Origins of variation in muscle cytochrome c oxidase activity within and between fish species

The ability to adjust mitochondrial oxidative capacity is critical to controlling energy production under changing physiological and environmental conditions. In mammals, mitochondrial content changes in response to factors such as exercise (Holloszy, 1967), electrical stimulation (Baar et al., 2002) and cold exposure (Puigserver et al., 1998; Wu et al., 1999). Fish respond in a similar manner to activity (e.g. Farrell et al., 1991), but they also induce mitochondrial proliferation in response to cold acclimation and winter acclimatization (Egginton and Sidell, 1989; Egginton et al., 2000; Guderley, 1990). Unlike the situation in mammals, where cold exposure induces hypermetabolism, cold treatment of fish depresses metabolism, leading to a suite of morphological, physiological, biochemical and genetic modifications (Egginton and Johnston, 1984; Johnston and Maitland, 1980; Johnston and Wokoma, 1986; Orczewska et al., 2010; Shaklee et al., 1977). The underlying cause of cold-induced mitochondrial proliferation remains elusive but it is thought to reflect a degree of thermal compensation to overcome the negative thermodynamic effects on metabolism and metabolic enzymes (Egginton and Sidell, 1989; Hazel and Prosser, 1974).

Regardless of the reason why fish remodel mitochondrial energy metabolism in the cold, the question of how such a change is achieved remains. Mitochondrial biogenesis is a complex process requiring exquisite coordination of nuclear and mitochondrial genes. Each oxidative phosphorylation (OXPHOS) complex is composed of multiple subunits, many of which have paralogues. All mitochondrial OXPHOS complexes, except complex II, have subunits encoded by both nuclear genes and mitochondrial DNA (mtDNA). Transcriptional regulation plays a very important role in the control of mitochondrial biogenesis (Hock and Kralli, 2009). As described in recent reviews (Hock and Kralli, 2009; Scarpulla, 2010), central to this pathway are three transcription factors that have been identified as master regulators of mitochondrial biogenesis: nuclear respiratory factor 1 (NRF-1), NRF-2 and peroxisome proliferator-activated receptor-γ coactivator-1α (PGC-1α). The two NRFs are DNA-binding proteins that associate with elements in the promoters of many genes that encode mitochondrial proteins. Once localized at the promoter, they directly (NRF-1) or indirectly (NRF-2) bind PGC-1α, forming a complex that recruits the general transcription machinery, activating transcription. The coactivator PGC-1α is the most important member of the gene family including two homologues, PGC-1β and PGC-1-related coactivator (PRC), which have overlapping roles with functional specialization (Hock and Kralli, 2009; Scarpulla, 2010).

Each of these transcription factors that play a role in mammalian mitochondrial biogenesis is also present in fish, but there is some uncertainty about their respective roles in lower vertebrates. In a study on thermal acclimation of goldfish (Carassius auratus), cold induced increases in COX activity, COX4 mRNA and NRF-1 mRNA, but PGC-1α mRNA was decreased (LeMoine et al., 2008). Likewise, in three-spined stickleback (Gasterosteus aculeatus), cold acclimation induced increases in COX, COX4 mRNA and NRF-1 mRNA but not PGC-1α mRNA (Orczewska et al., 2010). The lack of a uniform pattern of correlation between PGC-1α and mitochondrial changes prompted the suggestion that the pathway of regulation in fish may differ not just from that of mammals but also between species. The diminished importance of PGC-1α may be due to mutations in the NRF-1 binding domain (LeMoine et al., 2010) and it has been suggested that PGC-1β may be more important in the control of mitochondrial biogenesis in some fish (LeMoine et al., 2008).

In this study, our first goal was to investigate the patterns seen in the seasonal response of a broad range of ecologically diverse freshwater fish species. Secondly, we wanted to better understand the regulatory mechanisms that underlie the compensatory effects in mitochondrial enzyme content in lower vertebrates compared with those in higher vertebrates.

Eight local fish species were caught in Lake Opinicon and its nearby marshlands, about 50 km north of Kingston, ON, Canada. Lake Opinicon is a large (787 ha, or 7.87 km²), but relatively shallow (∼50% is <5 m; maximum depth <11 m) and wind-exposed lake (Keast and Fox, 1992). Because of these physical conditions, it is classified as a polymictic lake (Agbeti and Smol, 1995), with water mixing in summer as well as spring and autumn, thereby reducing thermal variation regionally (Crowder et al., 1977; Keast and Fox, 1992). The four sunfish (Centrarchidae) representatives, pumpkinseed (Lepomis gibbosus, Linnaeus), bluegill (Lepomis macrochirus, Rafinesque), black crappie (Pomoxis nigromaculatus, Lesueur) and largemouth bass (Micropterus salmoides, Lacépède), as well as northern pike (Esox lucius, Linnaeus) were caught by hook and line. Brook stickleback (Culaea inconstans, Kirtland), northern redbelly dace (Chrosomus eos, Cope) and central mudminnow (Umbra limi, Kirtland) were caught in minnow traps in the marsh around the lake. No longer than 1 h after being caught, animals were anaesthetized in buffered (0.8 g l^–1 NaHCO₃) 0.4 g l^–1 tricaine methanesulphonate and killed by cutting the spinal cord. White muscle samples were taken from the epaxial muscle below the dorsal fin but above the lateral line and immediately frozen in liquid nitrogen. Fish were caught in summer 2008 and 2009 and in winter/spring 2009/2010 (Fig. 1). Morphometric data including mass, length and condition factor of the fish are summarized in Table 1. As these fish were sampled from their natural environment, experiencing the whole range of seasonally changing stimuli, it is not surprising that some of them showed significant differences in their condition factors between summer and winter.

Mudminnows, caught in the marsh area around Lake Opinicon during summer 2008, were held in a 723 l tank attached to a flow-through system in the Animal Care facilities of Queen's University, Kingston. Water was supplied by the city system taken from Lake Ontario and filtered through a carbon filter prior to serving the fish tank. Fish were acclimated to the ambient water temperature of Lake Ontario for 4 weeks before the first sampling event. During the beginning of the experiment, water temperatures of Lake Ontario were the same as those in the marshlands around Lake Opinicon, but decreased at a slightly lower rate than field water temperatures throughout the experiment. Samples were taken every 27–42 days (Fig. 2A). Acclimation temperatures were defined as the mean temperature over the 3 weeks prior to each sampling event (Fig. 2A).

For enzyme extraction, white muscle samples were powdered under liquid nitrogen. About 50 mg was homogenized in 20 volumes of cold extraction buffer (25 mmol l^–1 K₂HPO₄, 1 mmol l^–1 EDTA, 0.6 mmol l^–1 lauryl maltoside, pH 7.4) using a 7 ml Tenbroeck tissue grinder (Wheaton Industries, Millville, NJ, USA). Homogenates were not centrifuged and were mixed prior to enzyme measurements. Enzyme activity was determined spectrophotometrically (Molecular Devices, Sunnyvale, CA, USA) at 550 nm and 25°C in 96-well plates (Corning, Corning, NY, USA) using assay buffer (25 mmol l^–1 K₂HPO₄, 0.6 mmol l^–1 lauryl maltoside, pH 7.4) and reduced cytochrome c (0.05 mmol l^–1) as the substrate. Cytochrome c was reduced by ascorbic acid, then dialysed exhaustively in 25 mmol l^–1 K₂HPO₄ (pH 7.4) and frozen in aliquots. All samples were measured in triplicate.

RNA was isolated from powdered white muscle samples using a slight modification of the single-step method by guanidinium thiocyanate–phenol–chloroform extraction (Chomczynski and Sacchi, 2006). The purified RNA pellet was dissolved in diethylpyrocarbonate-treated water and quantified at 260 nm prior to storage at –80°C. Additionally, absorption at 230, 270 and 280 nm was measured to test salt, phenol and protein contamination, respectively. Samples showed a 260 nm/230 nm ratio of 1.86±0.2, a 260 nm/270 nm ratio of 1.43±0.0 and a 260 nm/280 nm ratio of 2.14±0.1. Reverse transcription of RNA and the removal of genomic DNA were carried out using the QuantiTect Reverse Transcription Kit (Qiagen, Mississauga, ON, Canada) according to the manufacturer's instructions.

All analyses were performed on an ABI 7500 Real Time PCR System (Foster City, CA, USA) using the following protocol: 10 min at 95°C followed by 40 cycles of 15 s at 95°C, 15 s at the annealing temperature (Table 2), 34 s at 72°C. The efficiency of each primer set for every species was determined by real-time PCR with an appropriate dilution series of cDNA prior to the sample analyses. Based upon these results, an appropriate cDNA concentration for each reaction was chosen. Samples were then assayed in duplicate in a 25 μl total reaction volume containing 5 μl cDNA [100 ng of NRF-1, NRF-2α, PGC-1α, TATA-binding protein (TBP) and 18S; 50 ng β-actin], 12.5 μl FastStart Universal SYBR Green Master mix (Roche Applied Science, Penzberg, Bavaria, Germany), 3.5 μl double-distilled H₂O and 2 μl each forward and reverse primer (final concentration, 0.58 μmol l^–1). Controls were run with water to ensure the absence of contamination. Results were analysed according to the ΔCt (cycle threshold) method using TBP, β-actin and 18S ribosomal RNA as housekeeping genes with their calculated geometric mean for each sample as standardized Ct (Pfaffl et al., 2004). Specific primers were designed to amplify single products of 91–175 bp in length (Table 2). Species-specific real-time PCR primers were constructed after larger amplicons were generated from consensus primers, cloned and sequenced.

All data are presented as means + s.e.m. for temperature comparison within one species and are plotted relative to summer values. The significance of differences between the two seasons for each species was tested by the non-parametric Mann–Whitney U-test for enzyme activity, relative mRNA levels of target genes and the morphometric data. The significance of differences between multiple groups in the acclimation experiment was tested by the Kruskal–Wallis test. Results were tested at a significance level of 95%.

White muscle COX activity was not affected by season in three out of eight species: bluegill, bass and pike. For simplicity, we term these species non-compensators. The remaining five species (pumpkinseed, crappie, dace, stickleback and mudminnow) showed significant increases (1.3- to 3.5-fold) in white muscle COX activity in winter (Fig. 3). These species were denoted as compensators.

There was no phylogenetic pattern that distinguished compensators and non-compensators. Within centrarchids, pumpkinseed and crappie compensated whereas bluegill and bass did not. Within Esociformes, mudminnow compensated but pike did not show thermal compensation.

Mudminnows were held at seasonal temperatures from autumn through to winter. Mean temperatures dropped by 14°C, from 19°C at the beginning to 5°C at the end of the experiment (Fig. 2A). In contrast to acclimatized fish, the lab-acclimated fish showed no significant changes in COX activity (Fig. 2B).

The responses of each species in terms of transcription factor mRNA levels are depicted in Fig. 4. A previous study on goldfish acclimated to 5 and 20°C (LeMoine et al., 2008) found that cold acclimation was accompanied by an increase in COX activity and NRF-1 mRNA but a decrease in PGC-1α mRNA (Fig. 4A). In extending this approach to eight other species, there was no clear relationship between COX changes and the patterns seen in NRF-1, NRF-2α and PGC-1α mRNA levels. Five species showed thermal compensation in COX activity: dace, stickleback, pumpkinseed, crappie and mudminnow.

Among the compensators, NRF-1 mRNA increased in pumpkinseed (3-fold) and crappie (5-fold) by a factor that was much greater than the increase seen in COX activity. In contrast, no change was seen in NRF-1 mRNA in dace, stickleback or mudminnow (Fig. 4), each of which showed a greater change in COX activity than did pumpkinseed and crappie. The level of mRNA for NRF-2α, a subunit of NRF2, did not change in any compensator, except for crappie, which showed a 2-fold increase in the cold (Fig. 4).

As with goldfish, PGC-1α mRNA decreased significantly in three compensators: stickleback (3-fold), pumpkinseed (10-fold) and mudminnow (2-fold). There was no significant change in PGC-1α mRNA in crappie. In contrast to these species, dace PGC-1α mRNA increased significantly (7.5-fold) in winter fish (Fig. 4).

As noted above, there were three species of fish that failed to show a seasonal response in COX activity (Fig. 3). In these species, NRF1 mRNA did not change in pike, but significantly increased in bluegill (5-fold) and bass (9-fold). NRF-2α mRNA increased in bass (3-fold) but did not change in bluegill or pike. PGC-1α mRNA did not change in any of these species between seasons (Fig. 4).

Low body temperatures lead to pleiotropic effects on metabolism and physiology, and long-term responses to temperature result in many physiological and morphological changes in ectotherms. A cold-induced increase in mitochondrial content, whether expressed in terms of morphometry (e.g. volume density, cristae density) or marker enzyme activity (e.g. COX, citrate synthase), probably corrects shortfalls in ATP production in the cold (Guderley, 2004). In mammals, the regulation of mitochondrial biogenesis is attributed to a set of master regulators of transcription that coordinate the expression of genes encoding mitochondrial proteins (Hock and Kralli, 2009). However, a survey of the literature on fish models suggests that the process in fish may differ from that seen in mammals (LeMoine et al., 2008). In this study, we focused on the expression patterns seen in white muscle mitochondria of a variety of species found regionally, exploring the underlying control of these patterns.

Cold-induced changes in mitochondrial markers, first shown in goldfish (Freed, 1965), have since been shown in many other ectotherms – crustaceans, amphibians, reptiles and annelids (Bullock, 1955; Heinrich, 1977; Sommer and Pörtner, 2004). In contrast, some temperate species fail to change mitochondrial enzyme content in relation to temperature or season. The reasons for the different responses among studies and species are not clear, nor are the mechanisms by which the changes arise.

The variation in response among species does not appear to have a simple taxonomic basis. As in a previous study on centrarchids (Tschantz et al., 2002), we saw increases in COX activity in some species (pumpkinseed, 2-fold; crappie, 1.3-fold) but not others (bluegill, bass). We found a striking difference in the response of two esociforms: northern pike and mudminnow. The lack of response in pike is similar to the situation in chain pickerel (Esox niger), which show a very modest response to cold acclimation (20% increase in COX activity) and no response to winter acclimatization (Kleckner and Sidell, 1985). The closest relatives to Esociformes are the Salmoniformes, which typically experience only a mild response to the cold, generally less than a 75% increase in COX activity (Battersby and Moyes, 1998; Blier and Guderley, 1988; Egginton et al., 2000; Guderley and Gawlicka, 1992; Guderley et al., 1997). Thus, the dramatic response of mudminnows is atypical within this broader group.

Other taxonomic groups appear to have a more uniform response to the cold. Sticklebacks (Gasterosteidae) all appear to show a pronounced increase in mitochondrial enzyme activities in the cold: brook stickleback (Fig. 3), three-spine stickleback (Orczewska et al., 2010; Vézina and Guderley, 1991) and nine-spine stickleback (Guderley and Foley, 1990). Cyprinids (carp and true minnows) also show increases in muscle mitochondrial enzyme activities: dace (Fig. 3), goldfish (Freed, 1965; LeMoine et al., 2008), roach (Rutilus rutilus) and tench (Tinca tinca) (Heap et al., 1985). The lack of a thermal response seen in tropical zebrafish (Duggan et al., 2011; McClelland et al., 2006) is perhaps not surprising as the lower lethal temperature (7°C) for this species (Lawrence, 2007) lies above the temperature at which thermal compensation is usually seen (<5°C).

Collectively, these studies suggest that there are few phylogenetic patterns that could indicate which species change their muscle mitochondrial enzymes with temperature. Differences are seen in species that are closely related and even show similar locomotor patterns. Both mudminnow and pike are ambush predators whereas both bluegill and pumpkinseed swim in a labriform mode, yet these species pairs differ in their responses to season.

One challenge in reconciling our results with those from previous studies is related to the nature of the thermal challenge and confounding effects of other seasonal factors that accompany the cold. Independent studies on the responses of rainbow trout (Oncorhynchus mykiss) to acclimatization and acclimation show a similar compensatory response in muscle oxidative capacity (Battersby and Moyes, 1998; Egginton et al., 2000; Guderley and Gawlika, 1992; Thibault et al., 1997). However, there are studies where different responses in acclimation versus acclimatization have been found. For example, we found winter acclimatization to increase muscle mitochondrial enzyme activity in black crappie, whereas thermal acclimation had no effect in another study (Tschantz et al., 2002).

The disparity between acclimation and acclimatization is most compelling when the same researchers compared the two treatments. For example, chain pickerel (Kleckner and Sidell, 1985) and three-spine stickleback (Vézina and Guderley, 1991) have elevated mitochondrial enzyme activity after cold acclimation but the differences diminish when fish are winter acclimatized. We found a marked difference in the response of mudminnow to lab acclimation versus field acclimatization. In lab fish, we saw no changes in muscle COX activity even after 35 days at 5°C (Fig. 2B). However, mudminnows captured from the field in early April showed a 3-fold increase in muscle COX activity relative to that of summer fish. Thus, it appears unlikely that the seasonal effects we see in this species are due to simple effects of temperature.

Apart from temperature, the metabolic phenotype could be affected by many of the factors that differ between seasons. Photoperiod can induce modifications in mitochondria in both Atlantic cod (Pelletier et al., 1993) and rainbow trout (Martin et al., 2009). Oxygen is well known to affect the glycolytic phenotype (Chippari-Gomes et al., 2005; Jørgensen and Mustafa, 1980; Lushchak et al., 1998), though its effect on mitochondrial enzymes is still debated (Hoppeler et al., 2003; Klimova and Chandel, 2008). Diet can have complex effects on metabolism, though the effects are most commonly seen in liver (de la Higuera et al., 1999; Hemre et al., 2002; LeMoine et al., 2008; Menoyo et al., 2004). Activity levels change during the season, and these may induce the equivalent of a training response. The significant decrease in the condition factor of the lab-acclimated fish (Table 1) did not have an impact on enzyme activity, nor was the significant increase in enzyme activity in the winter-acclimatized fish mirrored in the condition factors. Reproductive status may also affect muscle mitochondria, through hormone levels or locomotor behaviours, including migrations, mating, spawning activity and territory defence. In our study, stickleback, mudminnow and dace each showed an increase in muscle COX activity. These species begin spawning activity in spring (Abbott, 1870; Peckham and Dineen, 1957; Reisman and Cade, 1967), which includes spawning migrations, nest building and guarding as well as active mating behaviour. A combination of factors triggers spawning in fish, primarily temperature and photoperiod (Lam, 1983). Changes in the latter could play an important role in our observations of mudminnow and dace. Spawning itself induces mitochondrial modifications in rainbow trout (Kiessling et al., 1995).

In summary, this study and others demonstrate a wide range of patterns in the response of muscle aerobic enzymes to thermal history. We have shown evidence that one source of variation is a distinction between thermal acclimation and seasonal acclimatization. Nonetheless, even closely related species with very similar acclimatization histories can have divergent patterns of metabolic remodelling.

Whether or not a species changes muscle mitochondrial content, the levels of mitochondria are controlled by genetic mechanisms affecting the transcription of mitochondrial genes, including those encoding proteins of mtDNA replication and transcription, and protein import and assembly (Hock and Kralli, 2009). In mammals, the cascade from transcriptional master regulators to mitochondrial gene expression and mitochondrial biogenesis is well understood (Hock and Kralli, 2009; Scarpulla, 2010). The role of a master regulator of mitochondrial biogenesis was first ascribed to NRF-1 (Evans and Scarpulla, 1990); however, more recent studies support a greater role for coactivator PGC-1α (Scarpulla, 2010). Although a few studies have assessed the potential role of these factors in adaptive remodelling in fish, the extent of the parallel with mammals is not yet clear. In particular, there is little correlation between COX activity and PGC-1α mRNA level in different muscles, dietary states, activity levels and temperature treatments (LeMoine et al., 2008; McClelland et al., 2006; Orczewska et al., 2010). This prompted the suggestion that the PGC-1α paralogue PGC-1β may be the major coactivator for mitochondrial gene expression control in fish (LeMoine et al., 2008). Though the role for PGC-1 paralogues may differ in fish versus mammals, these studies have found stronger support for a role for NRF-1 as master regulator.

One goal of our study was to broaden the species comparisons to better clarify the potential role of PGC-1α in controlling mitochondrial changes with temperature and/or season. Consistent with previous studies on goldfish (LeMoine et al., 2008) and stickleback (Orczewska et al., 2010), we found in four species a cold-induced increase in COX activity but no increase in PGC-1α mRNA level; three of these species (brook stickleback, pumpkinseed and mudminnow) showed a marked decrease in PGC-1α mRNA level. Only in dace was an increase in COX activity (3-fold) accompanied by an increase in PGC-1α mRNA level (7-fold), which is what is expected from mammalian models (Scarpulla, 2010). The large variation in PGC-1α mRNA between species could be related to its role in other cellular processes, including fatty acid oxidation (Vega et al., 2000) and muscle fibre specification (Lin et al., 2002). Alternatively, changes in PGC-1α activity may be related to the coactivation of other DNA-binding proteins, including thyroid hormone receptors, peroxisome proliferator-activated receptors (PPARs), retinoic acid receptors and estrogen-related receptors (Puigserver and Spiegelman, 2003; Schreiber et al., 2003). Thus, even if PGC-1α plays a role in mitochondrial biogenesis in response to temperature, its regulation could be complicated by simultaneous changes in diet, fuel selection, activity level and reproduction (Dean, 1969; Miranda and Pugh, 1997; Suski and Ridgway, 2009). Nonetheless, there is little evidence from fish studies to support a role of PGC-1α in determining changes in COX activity with temperature.

Mammalian studies have shown NRF-1 to be an important regulator of all nuclear-encoded COX genes (Dhar et al., 2008). It acts by binding to NRF-1 promoter elements (Virbasius et al., 1993), where it recruits PGC-1α (Scarpulla, 2010) The interplay between NRF-1 and PGC-1α is an important regulatory axis in mammalian muscle. Furthermore, transcriptional regulation of NRF-1 appears central to the control of its activity (Hock and Kralli, 2009). Previous studies on fish have shown that increases in NRF-1 mRNA accompany increases in mitochondrial enzyme activities seen with diet, temperature and exercise (LeMoine et al., 2008; McClelland et al., 2006; Orczewska et al., 2010). In this study, pumpkinseed and crappie both showed a significant increase in COX activity and NRF-1 mRNA level. However, the pattern was not consistent across species. Dace, stickleback and mudminnow each achieved an increase in COX activity without a corresponding increase in NRF-1 mRNA. Furthermore, two non-compensators (bluegill and bass) showed significant increases in NRF-1 mRNA level with no change in COX activity. Despite this variation, NRF-1 appears to be the most likely candidate for a role as master regulator of mitochondrial biogenesis in fish. In mammals, post-transcriptional regulation of NRF-1 can play a role in determining its transcriptional activity (Hock and Kralli, 2009), so perhaps this level of regulation is more important in fish.

Like NRF-1, NRF-2 has an important role in the regulation of all nuclear-encoded COX genes in mammals (Ongwijitwat et al., 2006). Its regulation is more complex because the active transcription factor is a heterodimer of NRF-2α and either NRF-2β or NRF-2γ (Gugneja et al., 1995). There is also clear evidence for post-transcriptional regulation of NRF-2 (Vallejo et al., 2000), making measurement of mRNA less informative. Furthermore, apart from its role in the control of mitochondrial genes, it regulates diverse cellular processes (Imaki et al., 2003; Ristevski et al., 2004; Yang et al., 2007). The role of NRF-2 has been less studied in fish but there is little evidence from our study of a role in the thermal remodelling of muscle. Only two species showed seasonal changes in NRF-2α mRNA; its level increased in crappie, paralleling changes in COX activity, and bass, which showed no change in COX activity.

The present study has shown that there is no phylogenetic pattern in the response of COX content in muscle to cold acclimation and/or seasonal acclimatization (Table 3). Changes seen seasonally in free-living fish may be confounded by non-thermal cues, such as reproductive status. Furthermore, the role of transcriptional master regulators of mitochondrial biogenesis appears to be more complex (Table 3) than is expected from mammalian studies. The transcriptional factors themselves may act in fundamentally different ways, and there is a potential role for post-transcriptional regulation.

We would like to thank the staff of the Queen's University Biological Station for their support of these studies; in particular, Frank Phelan and Bruce Tufts for their fishing expertise as well as students Sarah Wyness, Katrinka Kocha, Trevor Snider, Christopher Monk, Paul Debes and Scott Taylor for their help in the field.