Cortisol response to stress is associated with myocardial remodeling in salmonid fishes

The salmonid heart demonstrates a high degree of plasticity, both anatomically and physiologically, in response to environmental changes (Gamperl and Farrell, 2004). It is composed of an inner spongious myocardium, which is supplied with oxygen from venous blood returning to the heart, and an outer compact myocardium that contains coronary blood vessels delivering oxygenated blood from the gills (Pieperhoff et al., 2009). Plastic changes can involve cardiomyocyte hypertrophy and hyperplasia of both compartments (Gamperl and Farrell, 2004), and this is associated with an increased risk of myocardial remodeling and dysfunction, which is an increasing problem in farmed salmonids (Poppe and Taksdal, 2000; Poppe et al., 2002; Poppe et al., 2003; Takle et al., 2006). Myocardial dysfunction is also a problem in other farmed animals, including broiler chickens, where myocardial remodeling and failure is a leading cause of death (Olkowski et al., 1996). The underlying causes of pathological remodeling in fish have not been determined.

Stress can be defined as a condition in which a threat to the biological functions of an organism is perceived by that organism and a set of physiological and behavioral responses are mounted to counteract this challenge. Severe stress is clearly associated with a poor prognosis in individuals with established cardiac pathology and disease, including in humans (Engel, 1971; Meerson, 1994; Maxwell and Robertson, 1998; Brocklebank and Raverty, 2002; Poppe et al., 2007). Still, the mechanism linking stress responsiveness to the development of cardiac disease is poorly understood in fish. In particular, the association between stress and cardiac remodeling has previously not been addressed in salmonid fish, of which approximately two million tons are processed per year in the rapidly developing global aquaculture industry (Food and Agriculture Organization, 2010).

Cortisol is the major steroid stress hormone in salmonid fishes and humans. It has diverse effects on several tissues, including the myocardium. The effects of cortisol in salmonid fish are mediated through both the mineralcorticoid receptor (MR) and glucocorticoid receptors 1 and 2 (GR1 and GR2) (Colombe et al., 2000; Bury et al., 2003). Previous work has shown these receptors to be abundantly expressed in the myocardium of teleosts, including the rainbow trout (Greenwood et al., 2003; Sturm et al., 2005). However, a direct relationship between cortisol stress response and myocardial morphology and function has previously not been addressed. In contrast to mammals, where aldosterone is an important hormone in myocardial remodeling (Funder, 2001; Rocha et al., 2002; Qin et al., 2003), most teleosts, including salmonids, do not produce aldosterone (Bern, 1967; Sangalang and Uthe, 1994), and mineralcorticoid functions are instead mediated by cortisol (Bern and Madsen, 1992; Wendelaar Bonga, 1997). Furthermore, in mammals, glucocorticoids like cortisol directly induce protein synthesis and hypertrophy of cardiomyocytes in vivo and in vitro (Nichols et al., 1984; Lumbers et al., 2005), and plasma cortisol levels have been found to represent an independent risk factor of cardiac events and death (Yamaji et al., 2009).

Accordingly, to examine the effect of stress and cortisol on myocardial morphology and function, we examined cardiac structure and gene expression in the hearts of two genetically distinct strains of rainbow trout (Oncorhynchus mykiss Walbaum 1792) that respond to stress with either a high [high responsive (HR)] or low [low responsive (LR)] cortisol production (Pottinger and Carrick, 1999). We hypothesized that adult HR fish, consistently responding to stress with high serum cortisol, would have bigger hearts than LR fish. However, as cardiac remodeling may either be physiological or pathological in mammals, we also aimed to assess whether stress responsiveness and hence consistently different cortisol exposure throughout life would induce signs of pathology in fish. We were especially interested in genes mediating the response to cortisol, particularly those that are linked to vascularization, fibrosis and cardiac hypertrophy. Finally, we investigated whether a trait correlation of post-stress cortisol response and heart size also existed outside of the selected LR and HR trout lines. To this end, we examined heart size and cortisol responsiveness in wild-type European brown trout (Salmo trutta Linnaeus 1758).

The selection regime initiated at Windermere Laboratory, NERC Institute of Freshwater Technology, UK, generating the two lines of rainbow trout (LR and HR), has been described in detail previously (Pottinger and Carrick, 1999). In short, the parental generation of HR and LR fish was established on the basis of consistent divergence in plasma cortisol responses following repeated stress testing (3 h of confinement stress once a month for five consecutive months). The crosses between the selected parents were carried out and the F1 generation hatched in 1997. The F4 generation of LR and HR rainbow trout was hatched in spring 2006 and transported to the Norwegian Institute for Water Research Marine Research Station (Solbergstrand, Norway) in December 2007. The fish were then mixed and reared in two 1000 l fiberglass tanks until sampled as fully adult sexually mature individuals 2 years later (January 2010). At the time of sampling, all LR and HR trout were 40 months old and had a mean (±s.d.) body length of 47.5±5.3 and 49.2±4.8 cm, respectively. Only mature females were available for this study. The transport and study procedures of the experimental animals were reviewed and approved by the Norwegian Food Safety Authority (www.mattilsynet.no) and the Norwegian Animal Research Authority (http://www.fdu.no/fdu/), respectively.

Female European brown trout hatched at Aqua Center Boracko Lake, Konjic, Bosnia and Herzegovina, were used for these experiments. The parent generation consisted of wild adult endemic brown trout caught in the River Neretva drainage. At the onset of the experiment, the female fish were 21 months old and had a mean body length of 25.9±1.7 cm. Brown trout were transferred to 50 l aquaria and overhead windows provided light. Light tubes situated 40 cm above each aquarium were turned on 15 min after sunrise every day to provide additional light for behavioral observations. The artificial light was turned off 15 min prior to sunset. During the experimental period (10 March–25 June), photoperiod increased by 10 min per day at the experimental location (Boracko Lake). The fish were hand fed commercial food pellets daily. On day 16 the experimental fish were subjected to standardized confinement stress essentially as described by Øverli et al. (Øverli et al., 2006). In brief, fish were placed in pierced 1.5 or 1.9 l plastic boxes (Gefrier Box Frosty, PLAST TEAM GmbH, Flensburg, Germany) adjusted to the size of the fish (1.5 l boxes for fish ranging from 20 to 25 cm and 1.9 l boxes for fish ranging from 25 to 30 cm). The boxes were submerged in water in the aquarium for 2 h. The experimental model was approved by the State Veterinary Office of Bosnia and Herzegovina (http://www.vet.gov.ba).

HR and LR trout were anaesthetized in benzocaine (1.5 ml l^–1 Benzoak®, A.C.D. SA, Braine-l'Alleud, Belgium) and body length was measured (cm) before the fish were killed by decapitation. Three LR and three HR hearts were surgically excised, the bulbus and atrium removed, and the ventricles placed in 4% paraformaldehyde (PFA) (Electron Microscopy Sciences, Hatfield, PA, USA) for histochemistry analysis. Six LR and six HR hearts were sampled for gene expression analysis and the ventricle was weighed before being cut into smaller pieces, which were put on RNAlater® (Ambion, Austin, TX, USA) and stored at –20°C.

Following confinement stress, 27 brown trout were anaesthetized in benzocaine (1.5 ml l^–1) and body length was measured (cm) before the fish were killed by decapitation. A blood sample was taken and the blood was immediately spun (3 min at 13,000 g). Plasma was frozen and stored at –20°C for subsequent cortisol analysis. The hearts were excised and the atrium and bulbus removed before the ventricles were blotted dry and weighed (g). The cardiosomatic index (CSI) was determined by calculating the ratio of ventricle mass to fish length (g cm^–1). Body length, rather than body mass, was used to calculate CSI because several fish had running eggs upon netting from the holding tank. Images of the ventricles were taken using a Canon EOS350 digital camera (Canon, Tokyo, Japan) and processed in Adobe Photoshop CS3 (Adobe Systems Inc., San Jose, CA, USA).

Plasma cortisol levels from the brown trout were analyzed using a specific radioimmunoassay (RIA) essentially as described previously (Pottinger and Carrick, 2001). In short, ethyl acetate (Merck Chemicals, Darmstadt, Germany) was used for cortisol extraction. Before the donkey anti-cortisol antibody was added (1:6000 dilution, AbD Serotec, Dusseldorf, Germany), 60 Ci mmol l^–1 [1,2,6,7–³H] cortisol (Amersham Pharmacia Biotech, Little Chalfront, UK) was added to all samples, standards and controls. Samples were tested against a standard curve made with inert cortisol (Sigma Aldrich, St Louis, MO, USA) in scintillation fluid (Ultima Gold, Perkin Elmer, Waltham, MA, USA) on a Packard Tri-Carb A1900 TR liquid scintillation analyzer (Packard Instrument, Meriden, CT, USA). Cortisol concentrations were calculated from the equation of a three-parameter hyperbolic function fitted to a plot of the percentage of [³H] cortisol bound against inert cortisol using Sigmaplot 11 (SPSS Science, Systat Software Inc., San Jose, CA, USA).

Rainbow trout hearts stored in 4% PFA were cut into thick slices using a razor blade. These slices were further fixed in fresh 4% PFA at 4°C before embedding in paraffin. Dewaxed sections of 5 μm thickness were subjected to Acid-Fuchsin-Orange G (AFOG) staining. AFOG sections were scanned or micrographed using an Axio Scope with a 5× objective (Zeiss, Jena, Germany), and images were processed in Adobe Photoshop CS3.

Scanned AFOG-stained ventricular sections from rainbow trout hearts were processed using ImageJ (NIH, Bethesda, MD, USA). In brief, color images were converted into 8-bit grayscale, pixels scaled to mm, and brightness and contrast adjusted before edge detection was applied, allowing area calculation in mm² of the two muscular layers – the compact and spongious myocardium. The area of compact myocardium was divided by total area, giving the relative area of compact myocardium.

The hearts stored in RNAlater® were thawed and refrozen in liquid nitrogen before they were freeze-fractured in a BioPulverizer (Biospec Products, Inc., Bartlesville, OK, USA). The pulverized heart was thoroughly mixed before 280 mg were put into a 15 ml plastic tube and stored at –80°C for a maximum of 4 days. RNA was extracted from heart tissue using TRIzol® reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. The quality and quantity of the RNA was assessed using a 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA) and a NanoDrop® ND-1000 UV-Vis Spectrophotometer (NanoDrop Technologies, Rockland, DE, USA), respectively. RNA quality was determined from RNA integrity numbers (RINs) calculated by the 2100 Bioanalyzer (range: 1–10). RINs for the heart samples ranged from 9.50 to 9.90 with a mean (±s.d.) of 9.7±0.03, confirming excellent RNA quality. First-strand cDNA was synthesized from total RNA treated with 2 ng DNase I (DNA-free™ Kit, Ambion Applied Biosystems) using Superscript III reverse transcriptase (Invitrogen) with oligo dT_12–18 primers synthesized by Invitrogen.

The selected cardiac marker genes used in this study are presented in Table 1. Gene-specific primers for rainbow trout β-actin, proliferating cell nuclear antigen (PCNA), ventricular myosin heavy chain (VMHC), slow myosin light chain 2 (SMLC2), muscle LIM protein (MLP, also called CRP), regulator of calcineurin 1 (RCAN1, also called MCIP), mineralcorticoid receptor (MR), glucocorticoid receptor 1 and 2 (GR1 and GR2), vascular endothelial growth factor (VEGF), collagen alpha 2 (1) (COL1a2), collagen alpha 1 (1) (COL1a1), A-type natriuretic peptide (ANP) and B-type natriuretic peptide (BNP) were designed using the web-based Primer3 program (http://frodo.wi.mit.edu/primer3/) and synthesized by Invitrogen. The housekeeping gene β-actin was chosen as the internal control gene. GenBank accession numbers for the genes whose sequences were retrieved from NCBI (www.ncbi.nlm.nih.gov/) are listed in Table 1. Rainbow trout RCAN1 was retrieved at the Dana-Farber Cancer Institute database (http://compbio.dfci.harvard.edu) based on a BLAST with Danio rerio RCAN1 found in the NCBI database (GenBank accession no. BC076439.1). A minimum of five primer pairs were designed at exon junctions for each gene and the primers showing the lowest crossing point values, a single peak melting curve and amplification of the right amplicon were chosen (Table 1). The quantitative real-time (qRT) PCR products were also sequenced to verify that the primers amplified the right cDNA.

Data are expressed at single observation points or as group means ± s.e.m. The mean CSI of LR fish was normalized to 100%. mRNA levels and differences in CSI and area of compact myocardium between the HR and LR groups were examined using the Student's t-test, and the association between CSI and post-stress plasma cortisol was assessed by linear regression analysis (least-squares method) with Pearson's product-moment correlation coefficient as a measure of the resulting linear relationship. mRNA expression levels are presented as normalized values to LR mean (fold change), and differences were tested using an unpaired t-test. P-values <0.05 were considered statistically significant. All statistical analyses were performed in GraphPad Prism 5 (GraphPad Software, San Diego, CA, USA).

To examine the relationship between selection for post-stress plasma cortisol levels and heart size, freshly excised ventricles from adult HR and LR fish were weighed and CSI was calculated. The CSI of HR fish was 33.70±6.26% higher than that of LR fish (100.00±7.67%, P=0.0035; Fig. 1A,B). To evaluate the structural basis for the increased relative heart size of HR fish, sections of the ventricles were converted into grayscale images to distinguish the compact and spongious muscular layers. The relative area of compact myocardium in HR ventricles was 48.40±2.95% whereas that of LR fish was 33.40±2.44% (P=0.0173; Fig. 1C,D).

To evaluate whether the increased heart size in the HR fish was a result of myocyte hypertrophy or hyperplasia, markers of the two processes were measured by qRT-PCR. PCNA expression was 1.39±0.15-fold higher in HR ventricles compared with LR ventricles (1.00±0.12); however, this result was not statistically significant (P=0.07; Fig. 2A). In contrast, there was a 2.40±0.29-fold higher expression of VMHC in HR ventricles compared with LR ventricles (1.00±0.14, P=0.0015) and a 3.31±0.73-fold higher expression of SMLC2 in HR ventricles compared with LR ventricles (1.00±0.40, P=0.0194; Fig. 2B,C). MLP expression was also 1.93±0.41-fold higher in HR ventricles than in LR ventricles (1.00±0.23); however, this finding was not significant (P=0.0763; Fig. 2D).

To further examine the cardiac hypertrophy of the HR fish, we assessed whether the calcineurin–NFAT pathway was activated, as this signaling pathway is one of the major ones involved in mammalian pathological hypertrophy. The mRNA expression level of the NFAT target gene RCAN1 was 2.96±0.38-fold higher in HR ventricles compared with LR ventricles (1.00±0.14, P=0.0007; Fig. 3).

Visual investigation of surface vessels of HR and LR ventricles clearly revealed a higher degree of vascularization of HR ventricles (representative image in Fig. 4A), but this difference was not quantified. In line with this observation, VEGF expression was 2.66±0.19-fold greater in HR ventricles compared with LR ventricles (1.00±0.14, P<0.0001; Fig. 4B).

Excessive fibrosis is a hallmark of pathological myocardial remodeling in mammals, and to examine fibrosis in the trout hearts we stained ventricular sections histochemically with a marker for collagen depositions (AFOG) and measured Col 1a2 and Col 1a1 mRNA expression levels. There was moderate staining for collagen in all examined ventricles of HR fish (N=3), while no staining was observed in the LR ventricles (N=3; Fig. 5A left panel). Magnification of the HR ventricle revealed areas with (Fig. 5A, inset 1) or without (Fig. 5A, inset 2) collagen depositions, indicating focal fibrosis and disruption of muscle structure in the HR heart. Further supporting enhanced fibrosis in HR hearts, COL1a2 mRNA levels were 2.16±0.36-fold higher in HR ventricles compared with LR ventricles (1.00±0.21, P=0.0207; Fig. 5B). We also observed a 1.84±0.39-fold higher expression of COL1a1 in HR ventricles compared with LR ventricles (1.00±0.17), but this was not statistically significant (P=0.0770; Fig. 5C).

To assess further signs of remodeling in HR hearts, we measured expression of ANP and BNP, which are sensitive markers of heart failure in mammals. We found no significant differences in either ANP (1.35±0.26 HR vs 1.00±0.30 LR, P=0.4019; Fig. 6A) or BNP mRNA expression levels (0.90±0.10 HR vs 1.00±0.17 LR, P=0.6401; Fig. 6B).

To assess whether differences in post-stress cortisol levels are accompanied by differences in the expression of myocardial cortisol receptors, we measured MR, GR1 and GR2 mRNA levels in ventricles of HR and LR fish. HR fish, which consistently respond to stress with high plasma cortisol (Pottinger and Carrick, 1999; Pottinger and Carrick, 2001; Trenzado et al., 2003; Schjolden et al., 2006), displayed higher levels of all three receptors in the ventricle compared with LR fish (Fig. 7A–C). Specifically, the expression level of MR was 2.08±0.24-fold higher in HR ventricles than in LR ventricles (1.00±0.05, P=0.0012; Fig. 7A), that of GR1 was 2.36±0.24-fold higher (vs 1.00±0.09, P=0.0003; Fig. 7B) and that of GR2 was 2.19±0.10-fold higher (vs 1.00±0.08, P<0.0001; Fig. 7C).

Finally, to validate our results from the HR–LR selection model in wild-type trout, plasma cortisol levels were measured and CSI was calculated in a European brown trout population using an established stress model (acute confinement stress). There was a positive correlation between CSI and post-confinement plasma cortisol in these wild-type fish (R²=0.347, P=0.0012, N=27; Fig. 8), supporting our finding of an association between stress responsiveness and myocardial remodeling in HR and LR fish.

Our results show that fish responding to stress with high post-stress plasma cortisol levels had significantly bigger hearts than fish responding with low cortisol levels, and this phenomenon was observed in both a genetic selection model of rainbow trout and in wild-type brown trout. Immunohistochemical studies suggested that this was a result of growth of the compact myocardium, as this area, relative to the spongious myocardium, was bigger in HR ventricles compared with LR ventricles. Such an increase in compact myocardium is also seen during ontogenetic growth in salmonid fishes, during which the compact layer increases in both amount and thickness (Poupa et al., 1974; Farrell et al., 1988).

In mammals, under certain physiological (i.e. training or pregnancy) or pathological (i.e. hypertension or myocardial infarction) conditions, cardiomyocytes may undergo hypertrophy, which leads to increased muscular mass. Although recent findings suggest that cardiomyocytes within the diseased adult human heart can proliferate (Beltrami et al., 2001), most evidence to date indicates that myocyte proliferation is not a significant component of the mammalian cardiac growth response (Pasumarthi and Field, 2002). In adult teleostean hearts, including in rainbow trout, previous studies have shown that ventricular growth occurs through both hyperplasia and hypertrophy (Farrell et al., 1988; Clark and Rodnick, 1998; Poss et al., 2002). In the current study, there was a consistent upregulation of genes involved in the hypertrophic gene program. For instance, mRNA levels of commonly used hypertrophy markers in mammals (RCAN1, SMLC2 and the fish homologue of β-MHC, VMHC) (Swynghedauw, 1999; Lim et al., 2001) were significantly higher in HR ventricles compared with LR ventricles. Furthermore, although the hypertrophy marker MLP did not differ significantly between LR and HR ventricles, it displayed a trend towards being higher in the HR ventricles. We also found a trend towards a higher expression of the proliferation marker PCNA (Yu et al., 1992; Köhler et al., 2005), indicating increased cell proliferation in the HR ventricles compared with LR ventricles. Nevertheless, based on the combined gene expression profile of the current study, we believe that the bigger relative heart size in HR fish is primarily due to hypertrophy of existing compact cardiomyocytes. In cold-acclimated rainbow trout, Vornanen et al. observed an increase in heart size of 33% (Vornanen et al., 2005) that was, similar to the results of our study, associated with increased expression of VMHC and SMLC2. Furthermore, they reported a fivefold increase in the expression of MLP mRNA. The authors interpreted these results as an activation of genes associated with adaptive cardiac hypertrophy without the pathological features that impair the function of the hypertrophic mammalian heart. Fish hearts can undergo hypertrophy (and hyperplasia) as a routine remodeling mechanism (Farrell et al., 1988). Therefore, as opposed to mammals, where cardiac hypertrophy is often associated with a failing heart (Lloyd-Jones et al., 2002), a bigger fish heart is not necessarily detrimental. For example, an increased ventricular mass in cold-acclimated rainbow trout serves to offset the reduction in cardiac power output observed with acute decreases in temperature, and also enhances stroke volume (Graham and Farrell, 1989). Also, in the growing eel (Anguilla anguilla L.), structural remodeling of the heart, such as growth of the compact myocardium, is associated with enhanced mechanical performance (Cerra et al., 2004).

The calcium/calcineurin-regulated NFATc family is thought to have arisen following recombination approximately 500 million years ago, producing a new group of signaling and transcription factors (the NFATc genes) found only in genomes of vertebrates. It has been proposed that recombination enabled Ca²⁺ signals to be redirected to a new transcriptional program, which provided part of the groundwork for vertebrate morphogenesis and organogenesis, including the cardiovascular system (Wu et al., 2007). Calcineurin–NFAT signaling was first shown by Molkentin et al. (Molkentin et al., 1998) to be important in development of cardiac hypertrophy, and today extensive evidence exists showing that this intracellular signaling pathway is essential and is activated in pathological hypertrophy only (Wilkins et al., 2004). As the NFATc proteins are transcription factors, their activity can be measured by expression of target genes. RCAN1 is known to be a direct NFATc target in the heart (Rothermel et al., 2003; Oh et al., 2010). In this study we showed that HR hearts, which are bigger than LR hearts mainly because of increased hypertrophy of the compact myocardium, had almost threefold higher RCAN1 expression. Such an increase in RCAN1 expression indicates increased NFAT signaling and hence that the growth of HR hearts has a pathological character. Alternatively, NFAT activation in fish may serve a more physiological role given their adaptive hypertrophic potential. To our knowledge, this is the first time NFAT activation, well known in the hypertrophic mammalian heart, has been shown to occur in fish. Our results thus imply that NFAT activation in cardiac hypertrophy is evolutionarily conserved between mammals and fish, indicating that fish can serve as models with biomedical relevance.

In mammals, catecholamines (CAs) are important mediators of stress-related cardiac remodeling as they induce myocardial hypertrophy and cardiomyocyte death. In fish, though, the effects of CAs on the myocardium seem less pronounced (Tota et al., 2010). As it has been shown that LR fish respond to stress with higher plasma levels of the CA epinephrine than HR fish (Schjolden and Winberg, 2007), and the HR hearts were bigger than LR despite lower post-stress plasma CA, we propose that cortisol constitutes the main stress-induced influence on the trout heart.

We found a higher expression of the angiogenic gene VEGF in the HR ventricles compared with the LR ventricles, which likely reflects the increased vascularization necessary to support the thicker layer of compact myocardium (Cerra et al., 2004), as this is the muscular layer vascularized by coronary vessels. Increased vascularization was indeed extensive when HR ventricles were visually examined.

The fibrillar collagen network serves as an elastic element in the heart. Structural remodeling of this network is, however, associated with pathological conditions such as fibrosis (Abrahams et al., 1987; Caulfield and Bittner, 1988). Fibrosis is the dominant histological reaction to injury in the mammalian heart and it adversely increases tissue stiffness and impairs normal cardiac function (Schnitt et al., 1993; Weber et al., 1994). In fish, though, collagen remodeling has been postulated to play an important role in the maintenance of mechanical cardiac performance (Cerra et al., 2004; Icardo et al., 2005). We also found enhanced collagen deposition in the HR ventricles compared with the LR ventricles, as evaluated by AFOG staining. Still, as the collagen depositions were non-homogenously distributed, revealing fibrosis and disruption of muscle structure in certain areas (see Fig. 5A), we believe this may still represent pathological collagen production, similar to what is found during pathological focal fibrosis in mammals. In adaptive cardiac hypertrophy in trout, however, collagen gene expression was reduced approximately fivefold (Vornanen et al., 2005). Thus, a higher expression of COL1a2 and a trend towards higher COL1a1 in the HR hearts compared with the LR hearts, as observed in our study, indicates a different response than in adaptive cold-acclimation, and could be interpreted as related to pathological fibrosis.

Because expression of genes regulating hypertrophy and fibrosis could indicate pathological myocardial remodeling in the HR trout, we wanted to investigate the expression of ANP and BNP, which are commonly used markers of heart failure in mammals (Lerman et al., 1993; Maisel et al., 2002). Generally, little data is available concerning secretion of natriuretic peptides associated with cardiac growth and remodeling in non-mammalian species (Tota et al., 2010), and the regulation and/or function of ANP and BNP can be different in fish and mammals (Loretz and Pollina, 2000). For instance, the interplay between CA and natriuretic peptides seems to differ. As LR fish respond to stress with higher plasma CA (Schjolden and Winberg, 2007), this could have predicted differences in ANP and BNP production. Yet, the expression of ANP and BNP did not differ between LR and HR ventricles. Alternatively, assuming an upregulation of natriuretic peptides with heart failure in fish similar to that in mammals, our results indicate that the HR fish had not developed heart failure despite the indications of pathological remodeling.

A limiting factor determining the sensitivity of a cell to glucocorticoids is the intracellular concentration of glucocorticoid receptors (Vanderbilt et al., 1987; Dong et al., 1989). Hence, altering the expression of these receptors renders a tissue capable of adjusting the biological response according to the requirements of the environment. In rainbow trout, cortisol actions are mediated through three different glucocorticoid receptors, GR1, GR2 and MR (Colombe et al., 2000; Bury et al., 2003). In the HR ventricles, expression of all three receptors was significantly higher than in the smaller LR ventricles. Thus, it is probable that increased cortisol receptor expression in HR fish makes them more sensitive to the actions of cortisol and that cortisol-dependent intracellular signaling is increased. In addition, these fish experience higher post-stress cortisol levels regularly throughout life, at least in stressful environments such as intensive aquaculture. This assumption implies a role for cortisol not only in cardiac hypertrophy of fish, but also as one of the causative factors of pathological heart conditions in the fish farming industry.

In mammals, the mineralcorticoid aldosterone has been shown to induce fibrosis (Funder, 2001). Also, mineralcorticoid-induced cardiac hypertrophy is associated with increased NFAT activity through increased calcineurin activity (Takeda et al., 2002; Tong et al., 2004). Salmonid fish, however, lack aldosterone (Sangalang and Uthe, 1994) and the main corticosteroid is thought to be cortisol, acting as both a mineralcorticoid and a glucocorticoid hormone (Bern and Madsen, 1992; Wendelaar Bonga, 1997). Thus, cortisol in fish might mimic the effect of aldosterone on the mammalian heart and induce fibrosis and increased NFAT activity, as reflected by increased RCAN1 expression in the current study.

In our study, the HR–LR selection model provided a tool to study the effects of transiently elevated cortisol levels. However, the HR and LR rainbow trout were bred for several generations during the selection regime. Thus, founder effects, unique mutations or random genetic drift could in fact account for the observed divergence in CSI. Nevertheless, a positive correlation between cortisol responsiveness following stress and relative heart size was also revealed in wild-type European brown trout. This finding suggests an evolutionary conserved correlation between heart size and cortisol responsiveness and that this trait correlation is not an incidental artifact of the HR–LR selection regime. Indeed, a link between high cortisol responsiveness to stress and increased cardiovascular morbidity and mortality has also been reported in humans (Pedersen and Denollet, 2003).

In conclusion, high post-stress cortisol levels seem to be associated with cardiac remodeling and altered gene expression in salmonid fish. Although the significance of cortisol-related cardiac remodeling for the fish remains speculative, both hypertrophy (Lloyd-Jones et al., 2002) and high serum cortisol (Yamaji et al., 2009) have been identified as independent risk factors for adverse cardiac events in humans. Thus, a follow-up study should involve experiments to determine whether HR fish have impaired cardiac function and are more prone to stress-induced mortality. In aquaculture there is increasing worry about stress-induced mortality of fish carrying cardiac anomalies (Brocklebank and Raverty, 2002; Poppe et al., 2007); we suggest that cortisol-induced cardiac myocardial remodeling may be one of the explanatory factors.

 
• AFOG

  acid-fuchsin-orange G

 
• ANP

  A-type natriuretic peptide

 
• BNP

  B-type natriuretic peptide

 
• CA

  catecholamine

 
• COL1a1

  collagen alpha 1(1)

 
• COL1a2

  collagen alpha 2(1)

 
• CRP

  cysteine rich protein

 
• CSI

  cardiosomatic index

 
• GR1

  glucocorticoid receptor 1

 
• GR2

  glucocorticoid receptor 2

 
• HR

  high responder

 
• LR

  low responder

 
• MCIP

  modulatory calcineurin-interacting protein 1

 
• MLP

  muscle LIM protein

 
• MR

  mineralcorticoid receptor

 
• NFAT

  nuclear factor of activated T-cell

 
• PCNA

  proliferating cell nuclear antigen

 
• PFA

  paraformaldehyde

 
• qRT-PCR

  quantitative real-time PCR

 
• RCAN1

  regulator of calcineurin 1

 
• RIA

  Radioimmunoassay

 
• SMLC2

  slow myosin light chain 2

 
• VEGF

  vascular endothelial growth factor

 
• VMHC

  ventricular myosin heavy chain

 
• β-MHC

  β-myosin heavy chain

We are thankful to Tom Pottinger who donated the HR and LR fish, and to Jenny Carolyn Shaw who tagged the fish. We thank Emir Habibovic and Samir Muhamedagic at the Faculty of Agriculture and Food Sciences, University of Sarajevo, Bosnia and Herzegovina, for rearing the brown trout and assisting in the experimental procedures and sampling. Furthermore, we thank Torstein Lyberg at the Center for Clinical Research and Tove Norén at Department of Pathology, Oslo University Hospital-Ullevaal, Norway, for histochemical staining of the fish hearts, and Christina Sørensen at the Institute for Molecular Biosciences, University of Oslo, Norway, for the cortisol RIA protocol. We thank Guro Sandvik at the Institute for Molecular Biosciences, University of Oslo, for assistance in primer design. We are grateful for comments on the manuscript from Dr Jonathan Stecyk.