Treadmill Locomotion in the Intact and Spinal Mouse

Step cycle of the intact mouse

The step cycle can be subdivided into stance and swing phases. The stance phase begins as soon as the foot contacts the ground and ends when the foot starts its forward movement. The swing phase begins, not necessarily with the paw lift (because of a possibility of a paw drag), but at the onset of forward movement of the foot and ends as it touches the treadmill belt again. The duration of the step cycle, as well as the duration of the stance and swing phases, were measured at different treadmill speeds in 10 intact mice of three different strains: 3 CD1 (average weight, 29 gm), 4 BALB/c (average weight, 19 gm) and 3 C57BL/6 (average weight, 20 gm). The average values for each individual are summarized in Table 1 and illustrated in Figure 1A for a treadmill speed of 0.15 m/sec. Figure 1A clearly shows that the average step cycle duration differs somewhat for each strain, ranging from 253 ± 29 (CD1) to 403 ± 55 msec (C57BL/6). However, for a given mouse, each consecutive step cycle duration was constant (small SD) at this treadmill speed, especially for the CD1 strain. Mice had difficulty maintaining a regular step cycle at treadmill speeds lower than 0.15 m/sec. Note that the swing phase was comparable between each strains, whereas the duration of the stance phase varied in accordance with the duration of the step cycle.

Figure 1B shows a typical example of the average step cycle obtained for a mouse (Mouse 8) walking at three different treadmill speeds (0.1, 0.15, and 0.2 m/sec). The duration of the stance phase decreased with speed, whereas the swing phase was almost invariant. These results illustrate not only a typical characteristic of locomotion but also show that the mouse can be trained to walk at different constant speeds on the treadmill. As the treadmill speed increases from 0.1 to 0.2 m/sec, the average duration of the step cycle (mean of more than six cycles) decreased from 484 ± 94 to 323 ± 35 msec (Fig. 1B). The larger SD at 0.1 m/sec denotes, as mentioned previously, that the animal had some difficulty maintaining the same pace at such a slow treadmill speed. The duration of the swing phase remained almost constant, ∼112 ± 23 msec, whereas the duration of the stance phase decreased as the mouse walked faster, from an average of 370 ± 62 msec at 0.1 m/sec to 211 ± 40 msec at 0.2 m/sec.

Kinematics of the intact mouse

Figure 2 illustrates the kinematics of the hindlimb of an intact mouse walking on a treadmill at two different treadmill speeds: 0.1 m/sec (Fig. 2A,B) and 0.2 m/sec (Fig. 2C,D). Walking is represented by (1) a stick diagram of one step cycle of the left hindlimb with the swing and stance phase drawn separately and (2) the angular displacement of the hip, knee, and ankle joints averaged over several consecutive step cycles (n > 10). The orientation of each joint measurement is indicated on the stick diagram (lower left-hand graph in A) and the display of angle excursions are synchronized on left foot contacts (downward arrows). The subdivisions of the step cycle defined by Philippson (1905) can be used to describe the various subphases of the murine step cycle. Note that downward deflections of the angular traces correspond to a flexion of the joints. The swing phase starts with a flexion (F) of the hip and ankle while the knee remains at a constant angle, lifting the foot from the ground (upward arrows). Whereas the hip continues its flexion, the knee and ankle begin an extension (E1), which brings the paw in contact with the ground. As the paw contacts the ground (downward arrows), the knee and ankle are flexed (E2) while the hip begins its extension. Then, at ∼20% of the step cycle, the ankle starts an extension (E3; push off), which propels the body of the mouse forward. For this mouse, the average peak-to-peak angular excursions were 16° at the hip, 34° at the knee, and 28° at the ankle. Similar angular excursions were observed at treadmill speeds of 0.15 (data not shown) and 0.2 m/sec (Fig. 2C-D). Comparison of the stick diagram of the mouse walking at 0.1 (Fig. 2A) and 0.2 m/sec (Fig. 2C) indicates that higher speed is obtained by reducing the length and the duration of the stance phase (i.e., there are fewer sticks during the stance phase in Fig. 2C than in 2A). This difference also indicates that it is possible to impose a walking speed on the mice and that the step cycle can be constant at the chosen speeds.

Interlimb coordination in the intact mouse

The gait diagram of a duty cycle (Hildebrand, 1976) of the four limbs of the intact mouse during walking is illustrated in Fig. 3A. The horizontal bars represent stance phases of each limb, and the spaces correspond to the swing phases of an intact mouse walking on a treadmill at 0.15 m/sec. To illustrate how the coupling between the different limbs was calculated, a single cycle is highlighted in gray. Within this reference step cycle, three different couplings are measured: (1) coupling of the homolateral limbs [i.e., limbs on the same side of the animal (right or left)], (2) coupling of the homologous limbs [i.e., limbs on the same girdle of the animal (forelimbs or hindlimbs)], and (3) coupling of the diagonal limbs. For example, to estimate the value of the homologous coupling between the hindlimbs, we measured the time of the right foot contact (vertical dotted line indicated by arrow 2 in Fig. 3A) with respect to step cycle of the left hindlimb. When expressed as a percentage of the step cycle, the coupling interval between the left and the right hindlimb is ∼48% (or phase 0.48). For each step cycle, the same measurement was made for the homologous, homolateral, and diagonal limbs. The average coupling values were calculated and are shown in Table 1, as well as in Figure 3B-D for all 10 mice studied.

Figure 3B clearly shows that the homologous coupling between the RHL and the LHL, as well as between RFL and LFL were ∼50% for all mice tested at this treadmill speed. These numbers mean that at 0.15 m/sec, the limbs of each girdle step in strict alternation: the swing phase is followed half a step cycle later by the swing phase of the contralateral limb. Figure 3C shows that the homolateral coupling between the LFL and LHL or between the RFL and RHL ranged from 25 to 45%. This variability is not surprising because the different mice walked at different speeds (Fig. 1), and it is well known that the coupling between forelimbs and hindlimbs is more variable and is also speed-related (Hildebrand, 1976). Indeed, walking is characterized by the forelimb contacting the treadmill at ∼25% of the hindlimb step cycle, whereas during trot, the forelimb touches the treadmill at 50% of the hindlimb step cycle. Figure 3D shows the diagonal couplings (i.e., coupling between LFL and RHL and coupling between RFL and LHL); the percentages ranged between 78 and 97%.

EMG activity during locomotion

We were able to study the kinematics together with the synchronized EMGs in one chronically implanted mouse walking on a treadmill. Raw EMG traces obtained during locomotion, synchronized with the corresponding duty cycle of the hindlimbs, are shown in Figure 4A. For this mouse, the mean duration for 45 consecutive bursts of activity was 157 ± 35 msec for LVL, 120 ± 28 msec for RVL, 58 ± 17 msec for LTA, and 72 ± 23 msec for RTA. Comparing the timing with duty cycles given below the EMG traces, we see that the recordings of these muscles give a good indication of the swing and stance phases of the limbs. The knee extensors (LVL and RVL) were active during the stance phases of the hindlimbs, whereas the ankle flexors (LTA and RTA) were active during the swing phase of the step cycle of the corresponding hindlimb.

In Figure 4B, the EMG bursts were rectified, averaged over 20 consecutive step cycles and then normalized (synchronization on LTA) to illustrate the overall reciprocity between flexors and extensors, as well as between right and left sides of the animal. Indeed, within the same limb, there is good alternation between extensor and flexor muscles. Right-left alternation is also clear between both TAs and VLs.

Figure 4C represents one step cycle of the implanted mouse during locomotion. Note that, even if the VL and TA muscles of both hindlimbs are implanted, the kinematics of the limb is similar to those of the nonimplanted mouse (compared with stick diagrams of Fig. 2), suggesting that the chronically implanted microwire used as well as the connector do not interfere with the ongoing movements. Thus, it will be possible in the future to routinely use chronically implanted EMG electrodes in the mouse to compare the locomotor pattern before and after a spinal lesion in the same animal.

Recovery of locomotion in the spinal mouse

The kinematics and EMGs of normal mouse locomotion

The small size of mice may appear to preclude the use of conventional equipment to study kinematics or EMG activity to study locomotion. Overall, mice walked with no difficulty on the treadmill, and could keep up the imposed speed, enabling us to record several consecutive step cycles necessary for averaging. The video recording system used in the cat (Bélanger et al., 1996) and rat (Orsal et al., 2002) could be used to detect the markers on the joints automatically. The use of a conventional 30 frames per second video camera, with the ability to split frames in their two component fields (final temporal resolution, 16.7 msec), is appropriate for treadmill speeds lower than 0.2 m/sec. At these speeds, the duration of the swing phase was ∼100 msec, corresponding to approximately six video fields (three video frames). However, this is at the limit of tolerance to reconstruct the angular movements correctly during the swing phase and the exact timing of foot lift and the foot contact. To study the kinematics at higher treadmill speeds, a high-speed video camera would be more adequate.

In most animals, it is difficult to determine the knee angle accurately because of skin slippage. To minimize this, the knee marker was corrected by triangulation using bone measurements of the femur and tibia. Nevertheless, some slippage could still explain why, in some cases, the extension of the knee joint, characteristic of the E3 phase, was not clearly seen (Fig. 2). Despite these difficulties, the terminology used to describe the step cycle in other animals (Philippson, 1905) can be applied to murine locomotion. Moreover, as is the case for most mammals (Grillner, 1981), the swing phase is rather constant, even with varying speeds, whereas the stance phase varies with speed.

Measurements of foot contact also allowed us to establish the mode of coupling between the limbs and to illustrate the locomotor pattern using gait diagrams (Hildebrand, 1976). The results show that the homologous limbs of both the pelvic and scapular girdles are out of phase by 0.5, but that the duration of stance of the forelimbs and hindlimbs are different (Clarke and Still, 1999). The gait formula generally follows a sequence of alternation between three and two limbs in contact with the ground. The usual sequence of LH-LF-RH-RF is followed at these low speeds.

Here, we have also shown, in one animal, that it is possible to record satisfactorily from some muscles of the thigh and shank with chronically implanted electrodes. The miniature connectors with the small-diameter wires did not impede locomotion. The overall weight of the connector and wires was ∼0.30 gm, which represents ∼1% of body weight. The clean signals from EMG recordings could even be used to determine the various phases of the locomotor cycle, as well as the various couplings described above, using kinematic analyses. Work is underway to compare EMG activity in the same mice before and after spinalization.

There are numerous tests designed to assess the locomotor function of rats after spinal cord injury, and all have their merits (Rivlin and Tator, 1977; Kunkel-Bagden et al., 1993; Basso et al., 1995). Most of them rank the locomotor behavior in an open-field situation. The most widely used is the BBB (Basso-Beattie-Bresnahan scale), which standardizes the ranking on a 21-point locomotor index of over-ground locomotion obtained by observing the locomotor movements of the rats in an open field (Basso et al., 1995). Such scores rely on visual determination of the quality of locomotion and can possibly lead to incorrect interpretations by the untrained eye. For example, we mentioned in our results (Fig. 8) that it is very difficult by visual examination to ascertain that there is a correct coupling between the forelimbs and the hindlimbs during walking after recovery of locomotion in the hindlimbs (see also, Broton et al., 1996). A related problem arose in studies of cats with large but incomplete ventral and ventrolateral lesions at T13 (Brustein and Rossignol, 1998). With precise measurements, we showed that the forelimbs and hindlimbs could actually walk at slightly different mean frequencies, which led to periodic stumbling of the animal.

Spinal locomotion in the mouse

An important result of the present study is that spinal mice can spontaneously recover hindlimb locomotion on a treadmill. This corroborates in vitro studies on neonate mouse spinal cord that demonstrate the existence of a central pattern generator (Bonnot et al., 1998). In the present work, we have ensured also that the recovery of locomotion could not be attributable to some spontaneous regeneration of axons through the lesion. This was achieved by performing a second spinal section at the level of the first lesion at a time when the animal had recovered spinal locomotion. This second spinalization did not affect spinal locomotion or its subsequent evolution.

As in the cat (Barbeau and Rossignol, 1987), the locomotor recovery in the mouse took time. During the first postlesion week, only occasional flexion movements of all joints were observed on the treadmill. At the end of the first week, some alternation between hindlimbs was present with occasional foot placements. Most of the locomotor recovery took place during the second week after spinalization. The normal kinematics of each individual limb was first restored and was followed, 2-3 d later, by a proper out-of-phase alternation between the hindlimbs. Thus, it appears that the spinal networks responsible for locomotion of a given limb can recover to some extent before the interlimb coordination is ensured. This course of evolution between intralimb and interlimb coordination is somewhat different from that in the cat, in which both intralimb kinematics and bilateral alternation appear at the same time after spinalization (Bélanger et al., 1996). After the recovery of locomotor rhythm, the patterns observed in spinal mice remained for the whole period of observation (3-4 months).

Another important feature is the fact that two mice re-expressed locomotion spontaneously without either training or pharmacological treatment. The ability of the spinal mouse to re-express locomotion was surprising because another rodent, the adult rat, is unable to recover locomotion with their hindlimbs after a complete spinal lesion unless neurochemical stimulation is provided either by drugs or cellular grafts (Yakovleff et al., 1989; Broton et al., 1996; McDonald et al., 1999; Gimenez y Ribotta et al., 2000; Antri et al., 2002; Orsal et al., 2002). However, rats spinalized as neonates, can re-express locomotion of the hindlimbs for several weeks after spinalization (Stelzner et al., 1975; Weber and Stelzner, 1977; Commissiong and Sauve, 1993; de Leon et al., 2002).

The differences between rats and mice have already been exposed in studies using isolated spinal cord of neonates in vitro. Indeed, it was shown, by recording EMGs in isolated spinal cord with hindlimbs attached in vitro (Hernandez et al., 1991) or by recording ventral root discharges in isolated low spinal preparation (Branchereau et al., 2000; Whelan et al., 2000), that the lumbar spinal cord of the neonatal mouse has the capacity to generate spontaneously rhythmic locomotor patterns, without any drugs. In contrast, it is not possible to evoke fictive locomotor patterns in neonatal rats without pharmacological activation (Cazalets et al., 1992, 1995; Kiehn and Kjaerulff, 1996). Thus, our results confirm in vivo that the locomotor pattern can be expressed spontaneously.

In the context of a study on spinal cord compression injuries in mice, some mice also had a complete transection of the spinal cord (Farooque, 2000). They were reported to be paraplegic for the observation period of 12 weeks. The mice exhibited some movements of the hindlimbs in the open field but had no coordination or weight support. These movements were attributed to local spinal cord reflexes. In contrast, we have shown in this study that a mouse with a complete spinal cord section can re-express locomotion. However, it is important to note that the spinal mouse, which has a proper pattern of locomotion on the treadmill at 14 d after spinalization, will look paraplegic if put in an open-field situation (i.e., the hindlimb will simply drag behind the animal and only some occasional movements will be observed). In fact, the animal has to be supported and a lateral stability has to be provided by the experimenter. This does not mean that the spinal cord circuitry does not have the ability to generate locomotor pattern by itself when properly stimulated.

It is hoped that this work will promote the use of standard video and EMG recordings to evaluate the outcome of therapies after spinal cord injury in the mouse by providing a more quantitative assessment of the locomotor movements and the underlying pattern of muscle activity.