Experiments were performed on ten male Wistar rats (body mass 312.9 ± 17.7 g, mean ± s.d.). All procedures were in agreement with the guidelines and regulations concerning animal welfare and experimentation set forth by Dutch law, and approved by the Committee on Ethics of Animal Experimentation at the VU University Amsterdam (Permit Number: FBW 12-01).

According to standard procedures in our laboratory (Maas et al., 2001) the animals were deeply anesthetized by intraperitoneally injected urethane (1.2 ml/100 g body mass). If necessary, additional doses were given to suppress any reflexes. During surgery and data collection, body temperature was maintained at ~37°C. To prevent dehydration of exposed tissues, saline solution was applied regularly. At the end of the measurements, animals were euthanized with an overdose of intracardially-injected pentobarbital sodium followed by a double-sided pneumothorax.

The surgical procedures have been described in more detail elsewhere (Bernabei et al., 2015a) and will, therefore, be described only briefly here. The posterior crural compartment of the right hindlimb was exposed by removing the skin and biceps femoris muscle and the femur was exposed to allow attachment of a metal clamp. Medial gastrocnemius (MG) was removed fully without damaging muscle fibers of LG. This was done to prevent unphysiological strain between MG and LG as only the MTU length of LG+PL was changed (see Experimental protocol). All structures surrounding the SO, PL, LG muscle group were cut, but myofascial connections between their muscle bellies were left intact.

The hindlimb was positioned at ankle and knee angles of 90°. Markers were placed on the lateral collateral ligament and distal tendon of the peroneus muscle, and served as proximal and distal reference markers, respectively (Figure 1A). Markers on the distal SO tendon, distal LG tendon and proximal LG aponeurosis were placed parallel to the position of the reference makers. As result, changes in marker positions could be expressed relative to the reference position (L_REF) at 90° ankle and knee angles.

LG and SO tendons merge distally into the Achilles tendon before inserting on the calcaneus. To measure individual tendon forces, these tendons were separated. The distal SO tendon was cut and connected, using Kevlar thread and a metal rod, to a force transducer (ALPHA load beam transducer, 25N maximum capacity, max output error <0.1%, compliance 0.0162 mm/N; BLH Electronics Inc., Toronto, Canada). Distal LG and PL tendons were tied together (LG+PL complex) and separated from the skeleton by cutting a small piece from the calcaneus. The proximal LG+PL tendon was separated by cutting a small bone fragment from the lateral epicondyle. Both tendons were attached to force transducers (Z6 bending beam load cell, 50N maximum capacity, max output error <0.1%, compliance 0.0048 mm/N; HBM, Darmstadt, Germany) using Kevlar thread and metal rods. The three force transducers were aligned to the muscles' lines of pull.

To excite SO, the sciatic nerve was partly dissected free for placement of a cuff electrode. In addition, all branches distally to the cuff electrode were cut, except the branch innervating SO muscle (Maas and Sandercock, 2008; Tijs et al., 2014). At the end of the experiment, it was checked whether indeed only SO muscle was excited. In five animals, a not previously reported additional branch to PL was found at the SO-PL muscle belly interface and, hence, SO was not excited exclusively. These animals were excluded from the analysis. LG has a tri-pennate structure with multiple compartments, which can be excited separately, and multiple motor endplates (Prodanov et al., 2005). Bipolar intramuscular wire electrodes were inserted near motor endplates located in the proximal region of LG. Therefore, LG was excited partially.

SO muscle was excited by supramaximal stimulation of the sciatic nerve (amplitude: 0.4–0.5 mA, frequency: 100 Hz, pulse width: 100 μs) via the bipolar cuff electrode connected to a constant current source (Digitimer DS3, Digitimer Ltd., Hertfordshire, England). LG muscle was excited partially via the intramuscular electrodes (amplitude: 0.8–3.0 mA, frequency: 100 Hz, pulse width: 100 μs). Because only one LG compartment was stimulated, not all muscle fibers were excited. At the highest length tested (L_REF+3 mm, see below), this yielded a force of ~3N. For the same length, but during maximal excitation of the LG+PL complex via nerve stimulation, a force of ~14 N has been found (Bernabei et al., 2015a). Because the ratio of maximal forces exerted by LG and PL is ~3:1 (Johnson et al., 2011), we estimated that 10.8 N was exerted by LG and 3.2 N by PL in the study of Bernabei et al. (2015a). Based on these data, we estimated that ~28% of LG fibers were excited in the present study.

The distal markers of SO and LG were kept at reference positions (L_REF, see above, Figure 1A). Thus, SO was kept at a constant MTU length corresponding to a 90° ankle angle. The MTU length of the LG+PL complex was changed proximally (i.e., simulating knee extension) from −3 mm to +3 mm relative to L_REF with increments of 1 mm. These length changes correspond to knee joint angles of ~45° and 130°, respectively (Bernabei et al., 2015a).

For each LG+PL length, four combinations of SO and LG muscle activation were applied (Figure 1B): passive SO and LG muscles, separate excitation of either SO or LG, and simultaneous excitation of SO and LG (SO&LG).

Forces exerted at the distal SO tendon (F_SO), as well as proximal (F_{LG+PL_PROX}) and distal LG+PL tendons (F_{LG+PL_DIST}) were assessed before (passive force) and during SO, LG, and SO&LG excitation (total force). Total force was assessed as the mean of the total force output of the last 50 ms of tetanic stimulation.

Intermuscular mechanical interaction was assessed for different SO and LG activation protocols using two measures: (i) changes in F_SO with changes in LG+PL MTU length; and (ii) changes in the difference between F_{LG+PL_PROX} and F_{LG+PL_DIST} (i.e., ΔF_LG+PL = F_{LG+PL_PROX} - F_{LG+PL_DIST}) with changes in LG+PL MTU length. ΔF_LG+PL was used because it is a direct indication of the net epimuscular myofascial force transmission between LG+PL and surrounding tissues (Huijing and Baan, 2001).

In addition, the time-varying force trace of F_SO was normalized to the mean total force for each LG+PL length during SO and SO&LG excitation. The duration of F_SO decrease from its peak value after the last 50 ms of tetanic stimulation to 50% of that value (half-relaxation time, HRT) was calculated. Because changes in HRT are an indication of changes in SO fascicle length (Wallinga-de Jonge et al., 1980; Maas and Sandercock, 2008), we can potentially distinguish between different mechanisms of intermuscular mechanical interaction: epimuscular myofascial connections may affect 1) SO muscle fascicle length and, therefore, the amount of force exerted by SO; or 2) the SO force distribution between intramuscular and epimuscular pathways, while the total amount of force exerted by SO is unaffected.

Two-way repeated measures ANOVAs (SPSS 21, IBM, USA) with “LG+PL length” and “muscle activation” as fixed factors and F_SO as dependent variable were performed for the combination of passive SO with either active or passive LG and for the combination of active SO with either passive or active LG. Two-way repeated measures ANOVAs with “LG+PL length” and “muscle activation” as fixed factors and ΔF_LG+PL as dependent variable were performed for the combination of passive LG with either active or passive SO and for the combination of active LG with either passive or active SO. In case of significant interaction effects, one-way repeated measures ANOVAs were used to test for effects of “LG+PL length” on F_SO and ΔF_LG+PL for the passive and active muscle conditions, separately. One-way repeated measures ANOVA with “LG+PL length” as fixed factor and HRT of F_SO as dependent variable was performed for the SO excitation and SO&LG excitation condition. Greenhouse Geisser correction was used if the assumption of sphericity was violated. Level of significance was set at p ≤ 0.05.

ANOVA indicated a main effect of LG excitation (p = 0.021), a main effect of LG+PL length (p = 0.009) and an interaction effect between LG excitation and LG+PL length (p = 0.017) on F_SO. For passive muscles, F_SO increased from 0.003 ± 0.002 N at L_REF−3 mm to 0.008 ± 0.006 N at L_REF+3 mm (p = 0.039, Figure 2A, see Table 1 for mean slope values). For LG excitation only, F_SO increased from 0.007 ± 0.005 N at L_REF−3 mm to 0.023 ± 0.012 N at L_REF+3 mm (p = 0.009).

For active SO conditions (Figure 2B), ANOVA indicated a main effect of LG+PL length (p = 0.001) and an interaction effect between muscle excitation and LG+PL length (p = 0.002) on F_SO. However, no main effect of LG excitation on F_SO was found (p = 0.053). For SO excitation only, F_SO ranged from 1.35 ± 0.17 N at L_REF−2 mm to 1.38 ± 0.16 N at L_REF+3 mm (p = 0.006, Figure 2B), which is a 2.4% change expressed relative to the force exerted at L_REF. During SO&LG excitation, F_SO increased from 1.38 ± 0.13 N at L_REF−3 mm to 1.44 ± 0.12 N at L_REF+3 mm (p = 0.001), which is a change of 4.3% relative to the force exerted at L_REF. These results indicate that effects of LG+PL MTU length changes on F_SO were enhanced by synergistic muscle activation.

ANOVA indicated no effect of LG+PL length on the HRT of F_SO (Figure 3), both during SO excitation (average across LG+PL lengths: 102.5 ± 9.9 ms, p = 0.102) and during SO&LG excitation (average across LG+PL lengths: 103.5 ± 10.5 ms, p = 0.081). These results suggest that SO fascicle length was not affected by LG+PL length.

ANOVA indicated a main effect of LG+PL length (p = 0.002), a main effect of SO excitation (p = 0.041), and an interaction effect between SO excitation and LG+PL length (p = 0.007) on ΔF_LG+PL. For passive muscles, ΔF_LG+PL changed from −0.017 ± 0.005 N at L_REF−3 mm to 0.077 ± 0.033 N at L_REF+3 mm (p = 0.002, Figure 4A). For SO excitation only, ΔF_LG+PL ranged between −0.112 ± 0.068 N at L_REF−2 mm and 0.012 ± 0.068 N at L_REF+3 mm (p = 0.002). These results indicate that, when LG is passive, SO muscle activation affected the relationship between LG+PL MTU length and ΔF_LG+PL.

For active LG (Figure 4B), ANOVA indicated a main effect of LG+PL length (p = 0.038) and a main effect of additional SO excitation (p = 0.024) on ΔF_LG+PL, but no interaction effect (p = 0.161). For LG excitation only, ΔF_LG+PL changed from −0.116 ± 0.162 N at L_REF−3 mm to 0.083 ± 0.028 N at L_REF+3 mm (Figure 4B). During SO&LG excitation, ΔF_LG+PL changed from −0.166 ± 0.143 N at L_REF−3 mm to 0.056 ± 0.030 N at L_REF+3 mm.

For both passive and active LG conditions, ΔF_LG+PL was more negative in response to SO excitation, which implies either an increased F_{LG+PL_DIST} or reduced F_{LG+PL_PROX}. Because passive forces are small at low LG+PL MTU lengths (e.g., L_REF−3 mm: F_{LG+PL_PROX}: 0.002 ± 0.001 N, F_{LG+PL_DIST}: 0.019 ± 0.004 N), these results indicate that active force generated by SO muscle fibers was exerted at the distal LG+PL tendon.

The force difference between the proximal and distal tendons of LG+PL is a direct measure of net epimuscular myofascial force transmission between LG+PL and its environment (Huijing, 2009; Maas and Sandercock, 2010). The change in F_SO with changes in LG+PL MTU lengths is a measure of the extent of mechanical interaction between LG+PL and SO. In the present study, we found that the mean slopes of ΔF_LG+PL with changes in LG+PL MTU lengths were higher than the mean slopes of F_SO (see Table 1). This difference in slopes can be caused by (i) force transmission from LG+PL to adjacent structures other than SO, such as the neurovascular tract supplying LG+PL, and/or (ii) changes in the amount of force transmitted between LG+PL and the proximal tendon of SO. Although these pathways could not be distinguished in the present study, the amount of force transmission via each pathway was recently estimated using a phenomenological lumped-parameter model (Bernabei et al., 2016). It was found that, for most of LG+PL MTU lengths tested, the stiffness of the neurovascular tract supplying LG+PL was higher than that of structures between LG+PL and SO. The results from that study suggest that the difference in mean slopes of F_SO and ΔF_LG+PL found in the present study is likely the consequence of force transmitted from LG+PL to adjacent structures other than SO instead of force transmission between LG+PL and the proximal tendon of SO.

A previous study reported that, during maximal excitation of whole SO, LG, and PL muscles, proximal lengthening the LG+PL complex increased F_SO by ~10% (Bernabei et al., 2015a). In the present study, a lower increase in F_SO with LG co-activated was found (4.3%). This can be explained by the fact that, in the present study, PL was passive and LG was excited only partially (≈28%, see methods). Nonetheless, the effects of LG+PL length on distal SO tendon force during SO&LG excitation were more pronounced than during excitation of SO only, indicating increased intermuscular mechanical interaction. The combined results of the present and that of a previous study (Bernabei et al., 2015a) confirm that the extent of intermuscular mechanical interaction is dependent on the amount of synergistic muscle activation: higher levels of synergistic muscle activation (passive vs. partial stimulation vs. whole stimulation) resulted in increased mechanical interaction between LG+PL and SO.

We found significant interaction effects between LG+PL length and muscle activation on F_SO and ΔF_LG+PL, which could be caused by increased stiffness of myofascial linkages (Maas and Sandercock, 2008). A recent study has found less relative displacement between human LG and SO fascicles during LG excitation than in passive muscle conditions (Finni et al., 2015). This may be explained by a higher stiffness of intermuscular myofascial linkages and, hence, an increase in the extent of intermuscular mechanical interaction.

Although effects of LG+PL length on F_SO increased when SO and LG muscles were excited, the highest changes in F_SO were still rather low (0.0101 N/mm, Table 1). Because the largest relative muscle displacement occurred proximally (i.e., LG+PL MTU length was increased proximally), the rather low mean slopes of F_SO suggests that proximally located connections at the interface between SO and LG+PL were slack or in the toe region of the stress–strain curve (Sandercock and Maas, 2009), and oriented such that they allow LG+PL muscle bellies to move proximally with minimal resistance.

During both SO and SO&LG excitation, LG+PL length did not affect the half relaxation time of F_SO, suggesting no changes in SO fascicle length. As SO muscle excitation was equal, the amount of force generated by SO muscle fibers can be assumed constant. Therefore, the increased slope of F_SO during SO&LG excitation compared to excitation of only SO should be explained by other mechanisms: with increasing LG+PL length proximally, (1) more LG muscle force is exerted on the distal SO tendon; (2) more force produced by SO muscle fibers is exerted on its own distal tendon instead of on the distal LG+PL tendon. We found that ΔF_LG+PL was more negative during SO excitation than during passive muscle conditions and more negative during SO&LG excitation than during LG excitation (Figure 4). A negative ΔF_LG+PL means that F_{LG+PL_DIST} is higher than F_{LG+PL_PROX}, most likely resulting from active fore generated by SO transmitted to F_{LG+PL_DIST} (see Results). The first mechanism cannot fully explain the increased slope of F_SO during SO&LG excitation, because the additional forces measured at the distal SO tendon during excitation of only LG (Figure 2A) were limited. Therefore, the second mechanism is the most probable explanation of the increased interaction during SO&LG excitation. These results also indicate that the net orientation of connective tissue linkages was such that force transmission occurred mainly from SO to LG+PL and not vice versa.

As pointed out earlier (Herbert et al., 2008), maximal excitation of SO without excitation of its synergists is rarely seen during normal movements, which puts into question the physiological relevance of such an excitation protocol. Although selective activation of ankle plantar-flexion muscles has been found during prolonged, low-level static plantar-flexion in humans (Tamaki et al., 2011) and during paw-shakes in cats (Smith et al., 1980), these muscles are indeed co-activated during most activities. Co-activation of SO, MG and LG has been reported during human gait (Ishikawa et al., 2005), cycling (Wakeling, 2009) and (sub)maximal isotonic, isometric and isokinetic heel rising tasks (Ball and Scurr, 2015), as well as in rat (Roy et al., 1991; Bernabei et al., 2015b) and cat (Maas et al., 2009; Markin et al., 2012) locomotion. More specifically, near-maximal SO activation combined with submaximal MG activation has been found in upslope walking in rats (Roy et al., 1991) and cats (Pierotti et al., 1989). This indicates that the relative co-activation levels of synergistic muscles as applied in the present study are found during normal movements.

In the present study, we found epimuscular myofascial force transmission between SO and LG+PL. In an intact hindlimb, the distal tendons of LG and SO merge into the Achilles tendon (Maas and Sandercock, 2008; Tijs et al., 2015), which can potentially affect LG and SO muscle functioning (Tian et al., 2012; Tijs et al., 2014). In contrast, shear strain between triceps surae muscles (Bojsen-Møller et al., 2004; Finni et al., 2015) and non-uniform deformations within the Achilles tendon (Franz et al., 2015) have been found in humans. The latter results suggest that LG and SO transmit their forces independently to the calcaneus. Regardless the level of independence within the Achilles tendon, SO and LG do have very similar line of actions. Therefore, the results of the present study suggest that the absence of effects of knee angle (i.e., proximal lengthening GA and PL) on the ankle moment exerted by SO in cats (Maas and Sandercock, 2008) and rats (Tijs et al., 2015) may be a consequence of the similar line of action of SO and LG: proximal LG+PL lengthening minimally affects SO force output, and, although force is transmitted between these muscles, both muscles transmit force to the calcaneus along a similar line of action.

In contrast to the triceps surae, tibialis anterior (TA) and extensor digitorum longus (EDL) muscles within the anterior crural compartment do not have similar lines of action (TA: ankle inversion, EDL: ankle eversion). Even though connections between TA and EDL muscles are capable of transmitting force (Huijing and Baan, 2003, 2008; Maas et al., 2004), net effects of epimuscular myofascial force transmission on the ankle moment exerted by TA and EDL were found to be very limited (Tijs et al., 2016). This implies that these muscles act independently. Because of their opposite mechanical effects at the ankle joint, this may be advantageous for accurate control of the ankle joint. The studies on the triceps surae described above (Maas and Sandercock, 2008; Tijs et al., 2015) suggest that the mechanical relevance of epimuscular myofascial connections at the ankle joint is limited. Therefore, the presence of epimuscular myofascial force transmission found in the present study implies that epimuscular myofascial force transmission may be relevant for other physiological functions. It has been suggested that linkages between triceps surae muscles may distribute stresses and strains over multiple muscles and tendons, which might reduce local stresses (Bojsen-Møller et al., 2010). Additionally, intermuscular force transmission may result in local length changes within a length-restrained muscle. Muscle spindles within these length-restrained muscles may detect such deformations, thereby affecting sensory encoding (Smilde et al., 2016).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.