The relationship between skeletal muscle mitochondrial citrate synthase activity and whole body oxygen uptake adaptations in response to exercise training

Abstract

Citrate synthase (CS) activity is a validated biomarker for mitochondrial density in skeletal muscle. CS activity is also used as a biochemical marker of the skeletal muscle oxidative adaptation to a training intervention, and a relationship between changes in whole body aerobic capacity and changes in CS activity is often assumed. However, this relationship and absolute values of CS and maximal oxygen uptake ($\stackrel{.}{\text{V}}$O_2max) has never been assessed across different studies. A systematic PubMed search on literature published from 1983 to 2013 was performed. The search profile included: citrate, synthase, human, skeletal, muscle, training, not electrical stimulation, not in-vitro, not rats. Studies that reported changes in CS activity and $\stackrel{.}{\text{V}}$O_2max were included. Different training types and subject populations were analyzed independently to assess correlation between relative changes in $\stackrel{.}{\text{V}}$O_2max and CS activity. 70 publications with 97 intervention groups were included. There was a positive (r = 0.45) correlation (P < 0.001) between the relative change in $\stackrel{.}{\text{V}}$O_2max and the relative change in CS activity. All reported absolute values of CS and $\stackrel{.}{\text{V}}$O_2max did not correlate (r =- 0.07, n = 148, P = 0.4). Training induced changes in whole body oxidative capacity is matched by changes in muscle CS activity in a nearly 1:1 relationship. Absolute values of CS across different studies cannot be compared unless a standardized analytical method is used by all laboratories.

Introduction

Cardiac output and not skeletal muscle enzymatic activity is the limiting factor to aerobic performance in healthy people [1]. Nevertheless, adequate muscle enzymatic activity in e.g. glycolysis and Krebs cycle is necessary for a high performance and maximal oxygen uptake ($\stackrel{.}{\text{V}}$O_2max). Enzymatic activity in human skeletal muscle, and in particular citrate synthase (CS) activity, has been used a marker of cellular oxidative capacity and mitochondrial density following a training regimen [2,3]. These enzyme activities are highly adaptable to aerobic training and during exercise a high enzymatic capacity is essential for optimal performance during aerobic exercise [4]. While these characteristics of oxidative enzymes have been known for decades, there is a lack of literature on the relationship between training induced changes in CS activity and whole body $\stackrel{.}{\text{V}}$O_2max. The relationship between $\stackrel{.}{\text{V}}$O_2max and CS activity may provide information on whether cardiovascular and local metabolic adaptations are coupled (i.e. do both systems adapt together), and in which subjects or training types does one change more than the other if one is more important to changes in $\stackrel{.}{\text{V}}$O_2max than the other?

A relationship between changes in $\stackrel{.}{\text{V}}$O_2max and changes in CS activity is assumed and often based on observations from classical endurance training (ET) studies with low intensity and long duration. Most of these studies have shown increased CS activity after training [5, 6] (Table 1), with seemingly similar effect in both genders (Coggan et al., 1992). In the last decade high-intensity interval training (HIIT) has received wide interest as a time-efficient training modality, using a very high intensity for a very short duration. HIIT has been shown to increase CS activity in most but not all studies [7-11].

Table 1.

Included studies

Reference				Group characteristics				Intervention		Aerobic adaptations
Author	Year	Reference [##]	Group number	Group	Group characteristics	n	Gender	Training type	Total time trained (min)	Baseline $\stackrel{.}{\text{V}}$O2max (ml·min-1·kg-1)	ΔCS activity (%)	Δ$\stackrel{.}{\text{V}}$O2max (%)
Allenberg et al.	1988	[64]	1	DI	Patients with type 2 diabetes	7	Males	ET	2232	N/A	36	7
Bakkman et al.	2007	[66]	2	CON	Healthy young untrained	8	Mixed	ET	480	45	21	40
Bangsbo et al.	2010	[67]	3	CON	Untrained running group	25	Females	ET	1920	33	11	15
			4	CON	Untrained football group	25	Females	ET	1920	36	12	10
Barnett et al.	2004	[51]	5	CON	Young healthy untrained	16	Males	HIT	54	N/A	42	8
Berthon et al.	1995	[68]	6	OLD	Healthy Elderly	14	Males	ET	1440	35	46	6
Blomstrand et al.	2011	[49]	7	CON	Healthy young sedentary	5	Mixed	ET	1500	48	29	31
			8	CON	Healthy young sedentary	4	Mixed	HIT	1080	48	7	26
			9	CON	Healthy young sedentary	5	Mixed	HIT	1680	48	32	36
Bruce et al.	2004	[5]	10	DI	T2DM patients	6	Males	ET	1440	28	73	26
			11	CON	Healthy control	7	Males	ET	1440	31	85	18
Bruce et al.	2006	[69]	12	OB	Obese	9	Mixed	ET	2400	24	68	26
Brønstad et al.	2012	[108]	13	DI	COPD patients	12	Males	HIT	288	20	28	-1
Burgomaster et al.	2005	[34]	14	CON	Healthy young recreationally active	16	Males	HIT	198	49	11	6
Burgomaster et al.	2008	[48]	15	CON	Healthy young recreationally active	10	Mixed	HIT	45	41	16	7
			16	CON	Healthy young recreationally active	10	Mixed	ET	1500	41	30	7
Bylund et al.	1977	[4]	17	CON	Young healthy untrained	20	Males	ET	5040	N/A	46	26
Carter et al.	2001	[6]	18	CON	Healthy young active but untrained	8	Males	ET	2100	42	40	17
			19	CON	Healthy young active but untrained	8	Females	ET	2100	32	43	24
Charifi et al.	2003	[65]	20	OLD	Elderly healthy untrained	11	Males	ET	2520	29	33	14
Coggan et al.	1992	[71]	21	OLD	Elderly healthy untrained	12	Males	ET	6864	27	29	24
			22	OLD	Elderly healthy untrained	11	Males	ET	6864	22	17	21
Dawson et al.	1998	[9]	23	TR	Young fit	9	Males	HIT	N/A	57	-32	6
Dubouchaud et al.	2000	[72]	24	CON	Healthy young sedentary	9	Males	ET	3240	44	75	15
Duscha et al.	2012	[43]	25	OLD	Obese healthy middle age/elderly	40	Mixed	ET	32578	28	23	7
			26	OLD	Obese healthy middle age/elderly	47	Mixed	ET	17784	29	39	11
			27	OLD	Obese healthy middle age/elderly	41	Mixed	ET	33670	28	48	20
Ferketich et al.	1998	[42]	28	OLD	Elderly	24	Females	ET	1080	18	11	24
Green et al.	1992	[103]	29	CON	Healthy young active but untrained	9	Males	ET	720	55	5	2
Green et al.	1999	[74]	30	CON	Yong healthy with low Δ$\stackrel{.}{\text{V}}$O2max	7	Males	ET	120	41	21	9
Green et al.	2000	[75]	31	CON	Healthy young sedentary	10	Males	HIT	96	N/A	-2	-2
Green et al.	2009	[76]	32	CON	Young healthy	9	Males	ET	600	48	14	4
			33	TR	Yong healthy with high Δ$\stackrel{.}{\text{V}}$O2max	7	Males	ET	120	51	30	2
Gurd et al.	2010	[77]	34	CON	Healthy young active but untrained	9	Mixed	HIT	720	45	31	11
Harmer et al.	2008	[78]	35	DI	Young T1DM	8	Mixed	HIT	56	N/A	11	-9
			36	CON	Young healthy	7	Mixed	HIT	56	N/A	42	-3
Heilbronn et al.	2007	[15]	37	OB	Insulin sensitive	9	Males	ET	960	48	5	17
			38	DI	Insulin resistant	9	Males	ET	960	48	5	11
Hiatt et al.	1996	[59]	39	DI	Intermittent claudication	10	Males	ET	2160	15	10	17
Houmard et al.	1993	[61]	40	CON	Sedentary healthy middle aged	13	Males	ET	2888	30	69	21
Howarth et al.	2004	[79]	41	CON	Young healthy M	8	Males	ET	2100	N/A	32	6
Iaia et al.	2009	[50]	42	TR	Young healthy trained	17	Males	HIT	68	55	-5	-2
Irving et al.	2011	[80]	43	CON	T2DM offspring	8	Mixed	ET+HIT	945	26	11	0
			44	CON	Healthy control	8	Mixed	ET+HIT	945	27	23	0
Jeppesen et al.	2006	[81]	45	DI	Patients with mtDNA mutations	20	Mixed	ET	1500	26	66	27
Jeppesen et al.	2012	[104]	46	CON	Healthy young sedentary	8	Males	ET	1560	38	36	15
			47	CON	Healthy matched subjects	11	Mixed	ET	1500	34	65	21
Kohn et al.	2011	[10]	48	TR	Young well trained	18	Males	HIT	194,4	67	-4	3
Lange et al.	2000	[82]	49	DI	Healthy elderly Women	8	Females	HIT	2160	22	35	17
LeBlanc et al.	2004	[107]	50	CON	Young healthy	8	Males	ET	2400	N/A	40	15
Linossier et al.	1997	[11]	51	CON	Healthy young recreationally active	8	Males	HIT	1800	52	7	3
Luden et al.	2011	[83]	52	CON	Active young	6	Mixed	ET	N/A	50	66	9
MacDougall et al.	1998	[84]	53	CON	Healthy young recreationally active	9	Males	HIT	84	51	25	3
Mandroukas et al.	1984	[105]	54	CON	Obese	14	Females	ET	1800	N/A	27	19
Martin III et al.	1989	[85]	55	CON	Healthy young sedentary	6	Males	ET+HIT	2520	46	40	20
Masuda et al.	2001	[86]	56	CON	Healthy young sedentary	7	Males	ET	1680	45	28	16
McKenzie et al.	2000	[45]	57	CON	Healthy young sedentary	6	Males	ET+HIT	1500	46	34	12
			58	CON	Healthy young sedentary	6	Females	ET+HIT	1500	38	27	18
Messonier et al.	2005	[87]	59	CON	Untrained young	8	Mixed	ET	2880	43	54	8
Mogensen et al.	2009	[88]	60	DI	Type 2 diabetics	12	Males	ET	625	27	58	11
			61	DI	Obese	11	Males	ET	625	29	37	15
Moore et al.	1987	[21]	62	CON	Healthy young sedentary	4	Mixed	ET	2310	45	59	22
			63	CON	Trained healthy	8	Mixed	ET	2310	N/A	-13	1
Murias et al.	2011	[53]	64	OLD	Active elderly	7	Males	ET	1620	29	48	27
			65	CON	Active young	7	Males	ET	1620	49	67	16
Ngo et al.	2012	[23]	66	OLD	Healthy elderly	5	Males	HIT	3360	38	43	9
Perry et al.	2008	[81]	67	CON	Healthy young recreationally active	8	Mixed	HIT	720	45	26	9
Perry et al.	2010	[90]	68	CON	Healthy young recreationally active	9	Males	HIT	420	N/A	28	12
Putman et al.	1998	[91]	69	CON	Healthy young recreationally active	7	Males	ET	840	45	5	3
Randers et al.	2010	[60]	70	CON	Young healthy	10	Males	ET	4992	40	18	8
Rud et al.	2012	[92]	71	CON	Healthy young sedentary	8	Mixed	ET	1960	N/A	14	6
Schantz et al.	1983	[28]	72	TR	Trained	6	Males	ET	15840	61	0	0
Sjödin et al.	1982		73	CON	Healthy young sedentary	8	Males	ET	280	N/A	11	2
Slivka et al.	2013	[93]	74	CON	Young trained	10	Males	ET	N/A	N/A	11	2
Stannard et al.	2010	[93]	75	CON	Young healthy untrained	7	Mixed	ET	1000	N/A	18	5
			76	CON	Young healthy untrained	7	Mixed	ET	1000	N/A	19	5
Starritt et al.	1999	[95]	77	CON	Healthy young active but untrained	7	Mixed	ET	600	44	26	9
Svedenhag et al.	1983	[54]	78	CON	Healthy young sedentary	8	Males	ET	1280	43	62	7
Svedenhag et al.	1983	[96]	79	CON	Healthy young sedentary	8	Mixed	ET	1280	N/A	75	7
Talanian et al.	2007	[97]	80	CON	Recreational active	8	Females	HIT	280	36	20	13
Tarnopolsky et al.	2007	[44]	81	CON	Healthy young active but untrained	5	Males	ET	2100	43	26	9
			82	CON	Healthy young active but untrained	7	Females	ET	2100	37	3	13
Tiidus et al.	1996	[52]	83	CON	Healthy young sedentary	7	Males	ET	840	48	25	12
			84	CON	Healthy young sedentary	6	Females	ET	840	37	50	22
Tonkonogi et al.	2000	[98]	85	CON	Healthy young untrained	8	Mixed	ET	960	39	47	24
Trappe et al.	2006	[99]	86	CON	Recreational active	7	Mixed	ET	5460	50	37	5
Tynni-Lenné et al.	1999	[47]	87	DI	Patients with heart failure	8	Mixed	ET	480	18	23	3
			88	DI	Patients with heart failure	8	Mixed	ET	480	16	45	19
Vogiatzis et al.	2005	[7]	89	DI	COPD patients	10	Mixed	HIT	1350	N/A	43	9
			90	DI	COPD patients	9	Mixed	ET	900	N/A	40	5
Wibom et al.	1992	[100]	91	CON	Healthy young untrained	9	Males	ET	864	44	43	10
Yfanti et al.	2010	[101]	92	TR	Moderately trained	10	Males	ET+HIT	5400	50	54	18
			93	TR	Moderately trained	11	Males	ET+HIT	5400	51	50	22
Zoll et al.	2005	[102]	94	TR	Young healthy trained	9	Males	ET	2013	64	1	5
			95	TR	Young healthy trained	6	Males	ET	2090	59	-17	3
Østergård et al.	2005	[106]	96	CON	Healthy untrained	29	Mixed	ET	1350	38	25	14
			97	CON	Healthy untrained	19	Mixed	ET	1350	41	25	15

All studies and intervention groups included from search. Group column describes categorization in Figure 5: CON; Young healthy sedentary subjects, DI; studies investigating training in patients with various diseases, TR; Endurance trained subjects at inclusion. Group characteristics column: The group as described by the authors. Training type column: ET; The subjects performed endurance training, HIIT; The subjects performed high-intense interval training. Inclusion $\stackrel{.}{\text{V}}$O_2max column: N/A; not reported clearly in the study. Baseline $\stackrel{.}{\text{V}}$O_2max: $\stackrel{.}{\text{V}}$O_2max reported before the intervention.

Lower CS activity has been reported in elderly compared to equally active young subjects [12]. This has also been observed in a cross-sectional study where CS activity was lower in both sedentary and active elderly subjects compared to young sedentary and active subjects matched for daily activity by the Baecke questionnaire [13] but with a lower $\stackrel{.}{\text{V}}$O_2max per kg fat free mass (FFM) in the elderly subjects [14].

CS activity have been shown to be lower in a group of obese, insulin resistant subjects compared to a group of obese insulin sensitive subjects matched for $\stackrel{.}{\text{V}}$O_2max per kg FFM, but in none of the groups an increase in CS activity was seen in response to 6 weeks aerobic endurance training despite increases in $\stackrel{.}{\text{V}}$O_2max per kg FFM [15]. Thus, the metabolic state of subject may challenge the relationship between training induced changes in CS activity and in $\stackrel{.}{\text{V}}$O_2max.

Analysis of CS in skeletal muscle requires relative small biopsy samples (approximately 15 mg w.w.) and the assay has a relatively low inter- and intra assay variation (below 5% in our laboratory), and the analysis can be done on frozen samples. However, methodological variations and differences in preparation of the biopsies between the different studies is a possible concern. CS activity is traditionally analyzed by the methods described by Lowry and Passonneau [16] or by Srere [17]. The latter is based on a reaction between the thiolgroup in acetyl-CoA which react with Ellman’s reagent (5, 5’-dithiobis-(2-nitrobenzoic acid (DTNB)), which is measured spectrophotometrically [17]. The method by Lowry and Passoneau is based on the conversion of malate to oxaloacetate by reduction of NAD+ to NADH, where the formation of NADH is linear to the CS activity [16]. In this method NADH may be measured both spectrophotometrically and fluorometrically. Different laboratories use these methods with various modifications, different reagents or temperatures (range: 25-37°C) resulting in possible differences in CS activity between laboratories. Furthermore, the analysis may either be done on untreated tissue (wet weight) or tissue that has been freeze-dried and dissected free of visible connective tissue, blood and adipose tissue (dry weight). Using dry weight ensures that the analysis is done primarily on muscle tissue, and not on adipose or connective tissue, which improves the validity of the result. In addition to the analytical considerations, the time from last exercise bout to the biopsy sampling is of importance. Tonkonogi and colleagues showed that CS activity is increased immediately after acute exercise (30 sec. after cessation of exercise) [18]. This finding was later confirmed in females, but surprisingly not in males [19], which is in contrast to another study including trained and untrained males [20].

In the present review we have collected and compared the previous studies in humans in which CS and $\stackrel{.}{\text{V}}$O_2max was measured before and after a training program with the purpose of characterizing the possible relationship between these two variables, and determine which factors that may influence this relationship. Such factors may include the training modality, age, gender, presence of metabolic or other diseases, initial fitness status, and methodological variations. Furthermore, the material allows for a direct comparison of absolute values of CS activity between the different studies with comparable study groups.

Methods

Data sources and search profile

A systematic search of literature on a bibliographical database PubMed published from 1983 to June 2013. We used the search profile: (citrate) AND synthase) AND human) AND skeletal) AND muscle) AND training) NOT electrical stimulation) NOT in-vitro) NOT rats.

Inclusion and exclusion

We included all available studies in which CS activity in skeletal muscle (vastus lateralis) was measured as a marker for improved skeletal muscle oxidative capacity. We limited the search to human studies that included measurements of whole body oxygen uptake ($\stackrel{.}{\text{V}}$O_2max) before and after a physical training intervention program. Studies were excluded if the subjects did not complete an incremental $\stackrel{.}{\text{V}}$O_2max test to exhaustion. Finally, cross-sectional studies and detraining studies were excluded (Table 1 and Figure 1).

Figure 1.

Flow chart. Flow chart of literature search and inclusion/exclusion process for Figures 2 and 3.

Data extraction

Two authors screened the retrieved articles and relevant studies were independently assessed. One author used a standardized form to extract data; a second author controlled the data for accuracy. Discrepancies were resolved by consensus or third-party adjudication. We constructed tables displaying: First authors, publication year, group characteristics, gender, number of subjects, $\stackrel{.}{\text{V}}$O_2max at inclusion, delta CS activity and delta $\stackrel{.}{\text{V}}$O_2max.

The subjects were characterized as described in the study and the groups were primarily stratified according to men/females, young/elderly, trained/sedentary, healthy/disease (Table 1). If not defined in the article we defined elderly as age above 60 years and trained as a $\stackrel{.}{\text{V}}$O_2max above 55 and 50 ml O₂ min^-1·kg^-1 for men and women, respectively.

We wanted to study the isolated effect of HIIT and ET, therefore we excluded studies where detraining and resistance training was used [21-24], where spinal cord injuries were studied [25], electrical stimulation was used as stimulation [26], and studies where other muscles (deltoid or triceps brachii) were biopsied and analyzed [23,27,28].

Furthermore, we excluded a study if the main estimate for changes in aerobic capacity were Watt_max [29-32], a time trial [33-35] or time to exhaustion [36]. This was done to allow a comparison of the relative improvement in $\stackrel{.}{\text{V}}$O_2max by using the same units for endurance performance.

Some studies only reported pre values of citrate synthase activity and/or $\stackrel{.}{\text{V}}$O_2max and hence it was not possible to calculate a relative change [37-41]. Furthermore, 5 studies reported values of CS activity that were more than a factor 10³ different from other studies, when the unit for CS activity was recalculated to the unit used in the present review, μmol·min^-1.g^-1. We assumed in those cases that the reported unit in the original article was erroneous, but accepted the reported value and included the data in Figure 2 [6,42-45]. 5 studies only reported relative changes, and no absolute values of either CS or $\stackrel{.}{\text{V}}$O_2max pre and/or post the intervention, and these studies were not included in Figure 2. Two studies [46,47] reported the same results from the same study and Gordon et al. was excluded.

Figure 2.

Absolute $\stackrel{.}{\text{V}}$O_2max and CS activity before and after interventions. 65 studies (n = 148 data points). The data points are divided by the sample preparation before analysis: wet (no preparation), dry (samples were freeze dried and dissected free of visible blood, fat and connective tissue) of not report (N/A, if the publication did not state clearly how the samples were prepared).

Various terms describing a HIIT training program is used in the included publications (i.e. High Intense Training (HIT), High Intense Interval Training (HIIT), High Intensity Intermittent Exercise (HIIE), and Sprint Training (SIT)). For the purpose of the present review, all of these are termed High Intense Interval Training (HIIT).

Statistics

All statistical analyses were performed in Sigma Plot 12.5 (Systat software, Inc., San Jose, USA). The level of significance was set at P < 0.05. For correlations between different variables Pearson’s product moment correlation coefficient (r) and corresponding P-value were obtained.

Results

Inclusion and exclusion

The literature search identified 180 articles. 110 articles did not meet the inclusion criteria and were excluded. In the remaining 70 articles 149 intervention groups were identified. But 52 intervention groups did not meet the inclusion criteria and were excluded. The main reasons for exclusion were: the groups performed strength training, studied other muscle groups or were control groups. A total of 97 intervention groups including 1000 subjects were included in this review (Table 1 and Figure 1).

Absolute CS activity values

There was no relationship between absolute measures of CS and $\stackrel{.}{\text{V}}$O_2max when we included all time points (n = 148) from studies (n = 65) that reported both CS activity as μmol· min^-1·g (wet or dry weight)^-1 and $\stackrel{.}{\text{V}}$O_2max kg^-1 (r = - 0.07, P = 0.4, Figure 2). 12 studies including 28 study groups reported CS activity relative to dry weight (freeze dried and dissected free of visible connective tissue, lipids and blood). CS activity normalized to dry weight as an isolated factor did not correlate to $\stackrel{.}{\text{V}}$O_2max (r = 0.11, P = 0.60). Neither did the 33 studies with 68 groups that normalized CS activity to wet weight correlate to $\stackrel{.}{\text{V}}$O_2max when analysed alone (r = 0.18, P = 0.17). 20 studies (52 groups) did not report (N/A) clearly how the biopsies were treated prior to analysis, here there was no correlation between CS activity and $\stackrel{.}{\text{V}}$O_2max (r = - 0.14, P = 0.33, Figure 2).

Relative CS activity values

The relative changes in $\stackrel{.}{\text{V}}$O_2max and CS activity in response to a training intervention in 97 intervention groups (Table 1 and Figure 3) were significantly correlated (r = 0.45, P < 0.001). The equation for the trend line is: ΔCS = 1.1 Δ$\stackrel{.}{\text{V}}$O_2max + 16.8. The significant correlation was present also when all the included study groups were stratified according to training type (Figure 4 and Table 1): ET (r = 0.42, n = 69, P < 0.001), and combined ET and HIIT (r = 0.81, n = 7, P < 0.05), but not with HIIT alone (r = 0.24, n = 21, P = 0.29). Stratification according to inclusion background (Figure 5 and Table 1) showed significant correlations in young sedentary subjects (r = 0.35, n = 63, P < 0.05), endurance trained subjects (r = 0.79, n = 9, P < 0.05), and in patients with various diseases and complications (r = 0.67, n = 14, P < 0.05), but not in elderly subjects (r = 0.20, n = 10, P = 0.57). By stratification according to gender (Figure 6 and Table 1) only males correlated (r = 0.63, n = 52, P < 0.001). In studies using females alone (r = 0.57, n = 10, P = 0.08) or groups of mixed gender (r = 0.31, n = 35, P = 0.07), there was only a tendency towards a correlation.

Figure 3.

Relative changes in $\stackrel{.}{\text{V}}$O_2max and CS activity. The relative $\stackrel{.}{\text{V}}$O_2max and CS increase pre and post a training intervention in the 98 included groups. Number refers to the group number in Table 1.

Figure 4.

Type of training. All included groups from search expressed as the relative aerobic improvement and relative CS increase pre and post a training intervention divided by type of training used in the intervention: Endurance training, High-intensity interval training or a combination of Endurance training (ET) and High-intensity interval training (HIIT). The training type is divided by the description by the authors.

Figure 5.

Subject background. All included groups from search expressed as the relative aerobic improvement and relative CS increase pre and post a training intervention divided the subject background included in the study: Young sedentary subjects, patients with various diseases, see Table 1, Obese subjects, Elderly subjects or endurance trained subjects. The characteristics are listed as described by the authors and can be seen in Table 1.

Figure 6.

Gender. All included groups from search expressed as the relative aerobic improvement and relative CS increase pre and post a training intervention divided by the gender included in the study.

Discussion

There is a clear positive and significant correlation between the relative change in $\stackrel{.}{\text{V}}$O_2max and in CS activity in response to physical training (Figure 3). There was almost a 1:1 relationship between the relative change in CS activity and change in $\stackrel{.}{\text{V}}$O_2max. Thus, a ≈ 9% increase in CS activity may be expected from a 10% increase in $\stackrel{.}{\text{V}}$O_2max. It is noteworthy that this relationship was not present when the correlation analysis was constrained to HIIT training or in elderly subjects alone. Oppositely, the relationship was intact when considering young sedentary subjects, trained subjects, and males alone. Likewise, both endurance training studies and studies combining HIIT and endurance training displayed a significant correlation between changes in $\stackrel{.}{\text{V}}$O_2max and CS activity. Absolute values of $\stackrel{.}{\text{V}}$O_2max and CS activity did not correlate, indicating that absolute measures of CS activity cannot be compared across studies and hence not be used for characterization of subject groups between different studies.

Training type (ET and HIIT)

One purpose of this review was to collect and analyze previously published studies in order to determine magnitudes and interrelationships in changes of $\stackrel{.}{\text{V}}$O_2max and CS in response to ET and HIIT. We found a positive and significant correlation between improvements in $\stackrel{.}{\text{V}}$O_2max and increases in CS activity in response to endurance training. This finding was not unexpected, but in contrast to this is, the lack of relationship between improvement in $\stackrel{.}{\text{V}}$O_2max and CS activity in response to HIIT was unexpected. The two forms of training elicited similar average improvements in $\stackrel{.}{\text{V}}$O_2max (ET: ≈ 13% and HIIT: ≈ 8%) but ET (≈ 33%) lead to higher improvement in CS activity compared to HIIT (≈ 19%). This underlines the major importance of cardiac performance for maximal oxygen uptake. Since CS activity in skeletal muscle is well correlated with mitochondrial volume in skeletal muscle [2,3], the lower increase in CS activity with HIIT also indicate that mitochondrial biogenesis may not be stimulated at the same level as ET. In the studies where HIIT did not lead to an increase in CS activity, a significant increase in $\stackrel{.}{\text{V}}$O_2max was found in two [8,9] of these five studies [7,10,11]. The differences in the CS response may be due a large variation in total training time ranging from 45 min [48] to 3360 min [23] and intensity ranging from 75-95% HR_max [23] to 150 % Δ$\stackrel{.}{\text{V}}$O_2max [49] in the included HIIT studies. Another factor is that it is inherent in the nature of HIIT that the time spent training is less than that with endurance training (ET: ≈ 53 hr/study vs. HIIT: ≈ 12 hr/study in the included studies). The high intensity exercise for a short period may apparently be sufficient to elicit a cardiac adaptation (primarily an increase in maximal cardiac output), but not an adaptation of an important enzyme in the Krebs cycle in skeletal muscle.

On the other hand, the lack of significant relationship between ΔCS activity and Δ$\stackrel{.}{\text{V}}$O_2max in the collective HIIT studies may also be due to three distinct studies (no 8, 9, and 23 in Table 1; the 3 triangles in Figure 4 located most low-right) where disproportionate responses were seen. With exclusion of these three studies, a significant correlation is seen (r = 0.48, n = 18, P < 0.05).

Some HIIT studies have been used to induce improvements in endurance performance in already highly trained athletes, measured as time to exhaustion or time trial [10,50]. But these athletes did not have further increases in $\stackrel{.}{\text{V}}$O_2max or CS activity. It is possible that these athletes had already reached a plateau in the metabolic adaptations from the prior ET.

From the data it appears that a 8 wk. HIIT protocol with 3 training sessions per week each consisting of two to six 30 second sprint intervals was a highly time-efficient study [51]. This resulted in a 42% increase in CS activity with a total of 54 min. effective training [51]. Similar improvements in response to HIIT were shown in elderly subjects but after a longer HIIT training period [7]. The largest relative improvement (50-75%) in CS activity was seen in studies with endurance training [5,21,52-54]. These studies are all characterized by a high volume of total training and inclusion of subjects with a relatively low initial whole body $\stackrel{.}{\text{V}}$O_2max. Even though it is highly speculative, it is possible that the nature of HIIT interventions is too short or extreme to allow mitochondrial biogenesis.

Ageing

The expected relationship between in improvements in $\stackrel{.}{\text{V}}$O_2max and CS activity was not observed in the studies (n = 10) with elderly subjects (Figure 5). A 20 % decline in CS activity has been reported with age independent of lifestyle in some studies [14,55], while others are inconclusive [56]. In contrast, other mitochondrial oxidative enzyme activities, for example the activity of complex I-IV are unaltered [14,57]. Therefore, it is possible that adaptability in CS activity is altered with aging independently of changes in mitochondrial respiratory capacity, which has also been shown experimentally [57,58].

A recent study by Duscha and colleagues reports a discrepancy between the relative improvement in CS activity and $\stackrel{.}{\text{V}}$O_2max in 3 groups (40-65 years) that performed different amount and intensity (low amount moderate-intensity, low amount-high intensity or high amount-high intensity training) of aerobic training (group # 25-27, Table 1). Only in the group that performed high amount-high intensity training (r = 0.304, n = 41) a positive correlation between relative $\stackrel{.}{\text{V}}$O_2max and CS activity was seen (group # 27, Table 1) [43,58]. Thus, these findings indicate that in middle-aged and elderly a high amount-high intensity training program is necessary for improvement in both CS and Δ$\stackrel{.}{\text{V}}$O_2max.

Gender

We observed that only studies that included males alone showed significant correlation between $\stackrel{.}{\text{V}}$O_2max and CS activity. In studies (n = 10) including women alone the relationship was only nearly significant (P = 0.08), which is probably due to lack of statistical power. Is has been suggested [19] that transcriptional, translational, and/or post-translational regulation of CS is different between females and males at rest and immediately after acute exercise. However, this notion is not based on sound physiological considerations, and it remains to be proven.

Methodological differences: dry or wet weight?

There was no correlation between absolute values of $\stackrel{.}{\text{V}}$O_2max and CS activity in the included studies. The freeze-drying and dissection procedure of the muscle samples should have eliminated some variation due to contamination with non-muscle tissue/cells, but even in these samples, there was no correlation between the absolute values of $\stackrel{.}{\text{V}}$O_2max and CS. Although the measurements and analytical variation of $\stackrel{.}{\text{V}}$O_2max is well standardized across different laboratories, some day-to-day variation must be expected. Less standardized is the biochemical analysis CS activity. This analysis requires relatively small muscle biopsies, approximately 2-3 mg d.w. corresponding to 10-15 mg w.w. In the authors laboratory CS activity is measured spectrophotometrically as described by Srere [17] at 37°C. The assay has a low inter- and intra assay variation. We find an inter-assay variation of 4.2% in the low range (27 ± 1 (mean ± SD) μmol·min^-1·g (d.w.)^-1) and 0.8% in the high range (613 ± 5 μmol·min-1·g (d.w.)^-1) and an intra-assay variation of 2.5% in the low range (28 ± 1 μmol·min-1·g (d.w.)^-1) and 4.8% in the high range (589 ± 5 μmol·min-1·g protein (d.w.)^-1) (unpublished data). These are lower than was has been reported for analyses in non-freeze dried and un-dissected tissue (4.9% [34], 5.4% [49] and 7.7% [6] in the low range of CS activity). This speaks for analyzing on dissected tissue. Another major factor for variation in absolute values of CS activity is the analytical temperature (usually 25-37°C), which is, unfortunately, not always reported. Increased activity with 37°C compared to 25°C must be expected. Finally, it would be possible to correct data for blood contamination with e.g. creatine correction or other methods, but this is very seldom reported.

Five studies recruited a non-training control group [43,50,59-61]. In these groups no statistical change in Δ$\stackrel{.}{\text{V}}$O_2max or ΔCS activity were reported. However, the ΔCS activity reported varies from 10% decrease (NS) [60] to 14% increase (NS) [61]. This indicates that some physiological time related variation should be expected when measuring.

Responders and non-responders

Despite a positive correlation between Δ$\stackrel{.}{\text{V}}$O_2max and ΔCS activity there is a considerable variation in the relationship (Figure 3). We have suggested that training regimes, subject background or methodological variation contributes to this. However, it has to be considered that there is a significant inter-subject variation in training induced adaptations in $\stackrel{.}{\text{V}}$O_2max, which increases the variation [62,63]. The molecular mechanisms underlying the variation in response to exercise training are still poorly understood, but it is possible that also adaptations in CS activity may be individual. A close inspection of Table 1 and Figure 3 reveals that group 23, 31, 42, 48, 63, and 95 reported a negative ΔCS activity.

Limitations

CS activity and $\stackrel{.}{\text{V}}$O_2max are not always reported both pre and post training. This excluded a large number of studies, and thus removes valuable information. Furthermore, we decided to remove measures of aerobic capacity that was not $\stackrel{.}{\text{V}}$O_2max, but measured as e.g. time trial or Watt_max. There is a large variation in how these tests are conducted, which increases the variation in the results between studies. Cross-sectional studies were not taken into consideration. Inclusion of the many cross-sectional studies in the literature may have provided additional information on the absolute values of CS activity across various studies (Figure 2). Finally, the lack of relationship in females and elderly subjects may be due to a low number of included studies studying these groups, which then may have provided a false negative result.

Conclusions

Most factors (young sedentary or trained subjects, males, ET and combined ET and HIIT) showed a positive and significant linear relationship between Δ$\stackrel{.}{\text{V}}$O_2max and ΔCS activity. This was not the case in publications studying HIIT, females and elderly subjects. CS activity as a marker of mitochondrial density should be used with care in studies using very short term HIIT. The lack of relationship in the females and maybe also in the elderly is most likely a statistical power problem. Finally, a large methodological variation in the analysis of CS activity between laboratories is probably the major reason for a lack of significant relationship in absolute values in $\stackrel{.}{\text{V}}$O_2max and CS activity.

Disclosure of conflict of interest

The authors declare that they have no conflict of interest.