effects of axotomy on of in sensory motor nerve fibres · effects onaxonsclassified according...

Journal of Neurology, Neurosurgery, and Psychiatry, 1981, 44, 485-496

The effects of axotomy on the conductionof action potentials in peripheral sensory andmotor nerve fibresT E MILNER AND R B STEIN

From the Department of Physiology, University of Alberta, Edmonton, Canada

SUMMARY Medial gastrocnemius and sural nerves in one hindlimb of the cat were transectedand prevented from regenerating. After periods ranging from 29-273 days, compound actionpotentials were recorded from axotomised and contralateral control nerves. The amplitude andintegrated area of action potentials decreased and conduction velocity slowed followingaxotomy. The area under compound action potentials generated by stimulating sensory fibresdeclined significantly faster than that generated by stimulating motor fibres. Analysis of changesin whole nerve conduction velocity distributions showed that the velocities of fast conductingsensory fibres decreased at the most rapid rate. The conduction velocities of motor fibres andslow sensory fibres declined at significantly slower rates. The loss of electrical activity in thelargest sensory nerve fibres following axotomy, may play a role in determining the faster rateat which their action potentials deteriorate.

The process of peripheral nerve atrophy followingaxotomy has been extensively studied in terms ofits morphological consequences (reviewed bySunderland'). However, less is known about theelectrophysiological changes which affect theability of axons to conduct impulses. Conductionvelocity falls as fibre diameter decreases;2 3 thetwo recover only if regenerating axon sproutsestablish functional connections with their endorgans.4 5 Prevention of functional recovery doesnot necessarily lead to complete loss of axonalfunction. Spontaneous and evoked neural activitycan be recorded long after a nerve has beensevered.6-8 Although the magnitude of the activitydeclines for some months following section of thenerve, it may eventually approach a steady-statelevel.8Do all axons in a severed nerve degenerate to

the same extent or are some fibres particularlyaffected? Indications that sensory fibres are moreseverely affected than motor fibres2 8 9-11 have beenconfirmed recently by Hoffer et al.'2 Compoundaction potentials were recorded from dorsal and

Address for reprint requests: TE Milner, Department of PhysiologyUniversity of Alberta, Edmonton, Alberta, Canada T6G 2H7.Accepted 21 January 1981

ventral roots following stimulation of normal andpreviously severed sensory and muscle nerves. Theintegrated area of dorsal root compound actionpotentials was found to decrease more rapidlythan that of ventral root compound action poten-tials over a period of approximately 250 days.Although distinguishing between sensory andmotor axons, this type of analysis did not providethe detail necessary to resolve possible differentialeffects on axons classified according to size or con-duction velocity.We decided to use conduction velocity distribu-

tions computed from compound action potentialsas a tool for examining differential effects ofaxotomy on various classes of nerve fibres. Thedetails of the method used in computing the dis-tributions is outlined in the preceding paper.13From the conduction velocity distributions theviability of different classes of axons could be de-termined at various times following axotomy.

Methods

Data were obtained from experiments conducted on26 adult cats of both sexes, nine of which were normalanimals and 17 in which the sural and medial gastroc-nemius (MG) nerves of the left hindlimb had been

485

Protected by copyright.

on Septem

ber 13, 2020 by guest.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.44.6.485 on 1 June 1981. D

ownloaded from

http://jnnp.bmj.com/

486

surgically sectioned some time prior to the experiment.Transection of the nerves was performed underaseptic conditions. The animals were initially anaes-thetised with Nembutal and maintained on Halothanefor the duration of surgery. The nerves were ligatedproximal and distal to the point of section and thencut cleanly with scissors. The MG nerve was severednear its point of entry into the MG muscle and thesural nerve distal to the midpoint of its course overthe lateral gastrocnemius muscle. The proximal stumpof the severed nerve was sutured to a Silastic sheetapproximately 1 cm square. Care was taken to preventreinnervation by reflecting the MG nerve away fromits muscle and suturing the Silastic sheet to moreproximal muscles. In the case of the sural nerve thesheet was turned over onto the nerve and sutured tothe lateral gastrocnemius muscle. In some of theearliest attempts the precautionary procedure wasnot as thorough and reinnervation did occur.Acute experiments under deep Nembutal anaes-

thesia were conducted after periods ranging from 29-273 days following initial surgery. Nerves and spinalroots of both hindlimbs were prepared for stimulationand recording. The right hindlimb served as a control.Spinal roots from the L6 to S2 levels were exposedbilaterally by laminectomy. Extensive denervation ofthe hindlimbs was required to minimise artefact dueto muscle activity when stimulating ventral roots.All branches of the sciatic nerve with the exceptionof the two nerves of interest were cut. The MG andsural nerves were then dissected free from surround-ing tissue over a length of 15-25 mm. The cat wasmounted in a stereotaxic frame with both hindlimbsextended and securely clamped at the knee and ankle.Paraffin oil pools fashioned from skin flaps bathed thespinal cord and the hindlimb nerves. Their tempera-ture was maintained at 360 +20 by radiant heat.Body temperature was kept in the same range bymeans of a thermostatically controlled heating pad.Just prior to recording, nerves were ligated and cutdistal to the ligature. Axotomised MG nerves werecut 5-10 mm proximal to the neuroma while thelonger sural nerves were cut farther from the neuroma.Recording and sampling procedures have been des-

cribed in some detail in the preceding paper.'3 Com-pound potentials from control nerves were sampledat a 20 kHz rate. Lower sampling rates (down to6 kHz) were used for compound potentials recordedfrom axotomised nerves because of their longer dur-ation due to slower conducting fibres. Generally, thelonger the time which had elapsed since axotomy thelower the sampling rate. In addition to recording com-pound action potentials from the nerves, the nerveswere stimulated supramaximally in order to recordcompound potentials from the spinal roots.

Results

Figure 1 compares the mean conduction velocitydistributions obtained from control and contra-

T E Milner and R B Stein

lateral axotomised sural nerves of seven cats inexperiments conducted 29-71 days followingaxotomy. The conduction velocity distributionswere computed as described in the precedingpraper'3 and have been plotted here in three forms.The first plot (fig lA) is a normalised histogram

on a linear scale of conduction velocity with binsof equal width. The conduction distance, thelatency to onset of the compound action potentialarid the sampling rate determine the bin width.Because of the inverse relationship between con-duiction velocity and time, conduction velocityin:rements corresponding to the sampling inter-vals of the digitised compound action potentialdecrease as the conduction time increases. Usinghistogram bins of equal width fails to take ad-vantage of the higher resolution at lower conduc-ticin velocities offered by this relationship.

In fig lB the bin width is allowed to decreaseas conduction velocity decreases providing a con-seqluent increase in the resolution of the percentageof slowly conducting fibres. This was accomplishedby a threefold increase in the number of histo-gram bins. Note that the histograms have beenplotted on a logarithmic scale of conductionve:locity and hence the bin widths appear approxi-mcately equal. The choice of a logarithmic scalewas prompted by the need for a simple method ofqualitatively comparing the relative effects ofaxotomy on fast and slowly conducting fibres, asexplained below.

iumulative histograms (fig IC) proved useful incomnparing the conduction velocity distributions ofcontrol and axotomised nerves. For any given con-duction velocity the cumulative histogram repre-serLts the percentage of fibres in the distributionwhich conduct at velocities less than or equal tothat velocity. By using a logarithmic scale of con-duction velocity it is possible to determine therelative effects of axotomy on the fast and slowlyconducting populations of fibres simply by exam-inin-g the shifts in the conduction velocity distri-but;ion. A parallel shift of the distribution to theleft: without a change in its shape would implythat both fast and slowly conducting populationshad slowed by the same relative amount. Differ-ential effects would be apparent if the shift wasnot parallel.The sural nerve conduction velocity distributions

in fig 1 show this quite clearly. From the left handside of fig 1A and lB it is evident that the controlconduction velocity distribution is bimodal. Thesanie information is conveyed by the plateau inthe cumulative distribution of fig IC which, inaddition, clearly shows that approximately 40%


on Septem


j.com/

J Neurol N


ownloaded from


Effects of axotomy on sensory and motor conduction velocities

20 40 60 80 100

5 10 50 100 5 10 50 100Conduction velocity (m/s)

Fig 1 The mean conduction velocity distributions for control (left) and axotomised(right) sural nerves 29-71 days following axotomy plotted as: A conventionalhistograms on a linear scale of conduction velocity; B conventional histograms ona logarithmic scale of conduction velocity; C cumulative distributions on a

logarithmic scale of conduction velocity. Refer to text for explanation.

of the fibres were slowly conducting fibres and60% were fast conducting. Within the first twomonths following axotomy (right hand side offig 1) the bimodal nature of the distribution waslost and the plateau disappeared from the cumula-tive distribution.The loss of the plateau could have been due to

a greater slowing of fast conducting fibres (fillingthe gap between the two populations), a relativeloss of slowly conducting fibres or a combinationof the two. By superimposing the cumulative dis-tributions of the control and axotomised nerves

(fig 2 (top)) one can see that there was a clear

decrease in the conduction velocity of the fastestconducting fibres (the leftward arrow indicatesthat the 80% point has shifted from about 60 m/sto 50 m/s). However, the two distributions actuallycross over which suggests that whereas initially40% of the fibres were slowly conducting, therelative number has decreased to about 20%(diagonal arrow).The differences may have been exaggerated by

a tendency for the single unit potentials recordedfrom slowly conducting fibres to be diphasic even

under monophasic recording conditions.'4 Slowlyconducting fibres sometimes had a substantial

A.

020 40 60 80 100B

25 -

20

c15O

w 10a-

5

10

8'or

cC

IU, 4'tL

2'

0

100'

1-

@ 80.O-

a

g 60E

, 40

C2bu 200-

0

-j

P-

487


on Septem


j.com/

J Neurol N


ownloaded from



8

S

4

2

>

U)8'C1

tn3 E

-0

aC

0~

8

2

30

S0

0

0

01O

30

10

'O

O

105-146 days

10 198-273 days

30

10 -

0 1 .,P . .,5 10 50 100Conduction velocity (m/s)

Fig 2 Progressive changes in the conduction velocitydistribution of sural nerves following axotomy.Conduction velocity distributions for axotomisednerves (X) examined in the indicated time periodswere averaged and superimposed on the meanconduction velocity distributions of the correspondingcontrol nerves (+). Refer to text for discussion.

negative phase while fast conducting fibres showedrelatively less negativity. As a result, there mayhave been some cancellation of positive and nega-tive phases among slowly conducting fibres ofslightly different conduction velocities, producinga compound action potential which underestimatedthe relative number of slowly conducting fibres.With increasing time following axotomy there

was a progressive decrease in the conduction vel-ocities of all nerve fibres (fig 2). The decline wasquantified for fast conducting fibres and slowlyconducting fibres respectively by determining theconduction velocities below which 80% and 20%of the total number of fibres were represented inthe conduction velocity distribution. An estimateof their initial rates of decay following axotomywas obtained by fitting the data with an exponen-tial decay curve of the form v=v0e-t/T where v0 isthe control conduction velocity and T the timeconstant. Fitting the points with a curve of theform v=v,e-t/T+v,, where v,+v2 is the controlconduction velocity and v2 the asymptotic value,gives an estimate of the endpoint of the decay

100-

50

20 J

0 30-

>1

u 02

c0

j 10

100

80

60

40

20

0-

AFast conducting fibres

v

BSlow conducting fibres

M(05

Fast conducting fibres

0 50 100Days after axotomy

- MG(m)

MG (s)S50200 250 300

150 200 250 300

Fig 3 Conduction velocity decay curves. A and B:plots of v=voe-t/T on a semi-log scale for fast andsloivly conducting fibres: sural (()), MG sensory (A)and MG motor (V). C: plots of v=v1eC/T+v2 on alinear scale for fast conducting fibres. The latter are notshcwn for slowly conducting fibres, since v2 was notsigntificantly different from zero, ie the secondeqL1ation reduced to the first. Note that the rate ofcorduction velocity decline for fast afferent fibres inA (S and MG(s)) is greater than for fast motor fibres(MG(m)) or for any of the slowly conducting fibres inB. Regression curves were computed from individualdata points measured in all 26 experiments, but onlyaveraged data appear in this and subsequent figures(6-8) for clarity. Each symbol represents the meanvalue from several experiments, grouped into the timeintervals shown in fig 2. Regression parameters arelisted in table 1.

process (fig 3). Regression curves were calculatedusinlg all data points, but for the sake of clarityonly the averages of groups of points (grouped intothe same time intervals as figs 2, 4, and 5) have

488


on Septem


j.com/

J Neurol N


ownloaded from



been plotted in fig 3. The method of obtaining theregression curves is described in the Appendix.

It was felt that curves of the form v=voe-t/Tgave a better estimate of T than those of the formv=vle-t/T+v2 since the latter required the intro-duction of a third parameter v2, which influencedthe accuracy with which v1 and T could be de-termined. Furthermore, v2 is really the limit of theconduction velocity as t becomes very large. Sincethe range of observations extended only to 273days the conduction velocities may have declinedfurther had the nerves remained in their axotom-ised state for a longer period of time. Values ofthe regression parameters are listed in table 1.The conduction velocity of the fast conducting

fibres declined significantly faster than that of theslowly conducting fibres (two-sided t-test,p<00005). However, while the fast conductingfibres had an asymptotic conduction velocity v2,which was significantly greater than zero (t-test,p<00005), the slowly conducting fibres tendedtowards a final velocity near zero. However, thecomputed time constant was so large that it wasbeyond the period of the experimental obser-vations. Hence, it cannot be concluded that these

3E3

UN

.n

0R0C

0u

MG sensory29 -71 days

Ej

cno~~~~~~~~~~~~~~~

0

0~

Fig 4 Progressive changes in the conductionvelocity distribution of the sensory fibres in MG nervesfollowing axotomy. Conduction velocity distributionsare plotted as in fig 2. Refer to text for discussion.

fibres actually all died or stopped conducting.The MG nerve was separated into its sensory

and motor components by stimulating either dor-sal or ventral roots. The control conduction vel-ocity distribution of the sensory component wasalso bimodal, although to a lesser extent than thesural nerve. The fast conducting afferent fibresof the MG were approximately 50% faster thanthose of the sural. The tendency toward a uni-modal conduction velocity distribution followingaxotomy was much slower in the MG than thesural (fig 4). This may simply reflect a relativelygreater difference between the conduction vel-ocities of fast and slow MG afferent fibres, sincethe corresponding rates of conduction velocitydecline did not differ significantly from those ofthe sural. The same progressive reduction of con-duction velocity for all fibres was evident in theMG and the sural nerves. The fast conductingfibres slowed significantly faster than the slowlyconducting fibres (two-sided t-test, p<001) butapproached an asymptotic value which was sig-nificantly greater than zero (t-test, p<00005). Thefinal value projected for slowly conducting fibreswas near zero.

MG motor100 29-71 days80

60 -

40

20

100- 105-146days

80 -

60 -

40 -

20

O'100 203- 273 day

80 -

60

40

20-

5 10Conduction velocity (mis)

Fig 5 Progressive changes in the conduction velocitydistribution of the motor fibres in MG nervesfollowing axotomy. Conduction velocity distributionsare plotted as in fig 2. Refer to text for discussion.

489


on Septem


j.com/

J Neurol N


ownloaded from



The MG efferent fibres presented a differentpicture. While the conduction velocity distributionwas distinctly bimodal (a clear separation of alpha-motoneurones and gamma-motoneurones), thenature of the distribution did not change followingaxotomy (fig 5). Because the difference in conduc-tion velocities was so great, the fast conductingfibres approached their asymptotic conduction vel-ocity before they reached the slowly conductingfibre range. Although there was no significantdifference between the time constants for the con-

duction velocity decline of fast and slowly conduct-ing fibres, the time constant for slow fibres was

more than twice that for fast fibres (table 1). Thefailure of this difference to be significant is a re-

flection of the large uncertainty in the value ofthe time constant for slowly conducting fibres. Thisuncertainty was partially the product of an EMGartefact which occurred in a number of experi-ments. This artefact was recorded when stimulatingthe ventral roots of animals in which all tail andgluteal muscles had not been successfully dener-vated. In a few experiments the MG nerve wascrushed after recording in order to measure theartefact in isolation. It was found to be opposite insign to the neural potential and to occur with alatency corresponding to that of gamma-moto-neurones in control nerves. The artefact was small,but it tended to reduce the amplitude of the latercomponents of the motor compound action poten-tial. Hence, it probably had a more substantialeffect on the potentials of axotomised nerveswhich were smaller and more slowly conductingthan those of control nerves.The asymptotic velocity of the fast conducting

efferent fibres was significantly greater than thatof the fast conducting afferent fibres in the MGnerve (two-sided t-test, p<0 0025). This confirmedthe earlier finding of Hoffer et all2 that the fastest

conducting afferent fibres were affected more byaxcltomy than the fastest efferent fibres. The rateof slowing was also significantly less in efferentfibres (two-sided t-test, p<0 0025). It thereforeappears that degeneration progresses faster andmay also continue longer in fast conducting affer-ent fibres than in alpha-motoneurones.

Blecause of the degree of variability in the con-duc:tion velocities for the 20% level of efferentfibres, it is difficult to infer whether gamma-motoneurones were affected to a lesser extent thanalp:ha-motoneurones or slowly conducting afferentfibres.As noted by Hoffer et al," the integrated area

uncder a compound action potential (referred to ascharge because of its dimensions) was dramaticallyreduced following axotomy. However, it proved tobe less straightforward to quantify compoundaction potential charge decay than conductionvelocity decline. There was inevitably some pro-gressive deterioration of compound action poten-tialS during recording which could produce a re-duction in amplitude without significantly alter-ing the shape. Consequently, it was possible toextract information about changes in the relativedistribution of conduction velocities (determinedby the shape of the compound action potential)more reliably than changes in the absolute numberof conducting fibres (determined by the magnitudeof the compound action potential). Moreover, theabsolute number of axons and hence the magni-tud- of the compound action potential varied frompreparation to preparation.

Analysis similar to that for conduction velocitywas carried out for compound action potentialcharges recorded both from the spinal roots andthe peripheral nerves. Charge was plottedaga:inst time after axotomy and the values werefitted with curves of the form Q= Qoet/T and

Table 1 Regression parameters for conduction velocity decay curves

Nerve Fibre type N Control velocity (mls) Time constant(days) T±SE

log (V,)±SE Vs

Sural Slow 36 1-3168±0-0829 20-74 556±101Fast 36 1-7946±0-0517 62-32 264± 14

MG(s) Slow 34 1-4469±0-1413 27-98 424±107Fast 34 1-9681±0-0645 92-92 249± 17

MG(m) Slow 30 1-4496+0-1688 28-18 973±761Fast 30 1*9529±0-0768 89-72 400 1 59

Nerve Fibre type N Control velocity (m/l) Time constant Asymptotic velocity(VI+ Vo)±SD (days) T±SD (mls) V2 ±SD

Sural Fast 36 63-87±6-99 119±30 20-55±4-98MG(s) Fast 34 97-14±7-16 73±16 37 54±4 99MG(m) Fast 30 95 97±5-67 38±12 56-74±3-80

N=number of cases, SE=standard error, SD=standard deviation, s=sensory mr=motor.

490


on Septem


j.com/

J Neurol N


ownloaded from



Q=Qleet/T+ Q2, where QO and Q1+Q2 are thecontrol charge values and Q2 is the asymptoticcharge value (table 2).Comparison of the MG sensory and motor com-

pound action potentials, recorded from the dorsaland ventral roots, indicated a significant differencebetween the rates of decay (two-sided t-test,p<0005), the sensory compound action potentialcharge declining faster than that of the motorcompound action potential (results for root chargeswere combined with the results obtained byHoffer et al12 in fig 6). Surprisingly though, therewas no significant difference between the ratesof decay as determined from nerve compoundaction potentials. From the relative changes inafferent and efferent conduction velocities, aslower rate of charge decay would have beenexpected for the motor compound actionpotentials.The EMG artefact may have reduced the motor

compound action potential charge by a significantamount, affecting axotomised nerves more thancontrol nerves. Since the asymptotic compoundaction potential charge was less than 40% of thecontrol value, an EMG artifact which reducedthe charge on the control nerve by 10%, for ex-ample, could have reduced that of the axotomisednerve by 25%.The rate of charge decline of sural compound

action potentials did not differ significantly fromthe corresponding rates of either MG sensory or

motor compound action potential charges, whether

recorded from the spinal roots (fig 6) or the nerve(fig 7). Although not significantly different, therate of sural charge decay was slower than thatof the MG sensory charge. This is probably a con-sequence of there being a relatively larger numberof slowly conducting fibres contributing to thesural compound action potential, along with thefact that the conduction velocity difference be-tween the fastest and slowest conducting fibres isnot as great as in the MG. The decline of suralcompound action potential charge, therefore, re-flects the rate of fast conducting fibre slowing toa lesser extent than does the MG sensory charge.Conduction velocity distribution values were

normalised, scaled by the nth power of the con-duction velocity (where n is the slope obtainedfrom the relationship between the integrated areaand conduction velocity of single unit actionpotentials'3 and summed in order to obtain thecompound action potential charge value expectedfrom a particular conduction velocity distribution.The actual charge decline as recorded fromaxotomised nerves could then be compared withthe expected decline resulting from changes in theconduction velocity distribution alone, making theassumption that the total number of conductingfibres had remained unchanged. A significant dif-ference between the rate constants of recorded andcomputed charge decay would indicate that otherfactors had contributed as well.There was a difference between these two rate

constants in all cases (fig 7), the recorded charge

Table 2 Regression parameters for compound action potential charge decay curves

Nerve Potential type N Control charge (pC) Time constant(days) T±SE

log(Qo) ± SE QO

Sural Root 33 1-8463±0-1746 70 19 180±23Nerve 34 1-9119±0-2087 81-64 155±21Computed 36 81-64* 313±19

MG(s) Root 33 1-7792±0-2015 60-15 137±14Nerve 33 1-4255±0-2825 26-64 142±26Computed 34 2664* 221±15

MG(m) Root 35 2 2216±0 1720 166-57 217±30Nerve 32 2 0485±0 1932 111-81 149±23Computed 30 111-81* 246±41

Nerve Potential type N Control charge (pC) Time constant Asymptotic charge (pC)(Qs+Qs)±SD (days) T±SD Q2±SDSural Root 33 78-58±13-17 62± 29 21-98± 9-02

Nerve 34 88 95±54-66 124±123 7-73±38 90Coniputed 36 88-95±24.5C* 195± 86 22 09±17-46

MG(s) Root 33 65 52±13*38 83± 38 9-58±+ 9-20Nerve 33 32-93± 5 03 31± 15 7 85± 3*36Computed 34 32-93± 1-14* 93± 32 9-87± 3-33

MG(m) Root 35 183-95±26-16 52± 28 69-09±17-52Nerve 32 133*51±16-90 24± 12 37-58±11-28Computed 30 133-51±22-87* 54± 30 54-23±15-69

*Initialised to control nerve values.N=number of cases, SE=standard error, SD=standard deviation, s=sensory, m=motor.

491


on Septem


j.com/

J Neurol N


ownloaded from



200

100

10

u

5" 200

150

100

50

A

V

V

vit ~~~~~~~~~MG(m)MG(s)

B

XMG(m)

MS~~~MG(s)0 50 100 150 200 250 300Days after axotomy

Fig 6 Charge decay curves for compound actionpotentials recorded from the spinal roots. A: plots ofQ=QOe'IT on a semi-log scale for compound actionpotentials recorded from L7 and SI dorsal roots whilestimulating the sural (O ) or MG (A) nerves and fromthe ventral roots while stimulating the MG nerve (V).Each symbol represents the mean from severalexperiments. Note that the decay of MG sensorycharge is faster than MG motor charge. B plots ofQ=Q1eIt/T+Q2 on a linear scale. Note that MGmotor charge approaches a higher level relative to itscontrol level than MG sensory charge. Regressionparameters are listed in table 2.

decaying at a faster rate than the computedcharge. Except for the MG sensory compoundaction potential charge, the difference was signifi-cant (two-sided t-test, p<00005 (sural), p<0025(MG motor), 0 025<p<005 (MG sensory)). Itwould have been significant there as well, hadthere been slightly less variability in the recordedcompound action potential charge.The discrepancy between the two rate constants

must be interpreted as a loss in conducting fibresfollowing axotomy. Fibres may have stoppedconducting as the result of degenerative changesfollowing axotomy or they may have stopped con-ducting as the result of trauma suffered duringpreparation for recording. Changes definitely did

100 Sural A

Computed

10 Observed

5.40 - MG sensory

u~~~~~~Computed10

0 .

UObserved

2-C

200 - MG motor

00-0Computed

Observed10l 0

Days after axotomy

Fig 7 Decay curves of the form Q=Q0e-tIT forcompound action potential charge recorded from thenerves. Observed values are compared with the valuesexpected on the basis of changes in the computedconiuction velocity distributions following axotomy.A: ;.ural: observed charge (<) and computed charge(Cl). B: MG sensory: observed charge (A) andcornputed charge (El). C: MG motor: observedcha-ge (V) and computed charge (El). Points areplotted as in fig 6A. Note that the difference betweenthe slopes of recorded and expected charge is similarfor all cases. Regression parameters are listed intable 2.

occar while recording from the nerves since com-pound action potential charge was sometimes re-duced by as much as 20-30% over periods of 15-30 minutes. This was presumably due to loss ofaxoplasm from the cut nerves combined withconcentration changes in the, intracellular spaceand the restricted extracellular space formed bybringing the nerves into paraffin oil for recording.The: nerves were ligated to minimise these changesand the compound action potentials generallystatilised, remaining relatively constant for hoursafterwards.The decline of charge during the course of re-

cordling compound action potentials occurred in

0 50 100 150 200 250 30(

492


on Septem


j.com/

J Neurol N


ownloaded from



both control and axotomised nerves to more orless the same degree. Therefore, the rate constantsof the exponential charge decay should not havebeen altered significantly. Based on this assump-tion, it must be concluded that a significant num-ber of fibres stopped conducting in response toaxotomy, although the fact that the recordednerve charge values appear to reach non-zerolimits (fig 8) implies that a certain population offibres retain the ability to conduct action poten-tials, perhaps indefinitely.

Calculations of the total number of fibres in theconduction velocity distributions support this con-clusion. Nerves were separated into three groups:control, 0-105 days after axotomy and 126-273days after axotomy. The mean number of fibres

100 rural A

80

60\ z ~~~~Computed

40

20Observed

040- MG sensory

30 -

20 -Computed

10 -

Observed

150 MG motor C

120

90 \60 - Computed

30 Observed

50 100 150 200 29D 300Days after axotomy

Fig 8 Decay curves of the form Q=Qlet'IT+Q2 forcompound action potential charge recorded from thenerves. Observed values are compared with the valuesexpected on the basis of changes in the computedconduction velocity distributions following axotomy.A: sural: observed charge (y) and computed charge(El). B: MG sensory: observed charge (A) andcomputed charge (EJ). C: MG motor: observedcharge (V) and computed charge (l). Points are

plotted as in fig 6B. Regression parameters are listedin table 2.

Table 3 Computed number of conducting nerve fibres

Nerve Control 29-105 Days 126-273 Davs

Mean±SE N Mean±SE N Mean+±SE N

Sural 1135+104 (12) 847+ 99 (8) 365+±77 (6)MG (s) 277± 24 (14) 164+ 37 (8) 156+40 (6)MG (m) 745± 84 (14) 520+107 (8) 406±90 (5)

N=number of cases, SE =standard error, s=sensory, m=motor.

computed for each group is listed in table 3. Therewas no significant difference between the controlmean and the 29-105 day mean for either the sural(two-sided t-test, 0-025<p<0-05) or the MG motorfibres (two-sided t-test, p>0-05). Only the MGsensory fibres showed a significant difference (two-sided t-test, p<0-01), but the 126-273 day meanwas significantly less than the control in all cases(two-sided t-test, p<0-0005 (sural), p<0-025 (MGmotor), p<0-01 (MG sensory)).

Discusion

The results of the present series of experimentsindicate that, while changes in conduction velocitydistributions account for a large proportion of theloss of compound action potential charge follow-ing axotomy, there is a significant loss in the num-ber of conducting fibres. Fast conducting efferentfibres degenerate less rapidly than either fast con-ducting muscle or cutaneous afferent fibres. Incontrast, there is little difference between the ratesof degeneration of slowly conducting fibres in anyof these three categories. Furthermore, fast con-ducting afferent fibres of both types degeneratefaster than slowly conducting afferents.The present study not only confirms the finding

of Hoffer et al.12 "that following axotomy largemyelinated sensory fibres are substantially moreaffected than motor fibres in the same peripheralnerves," but also shows that muscle and cutaneousafferent fibres are affected to more or less thesame degree. The observed declines in conductionvelocity and compound action potential charge areto a large extent the products of nerve fibreatrophy. As the total fibre diameter decreases, con-duction velocity slows with a concomitant reduc-tion in the amplitude and charge of a single unitpotential. This is well-known from empiricaldata'5 and is predicted on the basis of theoreticalconsiderations.'" 17 Fibre diameter distributionsdetermined from histological sections, however,do not appear to be good measures of the conduc-tion velocity distributions of degenerating nerves.Cumulative fibre diameter histograms, derivedfrom measurements of total fibre diameter, showed

-a)

493


on Septem


j.com/

J Neurol N


ownloaded from



a much less pronounced shift toward the smallerend of the spectrum than the corresponding shifttoward slower conduction velocities in the conduc-tion velocity distributions (Gillespie and Hanley,unpublished observations).Arbuthnott et al18 made a careful comparison

of conduction velocity and ultrastructural charac-teristics of nerve fibres and found that conductionvelocity varied quite linearly with axon circumfer-ence over the entire spectrum of cat sensory andmotor nerve fibres. Since axon diameter tends tobe reduced relatively more than myelin followingaxotomy (references can be found in Sunderland'),the reduction of the total diameter would be anunderestimate of the reduction in axon circum-ference. Hence, changes in total fibre diameter(axon+myelin) could underestimate the changesin conduction velocity.Cragg and Thomas3 compared changes in con-

duction velocity with changes in axon diameter ofthe largest fibres in atrophying nerves. In reana-lysing the data in their fig 5B it appears that forchanges in conduction velocity between 0% and-20% there is a relationship which is approxi-mately linear (correlation coefficient=0-84; slope=0-86). However, in the same figure there are fibreswhose conduction velocity changed by up to-45% with little further change in axon diameter.Thus, an additional mechanism must be operatingin these extreme cases, for example a pulling awayof myelin from the internodal region. A shortregion of demyelination can produce a substantialslowing of conduction or even block conductioncompletely. 19-21 These mechanisms can now beexamined experimentally, so there is promise thatthe interaction between axons and their myelinsheaths following a distal nerve lesion can beunderstood in some detail.

Hoffer et al"2 were unable to determine theextent of cell death, if any, following axotomy.Carlson et a122 claimed that there was no signifi-cant loss in the number of motor axons or cellbodies 18 months after amputation of the hind-limb, although nerve fibres in the L7 ventral rootwere reduced in diameter. In contrast, there was asignificant loss of both dorsal root fibres andganglion cells (approximately 20%) in addition toa reduction in fibre diameter. Although such aloss of dorsal root ganglion cells might accountfor some of the difference between charge com-puted from the sensory conduction velocity distri-bution and the actual recorded charge, it seemsunlikely that it could explain the discrepancy com-pletely since they saw the greatest loss among thesmaller cell bodies. Furthermore, their findings

lead one to expect that the charge recorded fromeflerent fibres should not decay significantly fasterthan that computed from the conduction velocitydistribution. Since they do not describe the ap-pearance of any of the surviving fibres, no assess-ment can be made of the ability of these fibres toconduct impulses.

Recently, Jessell et al23 have shown that thereis a 75-80% depletion of substance P from thedorsal horn following sciatic nerve section. Theysuggest that this probably reflects the degenerationof substance P containing neurones. This is con-sistent with Carlson's findings mentioned above,sirnce so far substance P has been identified onlyin the smaller diameter myelinated afferents.24Thus, although much of the decay in compoundaction potential charge is due to slowing of con-duction in atrophying fibres, the percentage whichresults from failure of conduction or cell deathor both has still not been clearly resolved.

Support for the differential degeneration of fastand slow conducting afferent fibres may be de-rived from differences in the rates of regenerationand maturation following nerve crush. Devor andGovrin-Lippmann25 have shown that fast con-ducting fibres regenerate more quickly than slowlyconducting fibres and that recovery of conductionvelocity in regenerating sprouts occurs at fasterra tes in fast conducting fibres than slowly conduct-ingy fibres.26The present study offers no way of distinguish-

ing between the roles played by trophic factorsand ongoing electrical activity in maintaining theviaebility of nerve fibres. Czeh et al27 have shownthat while disuse of muscle afferents does causea small reduction in conduction velocity, thisslcwing is substantially less than that observedfollowing axotomy. Similar disuse of alpha-motoneurones28 leads only to changes in the elec-trical properties of the cell body without affectingaxonal conduction velocity. These findings suggestthat trophic factors play a primary role in pre-serving the functionality of an axon.We feel, however, that electrical activity may be

significant in determining the differential effects ofaxotomy on various classes of axons. Fast conduct-ings afferents should experience the greatest re-duction in impulse traffic following axotomy,relative to that present in normal fibres. Alpha-motoneurones and presumably gamma-motoneur-ones continue to be excited centrally,'9 while smallafferents terminate in relatively unspecialisedendings and continue to be excited by pressure orteinperature changes in a neuroma.30 Therefore,the most dramatic effects should occur in the large

494


on Septem


j.com/

J Neurol N


ownloaded from



afferent group of fibres, which only remain activefor a short period after ligation.6 The exact natureof any trophic role played by electrical activity isnot yet known though and awaits furtherinvestigation.

Appendix

Data points were fitted with curves of the formy=Ae-Bt or y=kvn using the transformations (ln(y)=ln(A)-Bt or ln(y)=ln(k)+nln(v) respectively. Cor-relation coefficients and standard errors were calcu-lated according to standard equations for linear re-gression.31 Regression coefficients from differentpopulations were tested for significant differences bytesting the null hypothesis. Decay curves of the formAeBt+C were obtained by computing parameterswhich minimised the residual sum of squares. A non-linear regression analysis programme was employed.This programme is available in the BMDP packageof programmes for biomedical applications developedat the Health Sciences Facility of the University ofCalifornia, Los Angeles.

This research was supported by grants from theCanadian Medical Research Council and the Mus-cular Dystrophy Association of Canada to RBSand by a Canadian Medical Research Councilstudentship to TEM.

References

1 Sunderland S. Nerve and Nerve Injuries. Edin-burgh: Livingstone, 1978.

2 Kiraly JK, Krnjevic K. Some retrograde changesin function of nerves after peripheral section. QJ Exp Physiol 1959; 44:244-57.

3 Cragg BG, Thomas PK. Changes in conductionvelocity and fibre size proximal to peripheralnerve lesions. J Physiol (Lond) 1961; 157:315-27.

4 Gutmann E, Sanders FK. Recovery of fibre num-bers and diameters in the regeneration of per-ipheral nerves. J Physiol (Lond) 1943; 101:489-518.

5 Cragg BG, Thomas PK. The conduction velocityof regenerated peripheral nerve fibres. J Physiol(Lond) 1964; 171:164-75.

6 Govrin-Lippmann R, Devor M. Ongoing activityin severed nerves: source and variation with time.Brain Res 1978; 159:406-10.

7 Davis LA, Gordon T, Hoffer JA, Jhamandas J,Stein RB. Compound action potentials recordedfrom mammalian peripheral nerves followingligation or resuturing. J Physiol (Lond) 1978; 285:543-59.

8 Stein RB, Hoffer JA, Gordon T, Davis LA,Charles D. Long-term recordings from cat per-ipheral nerves during degeneration and regen-eration: implications for human nerve repair and

prosthetics. In: Jewett DL and McCarrol HR,eds. Nerve Repair: Its Clinical and ExperimentalBasis. St Louis: CV Mosby, 1979; 166-76.

9 Fleming RA. Some notes on ascending degener-ation (so-called) and on the changes in nerve cellsconsequent thereon. Br Med 1 1896; 2:918-21.

10 Bucy PC. Studies in degeneration of peripheralnerves. J Comp Neurol 1928; 45:129-59.

11 Gutmann E, and Holubar J. The degenerationof peripheral nerve fibres. J Neurol NeurosurgPsychiatry 1950; 13:89-105.

12 Hoffer JA, Stein RB, Gordon T. Differentialatrophy of sensory and motor fibers followingsection of cat peripheral nerves. Brain Res 1979;178:347-61.

13 Milner TE, Stein RB, Gillespie J, Hanley B.Improved estimates of conduction velocity distri-butions using single unit action potentials. JNeurol Neurosurg Psychiatry 1981; 44:476-84.

14 Blair EA Erlanger J. A comparison of the charac-teristics of axons through their individual elec-trical responses. Am J Physiol 1933; 106:524-64.

15 Gasser HS, Grundfest H. Axon diameters inrelation to the spike dimensions and the conduc-tion velocity in mammalian A fibres. Am JPhysiol 1939; 127:393-414.

16 Stein RB, Pearson KG. Predicted amplitude andform of action potentials from unmyelinatednerve fibres. J Theor Biol 1971; 32:539-58.

17 Moore JW, Joiner RW, Brill MH, Waxman SG,Najaar-Joa M. Simulations of conduction in uni-form myelinated fibers. Relative sensitivity tochanges in nodal and internodal parameters.Biophys 1 1978; 21:147-60.

18 Arbuthnott ER, Boyd IA, Kalu KU. Ultra-structural dimensions of myelinated peripheralnerve fibres in the cat and their relation to con-duction velocity. J Physiol (Lond) 1980; 308:125-58.

19 Koles ZJ, Rasminsky M. A computer simulationof conduction in demyelinated nerve fibres. JPhysiol (Lond) 1972; 227:351-64.

20 Bostock H, Sears TA. The internodal axon mem-brane: electrical excitability and continuous con-duction in segmental demyelination. J Physiol(Lond) 1978; 280:273-301.

21 Waxman SG. Prerequisites for conduction indemyelinated fibers. Neurology (Minneap) 1978;28(2): 27-33.

22 Carlson J Lais A, Dyck PJ. Axonal atrophy frompermanent peripheral axotomy in adult cats. JNeuropathol Exp Neurol 1979; 38:579-85.

23 Jessell T, Tsunoo A, Kanazawa I, Otsuka M.Substance P: depletion in the dorsal horn of ratspinal cord after section of the peripheral pro-cesses of primary sensory neurons. Brain Res1979; 168:247-59.

24 Hokfelt T, Johansson 0, Kellerth J-O et al. Im-munohistochemical distribution of substance P.

c

495


on Septem


j.com/

J Neurol N


ownloaded from



In: von Euler US and Pernow B, eds. SubstanceP. New York: Raven Press, 1977; 117-45.

25 Devor M, Govrin-Lippmann R. Maturation ofaxonal sprouts after nerve crush. Exp Neurol1979; 64:260-70.

26 Devor M, Govrin-Lippmann R. Selective regen-

eration of sensory fibers following nerve crushinjury. Exp Neurol 1979; 65:243-54.

27 Czeh G, Kudo N, Kuno M. Membrane propertiesand conduction velocity in sensory neurones fol-lowing central or peripheral axotomy. J Physiol(Lond) 1977; 270:165-80.

28 Czeh G, Gallego R, Kudo N, Kuno M. Evidencefor the maintenance of motoneurone properties

by muscle activity. J Physiol (Lond) 1978; 281:239-52.

29 Gordon T, Hoffer JA, Jhamandas J, Stein RB.Long-term effects of axotomy on neural activityduring cat locomotion. J Physiol (Lond) 1980;303:243-63.

30 Wall PD, Gutnick M. Ongoing activity in per-

ipheral nerves: the physiology and pharmacologyof impulses originating from a neuroma. ExpNeurol 1974; 43:580-93.

31 Edwards AL. An Introduction to Linear Regres-sion and Correlation. San Francisco: WH Free-man, 1976.

496


on Septem


j.com/

J Neurol N


ownloaded from


effects of axotomy on of in sensory motor nerve fibres · effects onaxonsclassified according...

Documents