Thermally tolerant intertidal triplefin fish (Tripterygiidae) sustain ATP dynamics better than subtidal species under acute heat stress

Temperature is a key factor that affects all levels of organization. Minute shifts away from thermal optima result in detrimental effects that impact growth, reproduction and survival. Metabolic rates of ectotherms are especially sensitive to temperature and for organisms exposed to high acute temperature changes, in particular intertidal species, energetic processes are often negatively impacted. Previous investigations exploring acute heat stress have implicated cardiac mitochondrial function in determining thermal tolerance. The brain, however, is by weight, one of the most metabolically active and arguably the most temperature sensitive organ. It is essentially aerobic and entirely reliant on oxidative phosphorylation to meet energetic demands, and as temperatures rise, mitochondria become less efficient at synthesising the amount of ATP required to meet the increasing demands. This leads to an energetic crisis. Here we used brain homogenate of three closely related triplefin fish species (Bellapiscis medius, Forsterygion lapillum, and Forsterygion varium) and measured respiration and ATP dynamics at three temperatures (15, 25 and 30 °C). We found that the intertidal B. medius and F. lapillum were able to maintain rates of ATP production above rates of ATP hydrolysis at high temperatures, compared to the subtidal F. varium, which showed no difference in rates at 30 °C. These results showed that brain mitochondria became less efficient at temperatures below their respective species thermal limits, and that energetic surplus of ATP synthesis over hydrolysis narrows. In subtidal species synthesis matches hydrolysis, leaving no scope to elevate ATP supply.

Temperature exerts a profound effect on all levels of organization; with no single organism, capable of withstanding the full spectrum of temperatures across the biosphere 1,2 . This is true of ectotherms, which lack active thermoregulatory processes and especially true for intertidal species which experience large fluctuations in ambient temperature over the course of a day 1 . With anthropogenic climate change shifting mean and extreme temperature patterns, understanding the basis of thermal tolerance and thermal adaptation is essential [3][4][5] . Temperature typically promotes a near exponential rise in metabolic rate until a critical threshold is reached, which is then followed by a sharp decline 6 . What ultimately mediates this metabolic collapse at high temperature remains controversial, and oxygen limitation has been proposed to underpin thermal tolerance in aquatic ectotherms 7,8 . However, this concept is debated, as underlying literature finds support for and against its application [8][9][10][11][12] .
Perhaps the most central role of metabolism is to maintain a tight balance between ATP production and consumption. In almost all animals that live independent from a host, mitochondria are the predominant source of ATP, and oxidative phosphorylation (OXPHOS) provides ~ 90% of the cellular ATP, around 15 times more than that generated via fermentative glycolysis 13,14 . Considerable work has focused on the mitochondrion's role, specifically cardiac mitochondria, as the linchpin for thermal tolerance 15 . With rising temperature, the fraction of oxygen (O 2 ) flux and carbon substrate consumed by mitochondria escalates, yet the energy released is increasingly directed toward an apparent futile cycle, or "leak" of protons across the inner mitochondrial membrane [14][15][16] .
Rising temperatures increase cellular ATP demands 6 , yet mitochondrial efficiency and outright ATP production may decline, even with saturating O 2 15,17 . In sum, a mismatch between an organism's capacity to meet increasing ATP demands must occur. While considerable focus has been placed on respiration and ATP production, few have considered how these relate to rates of ATP hydrolysis at high ambient temperatures. This relatively simplistic concept of balance in cellular ATP-economics has yet to be followed in the contexts of temperature. We Animals and housing. The rock-pool exclusive and high intertidal species (Bellapiscis medius and Forsterygion lapillum, respectively) were caught using bait traps and hand nets in rock pools and off piers. Subtidal species Forsterygion varium were caught by SCUBA. All fish were collected around the greater Auckland region (− 36.081588, 174.598804). Fish were then transported to the facilities of the University of Auckland and held in 30 L aerated tanks with recirculating seawater at 18 ± 0.5 °C, 200 µm filtered, and 35 ± 1 ppt salinity. Fish were checked daily and fed ad libitum every 3 days with raw prawn meat. After at least a week of acclimation,  34 . A substrate-uncoupler-inhibitor-titration assay (SUIT) was used to test (i) mitochondrial components of the electron transport system (ETS) in a stepwise fashion, measured with the maximum O 2 flux mediated by each titration, (ii) ATP production through OXPHOS and (iii) ATP dynamics of triplefin brain mitochondria (Fig. 2). Both assays differed by the addition of either ADP or ATP. Brain homogenate (~ 10 mg wet weight equivalent) was distributed equally between two parallel respirometry chambers and left until steady-state respiration was reached. Then, 2 mM final concentration of ADP (assay 1, one chamber) or ATP (assay 2, second chamber) was added to measure respiration rates supported by endogenous substrates (Routine state). The addition of ADP primed OXPHOS, while the addition of ATP forced hydrolysis rates. The NADH 2 -generating substrates pyruvate (5 mM), malate (2 mM), and glutamate (10 mM) were then added to stimulate CI-OXPHOS. Subsequently, succinate (10 mM) was added to stimulate complex II (CII) and allow the measurement of OXPHOS with the combined inputs of CI and CII (CI&CII-OXPHOS). Respiration attributed to proton leak (LEAK) was assessed with the addition of the ATP F0-F1 synthase inhibitor oligomycin (2.5 μM). The contribution of adenine nucleotide translocator (ANT) to LEAK was calculated as the difference prior and after the addition of the ANT inhibitor carboxyatractyloside (cAtr; 5 μM). Mitochondria were then uncoupled from OXPHOS with repeated titrations of carbonyl cyanide m-chloro phenyl hydrazone (CCCP; 0.5 μM titration steps) to determine the maximum ETS capacity without the limitation of the phosphorylating system (ATP F0-F1 and ANT). Finally, non-mitochondrial respiration was determined by the addition of the complex III inhibitor antimycin A (Ama; 2.5 μM) was titrated to inhibit respiration (Fig. 2). www.nature.com/scientificreports/ ATP dynamics: assay and kinetics. Standard respiration media (MiR05) does not contain calcium (Ca 2+ ) or sodium (Na + ). As Na + and Ca 2+ salts are required to activate cellular ATPases; NaCl (5 mM) and CaCl 2 (0.25 mM) were added before homogenate to reach maximum ATP hydrolysis rates. Maximum OXPHOS rates cannot be achieved in the absence of Mg 2+ , which is required for proper cellular function and stabilisation of ATP and ADP 35 . ATP and ADP kinetics were measured fluorometrically, as described elsewhere [36][37][38][39] . Briefly, as both ADP and ATP require Mg 2+ for stabilisation, free Mg 2+ [Mg 2+ ] free within the O2K chamber was monitored using Magnesium-Green™ (5 mM). The fluorescent signal (470/530 nm; Ex/Em) was calibrated by two subsequent titrations of MgCl 2 (1.25 mM each) so that (i) Mg 2+ dependent reactions can be achieved and (ii) Mg 2+ -free ADP and ATP are stabilised. As binding varies with temperature, independent assays were run at each experimental temperature without sample to determine binding kinetics between ADP-Mg 2+ , ATP-Mg 2+ and Mg 2+ -MgG. Post fluorescent signal calibration, ADP and ATP (Mg 2+ -free) were titrated stepwise (1.25-2.5-3.75-5-6 mM) ( Supplementary Fig. S1). This allowed the construction of ADP and ATP binding curves to Mg 2+ . The ratio of the slopes between the two curves at working ATP/ADP concentrations for assays 1 and 2 were used to determine a fluorescence correction factor for each assay at each experimental temperature ( Supplementary  Fig. S1). The experimental ADP signals were subsequently multiplied by the correction factor to determine rates of net ATP production 36,40 . ATP hydrolysis rates were determined during the SUIT assay protocol following the addition of Oligomycin to inhibit the ATP synthase. Using this rate, we could calculate the overall ATP production rate through the sum of the net ATP production rate with the ATP hydrolysis rate. Overall production and hydrolysis rates provide insights into the condition of the tissue, but more physiologically relevant information is acquired from P/O ratios; the amount of ATP produced per molecule of oxygen consumed. Changes in this ratio across temperature are calculated by dividing the rate of overall ATP production by O 2 consumption during CI&CII-OXPHOS.

Calculations and statistical analyses.
Respiration data is presented as mean (n = 10 individuals) ± s.e.m (alternatives are otherwise stated). Respiratory control ratios (RCR) are calculated as (State 3-State 4)/State 3. State 3 corresponds to a more recently defined OXPHOS state when mitochondria are exposed to sufficient substrates and ADP, whereas state 4, or LEAK, is measured in the absence of ADP, or also in the presence of oligomycin (state 4 o ). In this present study on homogenates RCRs were calculated as (CI&CII-OXPHOS-LEAK)/ CI&CII-OXPHOS (e.g., (State 3-State 4 o )/State 3). The reserve respiratory capacity is calculated as CI&CII-OXPHOS-ETS. Prism (Vers.8) was used to conduct independent t-test between mitochondrial states when homogeneity of variance was verified. Two-way ANOVAs were performed to analyse the effect of temperature and species on respiration rates and ATP dynamics with a Greenhouse-Geisser correction when sphericity was not assumed. Tukey-Post Hoc tests were performed for pairwise comparison.

Results
Respiration assays. A SUIT protocol was applied to stimulate mitochondrial respiration in brain mitochondria. The respiratory flux at 15 °C, showed no differences among species for any of the mitochondrial states (Fig. 3). Differences were seen across all states at 25 °C (Fig. 3). Both F. lapillum and B. medius showed higher flux during CI-OXPHOS and CI&CII-OXPOS compared with F. varium (Fig. 3a,b (Fig. 3a,b). Complex II contribution to mitochondrial flux was calculated as CI-OXPHOS subtracted from CI&CII-OXPHOS ( Fig. 4a- ATP assays. At all temperatures ATP production rates differed substantially from hydrolysis rates for all three species (Fig. 5a; p < 0.0001). At 30 °C overall ATP production rates decline from those at 25 °C for F. lapillum and B. medius but remain above the increasing ATP hydrolysis rates ( Fig. 5a; 30 °C: p < 0.0001). Net ATP production rates decline sharply to closely match hydrolysis rates at 30 °C, and only remained marginally higher for B. medius

Discussion
We demonstrate that brain mitochondrial efficiencies likely play a key role in thermal tolerance and that as temperature increases respiration and ATP production decreases leading to a tight energetic balance. Notably our approach permitted some comparison of ATP synthetic to hydrolysis rates, and the amount of ATP consumed increases as the amount of ATP produced declines. For the subtidal F. varium at 30 °C the balance between production and consumption of ATP sits on a knife's edge (Fig. 5). While the intertidal species, B. medius and F. lapillum, showed declines in ATP production and consumption as temperature increased, the energetic surplus of ATP synthesis narrows yet remains in excess for these intertidal species at 25 and 30 °C.
Energetic balance in the brain. The role of neural function in setting the upper thermal limits has been largely ignored. Similar to the heart, the brain is an excitable tissue with high basal energetic demands [23][24][25][26][27] . The changes in mitochondrial efficiency and ATP dynamics experienced during heat stress will mediate dysfunction across the neural system and alter brain function. Early investigations have argued that neural function at the  www.nature.com/scientificreports/ organ and cellular level may limit upper thermal tolerance 11,16,30 . Decreased control of ventilation and blood circulation within the brain was shown as temperature increases 41 , while at the cellular level, action potential conduction and synaptic transmission is sensitive to increasing temperatures in several studies [42][43][44][45] . Other recent studies have compared the acclimation responses of heart and brain in eurythermal teleost species. While both organs showed significant acclimation responses, a larger acclimation response was found in the brain 30 . In two crustacean species (Penaeus monodon and Astacus astacus), the cardiorespiratory system maintained O 2 supply up to T crit suggesting temperature resistance of the heart; however, the generation and conduction of neuronal action potentials failed approaching T crit , suggest a greater thermal sensitivity of the nervous system relative to the heart in these species 12 .
The high energetic demands of the brain require a balance to be maintained between the production of sufficient ATP and its rapid consumption. This study is the first to attempt to formulate a measure of the dynamic energetic balance of ATP production and use, specifically, in terms of the mismatch between energy supply and demand during heat stress. Generation of action potentials involves rapid plasma membrane depolarization and requires repolarization using ATP-dependent ion pumps 46 . With alterations in channel properties, neuronal membranes become leakier with increasing temperature, and active ion repartition becomes more energetically costly 47,48 . Action potentials must also exceed a threshold and increased temperature decreases action potential amplitude and duration, decreasing propagation in excitable tissue 48 . An additional large ATP sink involves the extrusion of neurotransmitters into synaptic clefts to transmit signals, and this increases with heat stress 44,49 .
Here, we determined mitochondrial function and specifically, relatively simple ATP dynamics. These changed with temperature for all three species, and we observed a clear decrease in mitochondrial stability and capacity to meet ATP requirements as temperature increases. While we used brain homogenates, which may not represent in vivo ATP demands, it provides a proxy measurement and system to measure ATP synthesis and hydrolysis under equivalent conditions. Regardless, the subtidal species' F. varium is the most sensitive to temperature and is at a point of no reserve ATP capacity, while the intertidal B. medius shows greater resilience to elevated temperatures. www.nature.com/scientificreports/ Coupling of OXPHOS. OXPHOS rates increased with temperature for all species (Fig. 3a,b) and indicates increased O 2 consumption. Leak rates from all three species also increased significantly from 15 to 30 °C with the greatest proportion of respiration flux resulting from LEAK in F. varium at 30 °C (Fig. 3c). These results align with previous work on permeabilized heart 33 and skeletal muscle fibres 50 from triplefin fish, where LEAK O 2 flux also elevates at high temperatures. This supports the view that mitochondrial function and stability declines prior to CT max 15,17,33 . Elevated O 2 demand as OXPHOS efficiency decreases may become increasingly detrimental if sufficient O 2 cannot be extracted from water. Notably, F. varium and F. lapillum, have lesser capacities to extract O 2 than B. medius at low O 2 partial pressures 51 . The synergistic effects of increasing temperature will increase the drain on O 2 and substrate (e.g., glucose) to maintain nervous function.
We tested the integrity of the ETS complexes, at 15, 25 and 30 °C. Consistent with previous work 33 , ETS rates increased significantly from 15 to 30 °C in all three species (Fig. 3d); however, the greatest increase in flux was seen in the mid tide species F. lapillum (Fig. 3d). Reserve respiratory capacity, calculated as the difference between maximal O 2 consumption (ETS) and basal respiration (CI&CII-OXPHOS) has been employed to provide an estimate of a tissues capacity to cope with increases in ATP demand 52 . Under "normal" conditions, mitochondria are thought to operate at a lesser fraction of their energetic capacities. For cells such as neurons that experience large fluctuations in energy demand on short timescales, the capacity to increase supply to meet demands are essential. The mid-tide species, F. lapillum, had the greatest reserve respiratory capacity at 25 and 30 °C compared with both F. varium and B. medius (Fig. 4e). Theoretically, this greater reserve capacity will provide F. lapillum a wider window to defend and buffer ATP supply following s conditions of stress. However, the elevated LEAK rate at both 25 and 30 °C for F, lapillum (Fig. 3c) indicate that less of the consumed oxygen is directed towards ATP production. Our measures of ATP synthesis reflect this effect in the contexts of P/O ratios (Fig. 5b).

ATP dynamics.
Oxygen consumption is a traditional proxy or indirect measure of energy expenditure in aerobic organisms, as most ATP production occurs aerobically [53][54][55] . However, this assumption is compromised if OXPHOS is uncoupled. Only measures of ATP dynamics, or balance, can determine ATP production rates at elevated temperature and provide insight into function under physiological conditions. This study revealed www.nature.com/scientificreports/ that at 25 °C ATP production rates are elevated for the two intertidal species while it was diminished for F. varium. At 30 °C, and all three species showed significantly diminished ATP production rates and elevated ATP hydrolysis rates (Fig. 5). Work performed on the common New Zealand wrasse (Notolabrus celidotus) showed that the capacity of isolated cardiac mitochondria to efficiently produce ATP decreased at 25 °C, prior to temperature induced heart failure at 27.5 °C 15,17 . Work with isolated mitochondria from rat cardiomyocytes also revealed diminished ATP production at elevated temperatures and the eventual reversal of the ATP synthase at 43 °C, making the mitochondria a consumer of ATP 17 . Use of Phosphorus-31 NMR (31P-NMR) during acute heat shock measured the rapid decline in ATP at elevated temperatures in Tetrahymena ciliate species 56,57 , and the same technique has been used by others and revealed an immediate fall in ATP after brief exposure to sublethal heating 26,57 . While the initial ATP levels were recovered after 48 h, a prolonged exposure to heat stress would likely lead to irreversible loss of ATP and the eventual activation of necrotic pathways and death. In B. medius and F. lapillum, the exposure to elevated temperatures in rock pools occurs over several hours within a day 34 . While these are sub-lethal regarding their respective CT max , the decreases in mitochondrial efficiency and ATP production can lead to the irreversible loss of function. Similarly, exposure to moderate heat stress in liver homogenates of juvenile Brown trout (Salmo trutta) induced lowered mitochondrial coupling and increased leak rates and the inferred insufficiency to maintain ATP homeostasis may have diminished food intake and suppressed growth 58 . As expected, ATP hydrolysis rates in this study were highest at 30 °C in all species (Fig. 5a). The temperature sensitive F. varium had the lowest rates of ATP production relative to ATP hydrolysis and this is reflected in the declining P/O ratios (Fig. 5a,b). However, at 30 °C all three species showed declines in the efficiency of brain homogenates to produce ATP above the ever-increasing hydrolytic rates of ATP. Prior to their respective CT max , all species have decreased mitochondrial efficiency to adequately produce ATP, leading to growing imbalance between ATP demand and ATP production. Mitochondrial efficiency can be defined by the organelles ability to efficiently transfer the free energy released from reducing pathways to ATP production. Traditionally P/O ratios have been used to quantify the number of ATP molecules produced per molecule of O 2 consumed. Mechanistic P/O ratios have been previously calculated using end-point protocols in respiration assays 55,59 ; while the use of MgG fluorescence has allowed the generation of more accurate "active" and "steady-state" P/O ratios. This technique was utilized in recent studies that have provided insight in terms of mitochondrial efficiency with changing temperatures across a range of species 15,17,37,39 . Traditional approaches were unable to produce P/O ratios that were informative about ATP dynamics, whereas another approach 17 enabled a more dynamic assessment of production and hydrolysis of ATP. Increases in temperature mediated decreases in P/O ratios as mitochondria became less efficient at producing ATP per O 2 consumed. At 25 °C there was a near 25% decrease in the P/O ratio from 15 °C in F. varium. This was further exacerbated at 30 °C where the P/O ratio was further decreased by up to 75% (Fig. 5b). This decline in P/O ratio means more O 2 is required to sustain ATP production, this also requires a concomitant increase in other metabolic fuels, such as glucose.
The brain is predominantly aerobic, relying on steady supply of O 2 and glucose to fuel its activity. With limited capacity for anaerobic metabolism, ATP production at elevated temperatures is time restricted and comes at a potentially greater long-term cost for the individuals. Comparing across species, B. medius has the greatest brain tissue glycogen stores compared with both F. lapillum and F. varium 51 . This paired with the higher P/O ratios at elevated temperatures provides B. medius with a greater capacity to function at elevated temperatures it may experience during the tidal cycle. Similar declines in P/O ratios have been shown in studies at elevated temperatures 15,17 . The common New Zealand wrasse (Notolabrus celidotus) showed decreased ATP production rate by cardiac mitochondria as temperature increased and showed depressed "active" P/O ratios at 32.5 °C 15 . The "steady-state" P/O ratios calculated by Ref. 17 showed a decline in P/O ratio from 2.5 at 37 °C down to a negative P/O ratio at 43 °C for rodent heart mitochondria. At 25 °C, all three species could maintain sufficiently high P/O ratios while at 30 °C only the rock-pool exclusive B. medius was able to sustain a P/O ratio above one. Compared with respiration data, declining P/O ratios are an accurate indication of mitochondrial efficiency and can be used to assess the dynamic shifts of ATP within the cells.
The greater mitochondrial efficiency seen in B. medius coupled with recent work 34,51 ; showing greater brain glycogen stores and increased O 2 extractive capacity, compared with F. lapillum and F. varium, will aid greater survival at elevated temperatures in the rock pool environment but the closely matched production and hydrolysis of ATP at 30 °C will be further exacerbated as temperatures increase.

Conclusions
Assessment of the respiration rates showed declines in mitochondrial stability and function at elevated temperatures in the subtidal species, F. varium, which do not experience high temperature fluctuations. However, the higher intertidal species B. medius had greater mitochondrial efficiency and stability at elevated temperatures as was expected. These results agree with the responses from other fish species, which show decreases in mitochondrial function prior to CT max and may be involved in setting upper thermal tolerance limits. Assessment of ATP dynamics in real-time showed that the mitochondrial capacity to produce ATP was diminished at elevated temperatures as ATP hydrolysis rates increased. This led to a closely matched supply and demand dynamic of ATP that may become further exacerbated at CT max and could be involved in underpinning the upper thermal tolerance limit.

Data availability
The dataset supporting the results of this manuscript is available from the University of Auckland repository Research Space: https:// figsh are. com/s/ e4d85 4208b 6c8bc c8af7.