Thirteen healthy male students [age = 22.9 ± 2 years, height = 1.77 ± 0.6 m and weight = 74 ± 6 kg (mean ± SD)] with no history of hamstring injuries gave written informed consent to participate in the study. Because estrogen may exert an skeletal muscle protective effect (Kendall and Eston, 2002; Enns and Tiidus, 2010), maybe by increasing membrane stability and lowering CK in women (Hicks et al., 2016), female subjects were not recruited. The participant's dominant leg was determined by asking them their preferred leg when kicking a ball (Hody et al., 2013). However, there is lack of consensus in the literature on how to define lateral dominance since it can be determined according to a variety of criteria, such as strength, functional use, or performance in specific skills (Hoffman et al., 1998). The fitness level of the subjects varied according to their daily physical activity; two subjects were sedentary and the other 11 were “moderately active” to “physically active” (exercising 2–5 days·week⁻¹; Table 1). The subjects had not performed systematic heavy-resistance strength training for at least 6 months prior to the experiments. Moreover, they were asked not to perform any exercise during the week before or at any time during the experimental period. The study complied with the code of ethics of the World Medical Association (Declaration of Helsinki) and was approved by the Ethics Committee of the Catalan Sports Council (Generalitat de Catalunya) (099S/690/2013).

Changes in muscle function were examined for 1 week following a bout of ELCs performed with each leg. The recovery of the muscle FGC was assessed by repeated tests of isometric MVC during a knee flexion from a prone position. The first MVC test was performed before the eccentric exercise to establish baseline values and the rest of the tests were repeated 24, 48, and 72 h and 7 days after the exercise.

To monitor muscle enzyme leakage, blood samples were taken immediately before assessment of the FGC at baseline and at 24, 48, 72 h and 7 days after the exercise. To conduct a non-invasive analysis of the physiological changes occurring in the muscles involved during the exercise, fMRI was performed just before assessment of the FGC at baseline and at 24 h and 7 days after the ELCs. Finally, perceived muscle soreness in the hamstring muscles was also assessed using a visual rating scale (VRS) before and 24, 48, and 72 h and 4, 5, 6, and 7 days after the exercise protocol (Figure 1).

Distributions were considered for each of the variables, with the normality of continuous variables assessed by the Shapiro-Wilk test. Asymmetrically-distributed variables were log transformed before analysis. One-way repeated-measures ANOVA was used to identify the effect of time on serum enzyme levels. If significant effects were found, a post-hoc Bonferroni-corrected paired t-test was applied to identify significant differences from baseline values. A two-way repeated-measures ANOVA (leg dominance × time) was performed to identify the main effects of leg dominance on the MVC. A two-way repeated-measures ANOVA [leg (leg that showed the larger and leg that showed the smaller decrease in FGC) × time] followed by a paired t-test with a Bonferroni correction was performed to identify statistically significant differences from baseline MVC and muscle soreness values as well as differences between legs following ELCs. A three-way repeated-measures ANOVA [leg muscle (leg that showed the larger and leg that showed the smaller decrease in FGC) × region × time] followed by a paired t-test with a Bonferroni correction was performed to identify statistically significant differences from the baseline T2 values and differences between the hamstring muscles and their respective measured regions (proximal, middle, and distal) after the exercise protocol. Differences between the groups of subjects (high and moderate responders) were assessed using the Mann-Whitney U test. Associations between variables of interest were assessed using Spearman's rank correlation coefficient. Data are presented as mean ± standard error of the mean (SEM), unless otherwise stated. The level of significance was set at P < 0.05. Statistical analysis was performed with SPSS version 20.0 (SPSS Statistics, IBM Corp., Armonk, NY, USA).

According to the MVC data, muscle damage was classified as severe or moderate following the criteria presented above (see Materials and Methods). Subjects were then grouped as “high” (n = 10) or “moderate” (n = 3) responders according to the degree of hamstring muscle [ST and BF (BFlh and BFsh)] damage. Except in subjects 5, 6, and 8, the leg with the larger decrease in the FGC of the hamstring muscles was the non-dominant one, but there was no significant effect of leg dominance (dominant or non-dominant) on measured variables. No baseline differences in the MVC were found between (high and moderate responders) or within (leg-to-leg comparison in the high responders) subjects (Figure 2, inset graphs). The time-related changes in the MVC of the high and moderate responders are shown in Figure 2. Legs showing the greater FGC reduction of the hamstring muscles in the high responders suffered significant MVC decreases of 34 ± 7%, 52 ± 7%, 39 ± 8%, and 38 ± 7% at 24, 48, 72 h and 7 days after the ELCs, respectively. Unexpectedly, among the high responders, the leg with the smaller reduction in the MVC of the hamstring muscles only presented statistically significant decreases of 18 ± 6% and 34 ± 9% at 24 and 48 h, respectively, the percentage of MVC reduction differing significantly between legs at 48 h and 7 days after the ELCs (Figure 2). The three moderate responders (subjects 1, 7, and 9) showed a surprisingly low FGC reduction in both legs, with significant differences in the MVC reductions between the high and moderate responders occurring 48 h and 7 days after the exercise regimen (Figure 2).

Perceived muscle soreness among the high responders during stretching and contracting (isolated unloaded knee flexion from a standing position) was significantly elevated over baseline values at every time-point analyzed. Muscle soreness increased after the exercise protocol and peaked at 72 h, with a value of 7.8 ± 0.3 arbitrary units (a.u.). Significant differences in muscle soreness were found between the high and moderate responders at 72, 96 and 120 h after the exercise (Figure 3). There were no differences within subjects (leg-to-leg comparison in the high responders). Correlations between muscle soreness and peak and long-lasting MVC reductions observed 48 h and 7 days after exercising are shown in Table 2.

Significant T2 increases of 146, 172, and 186% were found in all the sections of the ST muscle [proximal (30%), middle (50%), and distal (70%), respectively] among the high responders 7 days after the exercise Table 3. Moderate T2 increases of 67% revealed that the BFsh muscle in the high responders was also damaged 7 days after the ELC regimen. There were no T2 differences in the BFlh muscle with regard to baseline values. Significant differences between the high and moderate responders were found in the ST T2 values from both legs 7 days after exercising. Surprisingly, there were no T2 differences between the legs showing the larger and lower FGC reduction of the hamstring muscles following the ELC bout Table 3. Analyses of variance revealed no T2 differences among the ST muscle regions assessed. Finally, the T2 values obtained 7 days after the exercise protocol from all the ST sections of the leg with the larger FGC decrease of the hamstring muscles correlated with the reductions in the MVC, but there was no such correlation in the leg with the smaller FGC decrease in the hamstring muscles (Figure 6). Furthermore, there were no correlations between the MVC reductions and the BFsh T2 values.

We first classified the 13 subjects into two groups based on the leg that suffered the larger decrease in MVC following intensive ELCs. This is a reliable measure that is considered the best indirect marker of the extent of EIMD (Warren et al., 1999; Hubal et al., 2007; Paulsen et al., 2012). Two groups of subjects were identified according to the extent of EIMD, which was indicated by peak reductions in the FGC (reflecting the number of fibers with ultrastructural and myofibrillar disruptions (Lieber et al., 1994; Lauritzen et al., 2009; Raastad et al., 2010; Paulsen et al., 2012) and the time required for recovery. The high responders (n = 10) showed an average peak MVC loss of 52% and/or recovery that was not completed in 1 week (severe EIMD), while the moderate responders (n = 3) presented an average peak MVC loss of 21% and/or recovery that was achieved in 2–7 days (moderate EIMD) (Figure 7). Large reductions in the hamstring FGC among the high responders suggest that severe muscle damage can be induced in humans following voluntary contractions (Lauritzen et al., 2009).

There were significant differences in serum enzyme activities or concentrations following the ELC regimen between the high and moderate responders. The severe ST muscle damage in the high responders was further highlighted by the clear increases in serum muscle enzyme levels. Serum CK elevation, to some extent, reflects the amount of myofibrillar damage, especially when the damage is severe (Lauritzen et al., 2009; Raastad et al., 2010). Moreover, large increases in serum CK and myoglobin levels 4–5 days after exercising may indicate degradation of protein structures and segmental myofiber necrosis (Brancaccio et al., 2010; Paulsen et al., 2012). The correlation between CK serum levels and MVC (FGC decreases) 7 days after the exercise suggests that the long-lasting increases in CK levels were related to myofiber necrosis. The CK increases in the high responders 7 days after the intensive ELCs (range: 4,198–42,972 U·L⁻¹) were similar to the CK elevation observed in exertional rhabdomyolysis (Kenney et al., 2012). However, exertional rhabdomyolysis was excluded since eccentric exercises can produce marked elevations in serum enzyme levels without renal compromise (Clarkson et al., 2006; Scalco et al., 2016) and we also found no laboratory evidence of acute renal failure in any of the three subjects who reported dark urine following the ELC regimen. Although hypertransaminasemia was also observed, hepatic damage was also ruled out because the hepatic injury markers ALP and γGT remained unchanged after the intensive ELCs in all the subjects (Bessa et al., 2008). AST and ALT are ubiquitously present in most tissues, including skeletal muscle (Botros and Sikaris, 2013; Dahlqvist et al., 2013), and hypertransaminasemia is commonly present in patients with high CK levels resulting from extreme exercise regimens (Nathwani et al., 2005). We can therefore deduce that the observed hypertransaminasemia following ELCs is linked to muscle damage.

A small biphasic pattern in CK, myoglobin, AST and ALT levels was observed in the moderate responders. An earlier study reported a biphasic enzymatic response, but only in high responders who suffered severe muscle damage after eccentric exercises (Paulsen et al., 2010). This observed biphasic pattern could be related to an initial increase in sarcolemmal permeability produced by mechanical straining during an exercise, with the following peak in serum enzyme levels 7 days after the ELCs due to segmental myofiber necrosis (Lauritzen et al., 2009; Paulsen et al., 2010). The serum enzyme levels in the moderate responders 7 days following intensive ELCs could be due to minor myofiber necrosis since minimal sarcomere disruptions can occur without significant changes in the FGC (Gibala et al., 2000), probably because force can be transmitted laterally through neighboring myofibrils and the extracellular matrix (Bloch and Gonzalez-Serratos, 2003; Grounds et al., 2005).

Serum increases of sMtCK were of particular interest because, in contrast to the other enzyme markers, sMtCK levels were absent or only marginally increased in the serum of the moderate responders. Large increases of serum myoglobin might be linked to mitochondrial damage because the release of iron ions from the haem of myoglobin promotes the peroxidation of mitochondrial membranes (Plotnikov et al., 2009). Mitochondria are key regulators of many cellular processes like Ca²⁺ regulation and apoptosis (Newmeyer and Ferguson-Miller, 2003). The mitochondrial intermembrane space contains several pro-apoptotic proteins that can lead to cell death upon release into the sarcoplasm (Primeau et al., 2002). sMtCK acts as an energy sensor, coupling the cellular energy state to cell apoptosis (Schlattner et al., 2006). Apoptosis has been suggested to be involved in muscle remodeling and repair following electrically-induced eccentric contractions in rats (Wistar) (Biral et al., 2000). The significant increases in serum sMtCK found in the high responders are likely to be indicative of mitochondrial swelling and reduced muscle respiratory capacity (Chen et al., 2000). Given the correlations observed between serum sMtCK levels and MVC reductions 48 h and especially 7 days after intensive ELCs, FGC reductions could indicate the number of myofibrils that have died by either apoptosis (due to a process triggered by large increases in sMtCK) and/or necrosis. Moreover, prolonged exercise has been demonstrated to induce mitochondrial damage in rats (Sprague-Dawley), with the training status having a protective effect on the mitochondria against exercise (Chen et al., 2000) via a specific reduction of mitochondrial Ca²⁺ uptake (Bonner et al., 1976) and increased antioxidant capacity (Venditti and Di Meo, 1997). Although only observed in animal models (Sprague-Dawley rats) (Chen et al., 2000), it seems reasonable to assume that the training status of the moderate responders might confer a protective effect on the mitochondria against eccentric EIMD.

The high and moderate responders did not differ in any of their baseline measurements, although there were differences in the levels of sport activity and training status (Table 1). In this regard, the moderate responders participated in athletics (participants 1 and 9) and football (participant 7), which involve sprint accelerations over different distances. Since hamstring eccentric force and activation are important in human sprint running mechanics (Kyröläinen et al., 2005; Schache et al., 2011, 2014; Morin et al., 2015; Sun et al., 2015), the practice of athletics and football might have produced a “repeated-bout effect” (RBE), which describes a protective adaptation occurring after a single bout of unaccustomed eccentric exercise that induces muscle damage (McHugh, 2003). Although adaptive mechanisms underlining the RBE remains unclear, recent well-supported theories suggest that eccentric exercises induce changes to connective tissue structures, such as extracellular matrix remodeling (Hoffman et al., 2016), and improves mitochondrial Ca²⁺ homeostasis, thereby stabilizing mitochondrial respiratory function (Rattray et al., 2013). In the present study, an enhancement of mitochondrial metabolism could have occurred since moderate responders showed no serum increases of sMtCK. Moreover, the chronic exposure to eccentric-biased exercise (i.e., sprint accelerations in athletics or football) might have induced hamstring muscles architectural changes in the moderate responders. Because sarcomere lengthening is non-uniform (Moo et al., 2016, 2017) and fascicles may have a regional response (i.e., more pronounced mechanical stretch might have been applied to single sarcomeres and thus fascicles) (Franchi et al., 2017), increases in fascicle length might be expected as an adaptive response to eccentric exercise. Those changes in fascicle length theoretically imply the addition of sarcomeres in series (Franchi et al., 2017), which may had a positive effect on eccentric hamstring strength at longer muscle lengths (Brockett et al., 2001) and produced a protective effect on the hamstring muscles (Timmins et al., 2016). While speculative, the RBE as well as hamstring muscles longer fascicle lengths as an adaptive mechanism to the type of sport practiced (eccentric exercises involving the hamstring muscles, e.g., sprinting), and the activity level (training status) could explain the smaller decline in hamstring muscle function and the low serum sMtCK levels in the moderate responders. The extraordinary tolerance to intensive eccentric exercise shown by moderate responders makes it necessary to analyze, in future studies, the extent of muscle damage in a group of eccentric resistance-trained subjects, from which a mild response might be expected (i.e., mild MVC decreases of <20% of the baseline values and recovery completed on 24–48 h) (Paulsen et al., 2012).

Experimental designs based on an observational comparison of response using contralateral limbs in a single group have been applied to studies analyzing muscle damage (Mackey et al., 2008; Paulsen et al., 2010). This kind of experimental design is based on the assumption that changes in the markers of muscle damage between limbs are similar; however, our study also found significant differences in muscle function within subjects among the high responders. To the best of our knowledge, this is the first study comparing hamstring muscle damage within subjects (limb-to-limb comparison) in response to the same unilateral leg curl exercise performed with each leg. The high responders showed significant differences in the MVC reductions between legs after exercising, indicating that, contrary to our hypothesis, hamstring muscles (ST and, to a lesser extent, BFsh) from each leg suffered different degrees of damage. As observed in previous studies, these differences in the magnitude of EIMD were not influenced by lower limb dominance (Hody et al., 2013). One limitation in the present investigation was that no force or velocity data were recorded during the exercise. Therefore, although the same exercise load was placed on both legs and the subjects tried to maintain a lowering velocity that was as constant as possible during the 3-s eccentric phase, slightly different force levels could have been produced, particularly at the later stages of ELC due to fatigue. During the last set of the ELC regimen, the high responders could not maintain a constant lowering velocity over long muscle lengths. Since damage occurs when sarcomeres are actively stretched to lengths corresponding to that of the descending limb of the length-tension relationship (Morgan, 1990; Hoffman et al., 2016), the force and velocity production could be important, especially at long muscle lengths. Furthermore, no baseline differences were observed in the MVC between legs. However, baseline hamstring MVC was measured at an optimal sarcomere operating length (hip flexion, 40°; knee flexion, 30°) and baseline differences at long muscle lengths cannot be ruled out. Future investigations should carefully monitor force production during the range of motion or, at least, measure baseline MVC at longer muscle lengths.

Differences within subjects were not observed in the fMRI T2 results. The T2 data and MVC reductions correlated in all the ST regions measured in the leg with the greater FGC reduction of the hamstring muscles, but no such correlation was observed in the leg with the smaller FGC reduction. T2 values have been shown to remain elevated following eccentric exercises, even when other markers of muscle damage (i.e., FGC and serum enzyme levels) have returned to baseline values (Clarkson and Hubal, 2002). Indeed, T2 values have been found to remain elevated for as long as 75 days (Shellock et al., 1991) and 31 days (Nosaka and Clarkson, 1996) following (in both cases) eccentric exercising of the elbow flexor muscles. Foley et al. (1999) suggested that long-lasting T2 increases after other markers of EIMD have returned to baseline values might reflect an adaptive process rather than damage. In fact, T2 fMRI scans provide information on the metabolic state of the muscle fibers (Cagnie et al., 2011; Mendiguchia et al., 2013b), with the higher the T2 signal after an exercise, the higher the metabolic activity of the muscle (Schuermans et al., 2016). Since muscle function in the leg with the smaller FGC reduction of the hamstring muscles recovered 7 days after the exercise protocol (i.e., MVC returned to baseline values), it seems reasonable to link the long-lasting increases in T2 values to an enhanced remodeling of the ST and BFsh muscles, an adaptation consistent with protection against further damage. Although changes in T2 signals are a relevant marker of muscle damage (Larsen et al., 2007; Black and McCully, 2008; Fouré et al., 2015), our results indicate that fMRI must be accompanied by muscle function assessments to measure EIMD.

GR was employed by company FC Barcelona. JM was employed by company Zentrum Rehab and Performance Center. All other authors declare no competing interests.