Stepping Behaviors in Chronic Spinal Cats with One Hindlimb Deafferented

The Journal of Neuroscience, August 1987, 7(8): 2537-2546

Carol A. GiulianP and Judith L. Smith Laboratory of Neuromotor Control, Department of Kinesiology, University of California, Los Angeles, California 90024

Adult cats develop spontaneous airstepping (walking mo- tween hindlimb generators for stable interlimb coordination. tions without ground contact) 4-6 weeks after spinal tran- Future modeling of interlimb coordination should consider section (Giuliani and Smith, 1985). This unique preparation the role of afference. provides an in vivomodel for studying the effects of hindlimb deafferentation on stepping behaviors without rostra1 input Neuronal circuits within lumbosacralsegments of the cat’s cord, to lumbosacral segments. The primary purpose of this study called spinal pattern generators, produce many features of the was to characterize airstepping and, to a lesser extent, bi- locomotor pattern without motion-dependent feedback from pedal treadmill locomotion in chronic spinal cats after deaf- the hindlimbs (Grillner and Zangger, 1979, 1984; Fleshmanet ferentation. Five cats were spinalized at T12, and EMG elec- al., 1984). However, some details, such as the 2-burst pattern trodes were implanted in selected knee and ankle muscles. of the semitendinosusmuscle (Grillner and Zangger, 1984; Per- After spontaneous airstepping developed, one hindlimb was ret and Cabelguen, 1980), may be altered after limb deaffer- deafferented extensively. entation, while other aspects,such as rhythmicity and interlimb Movements of both hindlimbs were depressed following phasing, deteriorate easily and become highly variable (Baker unilateral deafferentation. By the third week, spontaneous et al., 1984; Chandler et al., 1984; Grillner and Zangger, 1984). airstepping recovered in the normal hindlimb (N-Limb), but Here, we further characterize the capacity of spinal pattern gen- in the deafferented limb (D-Limb) airstepping was not spon- erators of lumbosacralsegments by assessingstepping behaviors taneous during the 3-4 months of testing. By the seventh in the chronic spinal cat with one hindlimb deafferented. week, bilateral airstepping was facilitated by tonic extero- An adult cat with a transected cord at the low-thoracic level ceptive simulation (tail-pinch). During bouts of bilateral air- (T 12) provides a good in vivo model for the study of the effects stepping, assessed during the third month of deafferenta- of deafferentation on the neuromuscular patterns of stepping tion, D-Limb airstep cycles were characterized by erratic behaviors. Adult spinal cats develop airstepping(walking move- rhythm. Further, cycle periods and extensor burst durations ments without ground contact; Giuliani and Smith, 1985) and were prolonged, but flexor burst durations were unmodified. some recover hindlimb walking with weight support 4-6 weeks In contrast, N-Limb cycles were rhythmic and of normal du- after cord transection (Rossignolet al., 1982; Giuliani et al., ration, but extensor burst durations were short and unrelated 1984; Lovely et al., 1986). Airstepping, in particular, is an ex- to cycle periods. Interlimb coordination was markedly un- emplary behavior to study the natural capacity of spinal pattern stable, showing only brief periods of alternating steps. When generatorsbecause it occursspontaneously and is not paced by erratic bilateral airstepping was accompanied by micturition, the speedof the treadmill belt. Further, it requiresno facilitation rhythmic and alternate stepping emerged, with normalized by neuropharmacological agents such as L-DOPA (typical of intralimb synergies. During hindlimb treadmill locomotion, fictive preparations; Grillner and Zangger, 1979) or midbrain tested in 2 cats, the N-Limb took weight-bearing steps and stimulation (typical of mesencephalic-deafferentedprepara- followed changes in the belt speed. The D-Limb, however, tions; Grillner and Zangger, 1984). Thus, our primary aim was stepped erratically, making contact on the dorsum of the to characterize airsteppingpatterns in the chronic spinal cat with paw; consistent bouts by bilateral stepping were not ob- one hindlimb deafferented.Deafferentation by ganglionectomy tained. was chosen becausethe procedure is performed extradurally, Lumbosacral afference of some type, either from the hind- minimizing the possibility of cord damage(Wiesendanger, 1964), limb or from regions remote from the limb, such as the blad- and cells of ventral root afferents are excised, eliminating all der, appears to be essential for stabilizing the coupling be- sensoryinput to the segment(Coggeshall, 1980). In a continuation of our efforts to compare airstepping with hindlimb locomotion on a treadmill (Giuliani and Smith, 1985), we placed the chronic spinal cat with one hindlimb deafferented Received Oct. 17, 1986; revised Jan. 26, 1987; accepted Jan 29, 1987. on a motorized treadmill and facilitated hindlimb walking by This study was supported by NIH Grant NS 19864. The authors acknowledge pinching the tail or perineum. We comparedairstepping patterns Gail Koshland for her assistance in data collection and Scott Chandler and Alan Garfmkel for their comments on an earlier draft of the oaner. with those of treadmill walking in our spinal-deafferentedcats Correspondence should be addressed to Judith L. Smith, Department of Ki- and in the mesencephalic-deafferentedcats tested by Grillner nesiology, College of Letters and Science, 2863 Slitter Hall, University of Cali- and Zangger(1984). fornia, Los Angeles, CA 90024. a Present address: Department of Medical Allied Health Professions, School of Last, we wanted to determine whether the recovery of motor Medicine, University of North Carolina, Chapel Hill, NC 275 14. functions after limb deafferentation was similar in chronicspinal Copyright 0 1987 Society for Neuroscience 0270-6474/87/082537-10$02.00/O cats and normal adult cats. Since Sherrington’sobservations in 2538 Giuliani and Smith * Stepping in Chronic Spinal Cats After Deafferentation

19 10, several investigators have described the recovery of motor walking prior to deafferentation, and 2 months after the gandionectomy functions after hindlimb deafferentation in normal cats. Most they were tested on a motorized treadmill. The apparatus, described by Carter and Smith (1986b). allowed the cat’s forelimbs to support the cats recover some ability to locomote overground @prong, 1929; upper trunk by being positioned on a shelf 2 cm above the belt. For Wiesendanger, 1964; Goldberger, 1977) or on a motorized additional support, the animal wore a nylon-net vest with flaps attached treadmill (Wetzel et al., 1976); however, movement of the deaf- to the sides of the apparatus,and the experimenterheld the cat’stail ferented limb is seldom normal. Neural mechanisms underlying for lateral balance and to prevent collapse during periods of inadequate the recovery process are not understood, but they are generally hindlimb weight support. Airstepping and treadmill locomotion were assessed primarily by attributed to alteration in descending fibers or intact dorsal root EMG recordings that were synchronized by a 16-bit binary code with afferents, both contralateral and ipsilateral to the deafferentation a videorecording oflimb movements (Betts et al., 1976). Using methods (Sprong, 1929; Goldberger and Murray, 1974; Goldberger, 1977). described by Smith et al. (1977), bipolar fine-wire electrodes-were sur- It is not known to what extent recovery depends on the adap- gically implanted in 4 muscles of the deafferented limb: one knee extensor, the vastus lateralis (VL), 2 ankle extensors, the soleus (SOL) and tation of fibers projecting to lumbosacral segments or to neural lateral gastrocnemius (LG), and one ankle flexor, the tibialis anterior circuits intrinsic to the cord, and recovery of motor functions (TA). In the normal limb, the soleus alone was implanted. Myopotentials after deafferentation has not been studied in chronic spinal prep- were amplified ( x 1000), filtered with a high- _pass of 100 Hz.and recorded arations. Preliminary reports of our work have been published on FM tape (33/4in./sec) for subsequent analyses. Signals were sampled (Giuliani and Smith, 1983). off-line at 1 kHz, full-wave rectified, and stored on disk (PDP 1 l-23). FORTRAN programs used to determine burst duration, cycle period (de- fined here as the time interval between sequential SOL bursts), and Materials and Methods onsetlatency for eachmuscle (referenced to SOL and expressedas a Subjects and preparation. Five animals were selected from a group of percentage of the cycle period) have been described (Carter and Smith, adult cats spinalized at the T12 level, and were tested periodically for 1986a). 6-8 months to characterize airstepping (Giuliani and Smith, 1985). In Statistics. To determine the effects of deafferentation on airstepping, 4 spinal cats, ganglia from dorsal roots L3 to Sl on the right side were data from the normal limb (N-Limb) and the deafferented limb (D- removed, and in one cat the caudal extent included S2. Ganglionecto- Limb) were compared. When possible, similar data acauired under mies were performed under aseptic conditions with sodium pentobar- control conditions (i.e., airstepping elicited by tail pinch in spinal cats bital anesthesia (Nembutal; 35 mg/kg). The ganglia were exposed by a with normally afferented hindlimbs) were compared to N- and D-Limb hemilaminectomy that left spinous processes intact. Under the view of data. Control records, taken from our previous study (Giuliani and an operating microscope, an incision was made in the dural sheath Smith, 1985), consisted ofdata from the D-Limb before deafferentation. surrounding the ganglion. After subdural tissue was removed, separation For all statistical tests. the level of significance was D 5 0.05. Although between dorsal and ventral roots was visible, and the dorsal root was the airstepping of the’spinal cat witi the more caudal deafferentation cut proximal to the ganglion. The ganglion was excised after loose tissues (L3-S2) was similar to that of the other 4 cats, data from this animal connecting it to the ventral root were dissected. The surgical site was were not included in the statistical analyses. packed with Gelfoam, and deep and superficial tissues, except for the dural sheath, were sutured. Recovery from anesthesia was uneventful, Results and the animals were returned to their cages within 24 hr. Procedures Clinical observations for daily care of spinal cats have been described previously (Smith et al., 1982; Lovely et al., 1986). During the entire testing period following deafferentation, no The L3-Sl ganglionectomy resulted in an extensive, but incomplete, EMG responseswere elicited by brisk taps delivered to the hindlimb deafferentation. The rostra1 extent, chosen to replicate deaf- Achilles tendon (SOL, LG), patellar tendon for VL, or TA ten- ferented preparations used in recent studies of hindlimb locomotion don of the D-Limb. By contrast, tendon-tap reflexeswere easily (Grillner and Zangger, 1984), spared afferents from a hip flexor (ilio- elicited in the N-Limb. In addition, no myopotentials were psoas) that enter roots rostra1 to L3. The caudal extent was selected with 2 objectives in mind. We wanted to spare tail innervation (Reed, 1970) evoked in deafferented musclesin responseto passivestretch. and some cutaneous innervation to the hindlimb. Branches of the pos- Initially, muscleatonia was evidenced by muscle palpation and terior femoral cutaneous nerve enter S2 (Koerber and Brown, 1982). a marked hypermobility of passivejoint motion at the ankle. Intact cutaneous fields of the hindlimb were stimulated to elicit air- Although flaccidity decreasedduring the first postoperative stepping and to test for paw shaking, a stereoptyical behavior examined in another study (Koshland and Smith, 1983). More important, it was month, the D-Limb always assumedan extended position char- necessary that the extent of deafferentation eliminate proprioceptive acteristic of deafferented limbs (Sprong, 1929; Wiesendanger, input from distal hindlimb muscles. In the cat lumbosacral plexus, 1964). Thus, when the cat was sitting or lying down, the knee except when classified as “postfixed” (< 15% of cat lumbosacral plex- and ankle were extended and the toes were flexed. Sometimes us), ankle and digit muscle afferents enter segments rostra1 to S2 (Romanes, 195 1; Jefferson, 1964). Thus, lack of tendon-tap reflexes in the hip was flexed, thereby positioning the extended D-Limb ankle muscles was evidence against a postfixed plexus. However, be- forward, but at other times the hip was extended, positioning cause the effects of S2input from Ia afferentsmay have been too small the limb backwards. In general, D-Limb extension was not ac- to be detected by reflex testing, postmortem measurements to determine companiedby tonic EMG activity from SOL, LG, or VL unless plexus fixation were conducted. extensor responseswere evoked by tail pinching or other cu- Testing protocols and EMG analyses. To assess airstepping and reflexes after deafferentation, testing was conducted twice a week for the taneous stimuli. At the knee and ankle, the full range of motion first 6 weeks and weekly thereafter for 4 months. To test for airstepping, was maintained throughout the testing period. At the metatar- the cat was supported under the axillae, lifted, and held vertically with sophalangealjoints, however, a full range of toe extension was hindlimbs pendant. Initially, we looked for spontaneous airstepping, difficult to obtain; this appeared to be due to adaptations in and if it did not occur, we attempted to evoke the behavior by pinching articular tissues,rather than to the inability to passively stretch the tail or stimulating the intact cutaneous fields on the deafferented limb. Reflex-testing stimuli included the tendon tap for deep tendon the toe flexors. reflexes, paw pinch for flexion and crossed extension reflexes, and pin To assessintact cutaneous fields of the D-Limb, pin prick prick for flexion withdrawal reflexes. To record the stimulus delivery, was applied systematically to all surfacesof the limb, and re- the cantilever of a force disulacement transducer (Grass FTO3C) was sponsive areas were mapped on a schematic diagram of the fitted with a small plastic bar, used as a reflex hammer for tendon’taps, anda 28-gaugeneedle, used to prick the skin.The transducersignal was hindlimb. For the 4 cats with L3-Sl deafferentation, the intact recorded on FM tape with the EMG resnonses. cutaneous fields were consistent with S2 dermatomes (Brown Of the 5 spinal cats, 2 exhibited good weight-bearing during hindlimb and Koerber, 1978; Koerber and Brown, 1982). In particular, The Journal of Neuroscience, August 1987, 7(8) 2539

A. FIRST WEEK responsive areas centered around the skin overlying the femoral trochanter and fields projecting down the posterolateral aspect of the thigh. During the first postoperative week, pin prick to TA this area elicited a brisk flexor responsethat occurred most 1:: noticeably at the hip. Pinching the N-Limb paw produced either a crossed flexion or a weak crossed extension responsein the D-Limb, while paw pinch of the D-Limb failed to elicit any response.By the third and subsequentweeks, pin prick to intact fields of the D-Limb and paw pinch of the N-Limb elicited brisk flexion and crossedextension responses,respectively, usually involving all 3 joints. Withdrawal reflexes in both limbs ap- pearedto be exaggerated.By the fourth week, D-Limb receptive N-SOL fields to nociceptive stimuli had expanded distally to include lateral areason the shank just below the knee. -trq- ’ I J 1 ~-~+v- Onset and recovery of airstepping TP During the first postoperative week, when the cat was held vertically to test for airstepping, the D-Limb was flaccid and hung lower than the N-Limb, which was slightly flexed at the hip, knee, and ankle, the posture typical of a pendant limb in adult spinal cats (Giuliani and Smith, 1985). Muscle activity was B. THIRD WEEK absent in the D-Limb except for occasionalepisodes of spontaneous extensor EMG, however, tail pinching or twisting elicited tonic extensor EMG in the D-Limb and brief bouts of airstepping in the N-Limb (Fig. 1A). Attaching a small clip to q--+--4+ innervated skin about the posterolateral surfaceof the D-Limb evoked spasmodic movements in both hindlimbs for a brief period. By the third week, the N-Limb exhibited prolonged periods of spontaneousairstepping, while extensor musclesof the D-Limb, particularly the SOL, were tonically active. Tail pinching increasedthe level of extensor EMG in the D-Limb, while increasing the stepping frequency of the N-Limb. Occa- sionally, a few bilateral airsteps occurred with tail pinch (Fig. 1B) or immediately after lifting the cat to a vertical position. More often, micturition and bowel evacuation were associated with bilateral airstepping. During weeks 4-6 after deafferentation, pinching the tail or applying a small clip to the skin over the posteriolateral thigh of the D-Limb evoked airstepping movements in both hindlimbs. Continuous D-Limb stepping, however, was difficult to elicit, and only brief episodesof 5-l 0 cycles were recorded (Fig. C. FIFTH WEEK lc). By the seventh postoperative week, bilateral airstepping was easy to elicit with tail pinch in all animals, and records taken over the next few weeks were usedto characterize interlimb coordination and the intralimb synergy. After the eighth y++--++-l-~++-- postoperative week, D-Limb airsteppingoccurred spontaneous- 800 msec ly for short periods and then ceaseduntil the tail was pinched. Becauseepisodes of spontaneousairstepping were rare, we did not characterize them. Interlimb coordination after recovery of airstepping Interlimb coordination was judged primarily by SOL activity patterns in both limbs. Following unilateral deafferentation, in- - terlimb coordination during airsteppingwas severely disturbed, and this continued throughout the testing period. As reported in the previous section, the N-Limb often airstepped while the

c Figure I. Recovery of airstepping after unilateral hindlimb deafferentation in a chronic spinal cat. For all records, myograms from TA D-Limb. B, By the third week, there were prolonged periods of spon- and SOL of the deafferented limb and soleus of the normal hindlimb taneous airstepping in the N-Limb and a few steps in the D-Limb after (N-SOL) are shown.A, Duringthe first week,tail pinching(TP) facil- tail-pinch (steps I-3). C, By the end of the fifth week, continuous bi- itated airstepping in the N-Limb and tonic extensor activity in the lateralairstepping was elicited with tail pinch. 2540 Giuliani and Smith * Stepping in Chronic Spinal Cats After Deafferentation

600 msec 500 msec

N-SOL 1 2 3 4 5 6 7 8 9 10 11 .. 12 13 14 15 16

ALTERNATING VARIABLE IN-PHASE Figure 2. Interlimb coordination revealed by the phasing of ankle extensor bursts. EMG records from the D-Limb (SOL) and N-Limb (N-SOL), taken during 2 bouts of bilateral airstepping recorded during the third deafferentation month, show 3 patterns of interlimb coordination. During steps 1-4, the limbs stepped alternately; but subsequently, the rhythmicity of D-Limb airstepping deteriorated, and phasing between the limbs was variable for steps 5-11. During steps 12-16, both limbs airstepped in phase. The average cycle period for the N-Limb was 650 msec for steps I-11 and 295 msec for steps 12-16.

D-Limb did not. Further, when both limbs airstepped, they extended with 1 l%, but the majority (62%) of airsteps were often exhibited cycle periods of different durations, resultingin accompaniedby D-Limb stepping. variable phasingbetween hindlimbs (e.g., steps5-l 1 of Fig. 2). To quantify the phaserelationship between the limbs, records There were, however, brief bouts of airstepping in which both characterizedby rhythmical bilateral airsteppingwere analyzed. limbs steppedalternately (e.g., steps1-4, Fig. 2) while, in other The onset of SOL activity in the D-Limb was referencedto the bouts, the hindlimbs steppedtogether or “in phase” (e.g., steps onset of SOL activity in the N-Limb, and expressedas a per- 12-16, Fig. 2). centageof the airstep cycle. The frequency distribution of SOL To assesspatterns of interlimb coordination, a total of 1935 onsetsshowed that phasingduring bilateral airsteppingwas dis- airstep cycles from 4 animals were examined from bouts of 5 tributed rather evenly over the entire range of possiblevalues or more N-Limb stepstaken in sequence.The records assessed (DEAFF in Fig. 3). By contrast, spinal cats with normally af- were from the third deafferentation month. During eachN-Limb ferented hindlimbs exhibited a narrower range of interlimb- airstep, the status of the D-Limb was assessedby SOL activity phasing values and showed a strong central tendency toward and the posture or motions observed in video records. In most alternate phasing (CONTROL in Fig. 3). The 2 distributions cases,D-Limb status wascharacterized by one of the following: from spinal cat preparations (DEAFF and CONTROL, Fig. 3) (1) no SOL activity, with the limb pendant; (2) tonic SOL ac- were statistically different (X = 25.6, p < 0.05), suggestingthat tivity, with the limb held in extension; or (3) phasic bursts of interlimb coordination during bilateral steppingwas altered by SOL activity, with the limb airstepping. In total, the D-Limb unilateral hindlimb deafferentation. was pendant with 27% of N-Limb airsteps and was actively Cycle periods and intralimb synergiesafter recovery From each animal, 50 airstep cycles from bouts of at least 5 sequentialsteps, recorded during the third deafferentationmonth, 40 were selectedat random to characterize the intralimb synergy. u7 ul 0 CONTROL Average data for cycle period, flexor and extensor burst dura- 6 30 & I 0 DEAFF Table 1. Airstepping characteristics in spinal cats before and after unilateral hindlimb deafferentation

Airstep parameter N-Limb* D-Limb ControP Cycle period (msec)d 504 f 160 884 1- 366 565 k 62

10 20 30 40 50 60 70 80 90 100 Burst durations (msec) SOLd 102 + 48 601+- 282 273 + 60 PERCENT OF CYCLE PERIOD TA Nk 223 +- 142 132 f 6 Figure 3. Frequency histogram of SOL-SOL phasing for airstepping TA onset (% of cycle) NA 78 -I 12 77 f 15 in 2 preparations. In both plots, onset of the SOL muscle in one limb was referenced to the onset of the same muscle in the contralateral limb. aAll airstepping elicited by tail pinch, average t- SD. Onset latencies, expressed as a percent of cycle period, were obtained b N-Limb, normal limb, D-Limb, L3Sl deafferented limb, n = 200 airsteps (50/ from 4 spinal cats with one hindlimb deafferented (DEAFF, n = 169 cat). airstep cycles) and 5 spinal cats with hindlimb afferents intact (CON- c Control, airstepping data from spinal cats with normally afferented hindlimbs; TROL; n = 140 airstep cycles). Airstepping was recorded during tail n = 140 steps. pinch in both preparations. The control data distribution was redrawn d ANOVA F-value significant at p 5 0.05; see text for post hoc comparisons. from figure 2 of Giuliani and Smith (1985). e NA, not applicable; TA in the normal limb was not implanted for EMG recordings. The Journal of Neuroscience, August 1987, 7(8) 2541

B. A. 1800

CONTROL .I’ . 1600 xi 350 r = 0.89 N = 85 1 300

z 250 . 5 200 N-LIMB 800 t 5 n 150 r = 0.37 . N=168 600 L 100 400 . .* 50 ‘. 200

0 I I I I I I I 0 0 270 450 630 810 990 1170 1350 0 400 800 1200 1600 2000 2400 2800

CYCLE PERIOD (msec)

Figure 4. Relationship between burst duration of the ankle extensor and airstep cycle period for normal and deafferented limbs. Dots represent data from each airsteu cycle. and data were Dooled for 4 cats. For the N-Limb. SOL burst durations and cvcle ueriods were weaklv correlated (solid line, A), while under controlconditions, the-2 were highly correlated [dashed line, A (redrawn from fig. 4,4-of C&uliani and Smith, -1985; data points are not illustrated)]. For the D-Limb, the 2 variables were highly correlated (B), but the intercept and slope were different from those of the control data. tions, and onset latencies are given in Table 1. In addition to under control conditions, SOL activity occupied about 42% of data for N- and D-Limbs, data acquired under control condi- the cycle. tions (spinal cat without deafferentation) are presented. In the D-Limb, however, neither the TA burst duration nor On average, the D-Limb airstep cycle was 330 msec longer the average onset latency of the TA was altered by deafferen- and more variable than that of the N-Limb. During bilateral tation (Table 1). But as a result of the prolonged SOL burst, the stepping, therefore, D-Limb stepping frequency was usually silent period betweenthe end of the extensor burst and the onset slower than that of the N-Limb, and the N-Limb often took 2 of the flexor burst, typical of airstepping patterns, was virtually steps for each D-Limb step (Fig. 5 and steps5-l 1 in Fig. 2). absent in the D-Limb (Fig. 5). The average N-Limb cycle period was similar to the control Finally, fast-contracting extensor musclesof the D-Limb, spe- average; thus, there was no changein the stepping frequency of cifically the LG and VL, were not usually active during tail- the hindlimb contralateral to the deafferented limb. In the pinch-induced airstepping,but, when recruited, their burstswere N-Limb, however, SOL burst durations were reduced substan- coactive with the SOL burst (e.g., see VL and SOL in Fig. 6). tially, being only one-third as long as thoseunder control testing This finding is in marked contrast to control conditions, in which (Table 1). Further, the shortened SOL bursts were poorly correlated with their own cycle periods, as demonstratedby a low correlation coefficient (R = 0.37) betweenthe 2 parameters(Fig. 4A), suggestingthat the duration of extensor activity in the N-Limb was independent of cycle period. This is in marked contrast to control conditions, during which SOL burst durations were highly correlated (R = 0.89) with cycle period (see dashedline in Fig. 4A). In contrast to the briefextensor burstsofthe N-Limb, extensor activity in the D-Limb was prolonged, and, on average, SOL burst durations were nearly 6 times longer in the D-Limb than in the N-Limb. One rather extreme example of this discrepancy is illustrated in Figure 5 (but seealso Fig. 2). Here SOL activity in the D-Limb varied from 500 to 800 msecin duration, while each SOL burst of the N-Limb lasted no more than 100 msec. I 500 msec The long SOL burst was proportional to the prolonged airstep Figure 5. Example of dissimilar SOL burst durations and cycle periods cycle of the D-Limb, and there was a high, positive linear cor- of normal and deafferented hindlimbs during bilateral airstepping. relation (R = 0.85) betweenthe 2 parameters(Fig. 4B). Although This record, recorded during the third deafferentation month, illustrates the correlation for D-Limb data was similar to that reported for that the average cycle period for the D-Limb (SOL) was longer and more variable than that for the N-Limb (N-SOL). Also, SOL activity control testing (seedashed line in Fig. 4A), the intercept and in the D-Limb occupied a larger percentage of the cycle period than slope of the regressionlines were different. For the D-Limb, that of the N-Limb. Nonetheless, flexor (T’) activity was reciprocal SOL activity occupied about 75% of the airstep cycle, while with extensor (SOL) activity in the D-Limb. 2542 Giuliani and Smith - Stepping in Chronic Spinal Cats After Deafferentation

N-SOL

900 msec MICTURITION Figure 6. Influence of micturition on airstepping during the third deafferentation month. At the beginning of the record, the D-Limb airstepped alone at about 1 cycle/set, and the average SOL burst duration was 585 msec. During spontaneous micturition (see bar), bilateral airstepping with alternate phasing at 2 cycles/set was evoked (steps 5-9) and SOL activity was shorter (214 msec) and similar in duration to that of the N-SOL in the normal limb. As micturition terminated, D-Limb cycle period increased and the SOL burst was prolonged (steps 10 and I I). tail pinch usually elicited recruitment of the LG and VL during coactive (Fig. 7B). In records with relatively normal EMG pat- rhythmical airstepping (Giuliani and Smith, 1985). terns (such asthose in Fig. 7A), movementsof the D-Limb were seldom normal. The D-Limb flexed at the hip and knee to Influence of micturition on airstepping advance during the swing phase; then the knee and ankle ex- When spontaneousmicturition occurred during a period of air- tended vigorously, with the knee held stiffly in extension during steppingwith irregular interlimb coordination, a regular pattern stance. Contact with the belt was made on the dorsum of the of alternate stepping emerged.A sample record illustrating this digits, with the toes flexed. Also, thesestiff, staccato-like steps is shown in Figure 6. At the beginning ofthe record, the D-Limb were typical of D-Limb movements during airstepping. was airstepping without the N-Limb, and cycle periods were During the best treadmill records, when both limbs exhibited long, with prolonged SOL activity (e.g., steps1-3, Fig. 6). With someweight-bearing, myopotentials of the soleusfrom N- and the onset of micturition, the N-Limb began to airstep. During D-Limbs were markedly different. During N-Limb stance, the micturition, the hindlimbs airstepped alternately at a faster fre- EMG burst was often subdivided into packets occurring at a quency, and D-Limb extensor activity decreasedin duration frequency of lo-12 Hz; theseEMG packetswere not observed and was similar to that of the N-Limb (e.g., steps4-9, Fig. 6). in D-Limb records(Fig. 7A). In past reports from our laboratory With the cessationof micturition, however, D-Limb stepping (Bradley and Smith, 1982; Smith et al., 1982) we have noted returned to the slow frequency and extensor burst durations that theseEMG packetsare characteristic of extensor and flexor increased(e.g., steps10 and 11 in Fig. 6). Thus, in this record bursts in young cats, spinalized at 2 or 12 weeksof age, during and in many others, it appeared that spontaneousmicturition treadmill walking. The frequency of these neuromuscular os- facilitated bilateral airstepping and normalized the pattern of cillations is similar to that of clonus (Granit, 1959) or tremor intralimb synergy and interlimb coordination. (Nichols et al., 1978) of the cat’s ankle extensors.That the EMG packets were absent in D-Limb records and characteristic of Treadmill locomotion after deafferentation N-Limb recordsis consistentwith the hypothesisthat segmental Two of the 5 spinal-deafferentedcats were tested for hindlimb reflexes, particularly those evoked by input from spindle Ia af- walking at slow belt speeds(0.2-0.4 m/set) on the treadmill ferents, are responsiblefor neuromusculartremor. Not only was apparatus.Usually the dorsal surfaceof the D-Limb paw dragged the stretch on D-Limb ankle extensorsreduced becausecontact on the belt with the leg extended, while the N-Limb stepped occurred on the dorsum of the paw during stance, but Ia input with some weight support and followed changesin the belt was abolished by deafferentation, as demonstratedby the lack speed. With additional stimuli, such as pinching the tail or of EMG activity upon brisk taps of the Achilles tendon. perineum, the D-Limb stepped alternately for brief sequences of 5-10 steps(Fig. 7A), but deteriorated quickly, with periods Anatomical observations during which there was no discernible pattern of interlimb co- During the ganglionectomies,we observed that the dorsal root ordination (Fig. 7B). ganglia of lumbosacral segmentswere not located in the same The D-Limb exhibited EMG synergiestypical of treadmill position as those in normal adult cats. In general, gangliawere locomotion. SOL and VL extensors were coactive, and TA ac- located more rostrally and closer to the spinal cord, barely ex- tivity was usually reciprocal with extensor activity (Fig. 7A). tending through the intervertebral foramen. This was especially When stepping deteriorated, however, SOL and TA were often true for thoseganglia close to the level of cord transection (T12). The Journal of Neuroscience, August 1987, 7(8) 2543

500 msec

9. N-SOL t-

Figure 7. Treadmill locomotion at SOL 0.2 m/set in a spinal cat with one hindlimb deafferented. The 2 records, taken 30 set apart, were recorded during the fourth deafferentation month. A, N-Limb (N-SOL) stepped altemate- TA ly with the D-Limb (SOL). B, Interlimb coordination was disrupted, and the N-Limb completed 2 steps for one D-Limb step. The deterioration of in- I II terlimb coordination was accompanied by coactivation of flexor (TA) and ex- 500 msec tensor (SOL) activity in the D-Limb.

Also, the ventral and dorsal roots were rotated slightly about rootlets on the operated side. Thin strandsof translucent tissue their longitudinal axis, so that the ganglia were situated more attached to the cord dorsum at the level of the ganglionectomies ventrally; this made the separation between ventral and dorsal proved to be remnants of the dural sheath.Finally, ventral rami roots more difficult. for segmentsL6-S2 on the unoperated side were measuredto Reasons for these anatomical anomalies were apparent when ascertain the fixation of lumbosacral plexus (seeMaterials and the perfused spinal cords were removed. The complete transec- Methods). On the basisof the relative sizesof the rami and the tion of the dura and spinal cord at T12 caused lumbosacral criteria of Romanes (195 l), the 4 cords with the L3-Sl deaf- segments to retract and shift caudally within the vertebral canal. ferentation were classifiedas “regular” or “prefixed” cords. These The shift created sufficient tension on the roots below the tran- anatomical observations corroborated our finding of no tendon- section (particularly T13 and Ll-L4) to pull them in a rostra1 tap reflexes in the TA and SOL after L3-Sl deafferentation. direction, bringing the ganglia closer to the foramen, and in some cases into the foramen. The functional consequences of Discussion these anomalies to the sensorimotor capacity of the chronic Recovery following deaflerentation in normal and spinal cats spinal cat are unknown; however, tension and pressure applied We found that following unilateral deafferentation, recovery of to these structures over time may produce both anatomical and hindlimb reflexes in chronic spinal cats was similar to that re- physiological changes. These effects, secondary to cord transec- ported for normal adult cats (Sprong, 1929; Wiesendanger,1964; tion, have not been reported previously and are worthy of future Goldberger, 1977). In both preparations, the deafferentedlimb consideration. exhibits little or no spontaneousactivity 3-5 d after surgery. At the level of the ganglionectomies, the outward appearance This flaccidity has been attributed to the suddenloss of input of the cords was normal, except for the “atrophy” of dorsal to lumbosacral segmentsand is similar to that described for 2544 Giuliani and Smith l Stepping in Chronic Spinal Cats After Deafferentation spinal shock (Sprong, 1929; Goldberger, 1977). The deafferented spinal cats 2-l 0 weeks after a T 12 spinalization. They also noted limb, however, is not without reflexes. Within 24-48 hr after that airstepping depended on activation of bladder afferents, surgery, extensor reflexes are easy to evoke by pinching the tail since it did not occur when the bladder was empty. or contralateral paw. In normal animals with one deafferented We believe that the loss of input from brain stem locomotor limb, extensor responses predominate, and flexion reflexes are centers (reviewed by Grillner, 198 l), exacerbated by the loss of difficult, if not impossible, to evoke (Sprong, 1929; Wiesendan- input from hindlimb deafferentation, resulted in a general ger, 1964). The lack of flexion responses is not the result of depression of spinal pattern generators and consequently a slow extensor rigidity or contractures of extensor muscles, typical of recovery and a continued need for nonspecific facilitation to intradural deafferentation with cord damage (Ranson, 1928; evoke stepping behaviors in spinal-deafferented cats. Similarly, Gelfan and Tarlov, 1958; Wiesendanger, 1964), but is due to to evoke a “fictive locomotor pattern” in the isolated-spinal the fact that in the completely deafferented limb there are few, cord preparation, pairs of severed dorsal roots or bilateral dorsal if any, adequate stimuli to evoke flexor reflexes (Sprong, 1929). columns are tonically stimulated (Grillner and Zangger, 1979). In our preparation, however, flexor reflexes were elicited early Also, in mesencephalic preparations, increasing the current in the first postoperative week by applying a mildly noxious strength to stimulate the “midbrain locomotor region” is nec- stimulus (pin prick or clip) to intact receptive cutaneous fields essary to facilitate treadmill locomotion after hindlimb deaf- ofthe L3Sl deafferented hindlimb. We agree with Sprong (1929) ferentation (Grillner and Zangger, 1984). Together, these results that the extension posture, typical of deafferented limbs, results suggest that activation of spinal pattern generators can be non- primarily from an imbalance of extensor and flexor reflex ac- specific and provided by afferents from remote sites (such as tivity. the tail) when supraspinal input is unavailable. During the first recovery month, both N- and D-Limbs of chronic spinal cats developed exaggerated responses to cuta- Importance of sensoryfeedback for intralimb synergies neous stimuli, and intact cutaneous fields of the D-Limb spread Airstepping in the deafferentedlimb was characterized by erratic distally to include areas of the shank just distal to the knee. rhythms and longer-than-average cycle periods. Although it is Others who have assessed reflex recovery after extensive, but proposed that the rhythmicity of locomotion is regulated by incomplete deafferentation in normal cats, have noted the de- spinal generators independent of sensoryfeedback, both tonic velopment of hyperactive responses to cutaneous stimuli (Sprong, and phasic input modify rhythmicity (Grillner and Rossignol, 1929; Wiesendanger, 1964; Goldberger, 1977; Hnik et al., 198 l), 1978; Grillner and Zangger, 1979; Knifii et al., 1981). The while those assessing afferent projections to the dorsal horn after results of studieson fictive locomotion in the acute spinal cat cord transection (Brenowitz and Pubols, 198 1) or spared-root suggestthat, in addition to providing the excitation necessary deafferentation, have found that receptive fields expand or re- to activate spinal generators, sensory feedback contributes to organize (Pubols and Goldberger, 1980; Sedivec et al., 1983; the control of the cycle period (Grillner and Zangger, 1979). but see also Pubols and Brenowitz, 1982). Commonly, changes Similarities between fictive locomotion in chronic spinal cats in reflex status and rearrangement of afferent projections to the and airstepping in spinal-deafferentedcats provides additional dorsal horn after partial deafferentation are attributed to the evidence for the role of sensory feedback. Baker et al. (1984), sprouting of intact dorsal root afferents, which strengthen ex- for example, reported a loss of cycle period rhythmicity, longer isting synaptic connections or create new ones (reviewed by burst durations, and coactivity of nervesinnervating ankle flexor Mendell, 1984). That the time courseof long-term reflex changes and extensor musclesduring fictive locomotion of adult cats is similar after deafferentation in normal and chronic spinal cat spinalized at 2 weeks of age. In these sameanimals, Chandler (already adapted to supraspinal denervation) is intriguing and et al. (1984) observed highly irregular patterns of activity re- suggeststhat neural mechanisms, whatever their origins, are corded from extensor motoneurons during fictive locomotion. intrinsic to lumbosacral segmentsand unrelated to changesin That these animals exhibit well-organized neuromuscular pat- supraspinalinput. terns during hindlimb walking on a treadmill (Forssberget al., In general, the recovery of stepping behaviors after unilateral 1980a, b; Smith et al., 1982) suggeststhat motion-dependent deafferentation was slowerin chronic spinal than in normal cats. feedback may be particularly important for the stability of the Goldberger (1977) and Wetzel et al. (1976) reported that normal cycle period (Chandler et al., 1984; Smith, in press). cats attempted to locomote with the deafferented limb 2-7 d Hip afferents provide important peripheral feedback for the after surgery,and that the most successfulanimals were stepping regulation of the cat step cycle (Andersson et al., 1978; Grillner with someweight support by the end of the third postoperative and Rossignol, 1978; Andersson and Grillner, 1981). For ex- week. In chronic spinal cats, by contrast, the deafferentedhind- ample, Andersson et al. (1978) reported an entrainment of the limb did not participate in airstepping until the fourth or fifth fictive locomotor pattern in responseto passivehip movement. recovery week, and then constant facilitation by cutaneousstim- In the present study, neither the hip afferents of the N-Limb ulation of the tail or perineum was needed, whereasthe con- nor those of the D-Limb entering Ll and L2 dorsal roots were tralateral hindlimb steppedwithout the addition of remote stim- sufficient to stabilize the burst duration of deafferented muscles ulation. Micturition also facilitated bilateral airstepping; or stepping frequency in the deafferented limb. however, mechanismsby which vesicosomatic reflexes evoke The resultsof our study also suggestthat there may be different leg reflexes and steppingbehaviors have been largely unstudied. controlling mechanismsfor cycle period and extensor burst du- Recent work by Jolesz et al. (1982) and Ruenzel et al. (1985) ration during stepping behaviors. Normally, the extensor burst suggeststhat bladder afferents make synaptic contact with in- duration is highly correlated with the duration of the cycle period temeurons of lumbosacral segmentsto evoke hindlimb reflexes. during locomotion (Forssberget al., 1980a; Grillner and Zang- Further, Thor et al. (1983) reported that airstepping coincided ger, 1984) and airstepping (Giuliani and Smith, 1985). In the with micturition induced by naloxone (an opiate antagonist) in spinal-deafferentedcat, however, this relationship is disrupted The Journal of Neuroscience, August 1987, 7(8) 2545 in both limbs during airstepping. Whereas the duration of dently. Similar anomalies in interlimb coordination are com- N-Limb extensor activity decreased substantially and was un- mon in the mesencephalic-deafferented cat during hindlimb related to its cycle period, the duration of D-Limb extensor walking. In these animals, cycle period durations were variable activity increased significantly and was highly correlated with and the phasing between the normal and the deafferented limb its cycle period, occupying nearly 75% of the airstep cycle (cf. was inconsistent and deteriorated easily (Grillner and Zangger, Fig. 5). The reasons for these differences are not clear and war- 1984). Taken together, these results strongly suggest that stable rant further study. It is interesting that these asymmetrical effects interlimb coordination is dependent on motion-dependent feed- were not evident when 2 of the spinal-deafferented cats walked back from both hindlimbs. on the treadmill (cf. Fig. 7) or when mesencephalic-deafferented The regulation of interlimb coordination during locomotion cats walked on a treadmill (Grillner and Zangger, 1984). It is is poorly understood. In Grillner’s model (198 l), a switch from possible that feedback from the normal limb was augmented by mutually inhibitory connections (for alternate bilateral stepping) contact forces associated with weight-bearing locomotion, and to mutually excitatory connections (for in-phase bilateral step- that motion-dependent feedback, greatly reduced during air- ping) between hindlimb generators is needed to effect changes stepping, was insufficient to regulate cycle periods and extensor in interlimb coordination. Cohen et al. (1982) in another ap- activity. Conversely, normalization of rhythmicity and extensor proach, have treated the 2 hindlimb generators as coupled os- activity may be due to the sheer quantity of afferent input, which cillators, noting that the phase relationship of 2 coupled oscil- is independent of the pattern-generating circuits, since mictu- lators is dependent on the frequency of each and on the strength rition (which increases parasympathetic afference) also facili- (or sign) of the coupling between them. But Stafford and Bam- tated the normalization of bilateral stepping (cf. Fig. 6). well (1985) assume only mutual inhibition between limb gen- Other investigators have presented evidence that separate erators, and that limb phasing is regulated by the degree of mechanisms control cycle period and burst duration during ster- inhibition; no mutual excitation is assumed. They note, how- eotypical movements. In a study of guinea pig mastication, for ever, that small changes in limb-generator excitation easily af- example, Chandler et al. (1985) showed that following the fected the stability of the phase relationships. Our data on step- administration of strychnine (a glycine antagonist), the fre- ping behaviors in spinal cats with one hindlimb deafferented do quency of cortically induced rhythmical jaw movements was not suggest one model over another; however, the future mod- unaltered, whereas the burst duration of the digastric (jaw open- eling of interlimb coordination for cat locomotion should con- er) muscle increased. Also, in our earlier study of airstepping in sider the role ofafference. Under normal conditions, movement- spinal cats (Giuliani and Smith, 1985), cycle period and burst relative feedback adds to the excitability of spinal pattern duration were affected differentially by hindlimb immobiliza- generators and influences the phasing of the contralateral limb. tion in a plaster cast; both flexor and extensor burst durations With one hindlimb deafferented, interlimb coordination is un- increased, while the cycle period was unaffected. Further evi- stable and bilateral stepping deteriorates easily unless there is dence for separate mechanisms controlling cycle period and activation of bladder afferents. The relationship between mic- burst duration is seen in the paw-shake responses of spinal cats. turition and the apparent regularization of the step cycle and Following hindlimb deafferentation, extensor burst durations interlimb coordination is of interest and requires further study. were unchanged, while duration of paw-shake cycles increased substantially (Smith et al., 1986). If extensor burst duration and References cycle period during stepping behaviors are regulated by different Andersson, O., and S. Grillner (198 1) Peripheral control of the cat’s mechanisms, future efforts to model spinal generators for lo- step cycle. Acta Physiol. Stand. 113: 89-101. comotion must attend to this detail. At present, the mechanisms Andersson, O., S. Grillner, S. Lindquist, and M. Zomlefer (1978) Pe- regulating the duration of muscle activity and cycle period are ripheral control of the spinal pattern generators for locomotion in the cat. Brain Res. 150: 623-636. not known. Baker. L. L.. S. H. Chandler. and L. J. Goldbera (1984) L-Donainduced locomotor-like activity observed in ankle flexor and extensor muscle Importance of sensory feedback for interlimb coordination nerves of chronic and acute spinal cats. Exp. Neurol. 86: 5 15-526. One of the most striking effects of unilateral hindlimb deaffer- Betts, B., J. L. Smith, V. R. Edgerton, and T. C. Collatos (1976) Tele- entation in the chronic spinal cat was the loss of stable interlimb metered EMG of fast and slow muscles of the cat. Brain Res. I 17: 529-533. coordination during airstepping. In some bouts, either the Bradley, N. S., and J. L. Smith (1982) Neuromuscular oscillations N-Limb or the D-Limb stepped alone, but in the majority of about the ankle joint in chronic spinalized cat. Sot. Neurosci. Abstr. bouts, bilateral stepping was characterized by irregular phasing, 9: 526. and often the N-Limb stepped at a faster frequency, taking 2 or Brenowitz, G. L., and L. M. Pubols (198 1) Increased receptive field 3 steps to each D-Limb step. Such irregularities in interlimb size of dorsal horn neurons following chronic spinal cord hemisections in cats. Brain Res. 216: 45-59. coordination are typical of spinal cats walking on a split-belt Brown, P. B., and H. R. Koerber (1978) Cat hindlimb tactile der- treadmill (Forssberg et al. 1980b) or combining a hindlimb paw- matomes determined with single-unit recordings. J. Neurophysiol. shake response with locomotion, while keeping pace with the 41: 260-267. Carter, M. C., and J. L. Smith (1986a) Simultaneous control of two speed of the treadmill belt (Carter and Smith, 1986b). Variable rhythmical behaviors. I. Locomotion with the paw-shake response in interlimb coordination under these testing conditions, however, normal cat. J. Neurophysiol. 56: 17 l-l 83. is adaptive and permits the spinal cat to continue hindlimb Carter, M. C., and J. L. Smith (1986b) Simultaneous control of two walking under a variety of environmental and behavioral per- rhythmical behaviors. II. Hindlimb walking with the paw-shake re- turbations. sponse in spinal cat. J. Neurophysiol. 56: 184-l 95. Chandler, S. H., L. L. Baker, and L. J. Goldberg (1984) Character- In the spinal-deafferented cat, behavior is not adaptive, and ization of synaptic potentials in the hindlimb extensor motoneurons variable phasing between homologous limbs suggests that sep- during L-DOPA-induced fictive locomotion in acute and chronic spi- arate pattern generators for each limb are working indepen- nal cats. Brain Res. 303: 9 l-100. 2546 Giuliani and Smith * Stepping in Chronic Spinal Cats After Deafferentation

Chandler, S. H., S. A. Nielsen, and L. J. Goldberg (1985) The effects Effects of training on the recovery of full-weight-bearing stepping in of a glycine antagonist (strychnine) on cortically induced rhythmical the adult spinal cat. EXD. Neurol. 92: 421-435. jaw movements in the anesthetized guinea pig. Brain Res. 325: 18 l- Mendell, L. M. (1984) Modifiability of spinal synapses. Physiol. Rev. 186. 64: 260-324. Coggeshall, R. E. (1980) Law of separation of function of the spinal Nichols, T. R., R. B. Stein, and P. Bawa (1978) Spinal reflexes as a roots. Physiol. Rev. 60: 7 16-756. basis for tremor in the aremammillarv cat. Can. J. Phvsiol. Phar- Cohen, A. H., P. J. Holmes, and R. H. Rand (1982) The nature of macol. 56: 375-383. - the coupling between segmental oscillators of the lamprey spinal gen- Perret, C., and J.-M. Cabelguen (1980) Main characteristics of the erator for locomotion: A mathematical model. J. Math. Biol. 13: 345- hindlimb locomotor cycle in the decorticate cat with special reference 369. to bifunctional muscles. Brain Res. 187: 333-352. Fleshman, J. W., A. Lev-Tov, and R. E. Burke (1984) Peripheral and Pubols, L. M., and G. L. Brenowitz (1982) Maintenance ofdorsal horn central control of flexor digitorium longus and flexor hallucis longus somatotopic organization and increased high-threshold response after motoneurons: The synaptic basis of functional diversity. Exp. Brain single-root or spare-root deafferentation in cats. J. Neurophysiol. 47: Res. 54: 133-149. 102-111. Forssberg, H., S. Grillner, and J. Halbertsma (1980a) Locomotion of Pubols, L. M., and M. E. Goldberger (1980) Recovery of function in the low spinal cat. 1: Coordination within a hindlimb. Acta Physiol. dorsal horn following partial deafferentation. J. Neurophysiol. 43: Stand. 108: 269-282. 102-l 17. Forssberg, H., S. Grillner, and J. Halbertsma (1980b) Locomotion of Ranson, S. W. (1928) The role of the dorsal roots in muscle tonus. the low spinal cat. 2: Interlimb coordination. Acta Physiol. Stand. Arch. Neurol. Psychiatry 19: 20 l-24 1. 108: 283-295. Reed, K. H. (1970) Dermatomes and skin innervation density of the Gelfan, S., and I. M. Tarlov (1958) Intemeurones and rigidity of spinal cat’s tail. Exp. Neurol. 26: 1-16. origin. J. Physiol. (Lond.) 146: 594-617. Romanes, G. J. (195 1) The motor cell columns of the lumbo-sacral Giuliani, C. A., and J. L. Smith (1983) Onset and perturbation of cord of the cat. J. Comp. Neurol. 94: 3 13-364. airstepping in the chronic spinalized cat. Sot. Neurosci. Abstr. 9: 358. Rossignol, S., H. Barbeau, and J. Provencher (1982) Locomotion in Giuliani, C. A., and J. L. Smith (1985) Development and characteristic the adult spinal cat. Sot. Neurosci. Abstr. 8: 163. of airstepping in chronic spinal cats. J. Neurosci. 5: 1276-1282. Ruenzel, P. W., F. A. Jolesz, and E. Henneman (1985) Micturition Giuliani, C. A., M. C. Carter, and J. L. Smith (1984) Return ofweiaht- induced in paraplegic cats by reflex inhibition of the external urethral supported locomotion in spinal cats. Sot. Neurosci. Abstr. 10: 6>3. sphincter. Sot. Neurosci. Abstr. I I: 1164. Goldberger, M. E. (1977) Locomotor recovery after hindlimb deaf- Sedivec, M. J., J. Ovelman-Levitt, R. Karp, and L. M. Mendell (1983) ferentation in cats. Brain Res. 123: 59-74. Increase in nociceptive input to spinocervical tract neurons following Goldberger, M. E., and M. Murray (1974) Restitution of function and chronic partial deafferentation. J. Neurosci. 3: 15 1 l-l 5 19. collateral sprouting in the cat spinal cord: The deafferented cat. J. Sherrington, C. S. ( 19 10) Flexion-reflex of the limb, crossed extension- Comp. Neurol. 158: 37-54. reflex stepping and standing. J. Physiol. (Lond.) 40: 28-l 2 1. Granit, R. (1959) Observations on clonus in the cat’s soleus muscle. Smith, J. L. (1987) Hindlimb locomotion of the spinal cat: Synergistic Ann. Fat. Med. (Montevideo) 44: 305-310. patterns, limb dynamics, and novel blends. In Neurobiology of Ver- Grillner, S. (198 1) Control of locomotion in bipeds, tetrapods and tebrate Locomotion, S. Grillner, P. S. G. Stein, H. Forssberg, and D. fish. In Handbook of Physiology Sec. I: The Nervous System, Vol. Stuart, eds., Macmillan Press, Cambridge, UK. II. Motor Control (Pt. 2) V. B. Brooks, ed., pp. 1179-l 236, Williams Smith, J. L., V. R. Edgerton, B. Betts, and T. C. Collatos (1977) EMG and Wilkins, Baltimore, MD. of slow and fast ankle extensors of cat during posture, locomotion Grillner, S., and S. Rossignol (1978) On the initiation of the swing and jumping. J. Neurophysiol. 40: 503-5 13. phase of locomotion in chronic spinal cats. Brain Res. 146: 269-277. Smith, J. L., L. A. Smith, R. F. Zemicke, and M. Hoy (1982) Loco- Grillner, S., and P. Zangger (1979) On the central generation of lo- motion in exercised and nonexercised cats cordotomized at two or comotion in the spinal cat. Exp. Brain Res. 34: 24 l-26 1. twelve weeks of age. Exp. Neurol. 76: 393-4 13. Grillner, S., and P. Zangger (1984) The effect of dorsal root transection Smith, J. L., N. S. Bradley, M. C. Carter, C. A. Giuliani, M. G. Hoy, on the efferent motor pattern in the cat’s hindlimb during locomotion. G. F. Koshland, and R. F. Zemicke (1986) Rhythmical movements Acta Physiol. Stand. 120: 393-405. of the hindlimbs in spinal cat: Considerations for a controlling net- Hnik, P., R. Vejsada, and S. Kasicki (198 1) Reflex and locomotor work. In Development and Plasticity of the Mammalian Spinal Cord, changes following unilateral deafferentation of the rat hindlimb as- M. E. Goldberger, A. Gorio, and M. Murray, eds., pp. 347-362, sessed by chronic electromyography. Neuroscience I: 195-203. Liviana Press, Padua, Italy. Jefferson, A. (1964) Aspects of the segmental innervation of the cat’s Sprong, W. L. (1929) A study of reflexes in the deafferented leg of the hind limb. J. Comp. Neural. 100: 569-596. cat and their relation to tonus. Bull. Johns Hopkins Hosp. 45: 371- Jolesz, F. A., X. Cheng-Tao, P. W. Ruenzel, and E. Henneman (1982) 395. Flexor reflex control of the external sphincter of the urethra in para- Stafford, F. S., and G. M. Bamwell (1985) Mathematical models of plegia. Science 216: 1243-1245. central pattern generators in locomotion. III. Interlimb model in the KnitIki, K. D., E. D. Schomberg, and H. Steffens (1981) Effects from cat. J. Motor Behav. 17: 60-76. fine muscle and cutaneous afferents on spinal locomotion in cats. J. Thor, K. B., J. R. Rappolo, and W. C. deGroat (1983) Naloxone- Physiol. (Lond.) 319: 543-554. induced micturition in unanesthetized paraplegic cats. J. Urol. 129: Koerber, H. R., and P. B. Brown (1982) Somatotopic organization of 202-205. hindlimb cutaneous nerve projections to cat dorsal horn. J. Neuro- Wetzel, M. C., A. E. Atwater, J. V. Wait, and D. G. Stuart (1976) physiol. 48: 48 l-489. Kinematics of locomotion by cats with a single hindlimb deafferented. Koshland, G. F., and J. L. Smith (1983) Intralimb coordination of J. Neurophysiol. 39: 667-678. the paw-shake response. Sot. Neurosci. Abstr. 9: 63. Wiesendanger, M. (1964) Rigidity produced by deafferentation. Acta Lovely, R. G., R. J. Gregor, R. R. Roy, and V. R. Edgerton (1986) Physiol. Stand. 62: 160-168.