Ascending collaterals of cutaneous neurons in the fasciculus gracilis of the cat

Brain Research, 117 (1976) 1-17 © Elsevier/North-Holland Biomedical Press, Amsterdam - Printed in The Netherlands

1

Research Reports

A S C E N D I N G C O L L A T E R A L S OF C U T A N E O U S F A S C I C U L U S G R A C I L I S OF T H E CAT

NEURONS

IN

THE

K. W. HORCH, P. R. BURGESS and D. WHITEHORN*

Department of Physiology, University of Utah College of Medicine, Salt Lake City, Utah 84132 (U.S.A.)

(Accepted March 29th, 1976)

SUMMARY Primary sensory neurons with myelinated axons in the sural nerve of the cat were found to be divisible into 3 systems on the basis of the length of their central collaterals in the dorsal columns. The short system consists of neurons that ascend only a segment or two in the fasciculus gracilis above their level of entry into the spinal cord. It is composed of all neurons with peripheral conduction velocities in the A6 range and thus includes both D hair and nociceptive neurons. Approximately 35 ~ of the A a neurons join the intermediate system and ascend 4-12 segments before leaving the dorsal columns. This system is composed of all sural type I neurons, as well as about 40 ~ of the G2 hair, 40 ~ of the intermediate field, and 50 ~ of the F2 field neurons in the nerve. Those nociceptive neurons conducting at A a velocities also contribute to the intermediate system. The remaining G2 hair, intermediate field, and F2 field neurons, together with almost all the sural type II, Ga hair, intermediate hair, and F1 field neurons, join the long system and ascend to the nucleus gracilis. Fibers in the intermediate system showed a relatively abrupt decrease in conduction velocity usually of 50 ~ or more (median 71 ~ ) a few millimeters rostral to their entry into the spinal cord. Members of the long system also decreased in conduction velocity at this point, but the magnitude of the change was typically less than 50 ~ of the peripheral velocity (median 3 6 ~ ) . In addition, the ascending collaterals of the long system underwent a second reduction in conduction velocity near the cervical enlargement.

* Present address: Department of Physiology and Biophysics, University of Vermont College of Medicine, Burlington, Vt. 05401, U.S.A.

INTRODUCTION Several different types of mechanoreceptors can be distinguished in mammalian skin on functional and morphological grounds (for a recent review, see ref. 7). This report is concerned with the way in which the central collaterals of neurons innervating different types of receptors in cat hairy skin are distributed within the fasciculus gracilis. In an earlier series of experiments 34, afferent neurons with collaterals reaching the nucleus gracilis were identified. It was also shown at that time that almost all the ascending collaterals from hindlimb neurons that reach the 8th thoracic segment (Ts) continue on to the medulla. However, a substantial number of hindlimb cutaneous neurons with myelinated fibers do not have collaterals that extend to Ts. The present paper examines how far rostrally the central collaterals of these non-projecting neurons extend in the fasciculus gracilis. Determinations were also made of the decreases in conduction velocity that the central processes of cutaneous neurons undergo as they enter the spinal cord and ascend in the dorsal columns. Such measurements are of interest because the magnitude of a fiber's decrease in conduction velocity may reflect the amount of branching occurring at that level z~. The data presented here serve as a foundation for the paper that follows, in which the effect of peripheral nerve section on the ascending collaterals is examined. METHODS To determine how far rostrally the ascending collaterals of primary sensory neurons extend in the dorsal columns, electrical stimuli were delivered to the fasciculus gracilis while recording from individual fibers in hindlimb cutaneous nerves. I f an antidromic impulse was not recorded from the fiber when the fasciculus gracilis was stimulated, this was taken as evidence that the neuron had no collateral in the dorsal columns at that level. Microelectrodes were used to make the unit recordings from the nerves so that the neurons could be kept in continuity with the skin, allowing identification of their receptor properties. After the procedures had been perfected and standardized on several practice animals, recording experiments were performed on 14 cats. The animals were anesthetized with an intraperitoneal injection of sodium pentobarbital. Anesthesia was maintained with subsequent intravenous injections of the same anesthetic, and the animals were paralyzed with gallamine triethiodide. Expiratory CO2 content was monitored and maintained between 4 and 5 ~ by positive-pressure artificial respiration. Carotid arterial blood pressure was measured and had a mean value in excess of 75 mm Hg in all experiments. Rectal temperature was maintained between 37 and 39 °C with external heat. Anesthetic depth was evaluated after the paralyzing drug was given by observing the blood pressure and the size of the pupil; the adequacy of these measures was confirmed by occasionally allowing the effect of the paralyzing drug to wear off. The sciatic nerve was exposed in the upper thigh, and the sural nerve was exposed about halfway between the knee and the ankle. Skin flaps were raised at both sites

to retain the mineral oil used to prevent drying of the tissues. A bipular stimulating electrode was placed on the sciatic nerve, and the sural nerve was prepared for microelectrode recording. The nerve was stabilized by placing it in a plastic trough and a 1-2 m m long opening was made in the perineurium 6. In one animal, recordings were made from the posterior femoral cutaneous nerve without a peripheral stimulating electrode. Laminectomies were performed to expose the spinal cord so that primary sensory neuron collaterals in the fasciculus gracilis could be excited electrically. The dura was left intact, and the exposed areas were covered with mineral oil. Monopolar, ball-shaped stimulating electrodes about 1 m m in diameter were placed on the intact dura near the midline of the cord dorsum. The return lead was attached to muscle near the spinal column. In some cases, several short laminectomies were made, and a single stimulating electrode was placed at each exposure. In other experiments, one long laminectomy was done, and a longitudinal array of electrodes spaced 0.5-1.5 cm apart was placed on the cord. The location of each dorsal column stimulating electrode was determined with respect to the spinal segments by post-mortem dissection'~Z. To know when the microelectrode was within recording range of a sural axon, search stimuli 100-200 #sec in duration were delivered at a rate of 2-5 stimuli/sec to the sciatic nerve while the microelectrode was advanced through the sural nerve. The stimulus was set to a value sufficient to excite fibers conducting as slowly as 5 6 m/sec. When an axon was encountered and the recording stabilized, the strength of the sciatic stimulus was reduced to about 1.5 times threshold for that particular fiber and the latency of the potential noted so that the conduction velocity of the fiber could be calculated. In the first 8 experiments, stimulus-response latencies were determined visually from the oscilloscope display. Later, the recording apparatus was linked to a P D P 9 digital computer, which was programmed to measure the latency (to a fixed point on the rising phase of the spike) for 10 successive sciatic shocks and to store this data on digital tape. Once the sciatic latency had been measured, negative 100-200 /~sec duration rectangular stimuli were applied to the most caudal cord electrode at a frequency of 2-5 stimuli/sec. I f the collateral was not excited by the strongest stimulus available, the duration of the stimulus was gradually increased until the axon was excited or the stimulus duration reached 400-500 #sec. Except for the very highest threshold collaterals, the stimulus was increased to a value somewhat beyond that needed to reliably produce an antidromic impulse in the sural axon; and the latency was measured for several successive stimuli (10 when the computer was used). The procedure was then repeated for the remaining cord electrodes or until the collateral had failed to be excited by two adjacent electrodes. The stimuli were coded so that the computer automatically kept track of the level being stimulated. The experimental methods were the same in the single experiment on the posterior femoral cutaneous nerve, except that there was no peripheral stimulating electrode and the search stimulus consisted of stroking the skin of the posterior thigh. In order to confirm that the nerve impulses observed were due to direct activa-

tion of an ascending collateral, the response was examined for latency fluctuations and the fiber was stimulated at 50-100 stimuli/sec through the most rostral electrode to which it responded. Latency fluctuations of more than 0.5 msec or failure to follow this frequency were taken as indications of synaptic activation of the fiber (e.g., a dorsal root reflex, see ref. 35), and the projection pattern was reexamined. Use of the computer allowed further control of this factor after the experiment was completed by examination of the printout of the measured latencies. Since 10 values were taken for each fiber at each level, 0.5 msec latency fluctuations were easily detected by this method. Of over 1000 fibers tested in these experiments and those reported in the next paper, only 4 instances of apparent non-direct stimulation were found that had not been detected during the experiment by visual observation. Once the rostral course of the neuron in the fasciculus gracilis had been examined, the receptive field was located and studied using methods already describedT, s so that the sensory properties of the neuron could be related to its central projection pattern. Conduction distances were measured at the end of the experiment with a string laid along the neural path between the various stimulating electrodes and the recording site. RESULTS

Overall distribution of collaterals in the fasciculus gracilis On the basis of preliminary experiments, 4 levels were selected for spinal cord ACF: 5.. 16

1-60

4:E:3

D I

308

S

T

A

259

N C E

2-26

102

I

~'

ET

5

I

I

I0

15

I

.720

j

3.'5

LATENt ","

Fig. 1. Plots of distance (mm) versus latency (msec) for impulses conducted antidromically by sural afferent fibers. All 60 fibers recorded from in this experiment are included. The levels labeled on the distance axis correspond to electrode locations, in increasing order: sciatic nerve, and spinal segments LG, Lm T~3,and C3. The slope of the lines is the mean conduction velocity in m/sec between adjacent electrodes.

stimulation that were considered most likely to reveal the gross central distribution of sural myelinated sensory neurons. These levels were L6, L3, T13, and C3. Short laminectomies were made, and a single stimulating electrode was placed on the dorsal columns at each site. Experiments were done on 6 cats with these placements, and a total of 340 neurons were identified as to receptor type and tested for central projection. In Fig. 1, plots of conduction distances versus response latencies for all 60 neurons studied in one animal are shown on the same axes. On the ordinate are indicated the distances from the recording electrode to the 5 stimulating electrodes. The latencies of the antidromically conducted impulses are plotted on the abscissa without correction for stimulus utilization time. The slope of a line segment represents the average conduction velocity of that fiber between the two levels delimiting the ends of the segment. The segments between 0 and 102 m m thus show peripheral conduction velocities, and the expected division into Aa and A6 fibers is evident iv. Many of the 6 neurons could not be excited from the spinal cord electrode near L6, and those that could had conduction velocities between the sciatic electrode and the most caudal cord electrode that were less than between the sciatic electrode and the recording electrode in the sural nerve. It cannot be determined from measurements of this sort where the decrease in conduction velocity occurred; only that it occurred somewhere between the two electrodes. With the use of more closely spaced electrodes that also stimulated the dorsal roots (see below) and with a movable stimulating electrode that tracked the fibers in the roots and through their entry into the spinal cord, it was found that the first measurable decrease in conduction velocity occurred 4-5 m m rostral to where the central process entered the spinal cord. Since most sural neurons enter the spinal cord through the L7 and S1 roots 15, it is clear that many A6 neurons in the sural nerve did not have collaterals that extended for more than a segment or two above their level of entry; and none extended as high as L3. The A a neurons had a different pattern, since almost all projected for at least several segments up the cord (Fig. 1). However, like 6 neurons, they undergo a conduction velocity decrease about 5 m m rostral to their entry into the cord. The patterns of projection can be most easily understood when correlated with the receptive properties of the neurons. The receptors were characterized by their responses to hand-held stimulators with the hair full length, following the classification scheme of Burgess and Perl 7. There is some arbitrariness in the way that the continuum of properties in the hair and field groups is subdivided, and the intermediate hair (IH) and intermediate field (IF) categories include all hair and field receptors not clearly belonging to the phasic (G1, F1) or tonic (G2, F2) ends of the continua. The relative projection frequencies of neurons innervating the major types of sural mechanoreceptors are shown in Fig. 2. According to these data, which were combined from 6 animals, 49 ~ of the D hair neurons could be excited from the L6 electrode, but no D hair neurons projected as far as La. At the other extreme, almost all G1 hair, intermediate hair, FI field, and type ]I neurons projected to the medulla. As presently classified, about 60 ~ of the G2 hair, 60 ~,; of the intermediate field, and

50 ~ of the F2 field neurons projected the length of the spinal cord. Hence, the more tonic neurons in the hair and field groups are less likely to project to the medulla than are neurons transmitting more phasic information, in agreement with earlier findings 34. Type I neurons constitute a distinctive category among those with peripheral

GI

DH

CI-

IH

G2

/// ///

/// ///

i.i.I

/// /// /// /// /// ///

113[.I

/// ///

L7 100% 50% 0%

/// ) I

J

i i

I i

)

i

i

i

i

i

)

i

i

i

RELATIVE NUMBERPROJECTING

T2

TI

CI-

IF

FI

F2

V / / '

/// /// /// /// /// /// /// ///

IJJ L~ O~ _J

/

TI3kl

/

/

J ,/ ///

// //// /

1_7 100% 50 % 0%

't I

I I

~"

I ~"

I

I

I I

I

)

/

/

I

I

I

1 I

I"

I

I I

RELATIVE NUMBER PROJECTING

Fig. 2. Percentages of sural mechanosensitive neurons projecting to the L6, L3, T13, and C3 spinal segments. The upper portion of the figure shows D hair (DH) neurons; and guard hair neurons with phasic (G1), intermediate (IH), and more tonic (G2) properties. The lower portion of the figure shows field neurons with phasic (F1), intermediate (IF), and more tonic (F~) properties, and type I (TI) and type II (T2) neurons. The number of neurons of each type tested is given in the legend of Fig. 5.

MYS-$4-24

A.

G~NEURON

D.

sz

MYS-SI-23

.~'"/

321

2Z4B I / ~

.......

:zo

0 2 4 36-

8 I0 12 14

/

E24"

O I I

I

3

0 I

i

5

7 9 II 13 15 17 19 21 23 i

i

i

i

~

i

i

36.E

B IO "2 14

0I

LATENCY (msec) MY5-SB-6

3 5

L2 entry

II 13 15 17 I9

?

i

25

NOCICEPTIVE NEURON

I

E

i

LATENCY (msec)

=50m~ec L5..... 3 24~ . _I _ , , ~ - - ~,,xI~1 / @=~"

~'°

44 C

( ~ ) : 68 m/~e¢

MYS-SI-18 T1en,,y 3 2 8 3 2 ~

//

32-2B.

J @ P

~

6 LATENCY(msec) IdYS- $7-27 F2 NEURON

O.

L2enlry

- 1t2/t / e

:84m/see (~),,0. . . . .

t2] /

20i 6£ ~2 o1

TYPE I NEURON

9 LATENCY (msec)

, ,o,,,

21

23 25

G2NEURON

40 36-

TIO eIy

32 28-

~ 24-

36 F.

MYS-$7-22

32 28

TI3entry

L6enlfy

~ 20-

F2 NEURON

/

1612-

/' ~

@:50~/se

6-

4-

oi

~

~

;

,;

LATENCYlmsec)

,)

,~

of

~

5

7

. . J5. 9 H 13 LATENCY (reset)

. Ir. .

J9

2~

2a

z5

Fig. 3. Distance-latency plots for sural neurons, 3 with collaterals ascending the length of the fasciculus gracilis (A- C) and 3 leaving the columns in the upper lumbar region (D-F). The lines were fitted to the points by eye. The peripheral conduction velocity of each fiber (labeled 1) was calculated between the sciatic stimulating electrode and the recording site in the sural nerve. The other points were obtained from an array of electrodes on the spinal cord, the most caudal of which in A, D, and E excited the neuron in the dorsal roots. Segmental levels are specified by a line passing through the root entry zone halfway between its rostral and caudal boundaries. Each neuron's receptor type is indicated above the corresponding plot.

A a fibers, since they are the only ones that consistently fail to reach the cervical spinal cord. However, these neurons do not terminate immediately after entering the cord. Rather, most extend rostrally for several segments to leave the fasciculus gracilis between L3 and T13 (Fig. 2). The G2 hair, intermediate field, and F2 field neurons that do not reach cervical levels also ascend the spinal cord varying distances before leaving the dorsal column pathway. However, they disappear from the columns in a more uniform manner than type I neurons, leaving in about equal numbers between L6 and L3 and between La and T13 (Fig. 2).

The lumbar and lower thoracic fasciculus gracilis In order to study local distribution patterns in greater detail, as well as the conduction velocity changes associated with a fiber's entry into the spinal cord, 6 experiments were done in which only the sacral, lumbar, and lower thoracic (up to TT) portions of the dorsal column pathway were stimulated with electrodes spaced 0.5-1.5 cm apart. Fig. 3 shows distance-latency plots for 6 of the 166 neurons studied in this fashion. The central processes of all the neurons decreased in conduction velocity shortly after entering the spinal cord. The change in conduction velocity was least in the case of the neurons that projected the length of the spinal cord (Fig. 3A-C); and after the reduction near the entry zone, the conduction velocity remained unchanged as the fibers traversed the upper lumbar and lower thoracic portions of the cord. In constrast, the non-projecting neurons showed large decreases in conduction velocity just rostral to the entry zone (Fig. 3D-F). This was particularly dramatic in the case of the neuron in Fig. 3E, which had a fine ascending collateral that extended to about L4 before it could no longer be activated. This was a myelinated nociceptive neuron that could be excited by firm pressure (moderate pressure receptor, see Burgess and Perl6;) and as is usually the case for the lower threshold members of the myelinated nociceptive group, the peripheral conduction velocity was in the a range. No nociceptive neurons with peripheral axons conducting at 6 velocities were observed to project more than 2-3 segments above their level of entry; in this respect, they resembled D hair neurons. However, sural nociceptive neurons with peripheral conduction velocities between 40 and 65 m/sec, all of which were of the moderate pressure type, regularly projected to upper lumbar levels.

t B" 38

f A 38 ~5 [

rn~

[

~

I

'

I

~' 20 40 60 80 CONDUCTION VELOCITY(m/sec)

'

I

,, 20

, , -~---,,r--]~ 9¢ ~, 40 60 REDUCTION (%)

80

I00

Fig. 4. A shows the 'final' conduction velocities of 38 fibers determined with an array of electrodes spaced 0.5 1.5 cm apart that spanned the point where the fiber could no longer be excited. Velocities were calculated from the latency difference between the most rostral electrode that excited the fiber and the one just caudal to it and the measured distance between the two electrodes. The median final velocity is 8 m/sec (arrow). In B, each of the conduction velocities shown in A has been subtracted from the peripheral conduction velocity of the same neuron and the difference expressed as a percentage of the peripheral conduction velocity. The arrow indicates the median reduction of 87 ~.

A two-stage reduction in conduction velocity was common for fibers leaving the dorsal columns in upper lumbar and lower thoracic segments, and an example is shown in Fig. 3F. This F2 field neuron underwent the usual marked reduction in conduction velocity shown by non-projecting neurons near the entry zone. The ascending collateral, already quite fine, decreased further in diameter as it traversed the L3 segment; and the final portion continued for about 3 segments conducting at 5 m/sec. A 3-stage decrease in conduction velocity was mapped in a few cases. In all, the conduction velocities of 38 fibers were measured in the fasciculus gracilis just caudal to the level where they could no longer be excited (Fig. 4). The median was 8 m/sec, but a few fibers conducted more rapidly than 20 m/sec just prior to their apparent departure from the dorsal columns. The conduction velocity data from the closely spaced electrodes were compared with similar determinations made with the coarser electrode placements. Fig. 5 shows the conduction velocities calculated from these more separated electrodes for different types of primary neurons distinguished by their receptive properties. Non-projecting neurons are indicated with shaded squares. The more phasic hair and field neurons (G1, F1) tend to conduct more rapidly than the more tonic hair or field neurons (G2, F2). In addition, there is an obvious tendency for the non-projecting neurons of a particular type to conduct more slowly than projecting neurons of the same type after they enter the spinal cord. However, projecting and non-projecting a neurons cannot be distinguished on the basis of their peripheral conduction velocities. In Fig. 6, all projecting and all non-projecting neurons conducting at Aa velocities have been plotted regardless of receptor type. The peripheral conduction velocities of both projecting and non-projecting neurons are similar (Fig. 6C), but the non-projecting neurons tend to have lower L6-L3 conduction velocities than those that project (Fig. 6B). Thus, non-projecting neurons show a significantly greater reduction in conduction velocity upon entering the spinal cord than do projecting neurons (Fig. 6E). The conduction velocities of the non-projecting fibers continue to fall in the upper lumbar cord, whereas the projecting neurons have about the same conduction velocity between Lz and T13 as between L6 and Lz (Fig. 6D). The use of the closely spaced electrode array also allowed more precise definition of where fibers left the dorsal columns. Electrodes placed between T13 and T7 demonstrated that most of the fibers extending beyond T13 either left the columns below T9 or traversed the region without any detectable change in conduction velocity. Insufficient numbers of neurons of the different receptive types were examined with the more closely spaced electrodes over the L3-T13 portions of the cord to do more than confirm the distribution patterns already described. In the L6-Lz region, it was found that the collaterals of 6 primary sensory neurons rarely ascended more than 2-3 segments above their level of entry. On the other hand, the collaterals of the nonprojecting Aa neurons that left the dorsal columns below Lz almost always did so after ascending further than the ~ neurons could be traced and usually reached L4.

The upper thoracic and lower cervical fasciculus gracilis In almost every case tested, the conduction velocities of projecting sural neurons

I0 DH

GI

IH

J

~

G2

i

,

,

,

,~.,

i

cLr

,,

..A,, ~m

I0

o~

°o 3b do~b 3o ~

b 3'o 6'o96o 3o 6o 9o

CONDUCTIONVELOCITY(m/sec) t.) -.I

I

T2

TI ~I

,

i,

FPq,

I

f

I I

In

Fl 11

IF i

7,I--~

F2 ,

7,

I

-i

,

~ ')

I0 I

"1

I

I

I

I

II

[

--~n

i

rr~

~E

~z I~z

43 i

i

"]

i

i

I

i

, ~ r ~ ,

~,

I

I

0 30 60 90 0 3b 66 9oo 3b 6 o 9 o o 3o 60 9oo CONDUCTION VELOCITY(m/sec)

30 6o 90

Fig. 5. Peripheral and central conduction velocities of dorsal root neurons designated as in Fig. 3. Peripheral velocities were measured between a stimulating electrode on the sciatic nerve and the recording site in the sural nerve. Central velocities were determined over 3 different portions of the fasciculus gracilis: between Ln and L3 (lower lumbar), L3 and Ta~ (upper lumbar), and T13 and C3 (thoracicocervical). Shaded blocks show fibers not projecting to upper cervical levels. Data are given for 39 D hair neurons, 28 G1 neurons, 16 IH neurons, 23 G2 neurons, 57 type I neurons, 7 type II neurons, 15 F1 neurons, 20 IF neurons, and 15 Fz neurons.

were l o wer between T13 and C3 than between L3 an d T13 (Fig. 5). T o define where this decrease actually takes place, electrodes spaced 0.5-1.5 cm a p a r t were used to stimulate the fasciculus gracilis rostral to Tg. D i s t a n c e - l a t e n c y plots o v e r this region o f the cord fo r 3 different n e u r o n s with peripheral axons in the p o s t e r i o r femoral c u t a n e o u s nerve are shown in Fig. 7. In each case, the r e d u ct i o n in c o n d u c t i o n velocity occurred r a t h e r

11 A,

T,3 -L3

D,

B.~ 301 L3-L6

,

),

,

%REDUCTO IN BTOA

,

CONDVELOCT I Y D102

[i] 102 13

•

2O

cU ~ L z

3 5 - t,.

-C SO CN A IT C I,E -SLU R A 25DV O C T ILY|FG ~

E. I N ~ [] I02 O%REDUCTO C TO B

•

57

-22

m-

[ ] io2

5-

0 20 40 80 CONDUCT O I NVELOCI n/sec)

-2

REDUCTO I N(%)

80 I00

Fig. 6. Conduction velocities of projecting and non-projecting mechanoreceptive sensory neurons measured over various portions of their peripheral and central distribution. Non-projecting neurons are represented by shaded squares, projecting neurons by open symbols. All the neurons in Fig. 2 that could be excited from L3 are represented, as well as a few others for which only an approximate receptor identification could be made. C shows the peripheral conduction velocities of the neurons measured between the sciatic stimulating electrode and the recording microelectrode in the sural nerve. The median velocities (arrows) of the projecting and non-projecting neurons were not statistically significantly different (P > 0.05, median test). The conduction velocities between L3 and L6 are shown in B, and the difference between this velocity and the peripheral velocity of the same neuron is presented in E as a per cent of the peripheral velocity. The medians (arrows) indicate a significantly greater reduction in conduction velocity near the entry zone for non-projecting fibers than for those that project. A gives the conduction velocities between T13 and L3. D shows the differences between the T,3-L.~ and Lz L(~ conduction velocities of individual neurons as a percentage of the L~-L6 velocity. The median for the non-projecting neurons is 33 ~ ; for the projecting neurons, 3.8 ~. The latter reduction is not statistically significantly different from zero (P -- 0.1, t-test).

a b r u p t l y as the ascending collateral traversed a particular region o f the spinal cord. Fig. 7C shows an unusual p a t t e r n in which a two-stage r ed u ct i o n occurred, the m o r e caudal at a b o u t T5 and the m o r e rostral n e a r C8. A total o f 48 n e u r o n s were studied over the u p p e r t h o r a c i c an d cervical p o r t i o n s o f their co u r s e ; 32 with peripheral axons in the p o s t e r i o r f e m o r a l cu t an eo u s nerve in one a n i m a l a n d 16 with axons in the sural nerve in another. All showed some reduction in c o n d u c t i o n velocity as they traversed this region o f the spinal cord (Fig. 8). Th er e was no m a j o r difference in the levels at which the c o n d u c t i o n velocities c h a n g e d in these t w o animals. In both, 75 ~ o f the n eu r o n s u n d e r w e n t a r ed u ct i o n in c o n d u c tion velocity between T2 a n d C6, with the r e m a i n d e r decreasing between T5 and T2, except for one n e u r o n where the change o c c u r r e d at C4. As would have been predicted

12 from earlier work (Whitehorn et al.37), posterior femoral cutaneous neurons conducted appreciably more slowly in the fasciculus gracilis than sural neurons. DISCUSSION These experiments have provided two types of information about primary sensory neurons not previously available: (a) the magnitude and location of the conduction velocity changes that the central processes of different types of cutaneous hindlimb neurons undergo as they enter and ascend the gracile fasciculus, and (b) the levels at which the ascending collaterals of non-projecting cutaneous hindlimb neurons can no longer be excited by stimulation of the dorsal columns. The significance of the experiments thus depends on the interpretation of these two types of information, and each will be considered in turn.

Significance of conduction velocity changes of primary sensory neuron collaterals in the fasciculus gracilis There is good evidence that the relationship between fiber diameter and conduction velocity is similar in peripheral nerves and within the central nervous system 1,2, 3,19-21,'~4,30. Thus, it is reasonable to assume that fibers undergoing large reductions in conduction velocity after entering the spinal cord decrease more in diameter than fibers undergoing small reductions in conduction velocity. Since motor and sensory fibers in mammalian peripheral nerves are known to decrease in diameter as they divide lz, 14,36 (however, see Clough et a1.11), a larger decrease in diameter within the cord is likely to reflect relatively more branching of the neuron's central process. Although there may be a decrease in the thickness of a fiber's myelin sheath as it enters the spinal cord zT, we were not able to detect any change in the conduction velocity of myelinated fibers as they passed from the dorsal root into the central nervous system (see also ref. 10). Hence, any decrease in the conduction velocity of the central processes of these primary sensory neurons is probably due to branching.

Reliability of the antidromic mapping technique Although there is an obvious relationship between the changes in conduction velocity that a neuron undergoes after it enters the cord and the level at which it could no longer be excited by stimulating the dorsal columns, it is not certain that this is the level at which the neuron actually leaves the columns. While some fibers conducted very slowly before becoming inexcitable form the cord surface, others were still conducting at more than 15 m/sec (see Fig. 4). It is possible that processes of these larger collaterals actually continued to ascend the dorsal columns but that antidromic impulses were blocked at the branch points where these processes were given off. The fact that antidromic impulses can pass from very slowly conducting collaterals into large parent axons argues against this possibility but does not rule it out. Previous attempts to identify, by direct recording from the dorsal columns, fibers that had been found to be non-projecting by antidromic stimulation methods were uniformly unsuccessful~,34; but if such fibers are small, they could have been missed during

13

36-

MYS-S9-7

A

G2 NEURON

32-

C5 entry

28-

T4 entry

g24~2o-

T9 entry

c~

12

/ ( ~

(~= 20 m/sec : 14 m/~c

4-

LATENCY(mse¢) 36-

B

MYS-S9-3

FI NEURON

C5 entr~ TI entry

3228

~24.

Tgenlry

12. (~) 25 m/sec (2")= 16 m/see =

LATENCY (msec)

36 C.

MYS-S9-20

GI NEURON

28 -~ 2

TI entry ~

j

,

76 entry

I

T9entry

12 /

(~= 40 m/sec

8

(~= 28 mlsec ~ : 20 mlsec

4

0

~

~

6-

B

Ib

1'2 14 16 I'8 20

LATENCY (msec)

Fig. 7. D i s t a n c e - l a t e n c y plots in the region of the cervical e n l a r g e m e n t for 3 n e u r o n s with peripheral a x o n s in the posterior femoral c u t a n e o u s nerve. T h e lines were fitted to the points by eye. T h e G2 n e u r o n in A s h o w e d a 30 ~ reduction in c o n d u c t i o n velocity at a b o u t the T4 level. T h e FI n e u r o n in B showed a 3 6 % reduction in c o n d u c t i o n velocity near C8. The G1 n e u r o n in C u n d e r w e n t a twostage reduction in c o n d u c t i o n velocity; the first at T~, the second a r o u n d C8, for a n overall decrease of 5 0 ~ .

14 15-7 A.

20 u 140 60 REDUCTION (%l

80

15--]B.

60 80 CONDUCTION VELOCITY(m/see)

,57c.

• S0R ,6 OPOST

CUT. 2

2---~ 60 80 CONDUCTION VELOCITY(rn/sec) Fig. 8. Changes in conduction velocity occurring at upper thoracic and lower cervical levels for 32 primary sensory neurons with peripheral axons in the posterior femoral cutaneous nerve (open symbols) and for 16 neurons reaching the skin via the sural nerve (shaded symbols). C shows the conduction velocities in the thoracic cord before the reduction in conduction velocity occurred; B shows the conduction velocity rostral to the level where the reduction occurred; and A gives the per

cent reduction (the difference in conduction velocityfor each neuron between C and B as a percentage of the neuron's conduction velocity in C). Medians are indicated by arrows for the two populations in each histogram. The difference between the posterior femoral cutaneous and sural distributions is statistically significant in B and C (P < 0.001) and in A (P < 0.05) by the median test. dorsal column recording. Thus, although it cannot be stated with certainty that neurons no longer backfired by dorsal column stimulation have left or are leaving the columns, it seems likely that this is the case. Fibers leaving the dorsal columns presumably enter the underlying gray matter for the purpose of establishing synaptic connections. How far they may ascend in the gray matter, beyond the reach of our stimulating electrodes, before making synaptic contacts cannot be determined from this study 31.

Distribution of collaterals of cutaneous primary sensory neurons in the fasciculus gracilis Keeping in mind the above considerations the following scheme can be tentatively proposed regarding the central organization of myelinated cutaneous sensory neurons supplying hindlimb hairy skin. (1) Neurons with peripheral conduction velocities in the AO range, whether mechanoreceptive or nociceptive, typically extend rostrally in the fasciculus gracilis not more than 2-3 segments after they enter the spinal cord. These A6 neurons might be said to form a short gracile system. (2) A group of neurons with peripheral axons conducting throughout most of the Act range extends rostrally for 4-12 segments to upper lumbar and lower thoracic levels before leaving the dorsal columns. These Act neurons form a gracile system of intermediate length. Almost all

15 type I neurons belong to this system, as well as about 50 ~ of the Fz field, 40 ~ of the G2 hair and intermediate field, and most of the more rapidly conducting, lower threshold nociceptive neurons. (3) The remaining Aa neurons extend the length of the spinal cord to reach the caudal medulla. These neurons constitute a long gracile system. Included are almost all type II, G1 hair, intermediate hair, and F1 field neurons, about 60 ~ of the G2 hair and intermediate field neurons, and about half the F~ field neurons. Generally similar results on the composition of the long system were obtained in previous studies by Brown a,4 and Petit and Burgess a4, although Petit and Burgess incorrectly identified some intermediate hair receptors as Pacinian corpuscles. In the present experiments, only 48 ~ of the sample of Aa neurons in the sural nerve entered the long pathway. This is less than the value reported by Petit and Burgess a4 and serves to emphasize that many hindlimb cutaneous neurons with relatively large peripheral axons do not contribute to the long dorsal column system. Neurons of the long system exhibit some branching upon entering the cord and then traverse the upper lumbar and lower thoracic segments with little if any further decrease in diameter. However, they do undergo consistent decreases in conduction velocity in the region of, or just caudal to, the cervical enlargement and thus would be expected to influence neurons involved in forelimb functions. The neurons of the short system presumably exert all their actions near the level of entry. Those of the intermediate system make many of their connections at this level, as judged from the marked reduction in conduction velocity that occurs near the entry zone, but also apparently influence postsynaptic elements several segments rostral to their entry. Just what neuron populations are influenced by the ascending collaterals of the intermediate system cannot be determined from the present experiments; although by ascending several segments from the entry zone, they are potentially able to establish direct connections with the cells of Clarke's column18,25, aa. Consistent with this suggestion is the demonstration that cutaneous receptors excite the dorsal spinocerebellar tract 1a,23,a2.

Composition of the intermediate gracile pathway Our studies represent an extension to the single unit level of the type of experiment first done by Lloyd and McIntyre 26. Using whole nerve recording, they showed that the great majority of the muscle sensory neurons with rapidly conducting afferent fibers leave the fasciculus gracilis in upper lumbar and lower thoracic segments; i.e., they form part of the intermediate system. This has subsequently been verified for sensory neurons supplying hindlimb muscle spindles and Golgi tendon organs by Mclntyre 2s,29, Whitsel et al. 38, and Clark 10. Clark TM has also shown that tonic neurons supplying the capsule and ligaments of the knee joint consistently enter the intermediate system. The few articular neurons joining the long pathway were all rapidly adapting. Articular neurons of both the intermediate and long systems showed the characteristic conduction velocity decreases in the lumbar and lower thoracic segments that were found in the present study; and although more rostral levels of the cord were not examined, it is likely that articular neurons in the long pathway, like

16 c u t a n e o u s neurons, u n d e r g o a reduction in c o n d u c t i o n velocity in the forelimb region 16. T h u s , the intermediate p a t h w a y is composed of the more tonic n e u r o n s that supply the skin, muscles, a n d joints. Type 1I cutaneous n e u r o n s are the only tonic s o m a t o s e n s o r y n e u r o n s that were f o u n d n o t to j o i n this system. U n i f o r m l y absent from the intermediate p a t h w a y are n e u r o n s signaling transients (G1 hair a n d P a c i n i a n corpuscle). It is striking in the case of articular nerves how the Pacinian corpusclelike n e u r o n s consistently j o i n the long pathway 5, a n d this appears to be the case for deeper tissues generally 29. ACK NOWLEDGEMENTS This work was supported by G r a n t GB42643 from the N a t i o n a l Science F o u n d a tion a n d by G r a n t s NS08769, NS07938 a n d NS05244 from the U.S. Public Health Service. The a u t h o r s wish to t h a n k J o h n Fisher, G a r y Frederickson, T o m O ' L e a r y and Jane Burgess for their valuable assistance.

REFERENCES 1 Bishop, G. H. and Clare, M. H., Organization and distribution of fibers in the optic tract of the cat, J. comp. Neurol., 103 (1955) 269-304. 2 Bishop, P. O., Jeremy, D. and Lance, J. W., The optic nerve. Properties of a central tract, J. Physiol. (Lond.), 121 (1953)415-432. 3 Brown, A. G., Cutaneous afferent fibre collaterals in the dorsal columns of the cat, Exp. Brain Res., 5 (1968) 293-305. 4 Brown, A. G., Ascending and long spinal pathways: dorsal columns, spinocervical tract and spinothalamic tract. In A. Iggo (Ed.), Handbook of Sensory Physiology, Vol. I1, Springer, New York, 1973, pp. 315-338. 5 Burgess, P. R. and Clark, F. J., Dorsal column projection of fibres from the cat knee joint, J. Physiol. (Lond.), 203 (1969) 301-315. 6 Burgess, P. R. and Perk E. R., Myelinated afferent fibres responding specifically to noxious stimulation of the skin, J. Physiol. (Lond.), 190 (1967) 541-562. 7 Burgess, P. R. and Perl, E. R., Cutaneous mechanoreceptors and nociceptors. In A. lggo (Ed.), Handbook ¢~fSensory Physiology, Vol. 11, Springer, New York, 1973, pp. 29 78. 8 Burgess, P. R., Petit, D. and Warren, R. M., Receptor types in cat hairy skin supplied by myelinated fibers, J. Neurophysiol., 31 (1968) 833 848. 9 Chang, H.-T., Fiber groups in primary optic pathway of cat, J. Neurophysiol., 19 (1956) 224-231. 10 Clark, F. J., Central projection of sensory fibers from the cat knee joint, J. NeurobioL, 3 (1972) 101 110. 11 Clough, J. F. M., Kernell, D. and Phillips, C. G., Conduction velocity in proximal and distal portions of forelimb axons in the baboon, J. Physiol. (Lond.), 198 (1968) 167-178. 12 Cooper, S., The relation of active to inactive fibres in fractional contraction of muscle, J. Physiol. (Lond.), 67 (1929) 1-13. 13 Eccles, J. C., Sabah, N. H., Schmidt, R. F. and T~.bofikov~., H., Cutaneous mechanoreceptors influencing impulse discharges in cerebellar cortex. 1. In mossy fibers, Exp. Brain Res., 15 (1972) 245 260. 14 Eccles, J. C. and Sherrington, C. S., Numbers and contraction-values of individual motor-units examined in some muscles of the limb, Proc. roy. Soc. B, 106 (1930) 326-357. 15 Ekholm, J., Postnatal changes in cutaneous reflexes and in the discharge pattern of cutaneous and articular sense organs. A morphological and physiological study in the cat, Acta physiol, scand., Suppl. 297 (1967) 1-130.

17 16 Gardner, E., Latimer, F. and Stilwell, D., Central connections for afferent fibers from the knee joint of the cat, Amer. J. Physiol., 159 (1949) 195 198. 17 Gasser, H. S., Effect of the method of leading on the recording of the nerve fiber spectrum, J. gem Physiol., 43 (1960) 927-940. 18 Grant, G. and Rexed, B., Dorsal spinal root afferents to Clarke's column, Brahe, 81 (1958) 567-576. 19 Grundfest, H. and Campbell, B., Origin, conduction and termination of impulses in dorsal spinocerebellar tract of cats, J. Neurophysiol., 5 (1942) 275-294. 20 Hildebrand, C. and Skoglund, S., Calibre spectra of some fibre tracts in the feline central nervous system during postnatal development, Acta physiol, scand., Suppl. 364 (1971) 5 4 1 . 21 Hwang, Y.-C., Hinsman, E. J. and Roesel, O. F., Caliber spectra of fibers in the fasciculus gracilis of the cat cervical spinal cord: a quantitative electron microscopic study, J. comp. Neurol., 162 (1975) 195-204. 22 Krieg, W. J. S. and Groat, R. A., Topography of the spinal cord and vertebral column of the cat, Quart. Bull. Northwestern Univ. med. School, 18 (1944) 265 268. 23 Kuno, M., Mufioz-Martinez, E. J. and Randi6, M., Sensory inputs to neurones in Clarke's column from muscle, cutaneous and joint receptors, J. Physiol. (Lond.), 228 (1973) 327-342. 24 Lennox, M. A., Single fiber responses to electrical stimulation in cat's optic tract, J. Neurophysiol., 21 (1958)62 69. 25 Liu, C.-N., Afferent nerves to Clarke's and the lateral cuneate nuclei in the cat, Arch. Neurol. Psyehiat. (Chic.), 75 (1956) 67 77. 26 Lloyd, D. P. C. and Mclntyre, A. K., Dorsal column conduction of Group 1 muscle afferent impulses and their relay through Clarke's column, J. Neurophysiol., 13 (1950) 39-54. 27 Maxwell, D. S., Kruger, L. and Pineda, A., The trigeminal nerve root with special reference to the central-peripheral transition zone: an electron microscopic study in the macaque, Anat. Rec., 164 (1969) 113-126. 28 Mclntyre, A. K., Central projection of impulses from receptors activated by muscle stretch. In D. Barker (Ed.), Symposium on Muscle Receptors, Hong Kong University Press, Hong Kong, 1962, pp. 19-29. 29 Mclntyre, A. K., Cortical projection of impulses in the interosseous nerve of the cat's hind limb, J. Physiol. (Lond.), 163 (1962) 46-60. 30 Ogden, T. E. and Miller, R. F., Studies of the optic nerve of the rhesus monkey: nerve fiber spectrum and physiological properties, Vision Res., 6 (1966) 485-506. 31 Oscarsson, O., Primary afferent collaterals and spinal relays of the dorsal and ventral spino-cerebellar tracts, Acta physiol, scand., 40 (1957) 222-231. 32 Oscarsson, O., Functional organization of spinocerebellar paths. In A. Iggo (Ed.), Handbook c~f Sensor), Physiology, Vol. H, Springer, New York, 1973, pp. 339-380. 33 Pass, I. J., Anatomic and functional relationships of the nucleus dorsalis (Clarke's column) and of the dorsal spinocerebellar tract (Flechsig's), Arch. Neurol. Psychiat. (Chic.), 30 (1933) 10251045. 34 Petit, D. and Burgess, P. R., Dorsal column projection of receptors in cat hairy skin supplied by myelinated fibers, J. Neurophysiol., 31 (1968) 849-855. 35 Toennies, J. F., Reflex discharge from the spinal cord over the dorsal roots, J. Neurophysiol., 1 (1938) 378-390. 36 Weddell, G., The pattern of cutaneous innervation in relation to cutaneous sensibility, J. Anat. (Lond.), 75 (1941) 346-367. 37 Whitehorn, D., Bromberg, M. B., Howe, J. F., Putnam, J. E. and Burgess, P. R., Activation of gracile nucleus: time distribution of activity in presynaptic and postsynaptic elements, Exp. Neurol., 37 (1972) 312-321. 38 Whitsel, B. L., Petrueelli, L. M. and Sapiro, G., Modality representation in the lumbar and cervical fascieulus gracilis of squirrel monkeys, Brain Research, 15 (1969) 67-78.