Central nervous system plasticity after spinal cord injury in man: interlimb reflexes and the influence of cutaneous stimulation

ELSEVIER

Electroencephalography and clinical Neurophysiology 101 (1996) 304-3 ! 5

Central nervous system plasticity after spinal cord injury in man: interlimb reflexes and the influence of cutaneous stimulation B l a i r C a l a n c i e * , S t e p h a n i e L u t t o n , J a m e s G. B r o t o n The Miami Project to Cure Paralysis and Department of Neurological Surgery, University of Miami School of Medicine, Miami, FL 33136, USA

Accepted for publication: 12 March 1996

Abstract In persons who have sustained severe injuries to the cervical spinal cord, electrical stimulation of mixed peripheral nerves in a lower limb can evoke short-latency, bilateral motor responses in muscles of the distal upper limbs; such motor responses have been termed interlimb reflexes. In the present study, we investigated the role that cutaneous stimulation plays in evoking interlimb reflexes. Fifteen subjects with chronic injury (>1 year) to the cervical spinal cord were investigated. Single motor unit activity was recorded from a number of distal upper limb muscles. The lower limb cutaneous area within which stimulation recruited a given motor unit of the upper limb was defined as that motor unit's 'receptive field'. Activity from a total of 48 single motor units was analyzed. The majority of motor units responded to light touch, individual hair movement, and thermal (hot and cold) stimulation. Excitatory responses were observed bilaterally, although contralateral responses predominated. Stimulation occasionally resulted in inhibition of a spontaneously active motor unit. Receptive fields varied a great deal in size, with proximal locations being larger than those encountered in more distal lower limb locations (i.e. the toes). The spinocervical tract is a possible candidate for mediating some portion of these interlimb reflexes, the origin of which may be due to new growth (regenerative sprouting) in the spinal cord caudal to a severe injury. Keywords: Spinal cord injury; Plasticity; Reflex; Human; Regeneration

1. Introduction In a previous study, we described a novel reflex in subjects who had sustained an injury to the spinal cord at or caudal to the sixth cervical root level and who had not recovered significant motor function below this level (i.e. they were motor-complete) (Calancie, 1991). Using the 'Frankel' or ' A S I A ' nomenclatures, these subjects would all be categorized as either ' A ' or ' B ' , indicating an absence of voluntary motor function caudal to the injury (Frankel et al., 1969; American Spinal Injury Association, 1992). In such persons, transcutaneous electrical stimulation of the tibial and/or posterior tibial nerves in the lower extremity evoked short-latency motor responses (interlimb reflexes) in muscle groups in the distal upper limbs, particularly the intrinsic hand muscles. Conversely, subjects with injuries leaving them motor-incomplete in their * Corresponding author. The Miami Project Research Laboratories, R-48, 1600 NW 10th Avenue. Tel.: +1 305 5857970; fax: +1 305 5458347.

lower limbs (Frankel or A S I A ' C ' , ' D ' or ' E ' ) did not demonstrate such interlimb reflexes, nor did able-bodied subjects (Calancie, 1991). The presence of such reflexes in persons who fail to demonstrate significant recovery o f motor function below the level of injury may be due to either: (1) pre-existing neural circuitry which is ineffectual (subthreshold) under ordinary circumstances but which becomes functional after neural injury at a remote site (i.e. latent synapses) (Guth, 1976; Goshgarian and Guth, 1977; Wall, 1988; Yu and Goshgarian, 1993; Wall and McMahon, 1994); or (2) establishment o f new synaptic connections between lower extremity afferent fibers and a - m o t o n e u r o n s innervating distal upper limb muscles, all of which occurs at neurological levels caudal to the lesion. The latter possibility would represent an example of regenerative sprouting in the human central nervous system (CNS) (Steward, 1989). In the present study we describe the role of cutaneous input in the mediation of these interlimb reflexes. In a subsequent paper we will present the time course of de-

0924-980X/96/$15.00 © 1996 Elsevier Science Ireland Ltd. All rights reserved PII $0921-884X(96)951 94-6

EEM 95194

B. Calancie et al. /Electroencephalography and clinical Neurophysiology 101 (1996) 304-315

305

2. Methods and materials

Computerscope) and simultaneously processed through two 6-channel audio mixers (Rane RM 26) connected to a loudspeaker for audio feedback. Typically, 12 channels of EMG activity were recorded, at least 8 of which were from upper limb muscles, including different combinations of the wrist flexors (flexor carpi radialis; FCR), wrist extensors (extensor carpi radialis; ECR), flexor pollicis longus (FPL), extensor pollicis longus (EPL), abductor digiti minimi (ADM), abductor pollicis brevis (APB, including adductor pollicis), and different interosseous (INT) muscles, bilaterally. We also recorded EMG from soleus and abductor hallucis (AbH; an intrinsic foot muscle) on the side of the lower extremity being stimulated, for which the presence or absence of the M-wave or H-reflex served as a means of localizing the tibial or posterior tibial nerves for stimulation.

2.1. Subjects

2.3. Stimulation

Responses in upper limb muscles to stimulation of lower extremity cutaneous receptive fields (i.e. areas within which cutaneous stimulation could evoke responses from an identified motor unit in an upper limb muscle) were obtained from 15 subjects with motorcomplete injury to the cervical spinal cord. All subjects had sustained their spinal cord injury more than 1 year prior to examination (i.e. their injury status was considered to be chronic for the sake of this study). Subjects included 11 men and 4 women, whose ages ranged between 21 and 46 years. All subjects gave informed consent for their participation in this protocol, which was approved by the Institutional Review Board of this University.

All stimuli were delivered to the lower limbs at or distal to the knee. Stimuli included: (1) square-wave, constant voltage electrical shocks delivered transcutaneously via surface electrodes or subcutaneously via non-insulated needle electrodes to either mixed nerves (tibial or posterior tibial) or to patches of skin within identified cutaneous receptive fields; (2) light stroking of the skin with a blunt probe; (3) controlled cutaneous indentation via Semmes-Weinstein monofilaments (equivalent to 'Von Frey hairs'); (4) application of cold via either a controlled thermal device (Peltier device; Thermal Devices Inc.) or with ice, and radiant heat from a small lamp focused to a 5 mm spot; (5) pulling and bending of single hairs on the toe, foot or leg using a needle micro-holder; and (6) vibration of cutaneous receptive fields with varying frequency and displacement, using a Ling 203 linear motor. Electrical stimuli were delivered from a Grass $88 stimulator via an SIU5 stimulus isolation unit. Stimulus rates, intensities and durations were varied and are reported in Section 3. The influence of temporal summation was investigated by comparing the response to single shocks (typically 0.2-0.5 Hz) to those of brief trains of shocks at high frequency (2 or 3 pulses at 500 Hz). Stimulus rates of between 0.1 and 1 Hz were used to investigate the phenomenon of 'wind-up' (increasing neuronal discharge intensity to repeated stimuli at low frequencies) (Mendell, 1966; Xu et al., 1995) of the upper limb motor response.

velopment of these reflexes post-injury. Our evidence to date leads us to conclude that these reflexes are due to the establishment of new synaptic connections made by sensory afferents of lower extremity origin onto cervical motoneurons, caudal to the level of a severe cervical spinal cord injury. While of no apparent functional benefit to the individual, establishment of such connections does serve as an example of plasticity and the capacity for new growth within the adult human spinal cord, and helps account for the 'spread' of involuntary movements from lower to upper limb muscles during a typical spasm in such a subject. Preliminary findings have been presented in abstract form (Calancie and Broton, 1992; Calancie, 1995).

2.2. EMG recording

Muscle electromyogram (EMG) was recorded with pairs of gross EMG electrodes. These were either surface Ag/AgCI electrodes (1 cm diameter; 5-7 cm spacing) or non-insulated EEG-type subdermal wire electrodes (1 cm length). The recording characteristics (impedance, muscle volume sampled, frequency response) of the wire pairs did not differ appreciably from the surface electrodes used (Hugon, 1973). Because interlimb reflexes often involved recruitment of only one or two single motor units in a particular upper limb muscle, in many cases it was possible to clearly distinguish the action potential of that motor unit with such electrodes. However, in some cases for which multiple motor units were recruited by sensory stimulation to a lower limb, a concentric bipolar needle electrode was also used, in order to achieve more selective single motor unit recordings. EMG signals were preamplified close to the source, filtered and amplified (100Hz-5 kHz; 1K or 10K gain), and stored on VHS tape after being digitized (Vetter 4000AS). EMG signals were displayed on a computer monitor (RC Electronics

2.4. Procedure

Subjects were given single-channel auditory feedback of EMG and asked to attempt voluntary, isolated contractions of each muscle in turn to confirm whether or not it was under some degree of volitional control. Electrical stimulus pulses were then applied to the tibial nerve at the

306

B. Calancie et al. / Electroencephalograph)' and clinical Neurophysiology 101 (1996) 304-315

popliteal fossa, using single pulses alone and a series of 2 and 3 pulse stimulus trains, using a 2 ms interpulse interval (i.e. 500 Hz 'doublets' or 'triplets'). This highfrequency stimulus pattern had previously been found to be particularly effective for evoking upper limb motor responses in appropriate individuals (Calancie, 1991). This process was repeated for stimuli delivered to the ipsilateral posterior tibial nerve at the ankle, to characterize the electrical response properties of upper limb motor units to receptive fields localized in the foot. Careful visual examination of the distal upper limbs was made by one of the investigators during delivery of stimuli to the tibiai and posterior tibial nerves, in order to look for time-locked contractions in response to nerve stimulation. We also displayed on the computer screen EMG from all upper limb muscles being recorded from, and listened to the auditory output of these channels. This mechanical, visual and auditory feedback was used to confirm or rule out the presence of muscle contractions evoked by lower limb nerve stimulation. When recruitment in an upper limb muscle was evident following electrical stimulation (i.e. an interlimb reflex was present), detailed sensory exploration of the lower limbs was carried out. This involved a rapid, systematic cutaneous survey by lightly stroking a blunt probe along the skin within and across different dermatomes distal to the knee, and flexing and extending each toe, and the ankle, through the full range of motion of each joint, followed by slight over-pressure at the end of range in each direction. When upper limb motor responses to skin stroking were observed, the threshold for evoking responses for approximately 50% of the applications was determined with monofilaments at the most sensitive (i.e. lowest threshold) region. A slightly thicker monofilament was then used to 'map out' the skin area within which indentation caused upper limb motor unit recruitment. This region is referred to as the receptive field for that upper limb motor unit. The reflex responses to other stimulus modalities (hair movement, thermal, and cutaneous electrical stimulation) were then tested. For studying the effect of hair movement, a microsurgical needle-driver was used to grasp individual hairs and apply slight traction, both along and against the hair's natural orientation. For immediate but relatively 'gross' skin cooling, small pieces of ice or (on one occasion) a chilled spoon were applied briefly to the receptive field. More controlled thermal stimuli were delivered via a Peltier device with a surface contact area of approximately 1 cm 2. The working temperature range of this device was between 0.2 and 55°C, and its maximum rate of temperature change was approximately 10°C/s. Analog signals from the Peltier device were recorded on tape, and provided reliable indications of the temperature being applied to the motor unit's receptive field. In those cases where no reliable signal was available to indicate stimulus delivery (e.g. hair pull/bend; heat/cold application; skin

stroke with a probe), a manual switch producing a 5 V transient was closed by the investigator to coincide with delivery of that particular stimulus. This signal was useful for providing a better approximation of the time of delivery of the stimulus than was possible from the voice record alone. Electric shocks were applied at different sites within a given motor unit's receptive field to obtain estimates of the minimum response latency to stimulation. Stimulating electrodes were either surface disks mounted on a Plexiglas bar (anode-cathode separation 2.0-2.5 cm on center) or EEG-type needle electrodes placed subdermally and separated by approximately 3 cm. In either case, both anode and cathode were positioned within the receptive field with a rostral-caudal orientation along the long axis of the leg or foot (cathode proximal).

2.5. Analysis All data were analyzed off-line from the taped records. When discrete trigger pulses were available (e.g. electrical stimulation), post-stimulus time histograms (PSTHs) of motor unit discharge were constructed using a 0.1 ms binwidth. Such a brief binwidth allowed definition of the time-course of the motor unit's post-stimulus excitability change with high resolution. Discrimination of motor unit waveforms was done using a 'template-match' method (RC Electronics). The outline of a given motor unit's receptive field was traced with ink onto the subject's lower limb during the experiment. This outline was sometimes transferred at that time to a picture of the foot and lower limb showing dorsal and ventral surfaces, or (at a later time) an outline was produced from the tape-recorded description of the receptive field. 3. Results

3.1. Receptive field size There were 39 cutaneous receptive fields identified in the 15 subjects examined. The smallest receptive field included only the dorsal surface of the great toe, while the largest area within which stimulation led to upper limb single motor unit recruitment included all of the dorsal/ medial leg from the distal thigh to the ankle, the dorsum of the foot, and the dorsal and plantar surfaces of all 5 toes. Distal receptive fields including the toes were nearly always much smaller than those encountered more proximally. Many of the receptive fields included surface areas associated with more than one dermatome and more than one cutaneous nerve. Receptive fields were encountered for dermatomes associated with the L3 through $2 neurological segments. We did not examine cutaneous areas associated with dermatomes rostral to L3, or caudal to $2. There were no examples of discontinuous receptive fields (i.e. separate and distinct lower limb regions within which

B. Calancie et al. / Electroencephalography and clinical Neurophysiology 101 (1996) 304-315

cutaneous stimulation caused recruitment of the same upper limb single motor unit). 3.2. Distribution of upper limb responses

In response to stimulation of these receptive fields, a total of 48 single motor unit potentials in various distal upper limb muscles were activated which could be reliably differentiated from other motor unit potentials. Table 1 summarizes, for both ipsi- and contralateral muscles, the number of motor units recorded for receptive fields of different locations. There was a higher probability for recruiting motor units within intrinsic hand muscles (APB and AP, ADM, INT) with appropriate lower limb stimulation compared to muscles within the forearm (ECR, FCR, EPL, FPL). The probability of evoking motor unit discharge in a muscle contralateral to the side being stimulated was typically higher than for ipsilateral evoked responses. The resultant upper limb movements were entirely without pattern or functional consequence, often being limited to brief and unsustained tremulous contractions of the fingers and/or thumb. 3.3. 'Typical' upper limb motor unit response pattern to lower limb stimulation

Within a given receptive field, we found that in most Table 1 S u m m a r y o f the n u m b e r o f m o t o r units r e c o r d e d in a particular muscle, the a p p r o x i m a t e location a n d size o f that m o t o r u n i t ' s receptive field, a n d w h e t h e r the response is ipsilateral to the stimulus (ipsi) or contralateral to the stimulus (contra) Muscle

Receptive field location Toes only

E C R (ipsi) E C R (contra) F C R (ipsi) F C R (contra) E P L (ipsi) E P L (contra) A P B (ipsi) A P B (contra) F P L (ipsi) F P L (contra) A D M (ipsi) A D M (contra) lnteross (ipsi) Inteross (contra)

1 2 1 3 . . 1 1 . 1 2 I -

Heel only

1 . 2 . .

.

S h a n k __. others

1

1

1

-

-

1 3

1 4 1

1 1 1 -

1 3 2

. .

. .

.

1 3 . 1 1

Knee only

. .

.

Foot -+ toes

. 1 2 -

.

Muscles h a v e been g r o u p e d f r o m top to b o t t o m in o r d e r o f their moton e u r o n s ' n e u r o l o g i c a l level o f origin within the spinal cord (e.g. A D M m o t o n e u r o n s lie c a u d a l to those for the ECR). Receptive fields h a v e been g r o u p e d f r o m left to r i g h t in o r d e r o f i n c r e a s i n g size (surface area). ' O t h e r s ' includes the knee, foot a n d toes. O n l y 47 o f the identified 48 m o t o r units are s u m m a r i z e d , as the receptive field o f the unit not i n c l u d e d c o u l d not be a c c u r a t e l y localized f r o m the voice record.

307

cases a wide variety of inputs could cause discharge of an identified upper limb motor unit. An example is shown in Fig. 1, which depicts the discharge properties of a motor unit from the left ECR muscle in response to a variety of inputs to the right leg. As shown in a series of poststimulus time histograms (Fig. 1A) summarizing the first 120 ms of post-stimulus discharge, single-shock electrical stimulation of the right tibial nerve at the popliteal fossa resulted in delayed discharge of the motor unit (>80 ms latency), whereas two pulse stimulation (2 ms interpulse interval) at an identical intensity (120 V) resulted in early (-40 ms) and late response of the same motor unit. A similar effect of temporal summation on motor unit response latency was reported in the original description of interlimb reflexes after SCI (Calancie, 1991). For comparison, the H-reflex in the right soleus muscle to the identical tibial nerve stimulus had an onset latency of 31 ms in this subject (not shown). Comparisons between upper and lower extremity reflex responses to electrical stimuli are provided in greater detail in the original report of interlimb reflexes (Calancie, 1991). Below these histograms is shown the motor unit's receptive field (Fig. 1B), the border of which was welldefined by a line separating the S1 and $2 dermatomes (ventrally) and the tibia (dorsally). Surface electrical stimulation within this receptive field at the point marked ' X ' in Fig. 1B (i.e. no longer applying stimulation directly to the whole tibial nerve) resulted in a shorter response latency (Fig. 1C) to a single-shock stimulus intensity which was insufficient to evoke short-latency responses when stimulating the tibial nerve itself. It is likely that the afferent path mediating the reflex at the point marked 'X' in Fig. 1B was the anterior femoral cutaneous nerve, rather than the tibial nerve which was stimulated for the histograms depicted in Fig. 1A. This may account in part for the difference in the minimum response latency seen. Note that an increase in stimulus intensity from 80 to 100 V resulted in still-earlier discharge of this motor unit (Fig. 1C, bottom histogram). An EMG trace in Fig. 1D shows the response of the ECR motor unit to repeated application (n = 4 trials) of a threshold-level monofilament (2.35 g) to the receptive field, causing discharge in the first and third applications. Fig. IE illustrates the rate of ECR motor unit discharge (averaged every 400 ms) within a burst of action potentials in response to the application of various inputs (indicated by horizontal bars). In this representation, 'taller' bins correspond to more rapid discharge rates, while the width of the contiguous bins indicates the duration of the burst. For the hair bend, the motor unit's initial burst occurred while gripping the hair within the jaws of the needle-driver, whereas the more prolonged and intense discharge was due to the actual bending of the hair against its direction of growth. Touching a piece of ice to the receptive field caused immediate motor unit discharge that persisted after ice removal, arguing against a purely

B. Calancie et al. /Electroencephalography and clinical Neurophysiology 101 (1996) 304-315

308

A. Electrical stimulation of R. tibial nerve

D. Four skin indentations with 2.35 g monofilament

'081 single, 120V

/ ,.I I ....

~'OSI2@5O0Hz' 8OV B.

o"

--r.

IIJ

II

L JJ

•

I I,"

L ECR EMG •

I . I, lllLjir : t..

J L . . . . ,T

-*

II

.087 2@500Hz, 100V

i 0t 00

.L ,,, 40

2s

80

E. Other somatosensory stimuli

120

Time (ms) Stroke skin six times

B. Cutaneous receptive field

.iLL.

.-=

Bending of single hair "~" 25 1 Z,~

C. Cutaneous electrical stimulation within receptive field

,~ o ~

~>" .08q single, 80V

It

Ill ,

25

l"

.IO

eI~.

I~'

,

0/ .08-] single, 100V

i

j

~5

.i

,

o 251 Radiantheat

(D 0

'

0

4()

80

Time (ms)

i,

120

bin width = 0.4 ms

0 5s

bin width = 400 ms

Fig. 1. Responses of a single motor unit in the left ECR to different forms of cutaneous and electrical stimulation, and the area of the contralateral lower limb within which this stimulation recruits the motor unit (i.e. the motor unit's 'receptive field'). (A) Post-stimulus time histograms summarizing the first 120 ms of motor unit discharge in response to a number of electrical stimuli delivered to the right tibial nerve at the popliteal fossa with surface electrodes. Stimulation in these and subsequent histograms is applied at time 0; binwidth 0.5 ms. Stimulus intensity and pattern (single shocks or 'doublet' shocks) as shown. All stimuli am square-wave, 1 ms constant voltage pulses. Response probability represents the incidence of motor unit discharge for a given time period (bin) per stimulus applied. Thus, a unit responding to the stimulus at the exact same time for every stimulus would produce a bin 'height' of 1.0. This convention for presenting response probability is used for all subsequent post-stimulus time histograms. (B) Cutaneous receptive field for this motor unit was along the medial half of the leg, bordered by the tibial ridge (dorsal) and the midline between medial and lateral gastroenemii muscles (ventral). This receptive field is included under the heading 'Shank' in Table 1. (C) Post-stimulus time histograms following surface electrical stimulation at the 'X' indicated in (B). Stimulus intensity as shown; duration 1 ms. (D) EMG record of motor unit response to a series of four applications of a threshold-level monofilament to the receptive field. Rectangular pulses indicate the time of press for each trial, as produced by the experimenter using a manual trigger. The ECR motor unit discharged with a single action potential in response to the first and third application. (E) Summary of instantaneous discharge rate of this motor unit in response to: (1) repeated stroking of the skin; (2) pulling a single hair backwards; (3) application of ice; and (4) shining a focused spot of light to the receptive field to cause heating. The motor unit discharged a series of action potentials to each of these stimuli. The horizontal bar in the second, third and fourth records indicates the time of application of the respective stimuli. Horizontal time calibrations in (D,E) as shown. mechanical effect of the ice contact. The bottom record of

a prolonged delay, presumably reflecting the time neces-

F i g . 1E s h o w s t h a t t h e m o t o r u n i t d i s c h a r g e d in r e s p o n s e to r a d i a n t h e a t a p p l i e d to t h e r e c e p t i v e f i e l d , b u t o n l y a f t e r

sary f o r t h e s k i n to w a r m to its m a x i m u m with this stimulus (approximately 48°C).

temperature

309

B. Calancie et al. I Electroencephalography and clinical Neurophysiology 101 (1996) 304-315

l lJIILl

L ADM

units in the upper limb which were recruited by the same stimulation did not demonstrate such habituation.

Ltlllll l .......... lJ,

LIIII11111llll IIIllIl llll"

i,,i, 2s

R AbH

,

,

On

Fig. 2. Records of EMG from the left ADM (top) and right AbH (middle) in response to application of ice to the ventral (plantar) surface of the fourth, third and second toe of the right foot, respectively. The period of ice application at each site is indicated by the rectangular pulses (bottom), produced by the experimenter using a manual trigger.

Monofilaments were used to indent the skin in a semi-quantitative manner to estimate the sensitivity of cutaneous receptors to punctate touch. The threshold intensities to evoke upper limb responses with monofilament application ranged from a low of 0.22 g to a high of 20.9 g. The mean (_+SD) of these values was 6.35 _+ 6.24 g, with a median of 4.64 g. Discharge was also typically observed in response to hair bend for those receptive fields which included hair follicles (i.e. non-glabrous skin). However, the two conditions which led to the most intense activation of UE motor unit discharge were: (1) placement of needle electrodes under the skin for electrical stimulation; and (2) the application of cold stimuli to the identified lower limb receptive fields. In both cases, the response of an upper limb motor unit recruited by nociceptive or thermal stimulation tended to be slowly adapting (Westling and Johansson, 1987), as shown in Fig. IE. There were no ipsi- or contralateral lower limb segmental reflexes resulting from any of the cutaneous inputs to the subject whose data are shown in Fig. 1. However, in response to stimulation of receptive fields involving the feet, and especially the toes, we commonly observed brief contractions in muscles controlling toe movements. Such an example is shown in Fig. 2, which depicts activity in the right A b H following 3 episodes of ice application to the plantar surface of the fourth, third and second toes of the ipsilateral foot. Ice application to the fourth and second toes, but not the third toe, resulted in local (segmental) reflex activity in the A b H muscle. In contrast, the A D M muscle in the contralateral hand responded to application of ice to the fourth toe and third toe, the latter response occurring in the absence o f a reflex response in the A b H muscle, showing an independence of the two motor behaviors. Although not shown, it was commonly observed that the local (i.e. AbH) reflex response to any given stimulus became less pronounced or fell silent with repeated presentation of that stimulus, whereas motor

3.4. Discharge rates o f upper limb motor units to lower limb stimulation

It was common to observe sustained discharge rates o f upper limb motor units in excess of 20 Hz in response to cold stimulation to receptive fields in the lower limb, but much higher rates were sometimes observed. Fig. 3 shows records from a single motor unit in the right adductor pollicis muscle of a C5 quadriplegic subject in response to cold stimulation of a cutaneous receptive field in her contralateral lower limb. The threshold for monofilament stimulation was only 0.22 g for this unit. The convex surface of a tablespoon which had been chilled in ice for approximately 5 min was gently applied to the unit's receptive field for a 5 s period. This caused an immediate, rapid burst of action potentials from this motor unit followed by a somewhat slower discharge pattern that persisted for approximately 50 s. During the period of cold application, the mean discharge rate exceeded 70 Hz, while the instantaneous discharge rate of this motor unit during the first 40 ms following application of the spoon was 220 Hz (Fig. 3 inset, showing an expanded time scale at the exact moment of discharge onset after touching the cold spoon to the receptive field). Such intense and prolonged discharge is not typically associated with human spinal motoneurons in response to either afferent or supraspinal (i.e. volitional) (Bellemare et al., 1983) excitatory input.

OOot ,ce 0/ 0

10

20

30

40

50

Time (s)

Fig. 3. (Top) EMG (needle electrode) from AP of the right hand in response to application of a chilled spoon to a receptive field on the subject's left lower limb. The spoon was gently touched to the receptive field for approximately 5 s, as indicated by the horizontal bar below the initial burst of action potentials. (Inset) Horizontal expansion of the same spike train to illustrate the initial 200 ms of action potential discharge following application of the spoon to the receptive field. The horizontal bar represents 20 ms in this inset trace. (Bottom) Histogram of the motor unit's instantaneous discharge rate (impulses/s), grouped into 1 s bins. Thus, the mean rate during the initial 1 s period of discharge (~90 Hz) is considerably less than the instantaneous rate of -200 Hz during the first 60 ms of discharge, as shown in the inset.

310

B. Calancie et aL / Electroencephalography and clinical Neurophysiology 101 (1996) 304-315

3.5. Wind-up o f upper limb response to lower limb electrical stimulation

For trials in which stimuli were delivered at discrete time intervals (e.g. electric stimulation), the rate of stimulus delivery was typically between 0.2 and 0.5 Hz. When first stimulating, or when resuming stimulation after a long pause, such modest stimulus rates sometimes resuited in a progressively more vigorous upper limb evoked motor response with successive stimuli, both in terms of the total number of motor unit potentials which discharged to a given stimulus, and their rate of discharge. Such behavior is similar in many respects to that of wind-up (Mendell, 1966). By way of example, Fig. 4 illustrates left EPL motor unit responses to electrical stimuli delivered to a contralateral receptive field, whose distribution was similar to that depicted in Fig. 1 (different subject). A nerve stimulator bar (7 mm salinesoaked pads separated by 2.4 cm on center; cathode proximal) was used to deliver the stimulus pulses at a rate of between 0.2 and 0.3 Hz. Stimulus parameters were fixed at 150 V and 1 ms duration square waves in all cases. For site 1, the stimulation bar was oriented along the long axis of the leg, with the anode on the distal border of the receptive field. Each stimulation resulted in two discharges during the initial trials (1-3), increasing to 3 or 4 spikes during the 800 ms period after the stimulus in trials 4 and 5, respectively (Fig. 4, top). Site 2 stimulation involved movement of the bar electrode by 2 cm in a proximal direction within the receptive field. At this site, the identical stimulus intensity led to a greater number of EPL motor unit discharges, with a distinct shift in the latency of the earliest response from a value of greater than 100 ms in most responses to one of approximately 95 ms. Additional proximal movement of the stimulating bar by 2 cm (site 3) resulted in a vigorous discharge pattern to identical stimuli, possibly because a larger population of afferents contributing to this 'wind-up' was being stimulated at the more proximal electrode positions. Note that for traces shown in the middle and bottom records of Fig. 4, the shape of the spike potential was changing in a gradual and reproducible manner. This is because the tetanic contraction which resulted from motor unit activation caused the recording surface of the concentric bipolar needle electrode to shift with respect to the muscle fibers being recorded from.

rate-following we observed in this study, Fig. 5 illustrates records from the same motor unit as was described above for Fig. 3, although now a concentric needle electrode was used to record the spike train. A brief series of 2 or 3 square-wave electrical stimuli of differing frequency (50/~s duration; 70 V for 10-50 Hz, 75 V for 60 Hz) was delivered to the skin within the receptive field described for this contralateral adductor pollicis motor unit. The stimulus rate (left column) was varied between 10 Hz and 60 Hz. Arrows on each trace point to the stimulus artifacts associated with delivery of the second (and sometimes

Site 1 b-

2

~t

2 2

I~ ~.___~

3 4

Site 2

r,

Site 3

,

~.~19

--4. ~ ~

2 ~

o 11200 pV 100 ms

3.6. Synaptic strength between lower limb afferents and upper limb motor units

To estimate the strength of synapses formed between lower limb afferent fibers and motoneurons of the cervical enlargement, a motor unit's ability to respond in a one-to-one manner to high rates of stimulation (i.e. 'ratefollowing') was examined, as was the width of its poststimulus time histogram. As an example of the most rapid

Fig. 4. Spike train from a motor unit in the left EPL following surface electrical stimulation of a receptive field in the right lower limb. Site 1, most distal within the receptive field; sites 2 and 3, 2 cm more proximal for each site. Stimulus intensity (150 V, 1 ms duration) is constant in each case. Stimulus rate 0.2-0.3 Hz. Time between onset of stimulation at each site approximately 5 s. Numbers to the right of each trace indicate the total number of spikes which discharged in response to the stimulus during the 800 ms period illustrated. The shape of the spike potential changes in a reproducible manner for those contractions which lead to a complete tetanus.

B. Calancie et al. /Electroencephalography and clinical Neurophysiology 101 (1996) 304-315

10 H z 20 Hz 30 Hz 40 Hz 50 Hz 50 Hz 60 Hz 0

50

100

150

200

Time (ms) Fig. 5. Spike train of the same motor unit as shown in Fig. 3, in this case recorded with a concentric needle electrode. The unit's receptive field is electrically stimulated for brief trains (2 o r 3 shocks) at rates of 10-60 Hz (stimulus rate shown to left of each trace). The initial stimulus for each series is aligned at time 0, and subsequent stimuli are indicated by arrows. The motor unit is able to follow most stimuli in a reliable, phase-locked manner, and often fires a burst in response to a single stimulus. The asterisks for the two trials at 50 Hz indicate the time at which the motor unit discharge was expected, but did not occur.

third) stimulus. Approximately 5 s elapsed between successive stimulus trials, and two trials at 50 Hz were included in this series. The motor unit often responded to a single stimulus with a burst of action potentials, as seen for the response to the second stimulus at 10 Hz (Fig. 5, top trace). Whether or not the motor unit responded with a burst, examination of the first spike within each record of Fig. 5 indicates that the motor unit showed reliable rate-following up to 50 Hz. At this higher rate, there were two single 'misses' (indicated by asterisks), but again at 60 Hz the motor unit followed the 3 stimulus pulses applied. In this latter case, there was also an after-discharge of the motor unit approximately 150 ms following the initial stimulus (Fig. 5, bottom record). Two other motor units were found to have reliable 1:1 rate-following at 30 Hz, and 5 of the 11 motor units tested in this manner were able to follow stimulus rates of 10-15 Hz in a reliable, 1: 1 manner. Fig. 6 shows post-stimulus time histograms from a different subject, summarizing the discharge of two simultaneously recorded motor units from the contralateral ECR in response to electrical stimulation of the receptive field (medial to the knee) of these motor units. Surface stimulation was applied with a bar electrode as described above (140 V, 1 ms duration). Each of the two motor unit responses is shown in both a compressed time scale (large format) and expanded time scale (inset). In the upper records, what appears to be a single, 4 ms wide histogram when plotted with a resolution of 1 ms is actually made up of two distinct peaks whose durations are each less than 1 ms, separated by a 0.8 ms period during which the

311

motor unit did not respond to stimulation (inset, plotted with 0.1 ms resolution). There were numerous instances of such narrow histograms, most of which occurred at relatively short latencies after the stimulus, similar to the upper records of Fig. 6. Of the 20 motor units for which response histograms were generated to electrical stimulation of the receptive field, 10 had histogram durations of 3.5 ms or less. Because histograms were obtained from motor units which were not rhythmically active, there was no difficulty in defining the histogram's onset and offset. The bottom records of Fig. 6 illustrate an example of one of the widest histograms obtained in these studies following stimulation of a cutaneous receptive field. This histogram was 12.5 ms wide, approaching the maximum observed width of 13.5 ms. Given these few wide histograms, it is not surprising that the mean PSTH width of 4.9 ms (_+3.1 ms) was well above the median value of 3.5 ms as mentioned above. In the inset of the lower records of Fig. 6, it is clear that the response probability for any one bin at this 0.1 ms resolution is considerably lower than that for the other FCR motor unit recruited by this stimulus (Fig. 6, top).

3. 7. Inhibition of upper limb motor unit discharge by lower limb stimulation Many of the subjects examined in this series demonstrated relative quiescence in their distal upper limb muscles, but in some cases we observed muscles in which one or more single motor units was discharging in a continu.5- A.

.25 e~ 32

o n (/) tO

0

37 Time (ms)

I,

40

80

120

80

120

.5 B.

Q_

(D

r~

.26-

....... ,li 82

0

0

.... ,,,..,. 87 Time (ms)

92

40 Time (ms)

Fig. 6. Post-stimulus time histograms of the response of two m o t o r units (A,B) from the same ECR muscle recruited by surface electrical stimulation of a receptive field on the medial side of the contralateral knee. As these data were recorded simultaneously, all stimulus parameters are identical. For each motor unit, the response probability is plotted using a 1 ms binwidth, and then replotted on an expanded timescale with a 0.1 ms binwidth (inset for each of (A,B)).

B. Calancie et al. / Electroencephalography and clinical Neurophysiology 101 (1996) 304-315

312

L

4. Discussion APB

~31L,...,.a,. R ADM 30"

,r.......

0

l,,J,, .,,, ......... jilt,,

J~,,L

,,,

s21 0

30

60

90

120

150

Time (s) Fig. 7. EMG records and histograms of instantaneous discharge rate for two motor units recorded from the left APB and right ADM in response to thermal stimulation of a receptive field at the base of the third toe on the left foot. Thermal stimulation was applied via a Peltier device, whose surface temperature is depicted at the bottom of this figure. The left APB unit was rhythmically discharging throughout the experiment, except for those trials in which extremes of temperature resulted in either reduction or complete silence of discharge in this motor unit. Conversely, the identical stimulus resulted in the more commonly observed behavior shown by the right ADM motor unit, which increased its discharge modestly for cooling, but more vigorously in response to prolonged heating.

ous, rhythmic manner. Under these circumstances, we occasionally found that stimulation of cutaneous receptive fields in the lower limb was inhibitory to one of these tonically active units, while it was excitatory to other, non-tonic motor units. By way of example, a Peltier device was used to deliver controlled thermal stimulation (Fig. 7, bottom) to a small receptive field on the base of the third toe on the left foot of a quadriplegic subject, recruiting a motor unit in the contralateral A D M (Fig. 7, middle). This motor unit discharged upon extremes of both cooling and heating. Meanwhile, throughout the time period shown in Fig. 7 and in fact for prolonged periods throughout the entire experiment, there was a welldefined motor unit in the ipsilateral APB muscle which discharged continuously at a rate of approximately 12 Hz. However, we repeatedly observed that for this receptive field, skin cooling or prolonged exposure to the hottest temperature tested (55°C) led to a reduction in rate or complete silence of the ipsilateral APB motor unit's discharge (Fig. 7, top). Suppression of ongoing motor unit discharge in distal upper limb muscles following stimulation of lower limb receptive fields was observed in 3 subjects. However, it is possible that this phenomenon (apparent inhibition of cervical motoneurons by lower limb cutaneous stimulation) was more common but was not evident due to the infrequent instances in which we observed continuous, rhythmic discharge in distal upper limb motor units, a condition necessary to demonstrate the inhibitory action of lower limb stimulation using the methods described herein.

The original report of interlimb reflexes after spinal cord injury utilized electrical stimulation of mixed lower limb nerves to evoke upper limb motor responses (Calancie, 1991). In that paper it was concluded that interlimb reflexes were mediated by proprioceptive afferents originating from the lower limb, but were not influenced by lower limb cutaneous input. A more thorough examination of the influence of cutaneous stimulation, the results of which are detailed in the present report, leads us to revise our original conclusion. Instead, we have now confirmed that cutaneous stimulation, sometimes of remarkably low intensity (e.g. bending of a single hair), can cause recruitment of motor units in distal upper limb muscles of persons with cervical spinal cord injury. Specifically, such reflexes become evident only after severe injury to the cervical spinal cord resulting in a motorcomplete neurologic status (Calancie, 199l), also known as Frankel ' A ' or 'B' (Frankel et al., 1969). Details about the stability of responses once such reflexes are established, and the time period after injury before such reflexes become evident, will be presented in a forthcoming paper and are not the subject of the present study. From a clinical standpoint, the presence of interlimb reflexes in a subject with injury to the cervical spinal cord is, almost without exception, a poor prognostic indicator for a subject's potential for spontaneous recovery of volitional motor function in the hands or lower limbs. The only exceptions to this finding have been 3 individuals with motor-complete cervical spinal cord injury who were entirely denervated of motoneurons to their distal upper limb muscles. Sensory stimulation of the types described above applied to the lower extremities failed to evoke interlimb reflexes in these individuals, but neither could motor responses be evoked from hand muscles by direct stimulation of the ulnar or median nerves of these persons. Such widespread gray matter loss within the caudal portion of the cervical enlargement after injury to the cervical spinal cord is rare, in our experience (and see Peckham et al., 1976). Properties of lower limb cutaneous inputs adequate to evoke upper limb motor responses were as follows: (1) even the smallest of the identified receptive fields tended to be much larger than those reported for studies of primary cutaneous afferents originating in the hand or fingertips (Westling and Johansson, 1987); (2) distal receptive fields were smaller than those found more proximally; (3) a given receptive field often included multiple sensory dermatomes and areas innervated by different cutaneous nerves; (4) excitatory effects were seen bilaterally, although contralateral effects were more common; and (5) light touch, hair movement, nociception, heat, and cold stimuli were shown to be effective, implicating both low threshold and wide dynamic range afferent fibers (Cervero et al., 1977).

B. Calancie et al. /Electroencephalography and clinical Neurophysiology 101 (1996) 304-3 ! 5

Response characteristics of upper limb motor units to these stimuli were as follows: (1) many motor units tended to have highly-reliable synaptic inputs, as evidenced by high 1:1 rate-following capabilities and narrow PSTH durations; (2) some motor units demonstrated windup-like responses to repeated low-frequency lower limb stimulation; (3) some motor unit response latencies to lower limb stimulation were extremely brief (e.g. see Figs. 1 and 6), suggesting afferent conduction velocities approaching 50 m/s; and (4) inhibitory responses to cutaneous stimulation were seen on occasion in tonicallyactive upper limb motor units. The large receptive field sizes encountered in this study, particularly for those found more proximally, suggest that a given upper limb motor response was not mediated by activation of a single, primary afferent fiber, but instead was derived from a number of primary afferents. This observation, coupled with the finding that the same motor unit often responded to a wide range of sensory inputs, is consistent with extensive convergence of lower limb primary afferents onto a second-order afferent system. We propose that in many cases, activation of one or more neurons within this second-order system was responsible for evoking the upper limb motor unit activity described in this report. The alternative explanation, that upper limb motor responses are mediated by large numbers of primary afferents which have all made synaptic contact with the same cervical motoneuron, cannot be entirely ruled out, but is less attractive for two reasons. First, one would expect to see a continuum of PSTH response latencies to electrical stimulation, reflecting the anticipated wide range of afferent axon diameters associated with light touch through nociceptor fiber participation. Instead, post-stimulus time histograms of motor unit responses to lower limb receptive field stimulation typically showed discrete periods of discharge, often separated in time by 40-50 ms periods in which no single response to lower limb stimulation was seen (e.g. see Fig. 1A,C). We propose that these distinct periods of increased excitability reflect the contribution of different primary afferents with very dissimilar conduction velocities. Second, it is likely that many motor units of a given motoneuron pool would be innervated by such a large number of lower limb afferent fibers, yet in the present experiments we rarely saw more than 2-3 motor units within any given muscle recruited by stimulation of a given lower limb receptive field. The available evidence therefore points to a secondorder system for mediating the cutaneous effects described in this paper, such that each of the spinocervical, spinocerebeilar (dorsal and ventral), spinothalamic, and long ascending propriospinal tract fiber populations is a potential candidate for these findings. Based on many, but not all, of the receptive field and response properties as summarized above, the spinocervical tract is a likely candidate for mediating at least some aspects of the cuta-

313 •

neous-evoked interlimb reflexes described in this study. In man and based only upon pathological (i.e. degenerative) studies, the importance of the spinocervical tract is controversial (Kircher and Hongchien, 1968; Kaas, 1990), probably due in part to a paucity of research related to this pathway. Fibers within the spinocervical tract ascend in the dorsolateral fasciculus and terminate within the ipsilateral lateral cervical nucleus, at and rostral to the C4 level in the upper cervical human spinal cord (Kaas, 1990). Arguments in favor of the spinocervical tract mediating the interlimb reflexes described herein are based in large part on the functional properties of this pathway as described in a variety of cat studies. The spinocervical tract has been shown to have extensive convergence from a wide range of afferent fibers mediating light touch, hair bend, thermal, and nociceptive inputs (Brown and Franz, 1970; Cervero et al., 1977; Downie et ai., 1988). The synaptic reliability of this pathway is typically high, with reports of 1:1 rate-following of some inputs at stimulus rates exceeding 100 Hz (Harrison and Jankowska, 1984; Brown et al., 1987). Vibratory stimulation was shown to cause optimal phase-locking of lateral cervical nucleus cells (the target of the spinocervical tract) at frequencies between 10 and 30 Hz in the primate (Downie et al., 1988). Afferent input has also been shown to be inhibitory to some neurons within this pathway, causing a reduction or silence in their tonic discharge pattern (Hongo et al., 1968; Cervero et al., 1977), as we saw in the present study (e.g. see Fig. 7). Such inhibition of a rhythmically-discharging cervical motoneuron indicates that its source of excitation is likely from one or more spontaneously-active second-order afferent fibers (Brown and Franz, 1970; Cervero et al., 1977), whose discharge rate is attenuated by the sensory input. The contribution of nociceptors to spinocervical tract cells is also consistent with our observation of 'wind-up' behavior of some upper limb motor responses (Thompson et al., 1993). While many of our findings point to the spinocervical tract as mediating the cutaneous-evoked interlimb reflexes described herein, there are two findings which are not consistent with this interpretation. First, interlimb reflex activity is seen both ipsi- and contralateral to the lower limb being stimulated, whereas only limited projections of the spinocervicai tract fibers to their contralateral target nuclei have been demonstrated (Hongo et al., 1968; Cervero et al., 1977). The contralateral influence of lower limb afferent input suggests a role for long ascending propriospinal fiber and/or spinothalamic tract contributions to interlimb reflexes. Involvement of this latter tract could also better account for our second observation which is not consistent with spinocervical tract involvement, which is that intense responses to cold stimulation were commonly observed. Conversely, cold was found to be a relatively poor stimulus for spinocervical tract and/or lateral cervical nucleus activation in animal studies (Cer-

314

B. Calancie et al. / Electroencephalography and clinical Neurophysiology 101 (1996) 304-315

vero et al., 1977; Downie et al., 1988). Thus, there can be, and probably are, other lower limb afferent fibers which are capable of mediating the interlimb reflexes described in this report, and in the original description of this phenomenon (Calancie, 1991). Regardless of the origin of the excitatory input to cervical motoneurons, the proposed mechanism by which interlimb reflexes emerge after severe injury to the cervical spinal cord is as follows. Given that subjects who demonstrate interlimb reflexes have essentially no voluntary motor innervation to their hands or lower limb muscles, one can safely conclude that most or all of the corticospinai tract has been destroyed at the level of injury in these subjects. This would lead to extensive denervation of motoneurons which control movements of the hand and digits, virtually all of which lie at or caudal to the C6 neurological level (Kendall and McCreary, 1983). The extent of that denervation would be most pronounced for intrinsic muscles of the hand, which depend exclusively upon corticospinal input, whereas more proximal forearm muscles receive a relatively greater proportion of segmental innervation (Phillips and Porter, 1977). Moreover, because such a lesion would also result in severe disruption of the dorsal, dorsolateral and lateral spinal cord within which areas many of the ascending afferent fibers are found, these ascending afferents would also be damaged at this point. Therefore, we suggest that as a consequence of losing their cervical or brainstem target nuclei, at least some of the lower limb afferent axons survive and develop novel connections through the extension of new growth processes, to innervate the partially-denervated motoneurons of the caudal cervical enlargement. The term regenerative sprouting (Steward, 1989) best describes the changes which we are proposing to account for the appearance of interlimb reflexes in subjects after severe cervical spinal cord injury. This conclusion is supported by our preliminary studies of the time-course of interlimb reflex development as determined through repeated examinations of subjects beginning immediately after cervical spinal cord injury. In these subjects, interlimb reflexes do not appear until at least 8 months after spinal cord injury, long after extensor spasticity has appeared and the spinal cord has emerged from 'spinal shock'. The alternative explanation for interlimb reflexes, that they represent an example of 'latent synapses', is not consistent with their prolonged time course of appearance in the subjects studied (Guth, 1976; Goshgarian and Guth, 1977; Wail, 1988; Yu and Goshgarian, 1993; Wall and McMahon, 1994). To our knowledge, there is no direct evidence from human pathology studies of aberrant connections which have been shown between ascending, lower limb afferent fibers and motoneurons of the distal cervical enlargement. Indirect evidence for the preservation of such ascending afferent fibers comes from studies of Bunge and coworkers, who have found that immediately caudal to an

anatomically complete lesion of the human cervical spinal cord, the fasciculus gracilis of the dorsal columns appears to be relatively well preserved, without the degeneration typically seen in these afferent pathways rostral to the lesion (Becerra et al., 1995; R.P. Bunge, personal communication). Characteristics of the upper limb motor responses to stimulation of lower limb receptive fields point to the development of novel synaptic connections with properties quite different from those occurring during normal development. The limited variability in latency (low jitter) and high rate-following capabilities seen in the majority of motor units tested suggest that large excitatory postsynaptic potentials are being produced by the ascending afferent fibers (Berry and Pentreath, 1976; Knox and Poppele, 1977; Ashby and Zilm, 1978; Fetz and Gustafsson, 1983). The high rate-following observed in the present study is not normally associated with direct, segmental reflex pathways. For example, attenuation of successive H-reflexes in the monosynaptic stretch reflex pathway has been shown to occur at both low and high frequencies: while the low-frequency attenuation (depression) may reflect segmental inhibitory mechanisms invoked by repeated stimulation (e.g. presynaptic inhibition) (Ishikawa et al., 1966; Van Boxtel, 1986; Calancie et al., 1993), the higher-frequency depression (e.g. rates >5 Hz) is probably due to post-synaptic failure (Curtis and Eccles, 1960). In the present study, however, the majority of motor units could follow stimulus rates >10 Hz, suggesting more efficacious synaptic connections within this (probably) oligosynaptic pathway than are typically encountered in human studies (Burke et al., 1989). Finally, such extreme firing rates as those illustrated in Fig. 3 suggest that an alteration in the motoneuron's intrinsic membrane properties has occurred. This type of discharge has been reported after a variety of lesions to CNS neurons (Guatteo et al., 1994), although the possibility that recurrent inhibitory feedback is ineffectual or has not developed cannot be ruled out.

Acknowledgements This work was supported by grants from the NIH (NINDS-NS 28059, and NCMRR-HD 31240), and by the Miami Project to Cure Paralysis. The authors thank Dr. Natalia Alexeeva for her critical review of this manuscript.

References American Spinal Injury Association. Standards for Neurological Classification of Spinal InjuryPatients. ASIA, 1992. Ashby, P. and Zilm, D. Synaptic connections to individual tibialis anterior motoneurones in man. J. Neurol. Neurosurg. Psychiatry, 1978, 41: 684---689. Becerra, J.L., Pucken, W.R., Hiester, E.D., Quencer, R.M., Marcillo, A.E., DonovanPost, M.J. and Bunge, R.P. MR-pathologiccompari-

B. Calancie et al. / Electroencephalography and clinical Neurophysiology 101 (1996) 304-315 sons of WaUerian degeneration in spinal cord injury. Am. J. Neuroradiol., 1995, 16: 125-133. Bellemare, F., Woods, J., Johansson, R. and Bigland-Ritchie, B. Motorunit discharge rates in maximal voluntary contractions of three human muscles. J. Neurophysioi., 1983, 50: 1380-1392. Berry, M.S. and Pentreath, V.W. Criteria for distinguishing between monosynaptic and polysynaptic transmission. Brain Res., 1976, 105: 1-20. Brown, A.G. and Franz, D.N. Patterns of response in spinocervical tract neurones to different stimuli of long duration. Brain Res., 1970, 17: 156-160. Brown, A.G., Koerber, H.R. and Noble, R. Excitatory actions of single impulses in single hair follicle afferent fibres on spinocervical tract neurones in the cat. J. Physiol. (London), 1987, 382: 291-3!2. Burke, D., Adams, R.W. and Skuse, N.F. The effects of voluntary contraction on the H reflex of human limb muscles. Brain, 1989, 112: 417--433. Calancie, B. Interlimb reflexes following cervical spinal cord injury in man. Exp. Brain Res., 1991, 85: 458--469. Calancie, B. Reflex and central motor conduction studies after SCI in man: what have we learned about spinal cord plasticity and longterm outcome? J. Neurotrauma, 1995, 12: 361. Calancie, B. and Broton, J.G. Evidence for CNS ingrowth caudal to cervical spinal cord injury in man. Soc. Neurosci. Abstr., 1992, 18: 1404. Calancie, B., Broton, J.G., Klose, KJ., Traad, M., Difini, J. and Ayyar, DR. Alterations in presynaptic inhibition contribute to segmental hypo- and hyperexcitability after spinal cord injury in man. Electroenceph, clin. Neurophysiol., 1993, 89: 177-186. Cervero, F., lggo, A. and Molony, V. Responses of spinocervical tract neurones to noxious stimulation of the skin. J. Physiol. (London), 1977, 267: 537-558. Curtis, D.R. and Eccles, J.C. Synaptic action during and after repetitive stimulation. J. Physiol. (London), 1960, 150: 374-398. Downie, J.W., Ferrington, D.G., Sorkin, L.S. and Willis, W.D.J. The primate spinoeervicothalamic pathway: responses of cells of the lateral cervical nucleus and spinocervical tract to innocuous and noxious stimuli. J. Neurophysiol., 1988, 59: 861-885. Fetz, E.E. and Gustafsson, B. Relation between shapes of post-synaptic potentials and changes in firing probability of cat motoneurones. J. Physiol. (London), 1983, 341: 387--410. Frankel, H.L., Hancock, D.O., Hyslop, G., Melzak, J., Michaelis, L.S., Ungar, G.H., Vernon, J.D.S. and Walsh, J.J. The value of postural reduction in the initial management of closed injuries of the spine with paraplegia and tetraplegia. Paraplegia, 1969, 7: 179-192. Goshgarian, H.G. and Guth, L. Demonstration of functionally ineffective synapses in the guinea pig spinal cord. Exp. Neurol., 1977, 57: 613-621. Guatteo, E., Bacci, A., Franceschetti, S., Avanzini, G. and Wanke, E. Neurons dissociated from neocortex fire with 'burst' and 'regular' trains of spikes. Neurosci. Lett., 1994, 175:117-120. Guth, L. Functional plasticity in the respiratory pathway of the mammalian spinal cord. Exp. Neurol., 1976, 51: 414-420. Harrison, P.J. and Jankowska, E. An intracellular study of descending and non-cutaneous afferent input to spinocervical tract neurones in the cat. J. Physiol. (London), 1984, 356: 245-261.

315

Hongo, T., Jankowska, E. and Lundberg, A. Post-synaptic excitation and inhibition from primary afferents in neurones of the spinocervicai tract. J. Physiol. (London), 1968, 199: 569-592. Hugon, M. A discussion of the methodology of the triceps surae T- and H-reflexes. In: J.E. Desmedt (Ed,), New Developments in EMG and Clinical Neurophysiology. Karger, Basel, 1973, pp. 773-780. lshikawa, K., Ott, K., Porter, R.W. and Stuart, D. Low frequency depression of the H wave in normal and spinal man. Exp. Neurol., 1966, 15: 140-156. Kaas, J.H. Somatosensory System. In: G. Paxinos (Ed.), The Human Nervous System. Academic Press, San Diego, 1990, pp. 813-844. Kendall, F.P. and McCreary, E.K. Muscles: Testing and Function, 3rd edn. Williams & Wilkins, Baltimore, 1983. Kircher, C. and Hongchien, H.A. The nucleus cervicalis lateralis in primates, including the human. Anat. Rec., 1968, 160: 376. Knox, C.K. and Poppele, R.E. Correlation analysis of stimulus evoked changes in excitability of spontaneously firing neurons. J. Neurophysiol., 1977, 40: 616--625. Mendell, L.M. Physiological properties of unmyelinated fiber projection to the spinal cord. Exp. Neurol., 1966, 16: 316-332. Peckham, P.H., Mortimer, J.T., and Marsolais, E.B. Upper and lower motor neuron lesions in the upper extremity muscles of tetraplegics. Paraplegia, 1976, 14: 115-121. Phillips, C.G. and Porter, R. Corticospinal Neurons: Their Role in Movement. Academic Press, London, 1977. Steward, O. Reorganization of neuronal connections following CNS trauma: principles and experimental paradigms. J. Neurotrauma, 1989, 6: 99-151. Thompson, S.W.N., Woolf, C.J. and Sivilotti, L.G. Small-caliber afferent inputs produce a heterosynaptic facilitation of the synaptic responses evoked by primary afferent A-fibers in the neonatal rat spinal cord in vitro. J. Neurophysiol., 1993, 69:2116-2128. Van Boxtel, A. Differential effects of low-frequency depression, vibration-induced inhibition, and posttetanic potentiation on H-reflexes and tendon jerks in the human soleus muscle. J. Neurophysiol., 1986, 55: 551-568. Wall, P.D. Recruitment of ineffective synapses after injury. In: S.G. Waxman (Ed.), Advances in Neurology, Vol. 47, Functional Recovery in Neurological Disease. Raven Press, New York, 1988, pp. 387--400. Wall, P.D. and McMahon, S.B. Long range afferents in rat spinal cord, Ill: failure of impulse transmission in axons and relief of the failure after rhizotomy of dorsal roots. Phil. Trans. R. Soc. Lond. B, 1994, 343:211-223. Westling, G. and Johansson, R.S. Responses in glabrous skin mechanoreceptors during precision grip in humans. Exp. Brain Res., 1987, 66: 128-140. Xu, X.-J., Zhang, X., H6kfelt, T. and Wiesenfeld-Hallin, Z. Plasticity in spinal nociception after peripheral nerve section: reduced effectiveness of the NMDA receptor agonist MK-801 in blocking wind-up and central sensitization of the flexor reflex. Brain Res., 1995, 670: 342-346. Yu, X.J. and Goshgarian, H.G. Aging enhances synaptic efficacy in a latent motor pathway following spinal cord hemisection in adult rats. Exp. Neurol., 1993, 121 : 231-238.