Temperature during embryonic development has persistent effects on metabolic enzymes in the muscle of zebrafish

Environmental temperature has a profound influence on the biological function of ectothermic organisms. The survival and performance of populations are limited to a finite range of temperatures, the breadth of which is a crucial determinant of distribution and abundance (Angilleta, 2009; Schulte et al., 2011). Thermal optima and performance breadth can be altered by temperature acclimatization, and although the underlying mechanisms have been studied for decades (Fry and Hart, 1948), research interest has intensified of late in an effort to understand the potential ecological impacts of global climate change (Pörtner and Farrell, 2008; Somero, 2011). However, much of what we know about the effects of temperature on ectothermic vertebrates comes from studies of juveniles and adults. There has been relatively little emphasis on how temperature affects early life stages and how temperature can interact with the developmental programme to alter phenotypic plasticity in later life (Beldade et al., 2011; Overgaard et al., 2011).

Adjustments in the metabolic pathways that dictate energy production are important for temperature acclimation in adult ectotherms. Cold acclimation increases mitochondrial abundance and/or oxidative capacity in muscle and other tissues in a variety of fish species (e.g. Johnston and Maitland, 1980; Egginton and Sidell, 1989; McClelland et al., 2006; Orczewska et al., 2010). This response is thought to improve cellular function by counteracting the kinetic effects of temperature on the catalytic rates of mitochondrial enzymes or on the rates of oxygen/metabolite diffusion (Hubley et al., 1997; Guderley, 2004; Kinsey et al., 2011; O'Brien, 2011). However, the presence and magnitude of metabolic temperature acclimation in adult fish depends on the magnitude of temperature change and can vary between populations and species (Bremer and Moyes, 2011; Dhillon and Schulte, 2011).

Our recent work suggests that the thermal environment experienced in early life can have a profound influence on exercise performance (Scott and Johnston, 2012). We found that embryonic temperature (T_E) affected prolonged swimming performance (critical swimming speed, U_crit) in adult zebrafish in response to a short-term thermal challenge, such that fish raised at a particular T_E had reduced thermal sensitivity at its respective T_E. Similarly, U_crit after a prolonged period of warm acclimation was higher in fish raised at warmer T_E than in those raised at a cold T_E. The overall effects of developmental temperature on performance at warm temperatures therefore appear to be consistent with the beneficial acclimation hypothesis (Huey et al., 1999). In contrast, U_crit after cold acclimation was highest in fish raised at low and high T_E, and lowest in fish raised at intermediate T_E. The effect of T_E on cold acclimation was based on substantial variation in the expression of several metabolic genes in the swimming muscle (detected using high-throughput mRNA sequencing, RNA-Seq), but the mechanisms of the observed effects of T_E on short-term thermal plasticity and warm acclimation capacity was unclear.

The objectives of this study were to determine: (1) the thermal breadth of embryonic development in zebrafish [Danio rerio (Hamilton 1822)] and (2) how developing at the extremes of this range affects the capacity of energy metabolism pathways and the temperature acclimation response in the swimming muscle of adult fish. Zebrafish are well-suited to this study for a number of reasons. Zebrafish are native to slow-flowing rivers and streams in the Ganges and Brahmaputra River basins in southern Asia, where daily temperatures typically fluctuate by ~5°C and can range from as low as 6°C in winter to as high as ~38°C in summer (Spence et al., 2008;

López-Olmeda and Sánchez-Vázquez, 2011). Natural populations typically reproduce during the monsoon season, when food is readily available and temperatures range from ~23 to 31°C (Spence et al., 2008). We tested two hypotheses that could explain our previous observations on the effects of embryonic temperature on swimming performance (Scott and Johnston, 2012) based on differences in muscle enzyme function: (1) that embryonic temperature would shift the thermal optima of metabolic enzyme activity towards the T_E, and (2) that development at a low embryonic temperature would blunt the warm acclimation response of metabolic enzymes. We focused our efforts on enzymes that reflect the flux capacity of oxidative phosphorylation [cytochrome c oxidase (COX)], the citric acid cycle [citrate synthase (CS)], glycolysis [pyruvate kinase (PK) and lactate dehydrogenase (LDH)] and fatty acid oxidation [hydroxyacyl-coA dehydrogenase (HOAD)].

Embryonic development was successful from 22 to 32°C, as reflected by high proportional survival to hatching (≥88%) (Fig. 1A). Survival declined to ~43% at 34°C, and was negligible above 34°C and below 22°C. Temperature had a strong effect on the rate of embryonic development (Fig. 1A), which increased 2.33-fold from 22 to 32°C (i.e. Q₁₀=2.33). Developmental rate was highest from 32 to 34°C, beyond which the rate dropped.

Effects of T_E on growth persisted into later life, even though larvae were transferred to a common 27°C at hatching. Compared with fish raised at 27°C as embryos, fish raised at 22°C tended to be larger (as reflected by standard body length) and fish raised at 32°C tended to be smaller at 8 and 12 weeks after hatching (Fig. 1B). There were comparable differences in standard length and body mass in the separate group of fish that were sampled for enzyme measurements at 3 months after fertilization (Table 1) (see below). Standard length changed very little after 12 weeks of age in the 22°C and 27°C T_E groups, but the 32°C T_E group still exhibited significant growth beyond this age.

We examined the effects of T_E values that represent the cold (22°C) and warm (32°C) extremes for normal development, relative to near-optimal thermal conditions (27°C). T_E had persistent effects on the capacity of energy metabolism pathways in the muscle of adult zebrafish that were raised at a common 27°C after hatching, as reflected by differences in maximal enzyme activities assayed at 27°C (Fig. 2A). The maximal activities of COX and PK were significantly higher in both the 22°C and 32°C T_E groups than in the 27°C T_Econtrols. The activities of CS and HOAD were higher in the 22°C T_E group than in the 27°C and 32°C T_E groups. However, there were differences in body mass and length between embryonic temperature groups (Fig. 1B, Table 1), and we observed a significant positive effect of body mass on the activities of all enzymes except for PK (Fig. 2B, Table 2). Therefore, the residuals from these regressions were also used to compare T_E groups, the results of which suggested that the differences in COX and PK activity between T_E groups were independent of variation in body mass (Table 3). In contrast, there was no significant variation in body mass residuals for CS and HOAD, suggesting that the differences between T_E groups for these enzymes were caused largely by variation in body mass.

T_E also had persistent effects on the temperature optimum (T_opt) for COX activity in fish that were raised at 27°C and acclimated to that temperature as adults. T_opt for maximal COX activity was 27°C in the 22°C T_E group but was 32°C in the 32°C T_E group, whereas activity in the 27°C T_E controls was similar across all assay temperatures (Fig. 3). In contrast, there were no distinct T_opt detected for the other enzymes assayed, as activity generally increased progressively with warming across the full range of assay temperatures (22, 27, 32 and 36°C). T_E also affected the substrate affinity (K_m) of COX activity, as the 22°C and 32°C T_E groups had lower K_m for reduced cytochrome c when assayed at 22°C than the 27°C T_E controls (Fig. 4).

T_E had persistent effects on the thermal acclimation response for all enzymes assayed. We did not use a residual approach to correct for body mass in the following comparisons with T_E groups, because there were no significant effects of acclimation temperature on body mass (Table 1). Warm acclimation at 32°C reduced maximal COX activity in the 22°C and 32°C T_E groups, but did not have a significant effect in the 27°C T_E controls (Fig. 3). COX K_m was not significantly affected by acclimation temperature, but it generally declined with increasing assay temperature (Table 4). Similar effects of T_E were observed for the acclimation response of maximal PK activity, as acclimation to 32°C reduced PK activity in the 22°C and 32°C T_E groups, but had no effect (trending towards an opposite effect) in the 27°C T_E controls (Fig. 5). Maximal CS activity was also reduced by acclimation to 32°C in the 22°C T_E group, but CS activity was not significantly affected by acclimation in the 27°C and 32°C T_E groups (Fig. 6).

In contrast to the above enzymes, for which the predominant effect of 32°C acclimation was a greater response in the 22°C and/or 32°C T_E groups, LDH activity was only affected by warm acclimation in the 27°C T_E controls. Maximal LDH activity increased substantially after acclimation to 32°C in the 27°C T_E controls, particularly at warmer assay temperatures, but was unaffected by acclimation in the 22°C and 32°C T_E groups (Fig. 7). This was not associated with any effects of T_E or acclimation temperature on the substrate affinity of LDH for pyruvate (Table 5). Maximal HOAD activity also increased in the 27°C T_E controls after acclimation to 32°C, but was not affected by acclimation in the 22°C and 32°C T_E groups (Fig. 8).

Thermal sensitivities of maximal enzyme activity, as reflected by temperature coefficients (Q₁₀), were not generally affected by T_E (Fig. 9). The exceptions were significant effects of T_E on Q₁₀ for COX and PK in zebrafish acclimated to 27°C. In this regard, COX activity was less sensitive to cooler temperatures in the 27°C T_E controls than in the 22°C and 32°C T_E groups, concurrent with the lack of a definitive T_opt in the former group (Fig. 3). Interestingly, acclimation to 32°C increased the Q₁₀ for HOAD activity between assay temperatures of 22 and 27°C, suggesting that warm acclimation increased the sensitivity of the HOAD enzyme to cooler temperatures (bottom panels of Fig. 9).

Variation in muscle protein content (which was 39.0±0.5 mg protein per gram tissue overall) did not contribute to any of the results discussed above. There were no significant effects of T_E (F_2,54=0.99, P=0.38), acclimation temperature (F_1,54=3.60, P=0.06) or their interaction (F_2,54=0.89, P=0.42) in two-factor ANOVA. There was a slight tendency for muscle protein to be lower in 32°C-acclimated fish (38.2±0.6 mg g⁻¹ overall) than in 27°C-acclimated fish (39.9±0.7 mg g⁻¹ overall), but the magnitude of this non-significant effect was much lower than the significant effects of acclimation on the activities of many enzymes.

Our present findings suggest that the thermal conditions experienced during embryonic development have effects that persist into adulthood on the metabolic capacities of the muscle, the thermal optima of enzymes, and the thermal acclimation response. Zebrafish raised near the cold or warm extremes for successful development had higher COX and PK activities than fish raised at an intermediate (‘optimal’) temperature. T_E also had a positive association with the T_opt of COX. The acclimation response of mitochondrial oxidative capacity also appeared to be enhanced by development near the cold and/or warm extremes of embryonic survival, as reflected by stronger inverse relationships between acclimation temperature and the activities of mitochondrial (COX, CS) and glycolytic (PK) enzymes. The thermal environment experienced during early life may therefore have a strong influence on the physiological response to warming temperatures in later life.

Our finding that T_E had a persistent effect on the thermal optimum of COX activity, even after all fish were raised for several months after hatching at 27°C, supports our first hypothesis. Enzyme activity is generally expected to be exponentially related to assay temperature (according to the Arrhenius equation) if the effects of temperature on enzyme activity are strictly due to thermodynamic effects on molecular movement (Schulte et al., 2011). This is somewhat reflected in the present study, insofar as the activity of most enzymes increased with increasing assay temperature. However, this was not the case for COX. The activity of this enzyme varied little across temperatures, differing by at most 30% across a 14°C range of assay temperatures and generally peaking between 27 and 32°C (Figs 3, 9). Similar patterns have been observed in some fish, amphibian and reptile species (Godiksen and Jessen, 2001; Glanville and Seebacher, 2006; Niehaus et al., 2011), but not in others (Caldwell, 1969; Blier and Lemieux, 2001). Furthermore, there were clear differences in thermal optima between the 22°C T_E group (T_opt of 27°C) and the 32°C T_E group (T_opt of 32°C). These findings provide some insight into the underlying mechanism of our previous observation that T_E influences the short-term thermal sensitivity of swimming performance (Scott and Johnston, 2012). For example, zebrafish raised at 22°C as embryos maintain higher swim speeds after transfer to 22°C than fish raised at 32°C, possibly in part because their COX enzyme was better suited to function at cooler temperatures.

T_E also had a persistent effect on the maximal activities of multiple enzymes in the muscle, even after fish were raised for several months after hatching at a common 27°C temperature. After correcting for the effects of variation in body mass, both the 22°C and 32°C T_E groups had higher activities of COX and PK than the 27°C T_E controls (Fig. 1, Table 3). These results were probably not caused by differences in fibre type composition of the muscle, because there are no differences between T_E groups in the total transverse area of either slow oxidative (red) or fast oxidative (pink, intermediate) fibres in the axial musculature at 27°C (Scott and Johnston, 2012). However, embryonic temperature does affect the size of both red fibres (Scott and Johnston, 2012) and fast glycolytic (white) fibres (Johnston et al., 2009) in zebrafish: fibres tend to be larger but fewer in number in fish raised at 22°C or 31–32°C as embryos compared with fish raised at 26–27°C. The observation that muscle COX and PK activities are higher in concert with an increase in muscle fibre size (but not total area) in the 22°C and 32°C T_E groups provides correlative evidence that there may be metabolic consequences for the often observed effect of T_E on muscle cellularity (Vieira and Johnston, 1992; Rowlerson and Veggetti, 2001; Stoiber et al., 2002; McClelland and Scott, 2013).

The extent to which T_E affected enzyme activities through long-term alterations in muscle growth phenotype is unclear. The T_E treatments occurred at a time of rapid and significant development and growth of the muscle (Hinterleitner et al., 1989; Johnston, 2006; McClelland and Scott, 2013), but impacts persisted into adulthood when growth rate had slowed (Fig. 1). Nevertheless, zebrafish were sampled at 4 months of age, when there were small differences in body size between T_E groups (Table 1) and there was allometric variation in activity for most enzymes studied (Fig. 2). Accounting for this variation using a residual approach (Table 3) suggested that differences in size per se (but not necessarily instantaneous growth rate) did not cause the differences in COX and PK activities between T_E groups.

The persistent effects of T_E on muscle enzyme activities could affect the overall metabolism of the animal. The axial muscle powers swimming and makes up a large proportion of body mass, so the metabolic processes of this tissue will have a strong influence on whole-body metabolic rates. Previous work has shown that developmental temperature affects resting oxygen consumption in larval zebrafish (Barrionuevo and Burggren, 1999) as well as resting and maximal oxygen consumption in larval cyprinids (Wieser and Forstner, 1986; Kaufmann and Wieser, 1992). These observations should be partly due to the well-known kinetic effects of temperature on reaction rates, but could also result from effects on the various physiological factors that control metabolism. Metabolic rate and oxygen consumption are system-level processes that are influenced by all steps in the pathway of oxygen and fuel delivery and oxidation, including the activity of metabolic enzymes and the abundance of mitochondria (Suarez and Moyes, 2012). In this regard, T_E has been shown to alter mitochondrial abundance in the muscle of larvae from several fish species (Vieira and Johnston, 1992; Brooks and Johnston, 1993). Although we did not measure mitochondrial abundance, it is often inferred from COX activity (Duggan et al., 2011). T_E may therefore have persistent effects on whole-animal energy metabolism by altering COX activity and mitochondrial abundance in the muscle.

Our observation that the warm acclimation response of metabolic enzymes was enhanced by development near the cold and/or warm extremes of embryonic survival did not support our second hypothesis. We found in previous work that both 27°C and 32°C T_E groups swam better after acclimation to 34°C than the 22°C T_Egroup (Scott and Johnston, 2012), and had hypothesized here that this would be explained by an enhanced metabolic warm acclimation response in the 27°C and 32°C T_E groups. The discrepancy could reflect the effects of the small differences in acclimation temperature used between studies (32°C here versus 34°C previously). However, it is also possible that the 27°C T_E group uses an alternative but still effective strategy for acclimating to warm temperatures, such as increasing the capacity for anaerobic glycolysis (Fig. 7). Consistent with this idea, carbohydrate content of the muscle increases in response to 34°C acclimation in zebrafish raised under control conditions, but has no effect on total muscle lipid stores (Vergauwen et al., 2010). Differences between T_E groups in other processes that are important for warm acclimation, such as the induction of heat shock proteins (Vergauwen et al., 2010), may also contribute to the observed differences in performance.

Even though our second hypothesis was not supported, our current findings are generally consistent with our previous observation that T_E has a persistent effect on thermal acclimation capacity (Scott and Johnston, 2012). For example, we showed that the cold acclimation response for U_crit, slow oxidative fibre density and/or mRNA expression of metabolic genes were enhanced in both 22°C and 32°C T_E groups compared with the 27°C control T_E group. This observation is mirrored in the present study in that many of the metabolic responses to warm acclimation in the 22°C and 32°C T_E groups were similar and enhanced compared with the 27°C control T_E group.

The cellular mechanisms underlying developmental thermal plasticity are not well understood. It has been suggested that T_E alters the abundance and/or the balance between proliferation and differentiation of myogenic progenitor cells in the muscle (Johnston et al., 2003; Steinbacher et al., 2011). Having more undifferentiated myoblasts in the muscle should influence the potential for muscle plasticity during thermal acclimation, because it would affect the ability to recruit new muscle fibres or add nuclei to existing muscle fibres (Johnston, 2006). This could help explain the results of the present study, because the nuclear content of muscle is an important determinant of metabolic enzyme expression (Dalziel et al., 2005). Fibre-type composition of the muscle can change in response to thermal acclimation (Johnston and Lucking, 1978), and this could have differed between T_E groups, possibly related to the presence of myogenic progenitor cells. T_E also changes the developmental expression of muscle transcription factors (e.g. myogenic regulatory factors) (Macqueen et al., 2007), and could regulate genes important for muscle phenotype and plasticity through epigenetic mechanisms (Johnson and Tricker, 2010).

Changes in the lipid composition of mitochondrial membranes could act in concert with the changes observed here to mediate the effects of thermal acclimation and developmental plasticity on metabolic capacity in the muscle. Temperature acclimation has a well-known effect on the lipid composition of membranes (Hazel, 1995), and can thus affect the activity of membrane-bound enzymes (Seebacher et al., 2010). The effect of T_E on the changes in membrane composition during thermal acclimation is not explicitly known, but it is intriguing that the most prominent differences between T_E groups were observed for COX (an enzyme normally located in the inner mitochondrial membrane). Detergents are used in the COX assay to minimize the direct effects of membrane lipids on the measured activity, but it is possible that some of the variation observed was caused by differences in the impact of residual lipids that were not removed. Alternatively, COX activity may be altered in response to acclimation temperature and/or T_E in response to changes in the membrane-lipid microenvironment (Frick et al., 2010; Martin et al., 2013).

Although there were few effects of T_E on the temperature sensitivity of enzymes (with the exception of COX, see above), warm acclimation changed the temperature sensitivity of HOAD in all T_E groups. The Q₁₀ of HOAD activity between assay temperatures of 22 and 27°C increased from ~2.7 in 27°C-acclimated fish to ~3.6 in 32°C-acclimated fish (Fig. 9). The cause of this increase in the sensitivity of HOAD to cold is unclear, but it could result from a change in the predominant isoform that is expressed in the muscle.

The high embryonic survival we observed between 22 and 32°C is consistent with previous reports that cumulative hatching success was lower in zebrafish after embryonic development at 21°C (~60%) compared with 26 and 33°C (~80%) (Hallare et al., 2005). Our results agree with the general expectation that there are sharp transitions in survival at the upper and lower thermal limits for development, and that these thermal limits lie well within the tolerance range of adults (Angilleta, 2009). Abrupt thermal limits for development have been suggested to occur at the temperatures when cell division is no longer possible, caused by reversible effects of temperature (rather than irreversible effects, such as denaturation) on the activity of transcriptional enzymes and cell cycle proteins (van der Have, 2002), but empirical evidence for this suggestion is lacking. Temperatures within the survivable range did not lead to any noticeable increases in the proportion of developmental abnormalities, indicating that there is some robustness in the developmental programme to moderate temperature challenges (Braendle and Félix, 2008; Frankel et al., 2010). Temperature had a typical Q₁₀ effect on full embryonic development between 22 and 32°C, consistent with previous observations on the rate of somitogenesis (a process that occurs during the segmentation period between roughly 15 and 30% of development to hatching) (Schröter et al., 2008).

Developmental plasticity in response to temperature can have striking and persistent effects on metabolic pathways, muscle phenotype and exercise performance in ectothermic organisms. We have shown here that T_E influences the capacity for metabolic temperature acclimation and the thermal sensitivity of oxidative enzymes in zebrafish. However, we still know relatively little about the underlying mechanisms of developmental thermal plasticity, and what influence it might have on the success of different species as climate change progresses (Beldade et al., 2011). Understanding the capacity of different organisms for both developmental and adult phenotypic plasticity will be essential to appreciate which species and life stages are most sensitive to thermal changes in the environment.

The zebrafish (D. rerio) used in this study were an outbred stock purchased from a tropical fish wholesaler (Aquality; Mississauga, ON, Canada). Two independent breeding groups, each containing eight males and eight females, were established for use in all experiments. These fish were maintained in a semi-recirculating system at 27–28°C (12 h:12 h light:dark photoperiod) and were fed three times per day with an alternating diet of tropical flake food and fresh brine shrimp. Three series of experiments were performed to determine (1) the thermal breadth of embryonic development, (2) the effects of embryonic temperature on post-hatching growth and (3) the effects of T_Eon metabolic enzymes. Embryos from a minimum of two spawnings from each family were collected at the 1–8 cell stage and raised in Petri plates (50 embryos in 50 ml per plate) using standard protocols (Westerfield, 2007).

Embryos were raised at each of a range of temperatures (every 2°C from 18 to 38°C). The average developmental stage of each plate and the proportion of dead or abnormally developing embryos were recorded every 12 h (starting at 12 h post-fertilization) until all surviving embryos had hatched. The stage of embryonic development was scored from 0 (i.e. fertilization) to 100% using the expected temporal progression of zebrafish development at 28.5°C as a reference (Kimmel et al., 1995).

Embryos were raised at each of three temperatures (±0.5°C) that represent near-optimal (27°C), cold-extreme (22°C) and warm-extreme (32°C) temperatures for normal development. After hatching, larvae were maintained at a common temperature of 27±0.5°C in small tanks. Larvae were fed several times a day a mix of commercial larva food, spirulina and brine shrimp larvae (as appropriate). Fish were then transferred from static conditions to the semi-recirculating system at roughly 1 month of age, and were then gradually transitioned onto the same feeding routine as the parental groups. Standard body length was measured for a random sample of 26 to 64 fish from each embryonic temperature treatment group at 2, 4, 8, 12 and 20 weeks post-fertilization. This was accomplished by lightly anaesthetizing the fish in buffered MS-222 (0.08 g l⁻¹) and then imaging each individual under a stereomicroscope.

Embryos were raised as in Experiment 2, except that they were not anaesthetized and measured during development. Once fish reached 3 months of age they were acclimated to either 27 or 32°C for 30 days. Temperature acclimation treatments were conducted in static 50 l glass tanks. Tank water was filtered and aerated, and optimal temperatures (±0.5°C) were maintained using aquarium heaters. Approximately one-third of the tank volume was replaced with fresh dechlorinated tap water at the appropriate temperature at least twice a week to maintain water quality (nitrate, nitrite, ammonia and pH). Fish were euthanized (immersion in buffered MS-222; 0.8 g l⁻¹), weighed and measured, and then sampled at the end of the 30 day acclimation period. The majority of both the red and white axial muscle was isolated altogether by removing the head, skin, fins and organs. Muscle samples were quickly immersed in liquid nitrogen and then stored at −80°C.

The frozen muscle samples were weighed and homogenized on ice in 20 volumes of homogenization buffer (concentrations in mmol l⁻¹: 20 Hepes, 1 EDTA and 0.1% Triton-X-100) at pH 7.0 (this and all subsequent pH values refer to measurement at 20°C), and were re-frozen and stored at −80°C until assayed. The maximal activities (V_max) in muscle homogenates of CS, HOAD and PK were assayed at each of four assay temperatures (22, 27, 32 and 36°C) by measuring absorbance at 340 nm (HOAD and PK) or 412 nm (CS) for 5 min. Assays were performed under the following conditions (in mmol l⁻¹): CS, 50 Tris, 0.5 oxaloacetate, 0.3 acetyl-coA, 0.1 5,5-dithiobis-2-nitrobenzoic acid, pH 8.0; HOAD, 50 imidazole, 0.1 acetoacetyl-coA, 0.3 NADH, pH 7.2; PK, 50 MOPS, 5 phosphoenolpyruvate, 0.15 NADH, 5 ADP, 0.01 fructose-1,6-bisphosphate, 100 KCl, 10 MgCl₂, excess coupling enzyme (lactate dehydrogenase), pH 7.4. Preliminary experiments verified that substrate concentrations were saturating for all of the above assays.

The activity of COX and LDH were measured for several different starting concentrations of substrate at each of the four assay temperatures. COX activity was assayed with 100, 50, 25 or 10 μmol l⁻¹ of reduced cytochrome c (Cytc[Fe²⁺]) in 50 mmol l⁻¹ Tris (containing the desired cytochrome c concentration and 0.5% Tween-20, pH 8.0) by measuring absorbance at 550 nm for 5 min. Cytochrome c was reduced in the assay buffer using sodium dithionite (Spinazzi et al., 2012), and the latter was then oxidized by bubbling the buffer with air for at least 2 h on ice. The background oxidation rate of reduced cytochrome c was negligible. LDH activity was assayed in buffer containing 1000, 500, 250, 100, 50 or 10 μmol l⁻¹ of pyruvate, 0.15 mmol l⁻¹ NADH and 50 mmol l⁻¹ Hepes (pH 7.0) by measuring absorbance at 340 nm for 5 min. The reported V_max is the highest empirical measurement of activity at each assay temperature. Substrate affinity (K_m) was determined by interpolating the substrate concentration at 50% of the empirically determined V_max value.

Enzyme activities of each individual were measured in triplicate on a SpectraMax Plus 384 microplate reader with temperature control (Molecular Devices, Sunnyvale, CA, USA). COX was assayed after a single freeze–thaw cycle, and the remaining enzymes were assayed after two or more freeze–thaw cycles. Background reaction rates were measured in control reactions that did not contain one of each enzyme's substrates, as follows: COX, Cytc[Fe²⁺]; CS, oxaloacetate; HOAD, acetoacetyl-CoA; LDH, pyruvate; PK, phosphoenolpyruvate. Enzyme activities were determined by subtracting the background rate from the rates measured in the presence of substrate. For HOAD, LDH, PK and COX assays, we calculated extinction coefficients (ε) empirically by constructing standard curves of absorbance versus NADH or Cytc[Fe²⁺] concentration in the buffers appropriate for each assay. We used an ε of 13.6 optical density (mmol l⁻¹) cm⁻¹ for CS assays. Enzyme activities are expressed in international units relative to muscle protein content (μmol mg protein⁻¹ min⁻¹). Protein contents of samples were measured in triplicate using the bicinchoninic acid method (Sigma-Aldrich, Oakville, ON, Canada), following the manufacturer's instructions.

Data are shown as means ± s.e.m. error (except when data points from individual samples are shown). The enzyme activity value for each individual was the average of its triplicate measurements (minus background; see above), and the mean and standard error of these values are reported. One- and two-factor ANOVA and Bonferroni multiple comparisons tests were used as appropriate to assess the effects of T_E, acclimation temperature and/or assay temperature. A significance level of P<0.05 was used throughout.

The authors thank Grant McClelland and Joanna Wilson for the use of their facilities while our lab was being set up. Michal Gallus provided invaluable technical assistance. We also thank two anonymous referees for excellent comments on a previous version of the manuscript.