Simultaneous measurement of mitochondrial respiration and ATP production in tissue homogenates and calculation of effective P/O ratios

Abstract The use of tissue homogenate has greatly aided the study of the functioning of mitochondria. However, the amount of ATP produced per oxygen molecule consumed, that is, the effective P/O ratio, has never been measured directly in tissue homogenate. Here we combine and refine existing methods previously used in permeabilized cells and isolated mitochondria to simultaneously measure mitochondrial ATP production (JATP) and oxygen consumption (JO2) in tissue homogenate. A major improvement over existing methods is in the control of ATPases that otherwise interfere with the ATP assay: our modified technique facilitates simultaneous measurement of the rates of “uncorrected” ATP synthesis and of ATP hydrolysis, thus minimizing the amount of tissue and time needed. Finally, we develop a novel method of calculating effective P/O ratios which corrects measurements of JATP and JO2 for rates of nonmitochondrial ATP hydrolysis and respiration, respectively. Measurements of JATP and JO2 in liver homogenates from brown trout (Salmo trutta) were highly reproducible, although activity declined once homogenates were 2 h old. We compared mitochondrial properties from fed and food‐deprived animals to demonstrate that the method can detect mitochondrial flexibility in P/O ratios in response to nutritional state. This method simplifies studies examining the mitochondrial bioenergetics of tissue homogenates, obviating the need for differential centrifugation or chemical permeabilization and avoiding the use of nonmitochondrial ATPase inhibitors. We conclude that our approach for characterizing effective P/O ratio opens up new possibilities in the study of mitochondrial function in very small samples, where the use of other methods is limited.


Introduction
Mitochondria are responsible for oxidation of food molecules to convert energy in the form of ATP that fuels cell functions (i.e., muscle contraction, protein synthesis, gene replication, ion homeostasis). As mitochondria produce about 90% of the cellular ATP (Lehninger et al. 1993), variation in the amount of ATP produced per unit of oxygen consumed, that is, the effective P/O ratio, have significant consequences at all levels of biological organization. Variation in effective P/O ratio is known to affect energy homeostasis in cells, and thus tissues function (Pham et al. 2014;Layec et al. 2015) and ultimately, disease pathologies and normal aging (Brand 2005;Wu et al. 2007;Kadenbach et al. 2013;Conley 2016). In extreme environmental conditions such as cold, starvation and hypoxia, mitochondrial efficiency has been suggested to be a major physiological adaptation for regulating energy homeostasis in the organism (Gnaiger et al. 2000;Trzcionka et al. 2008;Monternier et al. 2014). However, while methods of evaluating mitochondrial efficiency have received much attention (Hinkle 2005;Van Bergen et al. 2014), measuring P/O ratios has proved to be challenging (Ferguson 2010).
The mitochondrial preparation required for assaying bioenergetics involves numerous steps encompassing differential centrifugation or bath transfer (Van Bergen et al. 2014), making it technically challenging to obtain a sufficient amount of mitochondria from tissues smaller than those of a mouse (Chinopoulos et al. 2011). This limitation has motivated current efforts to implement the high yield technique of homogenization that is capable of recovering 100% of the tissue sample for use in the mitochondrial analysis (Pecinova et al. 2011;Makrecka-Kuka et al. 2015;Rovenko et al. 2015;Ziak et al. 2015;Salin et al. 2016). Tissue homogenate offers the advantage of studying mitochondrial function close to physiological conditions, therefore differences in mitochondrial properties observed in homogenates may better reflect physiological processes that will occur in the animal (Kuznetsov et al. 2002;Burtscher et al. 2015). However, the problem emerges that homogenates contain ATPases that interfere with the ATP assay.
Techniques used to measure ATP production (JATP) that are compatible with simultaneous measurement of mitochondrial oxygen consumption (JO 2 ) have included: (1) estimation of mitochondrial ADP consumption (Estabrook 1967), (2a) direct and (2b) indirect assays of changes in ATP concentration (Ouhabi et al. 1998;Salin et al. 2010a, respectively), and (3) measurement of changes in free magnesium concentration, [Mg 2+ ] (Chinopoulos et al. 2009). Technique 1 can produce biased measures of JATP when using tissue homogenates due to the presence of nonmitochondrial ATPases. Technique 2b allows measurement of this ATPase activity so that the JATP value can be corrected, but this correction requires twice the amount of tissue and time (Salin et al. 2012a). The presence of ATPase may also affect JATP measurements using technique 3. Techniques 2a and 3 aim to block the action of ATPase so as to prevent ATP hydrolysis (Ouhabi et al. 1998;Chinopoulos et al. 2014), but the efficacy of inhibitors may depend on the tissue, study model or mitochondrial preparation.
Here, we describe a protocol for measuring effective P/O ratios that simultaneously quantifies both JATP and JO 2 in tissue homogenate, using a combination of techniques 2b and 3. As with technique 3, we use the magnesium-sensitive fluorescent probe, Mg Green, to estimate changes in [Mg 2+ ] (Szmacinski and Lakowicz 1996). Mitochondrial JATP is calculated from the rate of change in [Mg 2+ ] and is based on the unequal affinities of ATP and ADP for Mg 2+ (Leyssens et al. 1996). Nonmitochondrial ATPase activity is measured as in technique 2b by inhibiting the mitochondrial ATP flux (Salin et al. 2010a), allowing correction of JATP. Our modified technique facilitates simultaneous measurement of the rates of "uncorrected" ATP synthesis and of ATP hydrolysis, thus minimizing the amount of tissue and time needed. We validate the technique and calculate JATP, JO 2 and the effective P/O ratio, using liver homogenates from juvenile brown trout (Salmo trutta). We compare P/O ratios of fed and starved fish to determine whether our approach was a specific and sensitive assay to detect variation in P/O ratio in tissue homogenate. We chose liver because hepatic mitochondria are known to change their properties in response to food availability. Attempts to assess the role of mitochondria in the regulation of energy homeostasis during fasting have been addressed in mammals (Bobyleva-Guarriero et al. 1984;Brand et al. 1993;Dumas et al. 2004), birds (Bobyleva-Guarriero et al. 1984;Monternier et al. 2015), amphibians (Trzcionka et al. 2008) and fish (Savina et al. 2009;Bermejo-Nogales et al. 2015). Of note, the role of the liver in the control of carbohydrate and lipid homeostasis is essential for providing substrates to others tissues during fasting (Postic et al. 2004). In fish, there is a large body of evidence illustrating adaptive changes in enzymatic activities of the oxidative phosphorylation pathway (Guderley et al. 2003;Frick et al. 2008;Bermejo-Nogales et al. 2015), but whether their effective P/O ratio shows similar flexibility has so far not been determined.

Animal husbandry
Juvenile brown trout were obtained from a hatchery (Howietoun, UK) in summer 2015 and moved to the University of Glasgow. Here, the fish were kept in a communal tank and maintained under an 8 h light: 16 h dark photoperiod at 12°C and fed daily in excess (pellets EWOS, West Lothian, UK). In January 2016, twenty four fish were transferred to individual compartments of a stream tank. Because of the logistical constraints associated with assaying fresh mitochondria, transfers of the fish to individual compartments were staggered over 12 days (two fish per day). All fish were first acclimated for a week and fed daily to excess prior to the start of the food treatments. Half of the fish were randomly allocated to an ad libitum ration, while the other half was deprived of food (initial body mass: 12.14 AE 0.61 vs. 12.00 AE 0.57 g in the fasted and fed group). Fish were held on these treatments for 2 weeks. At the end of this period, all fish were fasted overnight. The following morning (mean AE 1 SE: 09:30 AE 00:01 h) a pair of fish was culled and their livers collected to determine mitochondrial parameters. Fasting led to reduction in body mass (final mass fasted group: 10.85 AE 2.14 g). At the same time, the fed trout had an increase in body weight (final mass fed group: 16.50 AE 3.80 g). All procedures were carried out under the jurisdiction of a UK Home Office project license (PPL 60/4292).
Estimation of K d of ATP and ADP for free magnesium (Mg 2+ ) The method for determining the affinity of the magnesium with nucleotides was adapted from Chinopoulos et al. (2014). Magnesium green fluorescence signal was detected using respirometry chambers equipped with fluorescent sensors and recorded using DatLab software (Oxygraph-2k high resolution, Oroboros Instruments, Innsbruck Austria, as in [Chinopoulos et al. 2014]). The main adjustment of the assay of the K d is a measurement in the absence of tissue homogenate and ATPase inhibitors (see below). Thus, assays were made in duplicate in a buffer containing 20 mmol/L Taurine, 10 mmol/L KH 2 PO 4 , 20 mmol/L HEPES, 110 mmol/L D-sucrose, 60 mmol/L K-lactobionate, 1 g/L BSA fatty acid free, pH 7.2 and at 12°C. The titration protocol was as follows: the buffer was supplemented with pyruvate (5 mmol/L) and malate (0.5 mmol/ L). Magnesium green (2.1 lmol/L), EGTA (0.1 mmol/L) and EDTA (15 lmol/L) were subsequently added to the chamber. Then, stepwise additions of MgCl 2 were performed for calibration of the fluorescent signal into Mg 2+ (Fig. 1). Succinate (10 mmol/L) was then added. Stepwise additions of ATP or ADP were then performed to determine the affinity of those nucleotides for Mg 2+ (25 additions of 0.2 mmol/L ATP or 0.25 mmol/L ADP) (Fig. 1). The two K d values were determined using the least squares method, described in detail in Chinopoulos et al. (2014). The binding affinity (K d ) of ATP and ADP for Mg 2+ was calculated, using the method of Chinopoulos et al. (2014); the values were K d-ATP = 0.1545 mmol/L and K d-ADP = 2.1333 mmol/L.
The method for determining the affinity of the magnesium with nucleotides in the presence of mitochondria and ATPase inhibitors (as in Chinopoulos et al. 2014) was not used here due to the insufficient effect of ATPase inhibitors. In particular, there was a lack of stability of the fluorescent signal ( Fig. 1) after ATP additions when in the presence of homogenate (5 mg wet mass liver per ml buffer) and inhibitors (25 lmol/L adenylate kinase inhibitor Ap5A, 2 lg/mL oligomycin, 4 lmol/L carboxylatractyloside (cATR), 0.2 mmol/L beryllium sulfate tetrahydrate, 5 mmol/L sodium trifluoride and 30 lmol/L sodium orthovanadate). We therefore omitted ATPase inhibitors when making measurements of K d , JATP and JO 2 . Thus, the insufficient effect of ATPase inhibitors on liver homogenate motivates us to further develop a method to quantify the rate of ATP hydrolysis.

Mitochondrial homogenate preparation
Liver homogenate was prepared as in (Salin et al. 2016). Briefly, tissues (mean AE 1 SE: 43.08 AE 2.02 mg) were minced to obtain a shredded solution, which was then homogenized with a Potter-Elvehjem homogenizer in a buffer containing 20 mmol/L Taurine, 10 mmol/L KH 2 PO 4 , 20 mmol/L HEPES, 110 mmol/L D-sucrose, 60 mmol/L K-lactobionate, 1 g/L BSA fatty acid free, pH 7.2. Homogenates were then diluted to obtain a final concentration of 5 mg/mL. The entire procedure was carried out at 4°C and completed within 30 min of the fish being culled. A sample of liver homogenate was then immediately added to one of the two respirometry chambers. The leftover preparation of liver homogenate was preserved on ice for use in a replicate trial.

Mitochondrial rates of oxidation and phosphorylation: measurement
Oxygen and magnesium green fluorescence signals were detected simultaneously, using respirometry chambers equipped with fluorescent sensors (as in Chinopoulos et al. 2014). Pure oxygen gas was added to the respirometry chambers to reach a concentration of 550 lmol/L. The titration protocol was as follows: respiration was stimulated by adding pyruvate (5 mmol/L) and malate (0.5 mmol/L). Magnesium green (2.1 lmol/ L), EGTA (0.1 mmol/L) and EDTA (15 lmol/L) were subsequently added to the chamber. Then, stepwise additions of MgCl 2 were performed for calibration of the fluorescent signal into Mg 2+ . Succinate (10 mmol/L) was then added. State 3 was reached by adding a saturating concentration of ADP (2 mmol/L). In this condition, changes in oxygen and ATP concentrations in the chamber are representative of raw fluxes in oxygen and ATP (JO 2-raw and JATP raw , respectively). JO 2-raw reflects the rate of oxygen consumption by the mitochondria but also by nonmitochondrial reactions; JATP raw is the balance between the rate of mitochondrial ATP production and the rate of ATP disappearance due to the activity of the various ATPase enzymes (phosphorylases, phosphatases and kinases; for further details see Chinopoulos et al. 2014). Addition of cATR (4 lmol/L), an inhibitor of ATP-ADP exchanger, allowed calculation of the rate of ATP disappearance due to ATPase activity . Output from a representative experiment, using homogenized brown trout liver, showing changes in oxygen and ATP content as a function of time. The substrates pyruvate, malate and succinate, as well as free magnesium, were present at the start of the experiment. State 3 respiration was initiated by adding 2 mmol/L ADP; the disappearance of oxygen (black line) was matched by the appearance of ATP (gray line). State 3 parameters stabilized 15 min after adding ADP, at which point rates of oxygen consumption (JO 2-raw ) and ATP production (JATP raw ) were measured for a further 5 min. The rate of ATP disappearance due to the activity of ATPase (JATP St4 ) was quantified by inhibiting the mitochondrial ATP-ADP exchange with carboxyatractyloside (cATR) (note the cessation of ATP production at this point, and reduction in oxygen consumption). Finally, the nonmitochondrial oxygen consumption (JO 2-ROX ) was estimated by inhibiting the respiratory chain with antimycin A (Ant A) and rotenone (Rot). Mitochondrial parameters were measured at 5 mg wet mass liver per ml at 12°C; Oxygen content in the respirometry chambers never dropped below 300 lmol/L, so maintaining the liver above sensitive levels (%100 lmol/L in our conditions, data not shown). under state 4 conditions (JATP St4 ). Addition of complex I inhibitor, rotenone (0.5 lmol/L) and complex III inhibitor, antimycin A (2.5 lmol/L) determined nonmitochondrial, or residual, oxygen consumption (JO 2-ROX ). At each step the rates of changes in ATP and/or oxygen were allowed to stabilize. An identical trial was run on the homogenate obtained from the same fish 2 h later, in order to determine the repeatability of measurements and the stability of measurements in relation to time since sample preparation.

Analysis of mitochondrial rates of oxidation and phosphorylation
Mitochondrial oxidation and phosphorylation were analyzed with Microsoft Excel 2010. Free Mg 2+ was converted into extra-mitochondrial ATP concentration, using standard binding equations previously described (Chinopoulos et al. 2014). The rate of ATP production was calculated from the linear regression of ATP concentration as a function of time (Fig. 2); JATP raw was averaged over a 5 min period, starting 15 min post-ADP addition while JATP St4 was averaged over a 5 min period after flux stabilisation. JATP St4 was then subtracted from JATP raw to obtain the state 3 rate of mitochondrial ATP production (JATP St3 ), which was expressed in pmol ATP/sec/mg wet weight of liver. Changes in oxygen concentration over time were calculated in a similar way (Fig. 2), JO 2-raw being averaged over 5 min, starting 15 min post-ADP addition, while JO 2-ROX was averaged over 1-2 min after flux stabilisation. JO 2-ROX was then subtracted from JO 2raw to obtain the state 3 respiration rate (JO 2-St3 ), expressed in pmol O 2 /sec/mg wet weight of liver. Finally, the P/O ratio was calculated as the ratio of JATP St3 to twofold JO 2-St3 ; the rate of oxygen consumption is doubled since each molecule of oxygen is comprised of two oxygen atoms.

Chemicals
Magnesium green was purchased from Thermo Fisher Scientific (Renfrew, UK), ADP was purchase from Merck  Figure 3. Changes in mitochondrial properties as a result of time since liver homogenization, shown as the difference between measurements made immediately versus 2 h after homogenization, and plotted against the mean of the two values. (A) Hepatic mitochondrial oxygen consumption JO 2-St3 (drift between trials: paired t = 8.322, P < 0.001). (B) Liver mitochondrial ATP production JATP St3 (drift between trials: paired t = 3.099, P = 0.005). (C) Liver mitochondrial estimated P/O ratio (drift between trials: paired t = À1.476, P = 0.154). N = 23 fish in all cases. Thick lines represent the mean value of the difference, and the thin line a difference of zero (i.e., no difference in measurement between the two trials).

Statistical analysis
The reproducibility of the estimates of JO 2-St3 , JATP St3 , and the P/O ratio of the two trials from the same homogenate was assessed, using Intraclass Correlation Coefficients (ICC). Paired t-tests were then used to test for any consistent shifts in the values of the measurements between the first and second trial (2 h later). Finally, linear mixed models were employed to test whether fed and fasted fish differed significantly in their hepatic mitochondrial properties (mean value of JO 2-St3 , JATP St3, and P/O ratio) while taking into account any effect of testing date (included as a random factor). All statistical analyses were performed in SPSS Statistics 21 (Chicago, IL).

Results and Discussion
The reproducibility of JO 2-St3 and JATP St3 measurements between the two replicate trials from the same liver homogenate was high (JO 2-St3 : ICC r = 0.953, df = 22, P < 0.001; JATP St3 : ICC r = 0.746, df = 22, P = 0.001). As a result, the P/O ratio was also highly reproducible (ICC r = 0.778, df = 22, P < 0.001). However, the mitochondrial activity of homogenates declined over time, as indicated by significantly lower JO 2-St3 and JATP St3 measurements in the 2nd assay that was conducted 2 h after the first ( Fig. 3A and B). This led to a tendency for the P/O ratio to also change over time, although this change was not significant (Fig. 3C). This lack of a significant change in P/O ratio with time posthomogenization was due to both the numerator and denominator changing in the same direction and at broadly similar rates, but this may not be the case in other tissues or for greater time intervals between assays. The temporal stability of mitochondrial properties should therefore always be assessed when using this method, and delays between sample preparation and measurement kept to a minimum.
The liver mitochondria of trout starved for 14 days showed significant differences in their properties compared to those of fed conspecifics (Fig. 4). The mitochondria from fasted fish exhibited significantly higher rates of oxygen consumption and a trend for an increase in rates of ATP production in comparison with those of fed fish ( Fig. 4A and B). As a result of this substantial increase in JO 2 but lesser increase in JATP, the P/O ratio of liver mitochondria from food-deprived fish was significantly reduced compared to that of fed fish (Fig. 4C). While an increase in coupling efficiency has been suggested to be a primary means by which animals reduce energy expenditure when faced with harsh conditions (Theron et al. 2000;Monternier et al. 2014;Salin et al. 2015), other studies have proposed that the P/O ratio is a major constraint in optimizing the efficiency of energy utilization (Konstantinov et al. 1976;Bobyleva-Guarriero et al. 1984;Abele et al. 2002;Bottje and Carstens 2009;Salin et al. 2012b;Conley et al. 2013). The effect of starvation on mitochondrial properties is also likely to depend on both the tissue (Trzcionka et al. 2008;Brown et al. 2012;Crescenzo et al. 2012) and the animal species involved (Emel'yanova et al. 2007;Bottje and Carstens 2009). A reduction in the apparent P/O ratio in the liver of the food-deprived fish is consistent with previous studies of chickens (Bobyleva-Guarriero et al. 1984), but contrasts with results found in the skeletal muscle of starving penguins (Monternier et al. 2014) and in the livers of starved rats (Jung and Henke 1997). The degree to which mitochondrial properties change may also depend on the severity of the energetic challenge, for example, the duration of food deprivation (Hung et al. 1997;Salin et al. 2010b). It would be profitable to test whether the effect of fasting on mitochondrial coupling efficiency is consistent across different tissues from the same animal, and whether the direction and magnitude of change in P/O ratio is dependent on the duration of food deprivation.

Perspectives and significance
The aim of this study was to measure simultaneously JATP and JO 2 and hence the P/O ratio in tissue homogenate. Our approach was to use the magnesium-sensitive fluorescent probe, Mg Green, to estimate changes in [Mg 2+ ] and in turn JATP, while correcting for ATPase interference. In our study the average P/O ratio for pyruvate, malate and succinate oxidation was 1.367, which is only slightly lower than maximal theoretical values reported for pyruvate and malate oxidation (2.727) and for oxidation of succinate (1.636) (Brand 2005;Watt et al. 2010). This supports our conclusion that the method is appropriate for measuring effective P/O ratios. In the conditions pertaining in our experiment, inhibitors for ATPase failed to fully inhibit ATP hydrolysis (Fig. 1). This limitation has been overcome by inhibiting the ATP-ADP exchanger with a highly specific inhibitor, the cATR, in order to quantify rates of ATP hydrolysis. This extra step of the protocol does not require any more tissue and only a minimal amount of time for the mitochondria to reach a steady state 4. Our method can even be used to analyze P/O ratios from permeabilized cells and isolated mitochondria. Thus, P/O ratio comparisons among species, individuals and tissues can be carried out, even across a range of mitochondrial preparations. The Mg Green method has previously enabled evaluation of ATP production in permeabilized cells and isolated mitochondria (Chinopoulos et al. 2009(Chinopoulos et al. , 2014. However, such samples are not always appropriate in studies of the energy metabolism of an organism. Moreover, ecology and evolutionary biology requires methods for measuring mitochondrial traits that can be applied to nonmodel organisms that potentially exhibit smaller tissues and a lower mitochondrial density than a laboratory mouse. We have for the first time demonstrated the feasibility of a method for measuring mitochondrial coupling efficiencies on homogenized tissue, showing that our method can be used to calculate effective P/O ratios from tissue samples as small as tens of milligrams. This protocol may facilitate the study of P/O ratios in further models, and shed light on the role of mitochondrial coupling efficiency in influencing fitness-related traits.