EXPERIMENTAL
65243-254
NEUROLOGY
(1979)
Selective Regeneration of Sensory Fibers following Nerve Crush Injury MARSHALLDEVOR'ANDRUTH Neurobiology
Unit, Received
Life October
Sciences
Institute,
3, 1978; revision
GOVRIN-LIPPMANN Hebrew
University.
received
February
Jeru.salem.
Israel
20, 1979
We investigated whether or not all classes of myelinated sensory fibers regenerate simultaneously after nerve crush injury and whether each parent fiber produces and maintains a regenerated branch. Electrophysiologic recordings were made from isolated lumbar dorsal root fibers in rats at various times after sciatic nerve crush injury. For all fibers that responded to electrical stimulation distal to the crush, the conduction velocity of the parent axon was established using the collision method. During the first few months following sciatic nerve injury, A6 parent fibers were under:represented in the sample of successfully regenerated fibers. Only later did sprouts of the bulk of A6 fibers grow out. In time, however, all fibers regenerated. Combined with histologic data indicating that fiber number was restored in the distal nerve stump, this finding indicated that after crush injury the majority of parent fibers maintained one, but only one, regenerated branch.
INTRODUCTION The process of nerve regeneration after traumatic injury has been investigated without letup for a century and more [for reviews see (11, 17, 18, 24)]. Nevertheless, many fundamental aspects of the problem remain unexplored. One class of such problems deals with the relationship between axon sprouts and the parent fibers that produce them. Do all fiber types have a similar capacity for regeneration? Measurements of nerve profiles proximal and distal to a lesion provide population statistics but they do not allow one to match individual fibers profile for profile and therefore do not permit conclusions as to which type of parent fiber produced which type of sprout. In due course after crush injury, for example, the number Abbreviation: CAP-compound action potential. ’ Supported by the Fritz Thyssen Stiftung and the Foundations Fund for Research in Psychiatry. We thank W. Calvin, I. Pamas, Z. Seltzer, and P. D. Wall for their advice. 243 0014-4886/79/080243-12$02.00/O Copyright All rights
0 1979 by Academic Press. Inc. of reproduction in any form reserved.
244
DEVOR
AND
GOVRIN-LIPPMANN
of myelinated axons distal to the point of injury approaches that proximally, suggesting that each parent fiber maintains one sprout to maturity (12). One cannot exclude the possibility, however, that some fibers maintain many distal branches and that others maintain none. We have used an electrophysiological approach to establish for a large population of axons at various times after crush injury the conduction velocity of individual regenerated axons together with that of their parent fiber. In this paper we address two main questions: (i) Do fibers of all classes emit regenerating sprouts simultaneously following nerve crush injury? and (ii) Do all parent fibers produce and maintain a single regenerated branch? MATERIALS
AND METHODS
Subjects and Surgery. Experiments were completed on 60 adult Sabra strain albino rats each 250 to 350 g at the time of initial surgery. Under ether anesthesia and using aseptic precautions, the sciatic nerve of the left leg was crushed in midthigh. The crush was made with a mosquito haemostat with flattened teeth (jaw width 2.5 mm) which was closed tightly on the nerve for 30 s. Carbon particles applied to the crushing jaws left a black mark on the epineurium indicating the position of the lesion. The completeness of crush lesions made in this way was established in three ways. (i) Nerve conduction across the crush was blocked immediately. (ii) Conduction between two points distal to the crush using stimuli sufficient to drive C-fibers ceased completely within 3 to 4 days postlesion. Muscle jerk upon such stimulation also ceased. (iii) The lateral toes were completely anesthetic for at least 2 weeks (9). The rats were maintained postoperatively for as long as 481 days before being killed after acute electrophysiological experiments. Electrophysiology. Anesthesia was induced and maintained with sodium pentobarbital(50 mg/kg i.p., then 20 mg/kg/h i.p.). The rats, in a suspended posture, were paralyzed with Flaxedil, artificially respirated, warmed to 36 to 37°C and infused with 5% dextrose saline via one carotid artery. The lumbosacral spinal roots were then exposed and placed under warmed (36 to 37°C) paraffin oil. Finally the sciatic trunk, under warmed oil, was exposed from above the sciatic notch to the bifurcation of the (posterior) tibial nerve near the ankle. Pairs of bipolar Ag- AgCl stimulating electrodes were positioned along the length of the nerve (Fig. 1); one (Sl) about 10 mm proximal to the crush and a second (S2) on the tibia1 nerve just below the gastocnemius-soleus branches and about 20 mm below the crush. In some preparations a third pair of electrodes (S3) was placed 20 mm distal to S2 in order to follow the progress of regeneration. Data from these experiments are presented in a companion paper (8). Recordings were
SELECTiVE
NERVE REGENERATION
245
FIG. 1. Top-Sketch of the experimental setup, Bottom-a typical collision experiment. The top trace shows two fibers responding (at R in sketch) to stimulation at S2. Without changing stimulation current at S2, current at Sl was slowly increased. When the local (Sl) threshold for the faster of the two fibers was reached (bottom trace) the fiber suddenly responded to the Sl stimulus and the spike driven at S2 disappeared from the record because of collision somewhere between Sl and S2. Threshold for the slower of the two fibers at Sl had not yet been attained and the fiber was still responding to the S2 stimulus. Calibration: 1 ms. 1 mV.
made from the dorsal roots associated with the sciatic nerve, L4-6, particularly from L5. For compound action potentials (CAPS), the root was cut centrally and layed on a pair of Ag- AgCl hooks. To record from single fibers, fine strands about 20 to 50 pm in diameter and 1 to 2 mm long were dissected from the root and placed on a single Ag-AgCl recording lead referenced to a nearby ground. The strands contained a maximum of about six myelinated fibers contributing to the tibia1 nerve. This was determined by gradually increasing current strength at S2 and counting the number of discreet all-or-none potentials recorded. The distance between stimulating and recording electrodes was based on the contour length measured along the course of the nerve. Conduction velocity was determined as the quotient of conducting distance and response latency with no account made for utilization time. Stimuli consisted of 0.05- or 0. l-ms, 0- to 4-mA square wave pulses photoelectrically isolated from ground and delivered across the nerve at 1.5 Hz. Latency was measured from the stimulus artifact to the beginning of the rising phase of the evoked potential. Histology. To make an estimate of bias in our sample of fibers we compared the conduction velocity of a population of intact tibia1 nerve sensory axons with the fiber diameter distributions of a similar axon population. First, the L4 through Sl ventral roots were severed unilaterally in two rats. After 3 1 days, sufficient time for degeneration of the motor fibers, the rats were perfused transcardially with saline followed by 10% formol-saline.
246
DEVOR AND GOVRIN-LIPPMANN
The tibial nerves at S2 were removed immediately, immersed 5 days in 2% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.2, and then postfixed overnight in 1% osmium tetroxide in cacodylate buffer. Paraffin-embedded sections cut at 5 pm were photographed and printed at a total magnification of 1000 x and assembled into a montage. The diameter (axon plus myelin) of each myelinated fiber was measured to the nearest 0.5 mm (0.5 pm). RESULTS Sampling Bias. Our method of dissecting small bundles of axons from dorsal roots did not introduce a substantial bias into the axon population sampled, at least not for myelinated fibers. The shaded portion of Fig. 2 shows the distribution of conduction velocities for 644 consecutively encountered dorsal root axons that responded to stimulation at S2 on the tibial nerve from 13 intact rats. A comparable population of 5372 fibers from deefferented tibia1 nerves of two rats (2681 and 2691 fibers, respectively) is marked with a dashed line. To facilitate comparison the AC@ peak of the electrophysiologic and histologic distributions were brought together by multiplying fiber diameter (pm) by 7.8. The generally accepted figure of 6.0 relating fiber diameter to conduction velocity [in cat (15) and rat (19)] did not hold because conduction velocity over the course used was greater than local conduction velocity on the tibia1 nerve where histologic sections were taken (8). Thus adjusted, the two histograms show the same relative proportions of AC@ and A6 fibers. Unlike Hunt (14), who performed a -C
%-
-4
15-1
IO -
-1
-I 5-
o-
12 10
20
3
4 30
5 6 7 40 50
8 60
910111213 70 80
90
I m/o\ !C
FIG. 2. Histograms comparing the distribution of sensory fiber diameter in the tibia1 nerve (dashed line, right scale) and conduction velocity of the parent fiber segments (Sl + R) of tibial fibers (shaded, left scale) in normal rats. The major peaks of the two histograms are brought together to facilitate comparison. The heights of the histograms are not adjusted except to compensate for different bin widths.
SELECTIVE
NERVE
REGENERATION
247
similar analysis, we did not find a substantial under-representation of smalldiameter fibers. We attribute the improvement to differences in techniques of dissecting fine dorsal root filaments. Which Fibers Sprout? To determine if sprout outgrowth is simultaneous in all classes of myelinated fibers we recorded from populations of single dorsal root fibers that responded to stimulation distal to the crush (at S2) and established for each the conduction velocity of the parent fiber. In a typical experiment a small strand containing one to six active sensory axons was placed on the recording lead. Stimulation current at S2, distal to the crush, was adjusted to just above threshold for a particular single axon within the strand. The current at Sl, proximal to the crush and preceding S2 by 0.0 to 1.0 ms, was then increased gradually until threshold was attained. At that point the spike, with its own peculiar amplitude and form, appeared in the record at a latency fixed to the Sl artifact and the spike previously generated at S2 disappeared, having collided with the spike from Sl (Fig. 1). Collision confirmed that the same axon was being driven from the Sl and S2 positions. Regenerating axons consist of a parent fiber from the recording point (R) to the point of injury and a sprout distally. The distance from Sl to R divided by the response latency from S 1 provided the parent fiber velocity; the distance S2 to R divided by the latency from S2 gave the overall velocity. The mean velocity of the sprout component (V,) from S2 to the crush was calculated as v, = (S2 - C)/(T,
- T,(C/Sl)),
where S 1 and S2 are distances and T, and Tz conduction times from the Sl and S2 stimulating electrodes to the recording electrode, respectively. C is the distance from the crush to the recording position. The derivation assumes that the fiber conducted at essentially the parent velocity as far as the position of the crush. The assumption is substantiated by conduction velocity contours of single regenerating fibers where the most sudden change in conduction velocity was found to occur very near the point of injury (8, 16). In intact control animals measurements were made of conduction velocity in proximal and distal segments of the nerve over lengths corresponding to the parent and sprout lengths of the experimental rats. Histograms of parent fiber, overall, and distal segment velocities for a sample of 644 fibers from intact rats, each responding to stimulation at S2 on the tibia1 nerve, are shown in Fig. 3 (top). All fibers are represented in each histogram. Peaks representing the Aa@ and A6 groups are marked by arrows. Corresponding data are given in Fig. 3 (2nd row) for 94 fibers from four rats sampled 7 to 17 days after crushing the sciatic nerve. Here, all sprouts conducted at less than 5 m/s (mean 1.52 + 0.55 ms). The dis-
DEVOR AND GOVRIN-LIPPMANN
248
SI*R
3.
S2*R
S2*C
t-l=94 50
15
7-vd
2 IdL
z” b 5
3.
s
I5
!O
n=lo9
IL
E! 30
0
A
--r
n--l13
15 0 !JldLk.
CONDUCTION VELOCITY (m/w) FIGS. 3 AND 4. Histograms of parent fiber(S1 + R), overall (S2 + R), and sprout (S2 --, C) conduction velocity for samples of myelinated parent fibers at various times after a sciatic nerve crush. Each fiber was detected in a lower lumbar dorsal root and responded to stimulation of the tibial nerve distal to the crush. The number(N) of single axons included in each set of three histograms is given in the left column. Each histogram includes data collected from at least three rats. Note the under-representation of low-velocity parent fibers at short survival times. Arrows in the (Sl + R) histogram for control rats indicate the AC@ and AS fiber populations.
tribution of parent fiber velocities remained much higher. It was, however, not normal. In particular, the major peak was slower than in intact nerve and the minor peak was missing. We argue that this reflected two processes, the well-known retrograde reduction in conduction velocity in injured axons (1, 7), and a real under-representation of A6 fibers in the population of parent fibers that contributed sprouts distal to the crush at this survival time. Figures 3 and 4 show corresponding data for postlesion
249
SELECTIVENERVEREGENERATION
Sl+R
0
30
60
S2+R
0
30
CONDUCTION
S2+C
60
0
30
VELOCITY
FIGURE
60 (m/set)
4
survival time to as long as 481 days. Note that between about 150 and 250 days the parent fiber distribution returned to normal. That is, the faster fibers had recovered from their retrograde change and the slower ones had finally begun to appear in the record. Retrograde Slowing. Retrograde slowing of parent fibers could be assessed independently of selective fiber outgrowth by sampling fibers that responded to stimulation proximal to the lesion at S 1 without asking which of those fibers sent a sprout distally across the crush line. This was done by recording the CAP elicited by supramaximal stimulation at Sl and plotting the conduction velocity of its peak as a function of time after the crush (Fig. 5). Conduction velocity decreased to about 65% of control values by 20 to 25 days after the lesion and returned to normal by about the 90th day. Interestingly, 20 to 25 days is also the period elapsed until
250
DEVOR AND GOVRIN-LIPPMANN
3 60. J 2 50. Y n: don ; 30 . T OL
L 0
’
’ 20
’
’ 40
’
’ 60
’
’ SO
Days
’
’ 100
after
’
‘ii’
200
1 250
300
Crush
FIG. 5. Temporary retrograde slowing of sciatic nerve fibers proximal to a crush lesion. The conduction velocity of the peak of the compound action potential (CAP) elicited by supramaximal stimulation at Sl is plotted as a function of time after crushing.
the regenerating tibial nerve first makes functional sensory connections in the skin of the foot (8, 9). The reason for plotting the peak of the CAP rather than its leading edge was that this avoided the danger that the spread of current proximally along the sciatic nerve might stimulate intact fibers in nerve branches that had not been crushed. The actual plot of the leading edge of the CAP, however, closely resembles Fig. 5 shifted up the Y-axis. A sample of single fibers responding to stimulation at Sl (but not necessarily S2) in control rats and rats surviving 21 to 35 and 292 to 321 days after sciatic nerve crush is shown in Fig. 6. Percentage Fibers That Cross the Crush. For this measurement the S2 stimulating electrode was moved proximally onto the sciatic nerve about 10 mm distal to the crush site (asterisk in Fig. 1). In small strands of dorsal roots associated with the sciatic nerve we compared the number of single axons excited by stimulation at Sl proximal to the crush with the number excitable at S2 distally. As long as the strand contained no more than about seven or eight responsive fibers, each could be counted as an all-or-none addition to the multiple fiber potential as stimulation current was gradually increased. In three experiments in intact nerves, all of 157 fibers sampled responded to stimulation both proximal and distal to the site at which we crushed the nerve in the experimental animals. The same 100% success was found in three experiments totalling 202 axons 156 to 219 days after crush injury. At shorter intervals of 68 and 73 days (two rats), however, 13.6% of the 250 fibers that responded to stimulation proximally (Sl)
SELECTIVE
NERVE REGENERATION
k
251
Od n=20i
15
292-321d n=142
lilwk 0
20 40 60
Conduction
80
Velocity
(m/W FIG. 6. Histograms of parent fiber (Sl --* R) conduction velocity of dorsal root fibers that responded to stimulation of the sciatic nerve proximal to a crush lesion. The sample taken 21 to 23 days after crushing illustrates the maximal extent of retrograde slowing of conduction velocity. Note that in contrast to Figs. 3 and 4 these histograms are an unselected sample of sciatic nerve sensory fibers and do not represent only those that sent a regenerating sprout into the tibial nerve.
failed to respond to stimulation distally (S2). Most of the parent fibers that failed to send a sprout across the crush had conduction velocities in the A6 range (mean 23.4 + 8.4 m/s). DISCUSSION Do fibers of all classes produce regenerating sprouts simultaneously after nerve crush injury? Data summarized in Figs. 3 and 4 indicate that they do not. For the first 150 to 250 days after crush injury, myelinated parent fibers of low conduction velocity are relatively under-represented among the population of fibers contributing a sprout distally in the tibia1
252
DEVOR
AND
GOVRIN-LIPPMANN
nerve. However, the conclusion requires that the sampling frequency of small fibers is representative during this period and that the stimulating conditions were not such as to favor sprouts of large-diameter parent fibers. Figure 2 shows that our recording method is not fundamentally biased against small-diameter fibers. Furthermore, a sample of parent fibers driven by stimulation proximal to the crush within this period (21 to 23 days) was not deficient in fibers of low conduction velocity (Fig. 6). Finally, we found that at 68 to 73 days, 13.6% of fibers sampled failed to send a sprout across the crush line and that these were predominantly axons of low parent-fiber conduction velocity. This latter result suggests that at least part of the delay in the arrival of sprouts at S2 occurred in the vicinity of the nerve lesion. As for the adequacy of the stimulation, current was routinely increased to about twice the threshold of the highest threshold fibers that responded. This did not recruit more units. In a few experiments stimulation of as much as 10 mA at 0.5ms pulse duration was used and this too failed to recruit additional myelinated parent fibers although unmyelinated parent fibers now appeared. We conclude that in crushed sensory nerves, sprouts of rapidly conducting, large-diameter parent axons tend to grow out earlier than those of slowly conducting, small-diameter axons. A hint of this result was noticed previously by Burgess and Horch [(5); see their Fig. 11. In motor nerves, a number of authors observed that reinnervation of muscle by y fibers required much longer than reinnervation by (Yfibers (2, 3, 21, 22). Do all parent fibers maintain a regenerated branch? Because we failed to find parent fibers proximal to a crush lesion of long standing that did not have a branch distally we conclude that all parent fibers eventually establish at least one stable branch. Morphological studies indicate that by 150 days after nerve crush the number of myelinated axon profiles distal to the lesion site equals that proximally (12), although some investigators reported a moderate excess [reviewed in (17)]. Combined, these data indicate that most parent fibers have one but only one branch even though many emit more than one initially (18,20). Some parent fibers are thought to maintain more than one branch after nerve transection (IO), perhaps in compensation for those that fail to maintain any. The histologic equivalent of the electrophysiologic technique we used would involve staining individual axons and following changes in their caliber for substantial distances. Though recent technical advances may soon change matters (4), this has not been practical until now. Progress in our understanding of the functional selectivity of peripheral axons (6) makes it important to describe regeneration as a function of parent fiber class and not simply as a mass phenomenon. Selectivity in the rate and
SELECTIVE
NERVE REGENERATION
253
ultimate success of regeneration must be taken into account in interpreting the clinical pattern of sensory return after nerve injury (13, 23). REFERENCES 1. ACHESON, G. H., E. S. LEE, AND R. S. MORRISON. 1942. A deficiency in the phrenic respiratory discharges parallel to retrograde degeneration. J. Neurophysiol. 5: 269213.
BESSOU, P., Y. LAPORTE, AND B. PAGES. 1966. Observations sur la r&innervation de fuseaux neuro-musculaires de chat. C. R. Sot. Rio/. (Paris) 160: 408-411. 3. BROWN, M. C., AND R. G. BUTLER. 1974. Evidence for innervation of muscle spindle intrafusal fibres by branches of alpha motoneurons following nerve injury. J. Physiol. 2.
(London) 4.
5. 6. 7. 8. 9.
238: 41P-43P.
BROWN, A. G., P. K. ROSE, AND P. J. SNOW. 1977. The morphology of hair follicle afferent fibre collaterals in the spinal cord of the cat. J. Physiol. (London) 272: 779-797. BURGESS,P. R., AND K. W. HORCH. 1973. Specific regeneration ofcutaneous fibers in the cat. J. Neurophysiol. 36: lo!- 114. BURGESS, P. R., AND E. R. PERL. 1973. Cutaneous mechanoreceptors and nociceptors. Pages 29-78 in A. IGGO, Ed., Handbook of Sensory Physiology; Vol. 2: Somatosensory System. Springer-Verlag, Berlin. CRAGG, B. Cl., AND P. K. THOMAS. 1961. Changes in conduction velocity and fibre size proximal to peripheral nerve lesions. J. Physiol. (London) 157: 315-327. DEVOR, M., AND R. GOVRIN-LIPPMANN. 1979. Maturation of axonal sprouts following nerve crush injury. Exp. Neural., submitted. DEVOR, M., D. SCHONFELD, Z. SELTZER, AND P. D. WALL. 1979. Two modes of cutaneous reinnervation following peripheral nerve injury. J. Comp. Neurol. 185: 211-220.
10. FULLERTON, P. M.,ANDR. W. GILLIATT. 1965. Axonreflexinhumanmotornervefibers. J. Neural. Neurosurg. Psychiat. 28: 1- 11. 11. GUTH, L. 1956. Regeneration in mammalian peripheral nervous system. Physiol. Rev. 36: 441-478. 12.
13. 14. 15. 16. 17. 18. 19.
GUTMANN, E., AND F. K. SANDERS. 1943. Recovery of fibre numbers and diameters in the regeneration of peripheral nerves. J. Physiol. (London) 101: 489-518. HEAD, H. 1920. Studies in Neurology. Oxford Univ. Press, London. HUNT, C. C. 1954. Relation of function to diameter in afferent fibers of muscle nerves. J. Gen. Physiol. 38: 117-131. HURSH, J. B. 1939. Conduction velocity and diameter of nerve fibers. Am. J. Physiol. 127: 131-139. JACOBSON, S., AND L. GLJTH. 1965. An electrophysiological study of the early stages of peripheral nerve regeneration. Exp. Neural. 11: 48-60. MIRA, J.-C. 1976. Etudes quantitatives sur la regeneration des fibres nerveuses myelinisees. II. Variations du nombre et du calibre des fihres regenerees apres un ecrasement ou une section du nerf. Arch. Anat. Microsc. Morphol. Exp. 65: 255-284. RAM~N Y CAJAL, S. 1928. in R. M. MAY, Ed., Transl., Degeneration andRegeneration in the Nervous System. Reprinted in 1959, Hafer, New York. SCHNEPP, G., P. SCHNEPP, AND G. SPAAN. 1971. Analysis of peripheral nerve fibres in animal of different body size. I. Total fibre count, fibre size and nerve conduction velocity. Z. Zellforsch. 119: 77-98. [in German]
254
DEVOR AND GOVRIN-LIPPMANN
20. SHAWE, G. D. H. 1955. On the number of branches formed by regenerating nerve fibres. Br. J. Surg.
42: 474-488.
21. TAKANO, K. 1976. Absence of gamma-spindle loop in the reinnervated hind leg muscles of the cat: “alpha muscle.” Exp. Brain Res. 26: 343-354. 22. THULIN, C.-A. 1960. Electrophysiological studies of peripheral nerve regeneration with special reference to the small diameter (gamma) fibers. Exp. Neurol. 2: 598-612. 23. WALSHE, F. M. R. 1942. The anatomy and physiology of cutaneous sensibility: a critical review. Bruin 65: 48- 112. 24. YOUNG, J. Z. 1942. The functional repair of nervous tissue. Physiol. Rev. 22: 318-374.
65243-254
NEUROLOGY
(1979)
Selective Regeneration of Sensory Fibers following Nerve Crush Injury MARSHALLDEVOR'ANDRUTH Neurobiology
Unit, Received
Life October
Sciences
Institute,
3, 1978; revision
GOVRIN-LIPPMANN Hebrew
University.
received
February
Jeru.salem.
Israel
20, 1979
We investigated whether or not all classes of myelinated sensory fibers regenerate simultaneously after nerve crush injury and whether each parent fiber produces and maintains a regenerated branch. Electrophysiologic recordings were made from isolated lumbar dorsal root fibers in rats at various times after sciatic nerve crush injury. For all fibers that responded to electrical stimulation distal to the crush, the conduction velocity of the parent axon was established using the collision method. During the first few months following sciatic nerve injury, A6 parent fibers were under:represented in the sample of successfully regenerated fibers. Only later did sprouts of the bulk of A6 fibers grow out. In time, however, all fibers regenerated. Combined with histologic data indicating that fiber number was restored in the distal nerve stump, this finding indicated that after crush injury the majority of parent fibers maintained one, but only one, regenerated branch.
INTRODUCTION The process of nerve regeneration after traumatic injury has been investigated without letup for a century and more [for reviews see (11, 17, 18, 24)]. Nevertheless, many fundamental aspects of the problem remain unexplored. One class of such problems deals with the relationship between axon sprouts and the parent fibers that produce them. Do all fiber types have a similar capacity for regeneration? Measurements of nerve profiles proximal and distal to a lesion provide population statistics but they do not allow one to match individual fibers profile for profile and therefore do not permit conclusions as to which type of parent fiber produced which type of sprout. In due course after crush injury, for example, the number Abbreviation: CAP-compound action potential. ’ Supported by the Fritz Thyssen Stiftung and the Foundations Fund for Research in Psychiatry. We thank W. Calvin, I. Pamas, Z. Seltzer, and P. D. Wall for their advice. 243 0014-4886/79/080243-12$02.00/O Copyright All rights
0 1979 by Academic Press. Inc. of reproduction in any form reserved.
244
DEVOR
AND
GOVRIN-LIPPMANN
of myelinated axons distal to the point of injury approaches that proximally, suggesting that each parent fiber maintains one sprout to maturity (12). One cannot exclude the possibility, however, that some fibers maintain many distal branches and that others maintain none. We have used an electrophysiological approach to establish for a large population of axons at various times after crush injury the conduction velocity of individual regenerated axons together with that of their parent fiber. In this paper we address two main questions: (i) Do fibers of all classes emit regenerating sprouts simultaneously following nerve crush injury? and (ii) Do all parent fibers produce and maintain a single regenerated branch? MATERIALS
AND METHODS
Subjects and Surgery. Experiments were completed on 60 adult Sabra strain albino rats each 250 to 350 g at the time of initial surgery. Under ether anesthesia and using aseptic precautions, the sciatic nerve of the left leg was crushed in midthigh. The crush was made with a mosquito haemostat with flattened teeth (jaw width 2.5 mm) which was closed tightly on the nerve for 30 s. Carbon particles applied to the crushing jaws left a black mark on the epineurium indicating the position of the lesion. The completeness of crush lesions made in this way was established in three ways. (i) Nerve conduction across the crush was blocked immediately. (ii) Conduction between two points distal to the crush using stimuli sufficient to drive C-fibers ceased completely within 3 to 4 days postlesion. Muscle jerk upon such stimulation also ceased. (iii) The lateral toes were completely anesthetic for at least 2 weeks (9). The rats were maintained postoperatively for as long as 481 days before being killed after acute electrophysiological experiments. Electrophysiology. Anesthesia was induced and maintained with sodium pentobarbital(50 mg/kg i.p., then 20 mg/kg/h i.p.). The rats, in a suspended posture, were paralyzed with Flaxedil, artificially respirated, warmed to 36 to 37°C and infused with 5% dextrose saline via one carotid artery. The lumbosacral spinal roots were then exposed and placed under warmed (36 to 37°C) paraffin oil. Finally the sciatic trunk, under warmed oil, was exposed from above the sciatic notch to the bifurcation of the (posterior) tibial nerve near the ankle. Pairs of bipolar Ag- AgCl stimulating electrodes were positioned along the length of the nerve (Fig. 1); one (Sl) about 10 mm proximal to the crush and a second (S2) on the tibia1 nerve just below the gastocnemius-soleus branches and about 20 mm below the crush. In some preparations a third pair of electrodes (S3) was placed 20 mm distal to S2 in order to follow the progress of regeneration. Data from these experiments are presented in a companion paper (8). Recordings were
SELECTiVE
NERVE REGENERATION
245
FIG. 1. Top-Sketch of the experimental setup, Bottom-a typical collision experiment. The top trace shows two fibers responding (at R in sketch) to stimulation at S2. Without changing stimulation current at S2, current at Sl was slowly increased. When the local (Sl) threshold for the faster of the two fibers was reached (bottom trace) the fiber suddenly responded to the Sl stimulus and the spike driven at S2 disappeared from the record because of collision somewhere between Sl and S2. Threshold for the slower of the two fibers at Sl had not yet been attained and the fiber was still responding to the S2 stimulus. Calibration: 1 ms. 1 mV.
made from the dorsal roots associated with the sciatic nerve, L4-6, particularly from L5. For compound action potentials (CAPS), the root was cut centrally and layed on a pair of Ag- AgCl hooks. To record from single fibers, fine strands about 20 to 50 pm in diameter and 1 to 2 mm long were dissected from the root and placed on a single Ag-AgCl recording lead referenced to a nearby ground. The strands contained a maximum of about six myelinated fibers contributing to the tibia1 nerve. This was determined by gradually increasing current strength at S2 and counting the number of discreet all-or-none potentials recorded. The distance between stimulating and recording electrodes was based on the contour length measured along the course of the nerve. Conduction velocity was determined as the quotient of conducting distance and response latency with no account made for utilization time. Stimuli consisted of 0.05- or 0. l-ms, 0- to 4-mA square wave pulses photoelectrically isolated from ground and delivered across the nerve at 1.5 Hz. Latency was measured from the stimulus artifact to the beginning of the rising phase of the evoked potential. Histology. To make an estimate of bias in our sample of fibers we compared the conduction velocity of a population of intact tibia1 nerve sensory axons with the fiber diameter distributions of a similar axon population. First, the L4 through Sl ventral roots were severed unilaterally in two rats. After 3 1 days, sufficient time for degeneration of the motor fibers, the rats were perfused transcardially with saline followed by 10% formol-saline.
246
DEVOR AND GOVRIN-LIPPMANN
The tibial nerves at S2 were removed immediately, immersed 5 days in 2% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.2, and then postfixed overnight in 1% osmium tetroxide in cacodylate buffer. Paraffin-embedded sections cut at 5 pm were photographed and printed at a total magnification of 1000 x and assembled into a montage. The diameter (axon plus myelin) of each myelinated fiber was measured to the nearest 0.5 mm (0.5 pm). RESULTS Sampling Bias. Our method of dissecting small bundles of axons from dorsal roots did not introduce a substantial bias into the axon population sampled, at least not for myelinated fibers. The shaded portion of Fig. 2 shows the distribution of conduction velocities for 644 consecutively encountered dorsal root axons that responded to stimulation at S2 on the tibial nerve from 13 intact rats. A comparable population of 5372 fibers from deefferented tibia1 nerves of two rats (2681 and 2691 fibers, respectively) is marked with a dashed line. To facilitate comparison the AC@ peak of the electrophysiologic and histologic distributions were brought together by multiplying fiber diameter (pm) by 7.8. The generally accepted figure of 6.0 relating fiber diameter to conduction velocity [in cat (15) and rat (19)] did not hold because conduction velocity over the course used was greater than local conduction velocity on the tibia1 nerve where histologic sections were taken (8). Thus adjusted, the two histograms show the same relative proportions of AC@ and A6 fibers. Unlike Hunt (14), who performed a -C
%-
-4
15-1
IO -
-1
-I 5-
o-
12 10
20
3
4 30
5 6 7 40 50
8 60
910111213 70 80
90
I m/o\ !C
FIG. 2. Histograms comparing the distribution of sensory fiber diameter in the tibia1 nerve (dashed line, right scale) and conduction velocity of the parent fiber segments (Sl + R) of tibial fibers (shaded, left scale) in normal rats. The major peaks of the two histograms are brought together to facilitate comparison. The heights of the histograms are not adjusted except to compensate for different bin widths.
SELECTIVE
NERVE
REGENERATION
247
similar analysis, we did not find a substantial under-representation of smalldiameter fibers. We attribute the improvement to differences in techniques of dissecting fine dorsal root filaments. Which Fibers Sprout? To determine if sprout outgrowth is simultaneous in all classes of myelinated fibers we recorded from populations of single dorsal root fibers that responded to stimulation distal to the crush (at S2) and established for each the conduction velocity of the parent fiber. In a typical experiment a small strand containing one to six active sensory axons was placed on the recording lead. Stimulation current at S2, distal to the crush, was adjusted to just above threshold for a particular single axon within the strand. The current at Sl, proximal to the crush and preceding S2 by 0.0 to 1.0 ms, was then increased gradually until threshold was attained. At that point the spike, with its own peculiar amplitude and form, appeared in the record at a latency fixed to the Sl artifact and the spike previously generated at S2 disappeared, having collided with the spike from Sl (Fig. 1). Collision confirmed that the same axon was being driven from the Sl and S2 positions. Regenerating axons consist of a parent fiber from the recording point (R) to the point of injury and a sprout distally. The distance from Sl to R divided by the response latency from S 1 provided the parent fiber velocity; the distance S2 to R divided by the latency from S2 gave the overall velocity. The mean velocity of the sprout component (V,) from S2 to the crush was calculated as v, = (S2 - C)/(T,
- T,(C/Sl)),
where S 1 and S2 are distances and T, and Tz conduction times from the Sl and S2 stimulating electrodes to the recording electrode, respectively. C is the distance from the crush to the recording position. The derivation assumes that the fiber conducted at essentially the parent velocity as far as the position of the crush. The assumption is substantiated by conduction velocity contours of single regenerating fibers where the most sudden change in conduction velocity was found to occur very near the point of injury (8, 16). In intact control animals measurements were made of conduction velocity in proximal and distal segments of the nerve over lengths corresponding to the parent and sprout lengths of the experimental rats. Histograms of parent fiber, overall, and distal segment velocities for a sample of 644 fibers from intact rats, each responding to stimulation at S2 on the tibia1 nerve, are shown in Fig. 3 (top). All fibers are represented in each histogram. Peaks representing the Aa@ and A6 groups are marked by arrows. Corresponding data are given in Fig. 3 (2nd row) for 94 fibers from four rats sampled 7 to 17 days after crushing the sciatic nerve. Here, all sprouts conducted at less than 5 m/s (mean 1.52 + 0.55 ms). The dis-
DEVOR AND GOVRIN-LIPPMANN
248
SI*R
3.
S2*R
S2*C
t-l=94 50
15
7-vd
2 IdL
z” b 5
3.
s
I5
!O
n=lo9
IL
E! 30
0
A
--r
n--l13
15 0 !JldLk.
CONDUCTION VELOCITY (m/w) FIGS. 3 AND 4. Histograms of parent fiber(S1 + R), overall (S2 + R), and sprout (S2 --, C) conduction velocity for samples of myelinated parent fibers at various times after a sciatic nerve crush. Each fiber was detected in a lower lumbar dorsal root and responded to stimulation of the tibial nerve distal to the crush. The number(N) of single axons included in each set of three histograms is given in the left column. Each histogram includes data collected from at least three rats. Note the under-representation of low-velocity parent fibers at short survival times. Arrows in the (Sl + R) histogram for control rats indicate the AC@ and AS fiber populations.
tribution of parent fiber velocities remained much higher. It was, however, not normal. In particular, the major peak was slower than in intact nerve and the minor peak was missing. We argue that this reflected two processes, the well-known retrograde reduction in conduction velocity in injured axons (1, 7), and a real under-representation of A6 fibers in the population of parent fibers that contributed sprouts distal to the crush at this survival time. Figures 3 and 4 show corresponding data for postlesion
249
SELECTIVENERVEREGENERATION
Sl+R
0
30
60
S2+R
0
30
CONDUCTION
S2+C
60
0
30
VELOCITY
FIGURE
60 (m/set)
4
survival time to as long as 481 days. Note that between about 150 and 250 days the parent fiber distribution returned to normal. That is, the faster fibers had recovered from their retrograde change and the slower ones had finally begun to appear in the record. Retrograde Slowing. Retrograde slowing of parent fibers could be assessed independently of selective fiber outgrowth by sampling fibers that responded to stimulation proximal to the lesion at S 1 without asking which of those fibers sent a sprout distally across the crush line. This was done by recording the CAP elicited by supramaximal stimulation at Sl and plotting the conduction velocity of its peak as a function of time after the crush (Fig. 5). Conduction velocity decreased to about 65% of control values by 20 to 25 days after the lesion and returned to normal by about the 90th day. Interestingly, 20 to 25 days is also the period elapsed until
250
DEVOR AND GOVRIN-LIPPMANN
3 60. J 2 50. Y n: don ; 30 . T OL
L 0
’
’ 20
’
’ 40
’
’ 60
’
’ SO
Days
’
’ 100
after
’
‘ii’
200
1 250
300
Crush
FIG. 5. Temporary retrograde slowing of sciatic nerve fibers proximal to a crush lesion. The conduction velocity of the peak of the compound action potential (CAP) elicited by supramaximal stimulation at Sl is plotted as a function of time after crushing.
the regenerating tibial nerve first makes functional sensory connections in the skin of the foot (8, 9). The reason for plotting the peak of the CAP rather than its leading edge was that this avoided the danger that the spread of current proximally along the sciatic nerve might stimulate intact fibers in nerve branches that had not been crushed. The actual plot of the leading edge of the CAP, however, closely resembles Fig. 5 shifted up the Y-axis. A sample of single fibers responding to stimulation at Sl (but not necessarily S2) in control rats and rats surviving 21 to 35 and 292 to 321 days after sciatic nerve crush is shown in Fig. 6. Percentage Fibers That Cross the Crush. For this measurement the S2 stimulating electrode was moved proximally onto the sciatic nerve about 10 mm distal to the crush site (asterisk in Fig. 1). In small strands of dorsal roots associated with the sciatic nerve we compared the number of single axons excited by stimulation at Sl proximal to the crush with the number excitable at S2 distally. As long as the strand contained no more than about seven or eight responsive fibers, each could be counted as an all-or-none addition to the multiple fiber potential as stimulation current was gradually increased. In three experiments in intact nerves, all of 157 fibers sampled responded to stimulation both proximal and distal to the site at which we crushed the nerve in the experimental animals. The same 100% success was found in three experiments totalling 202 axons 156 to 219 days after crush injury. At shorter intervals of 68 and 73 days (two rats), however, 13.6% of the 250 fibers that responded to stimulation proximally (Sl)
SELECTIVE
NERVE REGENERATION
k
251
Od n=20i
15
292-321d n=142
lilwk 0
20 40 60
Conduction
80
Velocity
(m/W FIG. 6. Histograms of parent fiber (Sl --* R) conduction velocity of dorsal root fibers that responded to stimulation of the sciatic nerve proximal to a crush lesion. The sample taken 21 to 23 days after crushing illustrates the maximal extent of retrograde slowing of conduction velocity. Note that in contrast to Figs. 3 and 4 these histograms are an unselected sample of sciatic nerve sensory fibers and do not represent only those that sent a regenerating sprout into the tibial nerve.
failed to respond to stimulation distally (S2). Most of the parent fibers that failed to send a sprout across the crush had conduction velocities in the A6 range (mean 23.4 + 8.4 m/s). DISCUSSION Do fibers of all classes produce regenerating sprouts simultaneously after nerve crush injury? Data summarized in Figs. 3 and 4 indicate that they do not. For the first 150 to 250 days after crush injury, myelinated parent fibers of low conduction velocity are relatively under-represented among the population of fibers contributing a sprout distally in the tibia1
252
DEVOR
AND
GOVRIN-LIPPMANN
nerve. However, the conclusion requires that the sampling frequency of small fibers is representative during this period and that the stimulating conditions were not such as to favor sprouts of large-diameter parent fibers. Figure 2 shows that our recording method is not fundamentally biased against small-diameter fibers. Furthermore, a sample of parent fibers driven by stimulation proximal to the crush within this period (21 to 23 days) was not deficient in fibers of low conduction velocity (Fig. 6). Finally, we found that at 68 to 73 days, 13.6% of fibers sampled failed to send a sprout across the crush line and that these were predominantly axons of low parent-fiber conduction velocity. This latter result suggests that at least part of the delay in the arrival of sprouts at S2 occurred in the vicinity of the nerve lesion. As for the adequacy of the stimulation, current was routinely increased to about twice the threshold of the highest threshold fibers that responded. This did not recruit more units. In a few experiments stimulation of as much as 10 mA at 0.5ms pulse duration was used and this too failed to recruit additional myelinated parent fibers although unmyelinated parent fibers now appeared. We conclude that in crushed sensory nerves, sprouts of rapidly conducting, large-diameter parent axons tend to grow out earlier than those of slowly conducting, small-diameter axons. A hint of this result was noticed previously by Burgess and Horch [(5); see their Fig. 11. In motor nerves, a number of authors observed that reinnervation of muscle by y fibers required much longer than reinnervation by (Yfibers (2, 3, 21, 22). Do all parent fibers maintain a regenerated branch? Because we failed to find parent fibers proximal to a crush lesion of long standing that did not have a branch distally we conclude that all parent fibers eventually establish at least one stable branch. Morphological studies indicate that by 150 days after nerve crush the number of myelinated axon profiles distal to the lesion site equals that proximally (12), although some investigators reported a moderate excess [reviewed in (17)]. Combined, these data indicate that most parent fibers have one but only one branch even though many emit more than one initially (18,20). Some parent fibers are thought to maintain more than one branch after nerve transection (IO), perhaps in compensation for those that fail to maintain any. The histologic equivalent of the electrophysiologic technique we used would involve staining individual axons and following changes in their caliber for substantial distances. Though recent technical advances may soon change matters (4), this has not been practical until now. Progress in our understanding of the functional selectivity of peripheral axons (6) makes it important to describe regeneration as a function of parent fiber class and not simply as a mass phenomenon. Selectivity in the rate and
SELECTIVE
NERVE REGENERATION
253
ultimate success of regeneration must be taken into account in interpreting the clinical pattern of sensory return after nerve injury (13, 23). REFERENCES 1. ACHESON, G. H., E. S. LEE, AND R. S. MORRISON. 1942. A deficiency in the phrenic respiratory discharges parallel to retrograde degeneration. J. Neurophysiol. 5: 269213.
BESSOU, P., Y. LAPORTE, AND B. PAGES. 1966. Observations sur la r&innervation de fuseaux neuro-musculaires de chat. C. R. Sot. Rio/. (Paris) 160: 408-411. 3. BROWN, M. C., AND R. G. BUTLER. 1974. Evidence for innervation of muscle spindle intrafusal fibres by branches of alpha motoneurons following nerve injury. J. Physiol. 2.
(London) 4.
5. 6. 7. 8. 9.
238: 41P-43P.
BROWN, A. G., P. K. ROSE, AND P. J. SNOW. 1977. The morphology of hair follicle afferent fibre collaterals in the spinal cord of the cat. J. Physiol. (London) 272: 779-797. BURGESS,P. R., AND K. W. HORCH. 1973. Specific regeneration ofcutaneous fibers in the cat. J. Neurophysiol. 36: lo!- 114. BURGESS, P. R., AND E. R. PERL. 1973. Cutaneous mechanoreceptors and nociceptors. Pages 29-78 in A. IGGO, Ed., Handbook of Sensory Physiology; Vol. 2: Somatosensory System. Springer-Verlag, Berlin. CRAGG, B. Cl., AND P. K. THOMAS. 1961. Changes in conduction velocity and fibre size proximal to peripheral nerve lesions. J. Physiol. (London) 157: 315-327. DEVOR, M., AND R. GOVRIN-LIPPMANN. 1979. Maturation of axonal sprouts following nerve crush injury. Exp. Neural., submitted. DEVOR, M., D. SCHONFELD, Z. SELTZER, AND P. D. WALL. 1979. Two modes of cutaneous reinnervation following peripheral nerve injury. J. Comp. Neurol. 185: 211-220.
10. FULLERTON, P. M.,ANDR. W. GILLIATT. 1965. Axonreflexinhumanmotornervefibers. J. Neural. Neurosurg. Psychiat. 28: 1- 11. 11. GUTH, L. 1956. Regeneration in mammalian peripheral nervous system. Physiol. Rev. 36: 441-478. 12.
13. 14. 15. 16. 17. 18. 19.
GUTMANN, E., AND F. K. SANDERS. 1943. Recovery of fibre numbers and diameters in the regeneration of peripheral nerves. J. Physiol. (London) 101: 489-518. HEAD, H. 1920. Studies in Neurology. Oxford Univ. Press, London. HUNT, C. C. 1954. Relation of function to diameter in afferent fibers of muscle nerves. J. Gen. Physiol. 38: 117-131. HURSH, J. B. 1939. Conduction velocity and diameter of nerve fibers. Am. J. Physiol. 127: 131-139. JACOBSON, S., AND L. GLJTH. 1965. An electrophysiological study of the early stages of peripheral nerve regeneration. Exp. Neural. 11: 48-60. MIRA, J.-C. 1976. Etudes quantitatives sur la regeneration des fibres nerveuses myelinisees. II. Variations du nombre et du calibre des fihres regenerees apres un ecrasement ou une section du nerf. Arch. Anat. Microsc. Morphol. Exp. 65: 255-284. RAM~N Y CAJAL, S. 1928. in R. M. MAY, Ed., Transl., Degeneration andRegeneration in the Nervous System. Reprinted in 1959, Hafer, New York. SCHNEPP, G., P. SCHNEPP, AND G. SPAAN. 1971. Analysis of peripheral nerve fibres in animal of different body size. I. Total fibre count, fibre size and nerve conduction velocity. Z. Zellforsch. 119: 77-98. [in German]
254
DEVOR AND GOVRIN-LIPPMANN
20. SHAWE, G. D. H. 1955. On the number of branches formed by regenerating nerve fibres. Br. J. Surg.
42: 474-488.
21. TAKANO, K. 1976. Absence of gamma-spindle loop in the reinnervated hind leg muscles of the cat: “alpha muscle.” Exp. Brain Res. 26: 343-354. 22. THULIN, C.-A. 1960. Electrophysiological studies of peripheral nerve regeneration with special reference to the small diameter (gamma) fibers. Exp. Neurol. 2: 598-612. 23. WALSHE, F. M. R. 1942. The anatomy and physiology of cutaneous sensibility: a critical review. Bruin 65: 48- 112. 24. YOUNG, J. Z. 1942. The functional repair of nervous tissue. Physiol. Rev. 22: 318-374.
