.5±1.3 nl h-1 in the DSW groups. Modification of calcium in the diet had no effect on gill FinCa2+ or DR.
Plasma analyses
Exposure to DSW for 3 h had no effect on the total and ionic plasma calcium level (Fig. 2A). When calcium was limited in water and/or diet for 30 days, the plasma total calcium concentration decreased in all experimental groups, whereas the ionic fraction decreased only when calcium was limited in both water and diet(1.13±0.05 mmol l-1 Ca2+ in group a and 0.93±0.07 mmol l-1 Ca2+ in group d).
The plasma sPTHrP level was not affected after exposure to DSW for 3 h(Fig. 2B), with 0.28±0.06 nmol l-1 in the controls and 0.26±0.12 nmol l-1 sPTHrP in the DSW fish. In the long-term experiment, the plasma sPTHrP level had slightly, but significantly, decreased when calcium was limited in both water and diet, from 0.13±0.02 nmol l-1 in the controls (group a) to 0.10±0.04 nmol l-1 in group d(Fig. 2B). The fish from the short-term experiment were transferred to another tank 3 h before sampling,which could have induced a stress response and thus increased cortisol levels. The measured cortisol levels, 42.1±10.1 nmol l-1(N=7) in the SW fish and 38.5±10.8 nmol l-1(N=7) in the DSW fish, do not represent stress levels in sea bream(Arends et al., 1999). We thus exclude stress-induced elevated cortisol levels as a cause of the difference seen in sPTHrP levels between the short and long-term experiment. These differences may relate to different body masses of the groups in the short and long-term experiment.
Immunohistochemistry
Immunostaining with antisera to sea bream (1-34)sPTHrP and trout(Oncorhynchus mykiss Walbaum, 1792) SL revealed immunoreactive sPTHrP and SL in pituitary gland pi cells (Fig. 4). The sPTHrP immunoreactivity was found in the PAS-positive SL cells and bordered the pars nervosa (pn). Controls with omission of the first antibody and pre-absorption with sPTHrP confirmed the specificity of sPTHrP immunoreactivity (data not shown).
Pthrp, pth1r and casr mRNA expression
Expression of pthrp, pth1r and casr mRNA was significantly downregulated in the pituitary gland of fish exposed to DSW for 3 h (Fig. 5). In kidney,intestine and gill, no effect on pthrp, pth1r or casr mRNA expression was observed (data for kidney and intestine not shown).
Exposure to DSW for 30 days resulted in significant downregulation of pthrp, pth1r and casr mRNAs in the pituitary gland when compared with the control group (Fig. 6A). In gills, a significant upregulation of pthrp and pth1r mRNA was found in the DSW fish, whereas expression of casr mRNA remained constant (Fig. 6B). In intestine and kidney, no differences in expression of pthrp, pth1r and casr mRNA were found (data not shown).
In the short-term experiment, a strong relationship between plasma Ca2+ and pituitary gland pthrp mRNA expression was found in the controls (Fig. 7; R2=0.71, N=7, P<0.01), whereas in the DSW group, this relationship was absent. No such samples were available for the long-term experiment.
Discussion
This study provides new key findings about the regulatory role of sPTHrP in the calcium balance of juvenile sea bream under calcium constraint.
Branchial FinCa2+ and DRdecreased in DSW; plasma Ca2+ had slightly, but significantly,decreased after long-term exposure to limited calcium in both water and diet. Apparently, DSW induces mild hypocalcemia that is not counteracted by increased Ca2+ uptake from the water or by drinking.
Plasma sPTHrP level decreased after long-term limitation of calcium in both water and diet and plasma sPTHrP correlated with plasma Ca2+ and the body mass of the fish. This points to a decreased calcium turnover under calcium constraint.
Immunoreactive sPTHrP co-localised with immunoreactive SL in PAS-positive cells of the pi in the pituitary gland, indicating that pituitary sPTHrP may be the source of the high plasma sPTHrP levels in the fish.
In the pituitary gland, downregulation of pthrp, pth1r and casr mRNA was observed after both 3 h and 30 days of calcium constraint. In gills, mRNA for pthrp and pth1r was upregulated, but only after 30 days, whereas casr mRNA expression was not affected by calcium restraint. Thus we have evidence for a branchial sPTHrP regulatory system, acting independently of endocrine pituitary gland sPTHrP actions. The branchial chloride cell, being a key factor in calcium uptake in fish (Flik et al.,1995), appears to be fitted with a para-/auto- or intracrine hypercalcemic hormonal control mechanism. The production of sPTHrP by the chloride cell may be adjusted via CaSR activity. In situhybridisation experiments are required for confirmation.
Gill FinCa2+ and DR
Seawater is a strongly hypercalcemic environment for fish (∼10 mmol l-1 Ca2+) and therefore, fish are forced to reduce Ca2+ influx or to actively secrete Ca2+ to compensate for excessive Ca2+ influx. To compensate for osmotic water loss,seawater fish constantly drink water, which at the same time represents a high Ca2+ load. The transepithelial potential in fish is always more positive than the equilibrium potential for Ca2+ across the integument and therefore, the electrochemical gradient for Ca2+(the driving force for passive Ca2+ movement across the gills), is directed outwards (+30 mV), causing a substantial passive Ca2+efflux over the leaky branchial epithelium(Flik and Verbost, 1993). The uptake of Ca2+ is therefore not by diffusion, but by active transport.
The DSW (0.7 mmol l-1 Ca2+) causes a decrease in osmotic exchange, which subsequently results in the measured threefold decline in DR (and a consequent decrease in Ca2+ intake through drinking) and a fivefold decline in gill FinCa2+. In a hypotonic solution such as DSW, the osmotic water loss reverses to water gain and the influx of Ca2+ occurs almost entirely via the gills(Flik et al., 1986). The 15 times lower ambient Ca2+ concentration combined with a just five fold decline in FinCa2+ points to an increase in Ca2+ influx capacity or efficiency. This could be achieved by an increase in chloride cell density in the branchial epithelium(Flik et al., 1986) and an enhanced prolactin secretion. Low salinity is known to increase prolactin secretion from the pituitary gland (Kaneko and Hirano, 1993) in salt water fishes. Prolactin is known to limit ionic losses and water permeability in osmoregulatory tissues in hyposmotic media and to stimulate Ca2+ influx through gills and Ca2+-ATPase activity in gill plasma membranes(Flik et al., 1994), thereby increasing the Ca2+ influx capacity. The hypercalcemic control by sPTHrP which is shown in this study may connect both factors. This is further strengthened by the observation that gene expression for pthrp in mammals is upregulated in response to increased plasma prolactin levels(Thiede, 1989).
Endocrinology
An interesting observation is the relationship between the body mass of the fish and the plasma sPTHrP concentration. In sea bream, the plasma sPTHrP level increases with the body mass and plateaus with increasing mass of the fish, which suggests a decreasing need for hypercalcemic control with increasing body mass. Apparently, hypercalcemic control in juvenile stages is critically dependent on sPTHrP. As the growth rate of fish decreases with age,the need for calcium to be incorporated into the skeleton and scales also decreases, as may the requirement for regulatory sPTHrP. Strong positive correlations between plasma sPTHrP and the whole body content of calcium,phosphorus and magnesium (the main minerals in bone) were found (W. Abbink, X. Hang, P. M. Guerreiro, T. Spanings, A. V. M. Canario and G. Flik, manuscript in preparation), which strengthens the assumption that sPTHrP is involved in skeletal calcium physiology. This was recently suggested by Redruello et al.(Redruello et al., 2005), who showed a downregulation of the unique matricellular calcium binding glycoprotein osteonectin by PTHrP, and by Rotllant et al.(Rotllant et al., 2005b), who reported that PTHrP induced osteoclastic activity in scale tissue, as indicated by its stimulation of tartrate-resistant acid phosphatase (TRAPC; a marker for osteoclastic activity in fish scales).
Immunohistochemistry
In the sea bream pituitary gland, sPTHrP staining was detected in cells of the pi that were near the pn, and these sPTHrP-positive cells were identified as SL-producing cells. This was confirmed in earlier studies, when Rand-Weaver et al. (Rand-Weaver et al.,1991) and Kaneko et al.(Kaneko et al. 1993) found SL staining in PAS-positive cells that border the pn, in several teleosts. Our data confirm an earlier claim by Ingleton et al.(Ingleton et al., 1998) who reported that in sea bream, sPTHrP and SL are both located in the PAS-positive cells and that some cells contained both sPTHrP and SL. SL is a hormone from the prl gene family and is structurally related to both PRL and growth hormone. Kakizawa et al. (Kakizawa et al., 1993) studied SL plasma levels and sl mRNA expression in rainbow trout and suggested a role for SL in calcium balance and an increased hormone turnover rate at low calcium levels. Changes in SL plasma levels and pituitary gland mRNA expression at low ambient calcium appear only after several days (Kakizawa et al.,1993), which makes short-term effects of SL on calcium balance unlikely. However, the activity of SL-producing cells may be affected indirectly, possibly by the action of PTHrP. Our data show sPTHrP immunostaining in sea bream pituitary gland SL-producing cells and activation of pthrp mRNA production in the pituitary gland 3 h after transfer from SW to DSW. It could very well be that, in cells co-expressing pthrp and sl, the pthrp upregulation precedes that of sl and therefore the activity of SL in the pituitary gland. Interestingly, the pituitary gland sPTHrP-producing cells colocalise with a sub-population of SL-producing cells, the SLα cells as observed in zebrafish (Zhu et al.,2004).
mRNA expression
Expression of pthrp and pth1r mRNA was found in all tissues examined, indicative of an auto-/para- or intracrine function of sPTHrP. However, circulating plasma PTHrP levels in teleosts and elasmobranches, as well as immunostaining(Trivett et al., 1999) and mRNA expression (Hang et al.,2005) for pthrp in pituitary glands have been established, pointing to an endocrine function for PTHrP as well.
Downregulation of pthrp and pth1r mRNA in the pituitary gland was established after 3 h of calcium constraint and remained reduced after at least 30 days of calcium constraint. Both a rapid activation of pituitary gland sPTHrP production and a long-term involvement of sPTHrP in the adaptation to hypocalcic media seem required to maintain calcium balance at hypocalcic conditions. This suggestion is supported by the correlations that were found between plasma Ca2+ and pituitary gland pthrpmRNA expression and between plasma Ca2+ and plasma sPTHrP protein. However, the downregulation of pthrp and pth1r mRNA in the pituitary gland was not accompanied by a change in plasma sPTHrP levels, which had only slightly decreased in the group that was held in DSW and fed a Cadiet for 30 days. The adaptive response to calcium constraint results in a reduced metabolic clearance of sPTHrP from the plasma (in contrast to the reported action of SL), with downregulated mRNA expression in the pituitary gland and unaltered plasma sPTHrP levels. This points to a differential regulation of release of sPTHrP and SL in the cells that co-express these proteins, or alternatively a differential regulation of the two pituitary SL cell populations, recently reported (Zhu et al., 2004).
The five- to eightfold upregulated peripheral pthrp and pth1r mRNA levels that were found in gills after 30 days exposure to low calcium may reflect an adaptive response, possibly as a result of a decreased environmental calcium concentration or the reduced sPTHrP metabolic clearance from the plasma, by an autoregulatory feedback of sPTHrP on its own secretion. In earlier studies on mammals, Fujimi et al.(Fujimi et al., 1991)suggested that PTH(1-34) directly inhibits PTH secretion in parathyroid cells. In contrast, Lewin et al. (Lewin et al.,2003) hypothesized that PTH has a positive auto-feedback on its own secretion under hypocalcic conditions in rats.
Flanagan et al. (Flanagan et al.,2000) showed PTHrP staining in the chloride cells of gills in teleosts and identified these cells as the principal location of PTHrP in gill tissue. The increase in chloride cell density in diluted seawater conditions could explain the increase in pthrp mRNA that was found in gills. Chloride cells are the site of branchial Ca2+ uptake in gills and they also contain CaSR; however, expression of casr mRNA was unaffected in gills after calcium restraint. CaSR is regulated and equipped to respond to the blood Ca2+ level. Fluctuations as small as 0.2 mmol l-1 are sensed (Lopez-Ilasaca et al., 1997), which enables the fish to tightly regulate the blood Ca2+ concentration. In the present study, the plasma Ca2+ level was remarkably constant and maintained within a maximal range of 0.2 mmol l-1 difference when compared with the controls and therefore upregulation of casr mRNA may not be relevant. In the pituitary gland, a significant downregulation of casr mRNA was observed after both short-and long-term calcium constraint, indicating that the need for calcium controlled processes had decreased or reflecting a need for desensitisation to Ca2+ signals. Flanagan et al.(Flanagan et al., 2002) located immunostaining for CaSR in cells bordering the pn in both the pi and the pars distalis in the sea bream pituitary gland and suggested a possible feedback between nerve axons from hypothalamic nuclei and pituitary factors affected by calcium. This is supported by the localisation of sPTHrP in cells near the pn in this study, thereby possibly controlling the upregulation of pthrpgene expression in the gills.
Acknowledgements
This research was carried out with financial support from the Commission of the European Union, Quality of Life and Management of Living Resources specific RTD programme (Q5RS-2001-02904). The authors also wish to thank Ms Joana Amaral from Viveiro Villanova, Portugal, for production of superb fish.