Exp Brain Res (1992) 90:557 566

Brain Research �9 Springer-Verlag 1992

Entrainment of the locomotor rhythm by group Ib afferents from ankle extensor muscles in spinal cats

K.G. Pearson, J.M. Ramirez*, and W. Jiang** Department of Physiology, University of Alberta, Edmonton, Canada

Received August 22, 1991 / Accepted March 16, 1992

Summary. 1. Previous studies have concluded that the timing of the locomotor rhythm can be strongly in- fluenced by input from group Ib afferents from leg exten- sor muscles (Duysens and Pearson 1980; Conway et al. 1987). The main objective of the present study was to obtain additional evidence for this conclusion by examin- ing the characteristics of entrainment of the locomotor rhythm by rhythmic stimulation of group I afferents and by rhythmic force pulses in the ankle extensor muscles. 2. A reduced, non-immobilized preparation was de- veloped in spinal cats that allowed isometric contractions of ankle extensor muscles to be elicited by ventral root stimulation during the expression of locomotor activity. The same preparation was used to examine the influence of electrically stimulating group I afferents from the ankle extensors and the effect of rhythmically stretching these muscles. The locomotor rhythm was initiated by sustained mechanical stimulation of the perineum following the administration of Clonidine and, in some preparations, Naloxone. 3. The timing of the onset of flexor burst activ- ity was examined during entrainment with saw-tooth and ramp-and-hold stretches of the ankle extensor muscles. Flexor bursts were initiated about 200 ms following the release from the stretch, and this latency was independent of the entrainment frequency. 4. The locomotor rhythm was readily entrained by rhythmic contractions of the ankle extensor muscles produced by ventral root stimula- tion provided the magnitude of the contractions was greater than about 10N. Repetitive stimulation of group I muscle afferents from the ankle extensors also entrained the locomotor rhythm, with the timing of motor activity being similar to that during entrainment with rhythmic muscle contractions. Burst activity in the ipsilateral exten- sors was coincident with the stimulus trains in both cases. This similarity argues for entrainment being produced

* Present addresses: Fachbereich Biologic, Universit/it Kaiser- slautern, W-D-6750 Kaiserslautern, Federal Republic of Germany ** CRSN, Physiologic, Faculte de Medecine, C.P. 6128, succursal A, Montreal, Quebec H3C 3J7 Correspondence to: K. Pearson

mainly by input from group Ib afferents. 5. The functional implication of the results of this and previous studies is that input from group Ib afferents during the stance phase of walking acts to inhibit generation of flexor burst activ- ity and to promote extensor activity. The proposal that a decline in Ib activity near the end of the stance phase is involved in regulating the stance to swing transition is discussed.

Key words: Locomotion- Cat walking Rhythmic motor activity - Entrainment

Introduction

There is now considerable evidence that sensory feedback has an important role in establishing some features of the motor pattern for walking in the cat (Smith 1986; Grillner and Dubuc 1988; Rossignol et al. 1988). The clearest example of afferent regulation of the step cycle is the adaptation of stepping in decerebrate and spinal cats to the speed of the treadmill upon which they are walking. One source of afferent information involved in regulating the stance-to-swing transition in the hind legs may be tendon organs in leg extensor muscles. This possibility was first suggested by the finding in walking decerebrate cats that the locomotor rhythm in the isolated ankle extensor muscles of one hind leg can be inhibited by maintained stretches of the muscles. The generation of rhythmic flexor burst activity is suppressed so long as the muscle force is maintained above about 40N regardless of the length of the extensor muscles (Duysens and Pearson 1980). Subsequently Conway et al. (1987) reported that the fictive locomotor rhythm in DOPA/Nialamide treated spinal cats can be entrained by rhythmically stretching the ankle extensor muscles with the ipsilateral extensor activ- ity coinciding with peak tension in the extensor muscles. Both these observations indicate that the initiation of flexor burst activity during walking requires unloading of

558

the leg extensor muscles. The afferents involved in signall- ing the un load ing of the extensor muscles are p r o b a b l y the g roup Ib afferents. The evidence for this conclus ion comes f rom obse rva t ions on D O P A / N i a l a m i d e t rea ted spinal cats in which electr ical s t imula t ion of g roup I afferents f rom knee and ankle ex tensor muscles a b r u p t l y te rminates flexor act ivi ty and ini t iates an extensor burst , while selec- tive ac t iva t ion of the g roup Ia afferents by v ib ra t ion has no effect on the l o c o m o t o r r h y t h m (Conway et al. 1987).

The aim of the present inves t iga t ion was to examine fur ther the influence of input f rom p rop r iocep to r s in the ankle ex tensor muscles on the l o c o m o t o r rhythm. In par - t icular we wished to ob t a in add i t i ona l evidence that input f rom the g roup Ib afferents of these muscles influenced the l o c o m o t o r r h y t h m in a m a n n e r consis tent with their p ro- posed role in regula t ing the s tance- to-swing t ransi t ion. The first object ive was to de te rmine in more detai l the character is t ics of the t iming of the l o c o m o t o r pa t t e rn dur ing en t r a inmen t p r o d u c e d by rhy thmica l ly s t re tching the ankle extensor muscles. In the ear l ier s tudy by Con- way et al. (1987) it was found tha t ips i la tera l f lexor burs ts were in i t ia ted when the ankle extensor muscles were re- leased f rom stretch but a quant i t a t ive analysis of the t iming of the onset of the flexor burs ts relat ive to the stretch was no t carr ied out. Thus cur ren t ly it is no t k n o w n whether the la tency between the decline in extensor muscle force and the onset of flexor act ivi ty in reduced spinal p r e p a r a t i o n s is s imilar to this la tency in n o r m a l walk ing animals .

The second object ive of our s tudy was to es tabl ish whether en t r a inmen t of the l o c o m o t o r rhy thm could be p r o d u c e d by genera t ing a rhy thmic sequence of force pulses in the ankle extensor muscles and whether the character is t ics of en t r a inmen t were s imilar to those of en t r a inmen t p r o d u c e d by rhy thmica l ly s t imula t ing the ankle extensor muscle nerve at g roup 1 strength. Since only the g roup Ib afferents are ac t iva ted synchronous ly with the en t ra in ing s t imulus by bo th methods , a f inding of s imilar t empora l character is t ics of en t r a inmen t in the two s i tua t ions would p rov ide add i t iona l evidence that Ib input can influence the t iming of the l o c o m o t o r rhythm. More - over, if Ib act ivi ty is funct ioning to facil i tate the p roduc - t ion of ex tensor act ivi ty and to inhibi t the genera t ion of flexor burs t act iv i ty as ind ica ted in ear l ier studies (Duysens and Pea r son 1980; C o n w a y et al. 1987), we pred ic ted tha t dur ing en t r a inmen t burs t act ivi ty in ipsi- la tera l ex tensor muscles would occur at the same t ime as the en t ra in ing stimuli, i.e. synchronous ly with force pulses in the ankle extensor muscles and with the electr ical s t imula t ion of the g roup I afferents f rom these muscles.

Materials and methods

The data presented in this paper were obtained from experiments on 18 adult cats of both sexes. The experimental procedures were carried out on two groups of animals: 1) spinalized at the time of the experiment (n = 8), and 2) spinalized between 2 to 5 days prior to the experiment day (n= 10). The surgical procedure for preparing the chronic spinal animals was similar to that used by Barbeau and Rossignol (1987). Animals were anaesthetized with pentobarbital (Somnotol 40 mg/kg) and under aseptic conditions the spinal cord

was completely transected at the T13 level. The cord was cut with blunt scissors and the two cut surfaces were spread apart by a few millimetres to ensure that the transection was complete. The space between the cut surfaces was packed with coagulating gauze (Sur- gicel). The animals were housed individually and received fluid and nutrient supplements during the first 2 days following surgery (saline and Aminolean subcutaneously). Penicillin (lcc Ayercillin) was ad- ministered prophylactically daily. Postoperative pain, if it occurred, was controlled by the administration of an analgesic (Buprenor- phine 0.01 mg/kg). The bladder was emptied 2 or 3 times daily either by manual pressure or catheterization.

Experimental procedure

On the day of the experiment all animals were anaesthetized with Halothane. The right carotid artery was cannulated to monitor blood pressure and the left carotid artery was ligated. A cannula was inserted into the right jugular vein for the administration of saline and drugs. The trachea was cannulated and the aesthetic sub- sequently administered via this cannula. In all preparations both hind legs were extensively denervated by cutting the femoral, ob- turator, sural and common peroneal nerves. The tibial nerve distal to the branch to medial gastrocnemius was cut, as were branches of the hamstring nerve to anterior biceps and semimembranosus. This procedure left intact innervation to some proximal muscles at the hip, to the posterior biceps and semitendinosus muscles, and to gastrocnemius-soleus muscles. Depending on the experimental ob- jective, cuff electrodes were placed on various peripheral nerves (posterior biceps-semitendinosus, sciatic or.tibial) to either monitor the activity in the nerve or to stimulate the nerve. Each nerve was placed in a preformed silicon cuff and two copper hook electrodes (insulated wires except for the hooks) were positioned under the nerves. Silicon impression material (Reprosil, DeTrey) was then spread over the electrode hooks and nerves within the cuff to provide electrical isolation. In those preparations in which the L7 ventral root was stimulated and/or recordings were made from dorsal root filaments, the spinal cord was exposed by a laminectomy from L5 to L7 and the nerves to the left posterior biceps and semitendinosus muscles were cut as were nerves to the left hip, tail and perineal regions.

Following this initial procedure the animal was mounted in a stereotaxic frame and decerebrated by a vertical transection of the brain stem immediately anterior to the superior colliculus. The entire forebrain was removed from the cranial cavity and the bleed- ing controlled by packing Surgical around the brain stem. The exposed brain stem was covered with saline soaked swabs and the skin closed above the skull. The aesthetic was discontinued immedi- ately after decerebration. Following recovery of the blood pressure to normal levels (about 15 rain after decerebration) the spinal cord was transected at T13 if the animals had not been chronically spinalized. The plasma expanding agent Dextran was administered if the blood pressure fell below 60 mm Hg.

Both hind legs were rigidly fixed at the knee and ankle joints and the ankle extensor muscles of both hind legs were attached to two separate muscle pullers. The Achilles tendon and the distal end of the muscles were freed from the skin and underlying tissue and the ealcaneus bone cut a few millimetres from the tendon attachment. The puller was attached to the muscle via a strong thread tied close to the bone fragment or through a hole drilled through the fragment. The muscle pullers were controlled via a Transduction DT2821 interface card installed in a Zenith 386 microcomputer. Software programs were written to control the frequency, amplitude and duration of the controlling signals sent to the pullers.

Electromyograms (EMGs) were recorded percutaneously from the ankle extensor muscles, gastrocnemius-soleus (GS), and the knee flexor/hip extensor muscle, semitendinosus (St), of both legs to monitor locomotor activity. These electrodes consisted of a pair of teflon coated stainless steel wires, insulated except for about 0.5 cm, inserted into the muscles.

In laminectomized animals the L7 ventral root was cut close to the spinal cord and the distal end placed on bipolar stimulating electrodes. Filaments were dissected from the L7 dorsal root to record from afferents arising from the GS muscles. Three criteria were used to identify group Ia muscle afferents: 1:1 following of vibratory stimuli to the GS muscles (150 Hz at 100/,m), silencing of the tonic activity when the ventral root was stimulated, and an irregular dynamic response to ramp stretches (10 mm/s) of the GS muscle. Group Ib afferents where identified by the absence of 1:1 following when the GS muscle was vibrated, strong excitation when contractions of the GS muscles were produced by ventral root stimulation, a weak regular response to stretch of the passive GS muscles, and conduction velocities greater than 80 m/s.

The afferent volleys produced by electrical stimulation of the motor nerves supplying the ankle extensor muscles was monitored by cuff electrodes placed on the sciatic nerve. The stimulus strength was calibrated relative to the threshold intensity for the largest afferents. Strengths less than 1.5 x T were considered to selectively activate group I afferents (Lundberg et al. 1987). The motor nerves to the ankle extensor muscles were mounted on bipolar stimulating electrodes and insulated in a paraffin oil bath formed at the knee joint. The strength of the stimulus applied to the L7 ventral root was calibrated relative to the threshold intensity required to produce motor activity in the ankle extensor muscles.

Initiation of locomotor activity

Rhythmic locomotor activity was generated by pinching the skin of the perineal region following the i.v. administration of Clonidine (Sigma Chemicals). In later experiments Clonidine treatment was preceded by an injection of Naloxone (25 #g/kg) (Sigma Chemicals). Naloxone pretreatment was found to reduce the dose of Clonidine required for rhythm generation and led to more sustained sequences of activity (Pearson et al 1992).

In the majority of animals (15 out of 18) sustained sequences of rhythmic locomotor activity lasting for 30 s or more could be gener- ated in response to perineal stimulation (Fig. 1A). Normally we limited the duration of each trial to about 20s to minimize adapta- tion of the sensory input from the skin. The pattern of locomotor activity shown in figure 1A was observed in all preparations, namely reciprocity of burst activity in the St and GS muscles in each leg and alternating activity in the two legs.

An important but uncontrolled parameter was the mechanical stimulus to the skin of the perineum. The skin was usually rubbed between the experimenter's index finger and thumb at a frequency of between 2 and 3 Hz. This was established by recording EMG's from the experimenter's adductor pollicis muscle. No correlation was found between the timing of skin stimulation and the timing of the locomotor rhythm. In males the most effective site was the skin of the scrotum, while in females it was the skin around the vulva (see also Sherrington 1910). Despite the uncontrolled nature of the stimulus we found that the characteristics of the rhythmic sequences in each animal remained quite similar from trial to trial (Fig. 1B), and that once initiated the characteristics of the rhythm were fairly independent of the intensity of skin stimulation.

Data analysis

Data collection was commenced about 3 4 h after decerebration. All data were stored on VHS video tape using a Vetter 4000A PCM recording unit. Subsequently the data were displayed on a Gould ES1000 electrostatic chart recorder and sections were selected from this display for quantitative analysis. Selected sequences were digitized (sample rates from 700 Hz to 10 kHz depending on the analysis) and stored on disc using the Axotape data acquisition system (Axon Instruments) and a Greff 386 microcomputer. In most cases EMG records were rectified and low pass filtered (time con-

A

559

R GS

R St

L GS

L SI

!ON

5s

B 2

o Q Cz (b

o

i , i i , i J

15 30

Time (s)

Fig. 1. Characteristics of the locomotor activity generated in Clonidine-treated spinal cats. A) Short sequence of rhythmic activity recorded from knee flexor/hip extensor, semitendinosus (St) and ankle extensor, gastrocnemius-soleus (GS), muscles in a chronic spinal cat ( 3 days post-spinalization) in response to perineal stimu- lation. Note the strict reciprocity of activity in St and GS muscles of each leg and the alternating activity in the two legs. The forces generated in the GS muscle were typically in the range of 10 to 15N. C) Superimposed plots of the cycle period versus time for three consecutive rhythmic sequences in an acute spinal cat. Note the similarity of the cycle period in all three sequences

stant 50 ms) before digitization. The data files were analyzed using interactive software developed by ourselves for displaying files stored with Axotape and for measuring cycle periods, burst dura- tions and the phase relationships between applied stimuli and bursts of activity in different muscles.

Results

T h e charac te r i s t i c s of e n t r a i n m e n t p r o d u c e d by musc le s t re tch, force pulses and ne rve s t i m u l a t i o n were s imi la r in acu t e and c h r o n i c spinal an ima l s (acute, n = 8 ; chronic , n = 10). T h u s d a t a f r o m b o t h g r o u p s of an ima l s h a v e been i nc luded t o g e t h e r in the fo l l owing sect ions.

1. Entrainment of the locomotor rhythm by muscle stretch

T h e effect of r h y t h m i c a l l y s t r e t ch ing the G S musc les of one h ind leg was e x a m i n e d in 10 an ima l s (8 c h r o n i c and 2 acu te spinal p repara t ions ) . In all an ima l s it was poss ib le

560

to entrain the locomotor rhythm with a sawtooth pattern of stretch (Fig. 2). Other patterns of stretch such as sinusoidal were also effective but these were not examined in detail. Provided the frequency of the applied stretch was close to the intrinsic frequency of the locomotor rhythm, entrainment usually began within a few cycles of commen- cing the rhythmic stretching and was often maintained for periods longer than 20s (this being the time limit of most sequences). The range of frequencies over which entrain- ment occurred varied from animal to animal and, in all but one animal, it was centred around the intrinsic fre- quency, i.e. the entraining frequency could be either higher or lower then the intrinsic frequency. In the one excep- tional animal a sustained locomotor rhythm could not be elicited by perineal stimulation but a coordinated rhythm was initiated by stretching the GS muscles. The widest range of entraining frequencies was from 0.4 to 1 Hz (examples from this animal are shown in Fig. 2) and the narrowest was from 0.6 to 0.8 Hz.

To graphically illustrate the entrainment phenomenon we plotted the phase of the onset of flexor activity (as measured by the EMG from the ipsilateral St muscle) within the stretch cycle versus time (Fig. 3). Periods of

No stretch

~ ~,~ ~ ~ ~ I~,, ,~,~ St

0.4Hz

"~',~"~ ......... , ~ " t ~' I,r~,~we' ' ~l,,,~'~r, ~&m,n~q, '" "~Ir'~l~ GS

, ~,, ,,~, ~k ~ ' " ~ ~ st

0.6Hz

. , , ~ . . . . . . ~ , L ~ L . , ~ . . . . . i&~ , , ~ .',/J~ j 'a~ GS

~ " '~" ~" ' r ~ " '~ ' ' ~ IPW' ~ St

0.8Hz

1"'m~ ~ m ~ ' ~ , ~ ' ~ " ~ ' ~ [ ~ .~1 ~.,- GS

~' I I [ W~ II~ 1' IW ~ I~ ~" [l~' I~ ~re' St

2s

Fig. 2. EMG recordings from the ankle extensor GS and knee flexor St muscles during entrainment produced by a 5 mm sawtooth stretch of the GS muscle at 0.4, 0.6 and 0.8 Hz. The top set of records (no stretch) shows the rhythmic pattern elicited immediately prior to the recording of the entrainment sequences. Note that during en- trainment flexor burst activity occurred immediately after the peak stretch

A

B

C

v "~ ~ 1'0 1 �9 r 0.5

1 3 _

0 ,

0.5

D_

o o

'0.6Hz

Q O � 9 1 4 9 1 4 9 1 4 9 1 4 9

u n

O.9Hz

i i n u a ~ i

i u u i i i i i i

5 10 15 20

Time (s)

Fig. 3. Graphical representation of entrainment. A Illustration showing measured parameters: b cycle period of entraining saw- tooth stretch of the GS extensor muscle, a time of onset of ipsilateral St flexor activity relative to the beginning of the stretch, q~ - phase. The St EMG has been rectified and low pass filtered. B) and C) Plots of the phase (a/b) versus time for two sequences of rhythmic activity when the entrainment frequency was 0.6 Hz (B) and 0.9 Hz (C). The locomotor rhythm was strongly entrained at 0.6 Hz as indicated by the almost constant phase value throughout the sequence. At 0.9 Hz the rhythm was only briefly entrained near the beginning of the sequence. The progressive increase in phase following the loss of entrainment indicates that the intrinsic fre- quency was lower than the entraining frequency

entrainment were indicated in these plots when the phase remained relatively constant. In the example shown in figure 3B the phase was maintained constant at a value of about 0.6 through the trial. In this case entraining fre- quency was close to the intrinsic frequency. A short period of entrainment in the same preparation is shown in figure 3C when the entraining frequency (0.9 Hz) was maximal, i.e. higher than the intrinsic frequency. Entrainment oc- curred only near the beginning of this sequence and was lost after about 5 seconds. The slower intrinsic frequency is illustrated by the progressive increase of phase with time.

The reason the phase value during entrainment was around 0.6 was that St flexor burst activity commenced close to, or soon after, the peak stretch (Figs. 2, 3a, 5). The delay from the peak stretch to the onset of St flexor burst activity was usually in the range of 100 to 300 ms (mean about 200 ms) and independent of the entraining fre- quency (Fig. 4). Thus the main temporal characteristic of entrainment was that the onset of the ipsilateral flexor burst activity occurred within a few hundred milliseconds of the muscle being released from the peak stretch. The

St EMG

GS Length

i

25

association between the time of muscle release and the time of the onset of flexor burst activity persisted when release was delayed by adding a plateau to the peak stretch (Fig. 5A, C). The addition of a plateau to each stretch cycle delayed the onset of flexor activity thus resulting in an increase in phase (Fig. 5B). This phase shift

%

Z)

(i) ~3

c-

O

400 -

2 0 0

1B 20

, i i , i i , i i i i

015 0,6 0 '7 0 8 0"9 1'0

Entrainment frequency (Hz)

Fig. 4. The time of onset of flexor activity following the peak of a sawtooth stretch to the ipsilateral GS muscles is independent of the entrainment frequency. In this preparation entrainment occurred over a wide range of frequencies (0.6 to 1 Hz). The plot shows the means and standard deviations of the delay between the peak stretch of the GS extensor muscle and the onset of the burst activity in the ipsilateral St flexor muscle. The number above each data point is the number of observations.

A

St E M 6

1 0 N I ~ GS force

GS length la

B

15 1 10 �9

EL- 0 0 0 o o o o

0 5

C 3 0 0 ]

_ o o o o "] �9 �9 200 o o �9 �9 �9

/ ~" t0O

2 ] 0 "J i

0 5 10 15

Time (s)

Fig. 5. Prolonging the stretch of the ankle extensor muscles during a period of entrainment shifts the timing of ipsilateral flexor (St) burst activity relative to the onset of the stretch. A) Records showing the rectified and filtered EMG recording from the St muscle when a plateau was added to the peak of a sawtooth stretch. The shift in the time of initiation of St activity is indicated by an increase in phase (filled circles in B). The time of onset of St activity relative to the time of release (c) was only slightly reduced by the addition of the plateau to the stretch

561

usually developed progressively over a few cycles (2-5). In the example shown in Fig. 5 it occurred quickly and was complete by the second cycle.

2. Entrainment of the locomotor rhythm by ventral root stimulation

The onset of flexor burst activity during entrainment with stretch of the ankle extensor muscles is associated with a decline in both muscle length and force (Fig. 5). To dissociate these two parameters we held the muscles at a constant length and evoked isometric force pulses in these muscles by stimulating the ipsilateral L7 ventral root (60 Hz for 500 ms). We predicted that if afferent signals related to the decline in force were involved in initiating flexor activity then rhythmic force pulses should entrain the locomotor rhythm and the ipsilateral flexor bursts should occur in antiphase with the force pulses.

The execution of this experiment required a more extensive denervation of the left hind leg in order to eliminate afferent feedback from muscles other than the ankle extensors. Thus the entire left hamstring nerve and the nerves to the hip and tail muscles were cut. Because the ipsilateral St was denervated and the ipsilateral GS was strongly activated by the stimulus train, the locomo- tor rhythm was monitored by E M G recordings from the contralateral St and GS muscles. Ipsilateral flexor activity could not be reliably monitored by recording from the cut flexor nerves because burst activity was often obscured by the activity evoked in motor axons by the stimulus trains. Using the contralateral muscles as monitors of the loco- motor rhythm the prediction was that during entrain- ment, if it occurred, the contralateral flexor burst activity should commence close to the onset of the stimulus train to the L7 ventral root. This follows from the fact that the St muscles of the two hind legs are activated alternately during the generation of the locomotor rhythm (Fig. 1A).

As expected rhythmic pulses in the ankle extensor muscles reliably produced entrainment of the locomotor rhythm with the onset of contralateral flexor burst activity coinciding with the onset of the force pulse (Figs. 6). Figure 6A shows the raw data for a short sequence of entrainment. Data similar to that shown in figure 6A were obtained in the 5 animals (1 chronic and 4 acute spinal preparations) in which this experiment was performed. For each of these 5 animals the phase-values for the onset of the activity in the contralateral St flexor muscle relative to the onset of stimulus train to the ventral root were determined for a minimum of 6 entrainment sequences and the average phase-value calculated. The average phase-values were 0.1, 0.12, 0.13, 0.18 and 0.23, yielding an overall mean phase-value of 0.15 + 0.05. The time of onset of ipsilateral flexor activity during entrainment was esti- mated by noting the time the ipsilateral extensor bursts ended (small spikes in LGS trace of Fig. 6A) and by assuming that these bursts occurred in antiphase with the contralateral flexor bursts (Fig. 1A) . Both these measurements revealed that during entrainment ipsilat- eral flexor burst activity occurred approximately in anti- phase with the force pulses. The time of onset of ipsilateral

562

A

R St

- L GS

L GS

B 0 5

&O �9 �9

0.5 ,

0 0 0 � 9 O � 9 �9 o � 9 �9 �9

C 0.5 ~ �9 �9

0 ,0 �9 �9 �9

[3_ �9

- 0 . 5 i , i , g , , ~ i i

0 2 4 6 8 I 0 12 14 16 18 20

Time (s)

Fig. 6. Entrainment of the locomotor rhythm by rhythmic force pulses in the ipsilateral ankle extensor muscles. A) Force pulses lasting 500 ms (third trace) were evoked in the left GS extensor muscles by electrical stimulation of the left L7 ventral root, and EMG recordings were made from the GS muscles of left (bottom trace) and right (top trace) legs and from the St muscle of the right leg (second trace). The rhythmic burst activity in the left GS muscles was almost completely masked by the electrically evoked EMG but the termination of burst activity was usually visible (arrow). During entrainment the onset of the contralateral (right) St activity occurred close to the beginning of the force pulse. Note that detaching the muscle near the end to the sequence (indicated by the absence of [i~rce pulses but presence of the evoked EMG in the left GS) led to an immediate slowing of the rhythm. B) C) phase plots showing en- trainment when the magnitude of the force pulses was 15N (B), and the absence of entrainment when the force pulses were reduced to 8N by reducing stimulus strength (C). The phase was calculated as the time interval between the onset of the stimulus train and the onset of activity in the contralateral St flexor muscle divided by the cycle period of the stimulus train

extensor activity during entrainment also had to be esti- mated because during most trials this was obscured by the evoked E M G (LGS trace in Fig. 6A). Since the evoked E M G was usually preceded by an absence of activity in the ipsilateral GS muscle for a period corresponding ap- proximately to the contralateral flexor burst duration, we estimated that the onset of ipsilateral extensor activity must have closely followed the onset of the force pulse.

The minimum force necessary for entrainment was in the range of 5 to 10N. Figures 6B and 6C show plots of the phase of onset of contralateral St burst activity relative to the onset of the stimulus train versus time for two trials: in

one the magnitude of the force pulses was 15N and there was perfect entrainment (Fig. 6B), while in the other the magnitude of the force pulses was 8N and no entrainment occurred (Fig. 6C). The force magnitude was varied by altering the stimulus strength to the L7 ventral root. Similar results were obtained when the magnitude of the force pulses was varied by reducing the length of the GS muscles without changing the stimulus strength. We also noted that entrainment was lost by detaching the muscle from the force transducer (Fig. 6A).

The ventral root stimulus necessary for entrainment was in the range of 1.3 to 1.4 x T. This was below the minimum intensity of 2 x T for exciting gamma mo- toneurons (Matthews 1972). This was confirmed by re- cording from numerous single afferents dissected from L7 dorsal root filaments. These recordings showed that spindle afferents were silenced during contractions that produced entrainment. Increasing the stimulus strength beyond about 2 x T to produce large force pulses (greater than 20N) recruited gamma motoneurons and caused some spindles to discharge during the contraction. This did not occur for all spindle recordings presumably be- cause the gamma motoneurons innervating these spindles were not located in the L7 ventral root. We also recorded from numerous Ib afferents from the Golgi tendon organs and observed that these afferents discharged intensely during the contractions with the magnitude of activity related to the magnitude of the contraction.

3. Entrainment of the locomotor rhythm by muscle nerve stimulation

In 4 animals (1 chronic and 3 acute spinal preparations) we rhythmically stimulated the GS nerve with trains of stimuli (100/s for 500 ms) and observed that the locomo- tor rhythm could be readily entrained by input from group I afferents (Fig. 7). Figure 7A shows the data from a long entrainment sequence when the left GS muscle nerve was stimulated at 1.5 x T, this strength being below threshold for group II afferents. Figure 7B shows the plot of the phase of occurrence of the contralateral (right) St burst relative to the stimulus cycle for the same sequence. After an initial period of 4 cycles when the intrinsic rate was higher than the stimulation rate, the locomotor rhythm became entrained with a phase value between 0.1 and 0.3. Qualitatively similar data were obtained in all animals. Another example is shown in Fig. 8A. The aver- age phase-values during entrainment for the 4 experi- mental animals were 0.04, 0.09, 0.14 and 0.14, yielding a mean of 0.10 + 0.05. These values were not significantly different for those occurring during entrainment with force pulses (see previous section). Assuming that the flexor burst activity in the two hind legs alternates (Fig. 1) it can be concluded that the onset of ipsilateral flexor burst activity occurs soon after the termination of the stimulus train. This was indeed found to be true in one preparation in which the innervation to the ipsilateral St muscle was left intact (in other preparations this muscle was denervated to mir~imize movements in the ipsilateral

563

A

B 1.0

0.5

0.0

~ S t i m

5s

t . 5 x T

e � 9 �9 �9 � 9

�9 o �9 �9 o � 9

C

(u

1.0-

0.5

�9 ~ ~ 1 . 2 x T �9

o o ~

0.0

0 10 20 30 40

T i m e (s)

Fig. 7. Entrainment of the locomotor rhythm by electrical stimula- tion of the GS nerve. A) EMG records from the contralateral (right) GS and St muscle showing entrainment when the left GS nerve was stimulated rhythmically at 0.7 Hz (train duration 500 ms, stimulus magnitude 1.5 x T). A section lasting 10 seconds has been removed from the middle of the record. Note that during entrainment the onset of the contralateral St flexor bursts occurred slightly after the onset of the stimulus train. B) Phase plot for the sequence shown in A). The rhythm became entrained after about 4 stimulus cycles and remained entrained for almost 30s (indicated by the relatively con- stant phase value). C) Phase plot showing the absence of entrain- ment when the stimulus strength was reduced to 1.2 x T on the next trial. The intrinsic frequency of the locomotor rhythm was initially higher than the entrainment frequency and later slower than the entrainment frequency as indicated by the initial progressive decline in phase followed by a progressive increase in phase. The phase was calculated as described in Fig. 6

leg that sometimes caused movements of the GS nerve on the stimulating electrodes and hence a variation in the strength of the afferent volleys during a rhythmic se- quence). Moreover, in this preparation we confirmed that stimulation of the GS nerve at group I strength during an ipsilateral flexor burst prematurely terminated the burst (not shown, see Conway et al. 1987).

Although there is almost complete overlap of the stimulus range for exciting group Ia and Ib afferents in the MG nerve (Matthews 1972) a much greater percentage of Ia relative to Ib afferents are activated at stimulus

strengths in the range of 1 to 1.2 x T (Eccles et al. 1957). Assuming this is also true for the LGS nerve we predicted that if group Ib afferents are primarily responsible for entrainment, decreasing the stimulus strength to below 1 .2xT during an entrainment sequence would cause a loss or weakening of the entrainment. This prediction was tested in one animal in which long stable sequences of entrainment occurred, and in which a clear short-latency reflex was produced in response to stimulation of the GS nerve at all stimulus strengths used (Fig. 8B). The latter indicated the group Ia afferents were stimulated on all trials since this reflex was most likely produced by mono- synaptic excitation of motoneurons by group Ia afferents. The main result was that reducing the stimulus strength to 1.3 x T and below caused a weakening of entrainment as indicated by an increase in the variability of the phase (Fig. 8A). In other trials in this and other preparations we usually could not elicit entrainment when the stimulus strength was set at 1.2 x T throughout the trial, but we regularly observed entrainment with stimulus strengths in the range of 1.3 x T to 1.5 x T.

4. Absence of group Ia effects on locomotor rhythm

The inability to entrain the locomotor rhythm by electri- cal stimulation of the GS nerve at low stimulus strengths indicated that input from group Ia afferents alone does not influence the locomotor rhythm. This is consistent with the observation of Conway et al. (1987) that vibration of the GS muscle does not reset the rhythm in immobi- lized DOPA/Nialamide treated preparations. To establish this negative conclusion more firmly we examined the effects of vibrating the GS muscles (150 Hz, 100 #m) after immobilizing the preparations (3 acute animals). Immo- bilization was necessary because vibration otherwise pro- duced a reflex contraction of the GS muscles. Immediately before immobilization we noted that the locomotor rhythm could be entrained by muscle stretch. This was important because we wanted to be sure that a negative effect of vibration was not due to a generalized insensiti- vity of the locomotor rhythm to afferent input. An addi- tional control was to monitor the activity of single Ia afferents to confirm that vibration effectively excited these afferents after immobilization (Fig. 9A). In one prepara- tion the effective activation of the Ia afferents was also indicated by a weak vibration reflex in the GS muscles when the immobilizing effect of Flaxedil was wearing off. In none of the 3 preparations were we able to entrain the rhythm (Fig. 9), and intermittent trains of vibration failed to reset the rhythm or alter the rate of the rhythm (not shown, see Conway et al. 1987).

Discussion

A previous investigation by Conway et al. (1987) on DOPA/Nialamide treated spinal cats concluded that in- put from group Ib afferents can strongly influence the locomotor rhythm generator. The results of this study provide further evidence in support of this conclusion.

A

03 09 t~

_C2 13-

564

1.0-

0.5

0,0

D O O O O O D D D~'~ ~

n n �9 �9 �9

OO �9 o O I oO I I oO

i i i i �9 i i u

0 10 20 30 40 50 60

Time (s)

B

sca tcn

5ms

A

m iMIJdhL J IJHI JltUl J lJi, Jl n,, i J, J ill J Jill ~U[IJl~lUl'U jr,,t,~n 'HI ~[j~u' 1,,,,~ilU II,H]'RNL I 1H ]UU~' ] [ ~ i ' * 1 I , ' H 1Pi~ , ['i~ '

(integ)

-~ -~ -~ "-~ -~ -~ -~ -~ - - - - - I - - - i ~ Vib. R GS

5s

B

1.0-

0.5

0.0

'0 1'0 2'0 ' 4o Time (s)

Fig. 9. Absence of entrainment of the fictive locomotor rhythm by vibration of the GS muscles. A) Records showing the absence of phase locking of the fictive rhythm as recorded from the left St flexor nerve (second trace, rectifed and filtered record) to a rhythmic vibration of the right GS muscles (third trace). Vibration amplitude 100 #m. The amplitude of the vibration and length of the muscle were set so the vibratory stimulus resulted in 1 : 1 following of spikes in group Ia afferents from the GS muscle (top trace. B) Phase-plot showing the absence of entrainment of the fictive rhythm by vibra- tion of the GS muscle. During this sequence the phase of the onset of burst activity in the contralateral St nerve relative to the onset of the vibratory trains progressively decreased

Fig. 8. Loss of strong entrainment of the locomotor rhythm when the strength of rhythmic stimulation of the GS nerve was decreased to 1.3 x T and below. A) Plot of phase of the onset of contralateral St flexor burst relative to the stimulus train versus time. During the initial part of this sequence the stimulus strength was progressively decreased. Initially the locomotor rhythm was strongly entrained at stimulus strengths of 1.7 x T (open circles), 1.55 x T (open squares) and 1.4 xT (open triangles). Reducing the stimulus strength to 1.3 x T (filled circles) and then to 1.15 x T (filled squares) caused a loss of entrainment as indicated by the progressive increase in phase. Returning the strength to 1.3 x T restored entrainment but this was lost when the strength was reduced again to 1.15 x T. B) Records from the ipsilateral sciatic nerve showing the afferent volley (large spike) evoked a short-latency reflex volley (small spike) at 1.4, 1.3 and 1.15 x T. These records show that when the entrainment in the sequence in A) was weakened and eventually lost the afferent volley was still effective in eliciting the reflex

Using a different preparation (non-immobilized Clonidine treated spinal cats) we have found that the locomotor rhythm can also be entrained by rhythmic force pulses in the ankle extensor muscles (Fig. 6) and by repetitive stimu- lation of group I afferents from these muscles (Fig. 7). The mean phase of the onset of the contralateral flexor bursts relative to the onset of the stimulus train was 0.15 +0.05 (n=5) when force pulses entrained the rhythm and 0.10_+0.05 (n=4) when entrainment was produced by electrical stimulation of the group I afferents. These values were not significantly different. For entrainment by elec- trical stimulation of the muscle nerve the afferents respon- sible could only have been group Ia and/or group Ib since the stimulus strength for entrainment was below group II threshold. We consider any major contribution from group Ia afferents unlikely because selective activation of these afferents by vibration did not entrain the locomotor rhythm (Fig. 9), and reduction of the stimulus strength to a level sufficient to evoke short-latency reflexes but ex- clude much of the Ib input caused a loss or weakening of entrainment (Fig. 8). Input from muscle spindle afferents is also unlikely to be involved in the entrainment of the locomotor rhythm by force pulses. If spindle afferents had been primarily responsible for entrainment in this situ- ation then the timing of motor activity relative to the force pulses should have been significantly different from that when entrainment was produced by electrical stimulation of the motor nerve, i.e. an expected phase-difference of about 0.4. This was not observed. Since Ib afferents were strongly excited by the force pulses, and electrical stimula- tion of these afferents entrains the locomotor rhythm (see above), we conclude that entrainment with force pulses is most likely produced by input from group Ib afferents. However, we have not excluded the possibility that input from smaller force sensitive afferents (Mense and Meyer 1985) could also contribute in this situation. Stimulation

565

of small muscle afferents can strongly influence the loco- motor rhythm in D O P A and Clonidine treated prepara- tions (Grillner 1973). Of potential significance is that ip- silateral flexor bursts occur after the termination of the stimulus trains to these afferents, this corresponding to the fact that ipsilateral flexor burst activity occurs following the return of the muscle force to low values during en- trainment with force pulses. However, before any role for small force sensitive afferents in regulating the locomotor rhythm can be accepted it must be shown that these and not other small afferents are responsible for the effects produced by high intensity stimulation of muscle nerves.

The timing of the locomotor rhythm relative to the entraining stimuli we used in this study was consistent with the hypothesis that during the stance phase of walk- ing group Ib afferents function to inhibit the generation of ipsilateral flexor burst activity and to promote the genera- tion of ipsilateral extensor burst activity. First, when using force pulses for entrainment the onset of ipsilateral exten- sor activity occurred close to the onset for the force pulse (Fig. 6A). Second, we assume that this also occurred when group I afferents were electrically stimulated because the timing of entrainment of the contralateral flexors using this method was the same as those when force pulses were used. This is consistent with data from paralysed DOPA/Nia lamide treated animals where electrical stimu- lation of group Ib afferents evokes EPSPs in ipsilateral extensor motoneurons (Conway et al 1987; Gossard et al 1990). Finally, during entrainment with saw-tooth stretches of the ankle extensor muscles activity in these muscles was coincident with the increase in force, and the onset of flexor burst activity followed the release of the muscles from stretch (Figs. 2-5). We have attributed en- trainment in response to muscle stretch to input mainly from group Ib afferents because we know that these affer- ents are strongly excited during the stretch and that input from group Ia afferents has no effect on the rhythm. Furthermore, we have been unable to establish that input from group II muscle afferents influences the locomotor rhythm since no significant phase shifts occurs when the stimulus strength to the GS nerve is raised (up to 5 x T) to include the group II afferents (unpublished observations).

These data, together with those from other studies (Duysens and Pearson 1980; Conway et al. 1987; Gossard et al. 1990) suggest that a decline in force from leg extensor muscles signalled by a decline in activity of group Ib afferents from these muscles may play a role in initiating the swing phase of walking. Force measurements from ankle extensor muscles in normal walking cats have shown that a marked decline in force precedes the onset of the swing phase by about 150 ms (Hoffer et al. 1989). Correspondingly, chronic recordings from single Ib af- ferents show that their activity declines about 130 ms prior to swing (Appenteng and Prochazka 1984). Assum- ing that flexor burst activity begins about 50 ms prior to the onset of swing it follows that during normal walking flexor burst activity begins approximately 100 ms follow- ing the decline in force in the ankle extensor muscles. A similar temporal relationship has been reported for knee muscles where the onset of activity in the sartorius muscle follows a sudden decline in force in knee extensors

by 50 to 150 ms (see Figs. 2 and 3 in Hoffer et al. 1987). It is clear, therefore, that in normal walking animals the inter- val between the decline in extensor muscle force and the onset of flexor burst activity is significantly smaller than the interval we measured for the onset of flexor burst activity following the release of the isolated ankle extensor muscles (Figs. 4, 5). The latter interval was highly variable with a mean of about 200 ms (Fig. 4). The issue now is whether this discrepancy forces us to abandon the idea that unloading of leg extensors is directly involved in triggering the stance to swing transition. One possibility is that the difference is due to our limiting Ib input from a single set of extensor muscles. Perhaps the sudden re- moval of Ib input from all leg extensors would cause an earlier onset of flexor activity. Alternatively, the swing to stance transition may normally be regulated by input from other afferents such as input from group II afferents from proximal muscles (Edgley and Jankowska 1987) and from joint afferents (Shimamura et al 1984). Regardless of whether or not a decline in Ib input from extensor muscles is directly responsible for the stance to swing transition it now seems likely that these afferents do function to pre- vent the initiation of swing when extensor muscles are loaded.

Acknowledgements. We wish to thank Dr. A. Prochazka for his valuable comments on a draft of this paper, and M. Wheatley and R. Gramlich for assisting in some of the experiments. This study was supported by a program grant from the Medical Research Council of Canada, and by Postdoctoral Fellowships to JMR from the Medical Research Council of Canada, and to WJ from the Rick Hansen Foundation.

References

Appenteng K, Prochazka A (1984) Tendon organ firing during active muscle lengthening in awake, normally behaving cats. J Physiol 353:81-91

Barbeau H, Rossignol S (1987) Recovery of locomotion after chronic spinalization in the adult cat. Brain Res 412:84 95

Conway BA, Hultborn H, Kiehn O (1987) Proprioceptive input resets central locomotor rhythm in the spinal cat. Exp Brain Res 68:643-656

Duysens JD, Pearson KG (1980) Inhibition of flexor burst genera- tion by loading ankle extensor muscles in walking cats. Brain Res 187:321-332

Eccles JC, Eccles RM, Lundberg A (1957) Synaptic actions on motoneurones in relation to the two components of the group I muscle afferent volley. J Physiol 136:527 546

Edgley SA, Jankowska E (1987) An interneuronal relay for group ! and II muscle afferents in the midlumbar segments of the cat spinal cord. J Physiol 389:647-674

Grillner S (1973) Locomotion in the spinal cat. In: Stein RB, Pearson KG, Smith RS, Redford JB (Eds). Control of posture and loco- motion. Plenum Press, New York. pp 515-536

Grillner S, Dubuc R (1988) Control of locomotion in vertebrates: spinal and supraspinal mechanisms. In: Waxman SG (ed) Func- tional recovery in neurological disease. Raven Press New York. pp 425-453

Hoffer JA, Caputi AA, Pose IE, Griffiths RI (1989) Roles of muscle activity and load on the relationship between muscle spindle length and whole muscle length in the freely walking cat. Prog Brain Res 80:75-85

Hoffer JA, Loeb GE, Sugano N, Marks WB, O'Donovan M J, Pratt CA. (1987) Cat hindlimb motoneurons during locomotion. Ill.

566

Functional segregation in sartorius. J Neurophysiol 57:554-562 Lundberg A, Malmgren K, Schomburg ED (1987) Reflex pathways

from group II muscle afferents. 1. Distribution and linkage of reflex action to alpha motoneurones. Exp Brain Res 65:271-281

Matthews PBC (1972) Muscle receptors. Edward Arnold, London Mense S, Meyer H (1985) Different types of slowly conducting

afferent units in cat skeletal muscle and tendon. J Physiol 363: 403-417

Pearson KG, Jiang W, Ramirez JM (1992) The use of naloxone to facilitate the generation of the locomotor rhythm in spinal cats. J Neurosci Methods In Press

Prochazka A, Trend P, Hulliger M, Vincent S (1989) Ensemble proprioceptive activity in the cat step cycle: towards a represent- ative look-up chart. Prog Brain Res 80:61 74

Rossignol S, Lund JP, Drew T (1988) The role of sensory inputs in

regulating patterns of rhythmical movements in higher verte- brates. In: Cohen A, Rossignol S, Grillner S (eds) Neural control of rhythmic movements in vertebrates. John Wiley and Sons, New York. pp 201-283

Sherrington CS (1910) Flexor-reflex of the limb, crossed extension reflex, and reflex stepping and standing (cat and dog). J Physiol 40:28-121

Shimamura M, Kogure I, Fuwa T (1984) Role of joint afferents in relation to the initiation of forelimb stepping in thalamic cats. Brain Res 297:225-234

Smith JL (1986) Hindlimb locomotion of the spinal cat synergistic patterns, limb dynamics and novel blends. In: Grillner S, Stein PSG, Stuart DG, Forssberg H, Herman RM (eds) Neurobiology of vertebrate locomotion. Macmillan Press, Hong Kong. pp 185-200