Equipment setup for eccentric contraction and torque measurement
The ankle torque of each animal was measured on a dynamometer, the mechanical setup of which is shown in Figure 1a (same as Nakazato et al., 2007). The torque of a stepping motor (RKD514HA, Oriental Motor, Japan) was transmitted to a footplate. The footplate and its angular velocity were adjustable at 5° intervals. The final deceleration and backlash movement of the footplate were damped using a magnetic powder brake (ZKB-0.3AN, Mitsubishi Electric, Japan). The footplate was positioned such that the anatomical axis of the ankle coincided with the axis of the dynamometer shaft. The plantar flexion force was measured by installing a strain-gauge force transducer (LTB-2KA, Kyowa Electronic Instruments, Japan). The angular position was measured with a potentiometer (LP06M3R1HA, Murata Manufacturing, Japan). Both the force and position signals were sampled at 4000 Hz by using a data acquisition system (PowerLab/16SP, ADInstruments, Australia). We confirmed that this system could evaluate linear relationships from 1. 225 to 2450 mNm. The coefficient of variance (CV) of 5 measurements in same subject was 0.0251.

Procedure for performing eccentric contractions on rat gastrocnemius muscles
Before we performed the ECs, all the animals were anaesthetized with sodium pentobarbital (1 mg/100 g body mass). The right lower leg of each animal was used for experimental intervention. As shown in Figure 1b (same as Nakazato et al., 2007), the anesthetized rats were firmly secured on the platform at ankle joint angle of 0°(defined as the angle at which the sole of foot and tibial bone are orthogonally positioned). The electrically stimulated contraction force was measured as follows. The gastrocnemius muscle was stimulated using pulses of 0.4-ms duration at supramaximal voltage (30 V). The stimulus voltage was adjusted to produce maximal isometric twitch force. The muscle was stimulated at 100 Hz to cause tetanic contraction. During tetanic contraction via a skin electrode, ECs were performed 10 times as a single bout with simultaneous forced dorsiflexion of the ankle joint (velocity, 15°/s; range of motion, from 0°to 45°).

Passive torque against ankle dorsi-flexion
The static PRT of the ankle joint was measured to evalu-ate joint stiffness as shown in our previous study (Ochi, E et al. 2007). The ankle joint of anesthetized rats was dorsi-flexed from 0° (defined as the angle at which the sole of the foot and tibial bone are orthogonally positioned) to either 30° or 45° at an angular velocity of 30°/s. The stress relaxation was allowed to proceed for >90 s until the passive torque reached an almost steady level. The values of torque measured 90 s after the stretch were used as static PRTs. A joint angle of 30 degrees was selected because this angle was the optimal angle observed in preliminary experiment. As for the reason for selecting PRT45 is to compare with previous study(Gillette and Fell, 1996). Six animals were measured at pretreatment, immediately after treatment, and days 1, 2, 3, 4, 5, 6, 8 and 10 after treatment.

Statistical analysis
Results are expressed as mean ± S.D. In the study of PRT, Dunnett's multiple comparison test was performed to compare the forces during pretreatment, immediately after treatment, and on days 1, 2, 3, 4, 5, 6, 8 and 10 after the treatment. The force during pre-treatment at day 0 was used as a control. In the study of the muscle passive tension, the Student's t test was used to compare the EC group and the control group. Significant differences were set at P < 0.05. Pearson's product-moment correlation coefficient was used to assess the relationship between parameters. Significant level was set at P < 0.05.The analysis program used was the statistical package for the social sciences SPSS software for Windows (SPSS Japan Inc., Japan.

Correlation between PRT30 and PRT45
We calculated correlation between PRT30 and PRT45. We confirmed that positive correlations existed between PRT30 and PRT45 (r = 0.839; p < 0.01).

Correlation between passive resistive torque and passive tension
In the study 2, we calculated that correlation between PRT30 and passive tension of gastrocnemius at 10 mm elongation (shown in Figure 5). Significant positive correlation was observed between these two parameters (r = 0.682, p < 0.01.

In this study, we confirmed that ECs raise joint PRT and are associated with reduction of muscle passive tension. The changes in joint stiffness and muscle passive tension after ECs have been examined independently and the direct relationships have not been examined previously (Nosaka and Clarkson, 1995; Whitehead et al., 2001). We experimentally showed that ECs increased both joint PRT and muscle passive tension and these two parameters were significantly correlated. Although the variation of each parameter existed, we think that it is due to the individual differences. Hereafter, we will discuss changes in PRT and muscle passive tension after ECs.

We measured PRT as rat ankle joint stiffness. PRTs of rat ankle joint have not been fully examined previously. Gillette and Fell, 1996 showed that passive resistive tension of rat ankle joint at 45° was about 50 - 60 g, but they did not show absolute torque. Given that the lever arm of the rat ankle joint is 1.5 cm in our present data, then the PRT at 45° is 7.35 - 8.82 mNm. We measured PRT45 and the obtained value ranged from 4 to 15 mNm. Thus, we confirmed that our experimental setups gave relational value for PRT. We also confirmed that significant correlations existed between PRT30 and PRT45 (r = 0.839, p < 0.01, shown in Table.1). Above all, we conclude that our employed parameters of PRT30 and PRT45 accurately reflect stiffness of rat ankle joint.

We confirmed that a significant correlation exists between gastrocnemius muscle passive tensions and PRT on day 4 after ECs. This is the first study to show a direct correlation between them after ECs. Gillette and Fell, 1996 showed that hindlimb uploading made rat ankle joint stiff, and this increased passive torque was due to musculotendinous units, especially gastrocnemius. Although another model was reported, Gillette's study and ours indicate that the viscoelastic properties of gastrocnemius are a major factor for determining ankle joint stiffness. On the other hand, thixotropic behaviors at a relaxed joint are attributed both to the joint structures and to short-range stiffness of muscles acting at the joint (Wiegner, 1987). Since thixotropic behavior is a normal joint characteristic, abnormal gastrocnemius extensibilities become apparent in the eccentric contracted joint.

Gastrocnemius muscle wet mass of the ECs group was similar to that of the sham-operated group. Chleboun et al., 1998 showed that elbow flexor volume increased when elbow joint stiffness increased after ECs of elbow flexor. Since elevated muscle wet mass and muscle volume after ECs suggest existence of edema, our results suggest that factors other than edema contribute to this phenomenon. Proske et al. (2005) proposed that the reason for the rise in passive tension is that after sarcomere disruption by the ECs, there is the likelihood of membrane damage, perhaps at the level of the t-tubules or sarcoplasmic reticulum. The consequent uncontrolled release of Ca²⁺ into the sarcoplasm activates the contractile filaments to develop an injury contracture (Proske and Allen, 2005). We cannot exclude the contribution of edema, but we consider that Ca²⁺ also plays an important role in joint and muscle stiffness after ECs.