Responses of single units in laminae 2 and 3 of cat spinal cord

Brain Research, 160 (1979) 245-260

245

© Elsevier/North-Holland Biomedical Press

RESPONSES OF S I N G L E UNITS IN L A M I N A E 2 A N D 3 OF CAT SPINAL CORD

PATRICK D. WALL, E. G. MERRILL and TONY L. YAKSH Cerebral Functions Research Group, Department of Anatomy, University College, London WC1E 6BT (Great Britain)

(Accepted May 4th, 1978)

SUMMARY 333 units were recorded in laminae 2 and 3 of lumbar cord in decerebrate cats. Recording of small-amplitude spikes was made possible by the use of platinumsurfaced tungsten microelectrodes, continuously variable filters and an analogue delay line display. Stimulation of the lateral Lissauer tract showed that a sample of the units sent axons into this tract. Using iron-plated electrodes, recording sites were marked and shown to be within laminae 2 and 3. Axons of peripheral afferent axons were excluded from the sample, as were long-range descending axons. By using one electrode placed close to a cell body in lamina 4 and a roving electrode in the dendritic region dorsal to the cell body, it was possible to show that the recorded units were not field spread of deeper cells. Of the units 94 ~ had peripheral receptive fields, RF; 30 had small RFs less than 2 sq.cm, intermediate in size between RFs of peripheral axons and RFs of large cells in dorsal horn. These small R F cells occurred in clusters and their RFs constituted a fraction of the larger RF of nearby large cells. Of the units 56~o responded to brush and touch, 1 9 ~ to brush, touch and pressure, while 1 9 ~ required pressure on skin to excite them. Latency of response to electrical stimulation showed that all cells were excited by myelinated afferents. While no cells were detected exclusively by C afferents, many may have been excited by both A and C afferents. Eighteen per cent of the cells showed a prolonged discharge lasting more than 5 sec after a single stimulus. Some of these long discharge cells continued firing for minutes. Another unusual class, 14 ~ of all cells, habituated very powerfully to intermittant natural or electrical stimuli, and remained unresponsive for many seconds after responding to the first stimulus.

INTRODUCTION We report here on the properties of units recorded in laminae 2 and 3 of cat lumbar cord. The substantia gelatinosa Rolandi (SG) is synonymous with lamina 229.

246 Since we found no difference in the properties of the units recorded in the two laminae and reported here, we will use the term SG unit as a convenient short phrase. This does not mean that more detailed analysis will not reveal important anatomical and physiological differences between the two laminae. The substantia gelatinosa Rolandi (SG) has until recently resisted attempts at single unit recording. The reasons are 2-fold: (l) the SG cells and their axons are small; (2) the region contains other elements which generate easily detected action potentials. These other elements are (a) axons, (b) dendrites originating from cell bodies in other laminae, and (c) displaced cells from other laminae. The axons include peripheral afferents, some of which terminate in the region, and others pass through to deeper laminaeg,24,eT,28,30,31. There are also axons in SG which originate from distant regions of the spinal cord and brain stem 1. There are dendrites in SG which rise dorsally from the large cell bodies in lamina 4 and some originating from lamina 112,31. Since the borders of the laminae are not precise, occasional cell bodies of the lamina 1 or 4 type are observed histologically within laminae 2 and 3 z2. It is obviously essential that we clearly establish that we have been recording from the small SG cells and not from the other elements. We have done this by a combination of 6 techniques. For recording, we surveyed all of the standard types of fluid and metal microelectrodes, and found that glass-covered platinum-plated tungsten electrodes 19 were far superior to other electrodes in their ability to record isolated units in the region. The isolation of the units was improved by use of continuously variable filters and an analogue delay line, so that the shape of action potentials could be continuously monitored. For marking the recording sites, iron was incorporated in the platinum surface, electrolytically deposited and histochemically identified 18. To differentiate between SG cells and the dendrites of lamina 4 cells, we placed one recording electrode near a lamina 4 cell and then probed with an independent electrode immediately dorsal to the cell body 20. In this way we could differentiate between the distant field produced by action potentials in the deep cell and spikes in other units whose firing was not time locked to the firing of the deep cell. Finally, and most crucially, we have developed a technique for stimulating the lateral Lissauer tract (LLT) on the surface of the spinal cord zl. This tract contains axons originating from the SG cells 31, and its stimulation allows the marking of SG cells with antidromic impulses. Microelectrode stimulation and recording in the LLT reveals the presence of substantial numbers of unmyelinated fibres with properties similar to those of the cerebellar parallel fibres 2t. A stimulus of less than 5/~A and 50/~sec evokes a compound action potential which travels at less than 1 m/sec in a narrow beam 150/~m in width and moving only 2.5 mm in a rostral and caudal direction. Single unit analysis shows that there are axons which are not peripheral afferents with conduction velocities of 0.3-0.7 m/sec, absolute refractory period 0.8-1.0 msec and partial refractory period of 1.5-5.0 msec. In the absence of satisfactory physiological techniques for the analysis of SG units, speculative hypotheses have been proposed in the past to suggest a role for the cells. The earliest suggestion was that SG was the site of termination of fine afferents and was a way-station in a pain pathway25, 26. This view still receives anatomical and physiological supportlS,1L Wall showed that the most intense activity associated with

247 the negative dorsal root potential was located in SG, and that the Lissauer tract was partially responsible for the transfer of this activity from one segment to another 32. These and other observations led Melzack and Wall to propose that SG cells were ideally situated to play the role of modulators of sensory transmission, the gate control 17. In updating this view, it has been stressed that the control might operate both pre- and postsynapticaUy in the region of the first central synapses a4. DennyBrown et al. supported these views by showing that the size and sensitivity of a dermatome was influenced by the LT 6. Now we can begin to move out of this speculative period, since 4 groups of workers have reported single or multiple unit recordings from SG 5,10,13,36. We have recorded from 333 single units, and find them to have a wide range of properties, some of which are not seen in the larger cells of the dorsal horn. METHODS Cats weighing more than 2 kg were anesthetized with ether and decerebrated surgically at the midcoUicular level with carotid occlusion. Ether was then discontinued and the animals were paralyzed with gallamine and artificially respired. The results to be reported required that the animals were in excellent condition with blood pressure above 80 mm Hg, blood gases in the proper range and a brisk cord circulation. To achieve this, body temperature and oil pool temperatures were monitored and controlled. Respiration was controlled by measuring the expired pCO2 and by monitoring the intra-arterial pO2 by inserting a Wyeth oxygen electrode catheter from a carotid artery into the aorta. Blood pressure was measured from the same catheter which was also used for infusion of 4 ~ dextrose saline. The lumbar enlargement was exposed and the LT and SG were placed in convenient position for recording by rolling the cord by some 40 ° by dural retraction. Mechanical stability was produced by making a hammock of the dura, by bilateral pneumothorax, and on occasions by stabilizing pins below the dura, and by agar on the cord. Sometimes the L T was exposed by arachnoid dissection to reveal it, without further dissection, between dorsal roots L4 and 5 or L5 and 6. On other occasions, the entire L7 dorsal root was everted intact in a dorsal direction so that the LT running below it became visible. Retaining sutures maintained the root in its rolled-over position. Stimulation of the LT was produced by placing a tungsten microelectrode under visual control so that the 20/~m tip of the glass-covered electrode just penetrated the surface of the L L T 50-100/,m lateral to the lateral edge of the nearest dorsal root as it entered the cord. A negative square wave pulse 0.05 msec, 1-5/~A was delivered to the microelectrode through an isolated circuit at l Hz. For special studies of the properties of LLT axons, recordings were made from the LLT using platinum black-covered tungsten microelectrodes with 15 /*m exposed tip. O n all occasions the nearest rootlet to the LLT stimulus point was dissected free and cut peripherally and placed on recording hooks. This monitoring had a double purpose. First it was designed to detect stimulus spread to involve afferent fibres. To be certain that the recordings were capable of detecting not only myelinated but also unmyelinated volleys, the stimulus

248 electrode was first placed directly on the root, and the recording electrodes adjusted until both A and C volleys were apparent. The stimulating electrode was then moved back onto the LLT and averaged recordings were made to insure that the stimulus produced no sign of an antidromic volley on the dorsal root. The second use of the recording electrodes was to monitor a delayed negative dorsal root potential which appears 15 msec after the LLT stimulus 35. This potential was used here simply as a monitor wave to show that the LLT was being stimulated. I f afferent fibres were stimulated, the classical early dorsal root potential 1~I was evoked, while dorsolateral funiculus stimulation failed to produce any dorsal root potential when low stimulus intensities were used. The electrodes used for recording units in the LLT and SG were glass-covered platinum-plated tungsten electrodes with a tip length of 15/~m 19. The operative noise level of such electrodes is below 40 #V. Each unit's spike height was optimized by electrode movement and by tuning continuously variable filters (NeurLog N L 125, Digitimer) to set the optimal frequency recording band, usually between 100 and 7500 Hz. The spikes recorded were between 50 and 150 #V, and were used to trigger a window discriminator circuit which in turn triggered an analogue delay line (NeuroLog NL 740, Digitimer), which allowed the display of the full spike shape before and after the trigger. Once a spike was satisfactorily isolated, a sequence of tests was started. The Lissauer tract was stimulated. Seventy-six cells were detected which responded antidromically and 92 responded transynaptically. A further 165 small-unit spikes were detected from the SG which either did not respond to stimulation of the nearby LLT or were not tested. Then a peripheral receptive field was sought on the leg or flank by brushing, rubbing or lightly pinching. Once located the area was examined with a camel hair brush for a brush response, and by light finger pressure for a touch response, and picking up a fold of skin with broad-toothed forceps for a pressure response. In some cases, a pair of stimulating needles was placed in the peripheral R F to determine threshold and latency of the peripheral response. With rotation of the spinal cord by dural retraction and with oblique lateral penetration with the recording microelectrode, it is possible to make electrode tracks running roughly parallel to the laminar boundaries. A marker electrode was left in position for later histological location and then search tracks were made parallel to the marker electrode. This allowed all recording points to be defined as being located within laminae 2 and 3. Representative examples of particularly interesting units were marked by deposition of iron using a current of 1 /~A for 30 sec and staining with ferrocyanide is. All such units were located within either the LLT or in laminae 2 or 3 (Fig. 1). Data was recorded either by direct display, or by averaging responses and plotting on a chart recorder, or by the dot display method using the unit spike to brighten a spot on the oscilloscope. RESULTS Using the methods described, we have isolated and studied 333 units in laminae 2 and 3 in lumbar 5, 6 or 7 segments of adult cat (Table I). All the units were recorded

249 electrode marks

Fig. 1. Camera lucida drawing of a spinal cord section at L7 showing the locations of 3 recording sites in SG, marked with iron-plated recording electrodes. One mark was just below the surface, in the Lissauer tract. The electrode was made to penetrate diagonally, more or less across the width of the SG.

when the electrode tips were in lamina 2 or 3. All were generated by small structures, since spike heights were sensitive to small movements of the recording electrode. The existence of units was detected in 3 ways: 76 units responded antidromically to LLT stimulation; 92 units responded transynaptically to LLT stimulation with a variable latency; 165 units were spontaneously active and responded to peripheral stimulation. An examination of Table I shows that the distribution of reported properties is approximately the same in all 3 groups with one obvious exception. All cells in the third group by definition responded to peripheral stimulation, while the first and second groups were found to contain 18 ~ and 5~o units which failed to respond to peripheral stimulation. In other respects, we found no significant differences between the 3 groups with respect to the properties reported here, and therefore we will consider all 3 groups together. Before describing the properties of these units, we must obviously first establish that they were in fact SG units and not one of the intruding structures.

(1) Were the units peripheral afferent axons? All units tested by electrical stimulation of their peripheral receptive field responded with a slightly variable latency of the first response, with repetitive firing and with an inability to follow stimulus frequencies above 200 Hz. Of course peripheral fibres were detected with small receptive fields, a fixed latency of response to peripheral electrical stimulation, high following frequencies, and, if they demonstrated a repetitive response (the dorsal root reflex), this failed to follow stimulus rates higher than 10 Hz. These units are excluded from our sample. Of our units 64 ~ had receptive fields larger than any reported for single peripheral afferents. Seventy-six units responded with a fixed latency to L L T stimulation, and this antidromic action

250 TABLE I SG cells. R F size, adequate stimulus and response characteristk's

RF, receptive field; BT, cell responds to brush and touch; BTP, cell responds to brush, touch and pressure; P, cell responds to pressure only ; Habit, cell response fades if stimuli are repeated at 1 Hz or less; Long, cells discharging longer than 5 sec after a single stimulus; Small, RFs less than 2.0 sq.cm ; 1, RFs on less than one part of the leg, i.e. upper leg or lower leg or foot or toes; 2, RFs extending over 2 parts; 3, RFs over 3 parts; 0, no RF found. RF

BT

BTP

P

81 85 17 3

7 42 9 6

I1 47 4 2

186 56

64 19

64 19

Total

%

Habit

Long

30 52 9 3 6

16 18 8 4

14 39 5 3

46 14

61 18

AH cells(n - 333)

Small 1

2 3 0 Total

19 19 6

99 174 30 I1 19

Antidromic cells (n -- 76)

Small

16

3

6

1

18

4

9

2 3 0

4 0

0 2

0 0

Total

14

25 31 4 2 14

33 41 5 4 18

0

1

6

5

I 2

0 0

14

38

9

15

9

6

18

50

12

20

12

8

Orthodromic cells' (n - 92)

Small

15

2

3

1

22

17

9

2 3 0

7 I

6 2

2 1

45 49

27 29

15 16

Total

5 5 5

20 48 15 4 5

22 52 16 4 5

2

4

4

9

6 0

2 1

12 13

16 17

p o t e n t i a l c o l l i d e d with an o r t h o d r o m i c p o t e n t i a l . I l l u s t r a t i o n s o f this p h e n o m e n o n are s h o w n e l s e w h e r e 21. T h i s establishes t h a t this s a m p l e o f units had a x o n s in t h e L L T . W h i l e r e c o r d i n g these units, we m o n i t o r e d t h e n e a r e s t d o r s a l r o o t l e t a n d failed to i d e n t i f y an a n t i d r o m i c

volley in t h a t r o o t l e t . O n o c c a s i o n s we did s t i m u l a t e a

p e r i p h e r a l afferent a x o n in t h e L E T a n d r e c o r d it in SG. S u r p r i s e has been e x p r e s s e d t h a t it is possible to s t i m u l a t e L L T w i t h o u t i n v o l v i n g large n u m b e r s o f p e r i p h e r a l afferents. T w o r e c e n t p a p e r s s h o w t h a t t h e r e is a m i x t u r e o f afferents a n d S G a x o n s in L T 12,14, b u t t h e y deal w i t h s e g m e n t s r o s t r a l to the l u m b a r e n l a r g e m e n t . In t h o s e s e g m e n t s , the L T lies as a b u r i e d b a n d v e n t r a l to t h e d o r s a l r o o t entry, s a n d w i c h e d b e t w e e n d o r s a l a n d d o r s o l a t e r a l funiculi. H o w e v e r , in t h e l u m b a r e n l a r g e m e n t , the lateral p a r t o f t h e L T e m e r g e s as a sheet o f fibres e x t e n d i n g 200 # m o v e r t h e surface o f the cord, while a m e d i a l p a r t exists v e n t r a l to the dorsal r o o t e n t r y zone. W e believe, with Earle s, t h a t t h e lateral p a r t c o n t a i n s v e r y few afferent fibres, while t h e m e d i a l r e g i o n c o n t a i n s m a n y p a r t i c u l a r l y fine d i a m e t e r afferents. W h e n we p e n e t r a t e d w i t h

251 the s t i m u l a t i n g electrode 100 # m below the surface o f the L L T to an electrode l o c a t i o n in the medial L T i m m e d i a t e l y ventral to the r o o t e n t r y zone, the same stimulus strength which h a d failed to stimulate afferents f r o m the L L T now e v o k e d an obvious volley (particularly in C fibres), which ran a n t i d r o m i c a l l y in the nearest d o r s a l rootlet.

(2) Were the units the axons o f distant cells? Since the r e c o r d i n g electrodes were c a p a b l e o f detecting a c t i o n potentials in axons, it could be possible t h a t we were recording f r o m axons whose cell bodies were m a n y segments f r o m the r e c o r d i n g site. W e can rule o u t the possibility t h a t these were ascending or descending axons r u n n i n g in the L L T , because we never detected a cell a n t i d r o m i c a l l y r e s p o n d i n g to L L T s t i m u l a t i o n m o r e t h a n 2.5 m m r o s t r a l or c a u d a l to the stimulus point. F u r t h e r m o r e , all o f the a n t i d r o m i c a l l y r e s p o n d i n g cells were limited to the lateral h a l f o f l a m i n a 2. Evidently the L L T axons h a d a relatively s h o r t length. This fits the o b s e r v a t i o n t h a t the c o m p o u n d a c t i o n p o t e n t i a l e v o k e d in the L L T c o u l d only be followed for 2.5 m m , even when 300 a v e r a g e d sweeps were used to a m p l i f y the signal 21. A few individual fibers were followed over longer distances. W e can c o m p l e t e l y discard the possibility t h a t we were r e c o r d i n g f r o m axons destined to

/~

Depth

50 S.G.

1.,o

1.10

6.3

1.20

12.5

1-25

[ 25

1.30'

Lamina IV

[ 50

J

135---J

100

1.40- ~

100 pV

mr.

i

I

2-5 mS Fig. 2. Averaged responses from dendritic regions of a selected lamina 4 cell. The soma was at a depth of 1.50 mm, where one microelectrode recorded its soma spike. A second microelectrode was withdrawn verticallyfrom the soma location, and potentials were recorded, delayed 1.25 msec and averaged, using the soma spike as a time base trigger. Each average contained 32 responses. At depth a of 1.10 mm, within the SG, the field potential due to the dendrites of the lamina 4 cell was only about 4/~V in amplitude, but SG units recorded at the same location had spikes from 20 to 100 #V. One such SG unit spike, captured as a single sweep, is shown in the top trace (note the different vertical calibrations for the SG spike and the lamina 4 potential at this point).

252 run in the dorsolateral funiculus, because we never recorded any cells antidromically invaded when this white matter was stimulated. Finally, there is the possibility that there is a type of cell lying outside lamina 2 or 3 which sends a short axon through these laminae into the LLT. No such cells have been described histologically or physiologically. We have located no cells in laminae 1,4, 5 or 6 antidromically invaded from the LLT. Furthermore, we and others have not recorded cells in these laminae with several of the properties we see in these SG units.

(3) Were the units dendrites of deep cells or electrical field spread from deeper cells ? As we shall show below, a number of the SG units had receptive field properties which have never been reported as properties of the easily recorded deeper cells. To differentiate field spread of deep cells from SG cells, we intentionally mapped the field spread of single cells whose cell bodies were in lamina 4 but whose dendrites reach up into laminae 2 and 3. A glass microelectrode was made to penetrate along a slanted track until it recorded 1 mV spikes extracellularly from a single lamina 4 cell. Next an independently manipulated metal microelectrode was made to penetrate on a vertical collision course towards the tip of the first microelectrode 2o. The roving electrode was withdrawn step by step away from the cell body in lamina 4 into the dendritic regions in laminae 2 and 3. An averager was triggered by the appearance of an action potential recorded by the electrode close to the cell. Recordings from the roving electrode were delayed and averaged (Fig. 2). It is evident that there is a steady decrease of the unit spike height as the electrode moves dorsally. When the roving electrode was in lamina 2 (second trace, Fig. 2), the distant cell produced only a small spike; larger independent spikes were, however, recorded by the roving electrode (top trace, Fig. 2). The firings of these separate units were not related in time to the deeper lamina 4 cell's firing, and had different receptive fields. The heights of these unit spikes were critically dependent on the electrode position, and were never seen to increase progressively if the recording electrode penetrated toward deeper laminae.

(4) Were the units displaced large cells from nearby laminae? Pearson has noted that large cells of the flattened Waldeyer marginal lamina l type are occasionally observed in lamina 222 . Similarly, it is often difficult to define the exact border between laminae 3 and 4 because occasional large cells are apparent in lamina 32s,zg. However, these are rare events. It will be seen that we had no great difficulty in isolating 333 units, so that we are not reporting rare events. Furthermore, having isolated a single unit, we could frequently see signs of numerous similar units in the background of the recording. On occasions we isolated 2 or 3 units within 50/~m on a single electrode track. All of this evidence suggests that SG units were being detected, and the spikes were not produced by the 4 components which intrude into the SG from the outside.

Receptive field size SG was searched in two ways. In the first, it was probed for spontaneously active units with small-amplitude, brief, localized spikes. The ongoing activity of such units

253 was usually 2-5 Hz, but some had much higher frequencies. For all such units, a peripheral RF could be detected. In the second method, units were located which responded antidromically or orthodromically to L L T stimulation. Since about onethird of these cells were not spontaneously active, this population may have included less excitable cells or different types of cells. Eighteen per cent of the antidromically driven cells and 5 ~ of the orthodromically driven cells had no apparent RF. The properties of the cells with RFs appeared the same no matter which search method was used. Thirty per cent of all cells (33 ~ of antidromically driven cells and 22 ~ of orthodromically driven cells) had strikingly small receptive fields measuring less than 2 sq.cm. They were located on distal, middle and proximal parts of the leg, and were particularly striking on the proximal parts where all other cells so far reported in dorsal horn have relatively large receptive fields 3,7. We were, of course, immediately suspicious that such units might be peripheral afferent axons, and we therefore repeatedly examined their response to peripheral electrical stimulation and satisfied ourselves that they were indeed central cells by the tests described above. Their response properties have not been described in peripheral axons. All failed to respond at fixed latencies and to follow high-frequency stimulation. These cells frequently occurred in small nests where several units appeared to be in close proximity. With vertical penetrations we encountered families of these small R F units whose receptive fields were very close to each other, often overlapping. On penetrating ventrally along the same track, a large amplitude unit would be encountered with a somewhat larger receptive field. This unit was always of the type described as characteristic of lamina 433. The RFs of the small R F units each made up a fraction of the larger RF. The commonest variety of SG unit, 5 2 ~ of all units, ( 4 1 ~ of those antidromically driven from LLT and 5 2 ~ of those orthodromically driven) had RFs restricted to one of the fractions of the leg, i.e. flank, upper leg, lower leg, foot or toes. Their sizes were those reported for lamina 4 or 5 cells 33. Nine per cent of all cells had RFs extending over two parts of the leg and 3 ~ had large fields extending to 3 parts of the leg. Fig. 3 shows typical RFs.

Fig. 3. Representative receptive fields of 3 substantia gelatinosa units. A: a unit with large receptive field even extending contralaterally. This type of cell is classified as RF3 in the table because its RF extended over 3 parts of the hind limb, upper leg, lower leg and foot. B: an example of the small field unit with an RF of less than 1.5 sq.cm, highly unusual especially in proximal parts of the limb. C: an example of an RF1 unit with an RF larger than 1.5 sq.cm, but limited to one of the 4 regions of the hind leg.

254

Adequate excitatory stimulus Six per cent of the overall sample and 18 ~o of the antidromically activated cells could not be excited by any applied peripheral stimulus; 56 ~ of the cells responded to brushing or touching in their R F and if the pressure of the stimulus was increased, they failed to increase their discharge frequency; 19 ~ of the cells responded over a wide range of stimulation intensities from light brushing up to pinching. As with lamina 5 cells, cells with RFs measuring more than 2 sq.cm often had a central area where brush, touch and pressure produced increasing discharge, and also a surrounding area where pressure was required to fire the celP 1. The existence of these surround pressure areas was not due to mechanical spread of the stimulus. A fold of skin was gently raised by forceps and intentionally moved laterally. This failed to fire the cell. However, if toothed forceps were used to squeeze the fold of skin, then the cell fired. There were a few cells which did not respond to brush but were excited by touch and pressure. Nineteen per cent of all cells required heavy pressure or pinching for excitation.

Response to peripheral electrical stimulation Hypodermic needles were placed in the centre of the cell's R F and stimuli delivered at 1 Hz. The latency of first response varied from 2 to 20 msec. Since the peripheral conduction distance was highly variable, we concentrated particularly on cells with RFs on foot or toes. Here 84 ~ of the ceils responded with latencies of 4-10 msec (mean 6-7 msec). There was no difference of latency distribution of the cells with small receptive fields compared to those with larger RFs. The variation of latency from stimulus to stimulus was more marked for those with longer initial latencies. Single stimuli produced repetitive firing and, in 18 9~oof the cells to be considered below, the discharge lasted more than 5 sec. A more typical cell is shown in Fig. 4, where the discharge lasts 65 msec and there is a very high rate of initial firing. As the strength of stimulus was increased, so did the duration of repetitive firing. Particularly in the case of cells responsive to brush, touch and pressure, or to pressure, the repetitive discharge increased for stimulus strengths and at latencies which could have been due to C fibers

z:'o t.D

20mS.

I

Fig. 4. A substantia gelatinosa cell response. The receptive field o f this cell was on the lateral foot. W h e n needle electrodes were placed in the R F a n d a stimulus was delivered (S), the cell r e s p o n d e d after 6 msec a n d t h e n gave a repetitive discharge. T h e low noise level a n d clearly isolated spikes should be noted.

255 in the afferent volley. However, for the initial action potential, no cell was discovered which could have been fired exclusively by C afferents, as judged by the stimulus strength or latency of first response. Foot cells with initial latencies of 4-6 msec would follow peripheral stimuli at over 100 Hz. More delayed responders followed only irregularly at lower stimulus frequencies. Fourteen per cent of the cells showed marked habituation on natural stimulation, and many of this class of cell failed to follow even at 1 Hz.

Cells which habituate and have variable size receptive fields We have previously seen large dorsal horn cells in decerebrate cats whose response faded to slowly repeated stimuli 33 but, in the SG, 46 of the units (14 ~ of the total) showed very marked habituation. In the extreme examples, some units responded with a brisk burst to the first brush to their receptive fields, and it was necessary to wait 5-10 sec before a repeated identical stimulus would evoke even a single impulse. In all habituating cells, a brush or pressure stimulus repeated at 1 Hz resulted in a stimulus by stimulus decline of the response until the cell failed completely to respond. Similarly, electrical stimulation produced a fading response. It was necessary to wait at least 20-30 sec before excitability was fully restored. In all cells with a sufficiently large RF for testing different parts of the RF, it was found that the habituation was selective to the part of the RF which had been stimulated. If a stimulus was repeated in one part of the RF until the cell failed to respond and then the stimulus site was shifted to another part of the RF, the cell immediately responded briskly to the new origin of impulses but then again habituated. Further evidence that the habituation is not produced by an overall inhibition of the units is provided by the fact that habituation was not accompanied by a decrease of the cell's ongoing activity. As might be expected of cells with such powerful associated inhibitory habituating mechanisms, the actual extent of the R F was sometimes difficult to determine, and appeared to vary with time and with the immediate history of past stimuli. Often when such a unit was isolated, the first survey of the leg would seem to reveal a large RF but, when the leg was re-examined with small repeated stimuli, the R F would shrink to a much smaller reliable area. It was necessary to rest for minutes between stimuli, and then the edges of the R F would seem to vary in an amoeboid fashion, apparently determined by some uncontrolled fluctuations in the state of the cell and not determined by the frequency of the immediately preceding stimuli. The habituation of cells responding to light brush could not have been due to habituation of the peripheral receptors, since this phenomenon has not been reported in low-threshold peripheral mechanoreceptors. Furthermore, peripheral electrical stimulation which generates impulses proximal to the receptors also produced habituation. A final reason for attributing the habituation to a central mechanism is that cells in lamina 4 continued to respond to repeated peripheral stimuli at a time when SG units immediately dorsal to them had habituated to the same stimulus. Prolonged discharge cells Sixty-one cells were detected which responded for seconds to minutes after a

256

Fig. 5. Prolonged discharge unit in SG. Each dot represents an impulse in this single unit. The vertical scale is interpulse interval, in msec. The time base is in sec. Before stimulation the unit had a low rate of ongoing activity with most pulse intervals more than 200 msec. At S, a single brush stroke moved across the cell's RF, and this was followed by a high-frequency firing with pulse intervals of less than 10 msec. This high rate of firing dropped to a pulse interval of about 75 msec, which was maintained until 45 sec after the stimulus. Then over a period of 5 sec the discharge rate returned to the resting level. single brief natural stimulus to their receptive field. In previous work ~3, deep cells were described which responded for as long as a second after a single peripheral stimulus involving C fibers, after which there was a period of facilitation lasting 1-2 sec, resulting in a 'wind up' of discharge if stimuli were repeated at 1 Hz. It has also been reported that peripheral C fibers may discharge for 3 - t 0 sec after a stimulus 2. However, the prolongations reported here were of a completely different order, partly because the prolonged firing was set off by a single threshold stimulus, a brush or a pinch depending on the unit, and partly because the discharge lasted from 5 sec to over 3 min. Electrical stimuli to the R F would set offsuch discharges, but when the unit was of the BT type a single brush was sufficient (Fig. 5). These cells occurred with all categories o f receptive field size and o f adequate stimulus intensity. For the majority of cells, the highest frequency occurred during the stimulus but, for some where the R F was complicated by overlaid excitatory and inhibitory zones la, the highest frequency was reached after the stimulus. The frequency slowly declined during the prolonged afterdischarge, sometimes asymptotically, until it reached the resting rate, but sometimes there would be a sudden cessation of discharge. In 8 of the 6l cells, the discharge lasted 1-3 min. In some units tested, a single brush stimulus applied outside the excitatory receptive field would interrupt or terminate the prolonged discharge. In one cell the discharge appeared to settle down to a steady continuous low-frequency firing, which was promptly abolished by a brush close to its RF, whereupon the cell became silent. This cell therefore could be switched between steady firing if the R F alone had been stimulated, and silence if the surrounding area had been brushed. In two tests, pentobarbitone 25 mg/kg was injected intravenously while recording from a prolonged discharge cell. This produced a complete abolition of the afterdischarge, while a brief response to R F stimulation remained.

257 DISCUSSION We have shown that it is possible to record from substantial numbers of SG units, and we have presented the evidence that these units are not afferent fibers or descending axons or dendrites of deep cells. We show that these cells have a number of unusual properties not previously described for units recorded elsewhere in spinal cord. Four other groups have reported SG units with different properties from those described here. There is no necessary contradiction between our results and others, since very different recording and stimulation techniques have been used. Histologically different types of cell exist in laminae 2 and 328,31, and it may be that different techniques select different fractions of a mixed population. Kumazawa and Per118 have described 19 cells in SG, 14 of which responded to Ab and C noxious inputs and 5 responded to C innocuous inputs. We detected 64 cells which responded only to pressure or pinching. None of these were fired only by C afferents. It is true that we encountered 19 cells for which we could find no RF, and it is possible that these might have had such high peripheral thresholds that our test stimuli did not reach threshold. Cervero et al. 5 have reported 31 cells. They used very small glass electrodes (60-200 Mf~). They searched for units which responded antidromically to stimulation of a bridge of tissue, including variable amounts of Lissauer tract and nearby white matter. The detected units had myelinated axons conducting at 4-16 m/sec over many segments. It is not known to us what system these axons represent. Fine tips of the electrodes used cannot be accurately localized, so we do not know the structure in which the units were located. They stressed inhibitory responses in units with high spontaneous activity. We too saw inhibitory fields, but always associated with excitatory fields which we report here. Wilcox et al. 36 lowered pairs of 10/~m platinum wires separated vertically by 50--100 /,m into SG of anesthetized rat cord. They recorded a correlated shift of general activity when both electrodes were in the SG, and report that this shift of activity matches the negative dorsal root potential in its time course. It is not known what is being recorded, and the time course of the dorsal root potential shown is very unusual. Much as we would like to confirm these results, since they fit a previous prediction 32, we did not encounter a class of cells whose envelope of repetitive discharge matched the DRP. As we have said, we did record from cells with repetitive discharges whose duration varied from tens of milliseconds to tens of seconds, and it is possible that the authors were recording a massed average of such discharges. However, we noted that barbiturate reduced the length of repetitive discharge, whereas barbiturate prolongs the DRP. During the negative DRP, there is an intense current sink in SG 82, and this might vary both membrane and electrode properties. Finally, Hental110 has recorded units in cat dorsal horn using metal electrodes with a thick covering of platinum black. His recording methods are closest to ours, and he reports a highly unusual class of cell with a prolonged discharge starting after the stimulus. We too saw a few such units as part of the group of 61 prolonged discharge units, and believed them to be the result of producing first inhibition and then excitation.

258 Turning to our units, the first surprising result is that we found no units which responded only to C afferents. It is of course possible that our recording or stimulus methods failed to reveal such cells. We cannot tell if the 19 non-responsive cells were in fact connected to high-threshold afferents in some region which was not adequately stimulated. On the positive side, it is clear from threshold and latency that 64 units were excited by both A6 and C high-threshold afferents. There is mounting evidence that fine afferents terminate in lamina 214,15,27. However, we also recorded from 186 units responding only to hair movement, and many of these had a latency of firing, suggesting that they were fired monosynaptically by Aft afferents, i.e. 4-6 msec from foot RFs. Some believe that Aft afferents terminate in lamina 3 and not 2 z, although others disagree 24. Some of our units of the low-threshold low-latency type appeared clearly to be located in lamina 2. It is possible that we were recording from axons of lamina 3 cells or that lamina 2 cells send dendrites into lamina 3. It does seem clear that SG cells respond to a wide variety of afferent fibre types and that convergence of different types of afferent onto single units may occur. Three properties of units appear special to this region and have not been reported elsewhere in the spinal cord a3. The first is the existence of units with small RFs. The size of these RFs is intermediate between those reported for peripheral afferents and for central cells 3,4. Only 3 % of all recorded units had large diffuse RFs which might have been expected in a neuropil. Not only did 3015oof all units have these small fields, but they appeared to be organized in nests, with each group concerned with closely related areas of skin. It also appeared that these groups of cells existed in compartments separated mediolaterally by quiet zones in which no units were recorded. Furthermore, on penetrating vertically into the region of large cells, the first such cell encountered had its RF encompassing all of the small RFs of the cells above. This suggests small cells among the dendrites of a large cell, like sparrows in a tree, where each small cell shares a fraction of the afferents which drive the large cell. The other highly unusual features of the units were the existence of very prolonged excitations and inhibitions. It is possible that the prolonged firing in 18 % of the units is linked to the prolonged periods of habituation seen in 14 % of the units. Repetitive firing has been seen in peripheral unmyelinated afferents after natural stimulation, but not lasting beyond 10 sec 37. Furthermore, we observed prolonged discharges after brief threshold peripheral electrical stimuli which evoke only one impulse in each stimulated fibre. Therefore, it is evident that these long discharges must be a property of central structures. Prolonged firing lasting seconds has been reported in large central cells 23, and careful analysis may show that some cells maintain an increased discharge for tens of seconds. However, we found here 18 o~ of the units obviously firing repetitively for tens of seconds and sometimes for minutes. One may guess that some sign of the effects of this long firing may become apparent in studies of excitability shifts of larger cells. Such shifts might help to explain the prolonged sensory effects of afterglow sensations in normal people and the hyperpathias in pathological states.

259 ACKNOWLEDGEMENTS W e are g r a t e f u l to M r . A . A i n s w o r t h , Ms. D . C o n w a y a n d M r . J. P. Patel f o r t h e i r t e c h n i c a l help. T h e w o r k was s u p p o r t e d b y t h e M e d i c a l R e s e a r c h C o u n c i l a n d t h e N a t i o n a l Institutes of Health.

REFERENCES 1 Basbaum, A. I., Clanton, C. H. and Fields, H. L., Opiate and stimulus produced analgesia: functional anatomy o f a medullospinal pathway, Proc. nat. Acad. Sci. (Wash.), 73 (1976) 4685-4688. 2 Brown, A. G., Rose, P. K. and Snow, P. J., The morphology of hair follicle afferent fibre collatelals in the spinal cord of the cat, J. Physiol. (Lond.), 272 (1977) 779-797. 3 Brown, P. B. and Fuchs, J. L., Somatotopic representation of hindlimb skin in cat dorsal horn, J. Neurophysiol., 38 (1975) 1-9. 4 Brown, P. B.,Fuchs, J. L. andTapper, D. M.,Parametricstudies of dorsal horn neurons responding to tactile stimulation, J. Neurophysiol., 38 (1975) 19-25. 5 Cervero, F., Molony, V. and Iggo, A., Extra- and intracellular recordings from neurones in the substantia gelatinosa, Brain Research, 136 (1977) 565-569. 6 Denny-Brown, D., Kirk, E. J. and Yanagisawa, N., The tract of Lissauer in relation to sensory transmission in the dorsal horn of spinal cord in the macaque monkey, J. comp. NeuroL, 151 (1973) 175-200. 7 Devor, M. and Wall, P. D., Dorsal horn cells with proximal cutaneous receptive fields, Brain Research, 118 (1976) 325-328. 8 Earle, K. M., The tract of Lissauer and its possible relation to the pain pathway, J. comp. Neurol., 96 (1952) 93-109. 9 Heimer, L. and Wall, P. D., The dorsal root distribution of the substantia gelatinosa, of the rat with a note on the distribution in the cat, Exp. Brain Res., 6 (1968) 89-99. 10 Hentall, I., A novel class of unit in the substantia gelatinosa of the spinal cat, Exp. NeuroL, 57 (1977) 792-806. 11 Hillman, P. and Wall, P. D., Inhibitory and excitatory factors controlling lamina 5 cells, Exp. Brain Res., 9 (1969) 284-306. 12 Kerr, F. W. L., Neuroanatomical substrates of nociception in the spinal cord, Pain, 1 (1975) 325356. 13 Kumazawa, T. and Perl, E. R., Differential excitation of dorsal horn and substantia gelatinosa neurons by primary afferent units with fine fibres. In Y. Zotterman (Ed.), Sensory Functions of the Skin, Pergamon Press, Oxford, 1976, pp. 67-89. 14 LaMotte, C., Distribution of the tract of Lissauer and the dorsal root fibres in the primate spinal cord, J. comp. NeuroL, 72 (1977) 529-562. 15 Light, A. R. and Perl, E. R., Central termination of identified afferent units with fine myelinated fibers, Neurosci. Abstr., 3 (1977) 1554. 16 Lloyd, D. P. C., Electrotonus in dorsal nerve roots, Cold Spr. Harbor Symp. quant. BioL, 17 (1952) 203-2 l 9. 17 Melzack, R. and Wall, P. D., Pain mechanisms: a new theory, Science, 150 (1965) 971-979. 18 Merrill, E. G., Iron-plated tungsten microelectrodes for tip marking, J. PhysioL (Lond.), 241 (1974) 68-69P. 19 Merrill, E. G. and Ainsworth, A., Glass-coated platinum-plated tungsten microelectrodes, Med. biol. Eng., 10 (1972) 662-672. 20 Merrill, E. G. and Wall, P. D., Factors forming the edge of a receptive field : the presence of relatively ineffective afferent terminals, J. Physiol. (Lond.), 226 (1972) 825-846. 2l Merrill, E. G., Wall, P. D. and Yaksh, T., Properties of two CNS unmyelinated fibre tracts: the lateral Lissauer tract, and the parallel fibres of the cerebellum, (1978) in press. 22 Pearson, A. A., Role of gelatinous substance, Arch. NeuroL Psychiat. (Chic.), 68 (1952) 515-529. 23 Price, D. D., Hayes, R. L., Ruda, M. A. and Dubner, R., Neural representation of tactile evoked after sensations by spinothalamic tract neurons, Neurosci. Abstr., 3 (1977) 1568.

260 24 Proshansky, E. and Egger, M. D., Staining of the dorsal root projection to the cats dorsal horn by HRP, Neurosci. Lett., 5 (1977) 103 110. 25 Ranson, S. W., The course within the spinal cord of the non-medullated fibres of the spinal dorsal roots, J. Comp. Neurol., 22 (1912) 159-175. 26 Ranson, S. W. and Billingsley, P. R., The conduction of painful afferent inputses in the spinal nerves, Amer. J. Physiol., 40 (1916) 571 584. 27 Rethelyi, M., Preterminal and terminal axon arborizations in the substantia gelatinosa of cat cord, J. comp. Neurol., 172 (1977) 511 528. 28 Rethelyi, M. and Szent~gothai, J., The large synaptic complexes of the substantia gelatinosa, Exp. Brain Res., 7 (1969) 258-274. 29 Rexed, B., The cytoarchitectonic organisation of the spinal cord of cat, J. comp. Neurol., 96 (1952) 415-495. 30 Scheibel, M. E. and Scheibel, A. B., Terminal patterns in cat spinal cord. Ill. Primary afferent collaterals, Brain Research, 13 (1969) 417-443. 31 Szent~igothai, J., Propriospinal pathways and their synapses. In J. C. Eccles and J. P. Schad6 (Eds.), Organisation of Spinal Cord, Progr. Brain Res., Vol. 2, Elsevier, Amsterdam. 32 Wall, P. D., The origin of a spinal cord slow potential, J. Physiol. (Lond.), 164 (1962) 508-526. 33 Wall, P. D., The laminar organisation of dorsal horn and effects of descending impulses, J. Physiol. (Lond.), 188 (1967) 403 423. 34 Wall, P. D., The gate control theory of pain mechanisms. A re-examination and re-statement, Brain, 101 (1978) 1-18. 35 Wall, P. D. and Yaksh, T. L., The effect of Lissauer tract stimulation on activity in dorsal and ventral roots, Exp. Neurol., 60 (1978) 570-583. 36 Wilcox, G. L., Luttges, M. W. and Mayer, D. J., Unit activity of substantia gelatinosa neurones, Neurosci. Abstr., 2 (1976) 1384. 37 Zotterman, Y., Touch, pain and tickling: an electrophysiological investigation on cutaneous investigation on cutaneous sensory nerves, J. Physiol. (Lond.), 95 (1939) 1-28.