Research Reports - Science Direct

2 downloads 0 Views 2MB Size Report
crayfish fast flexor motor neurons and lateral and medial giant fibers (see also ref. ... (2) Do the various postsynaptic cells of a given presynaptic neuron vary in the ...... with particular reference to gamma-aminobutyrate and glutamate, J. Neuro-.
Brain Research, 112 (1976) 221-249

221

© Elsevier ScientificPublishing Company, Amsterdam - Printed in The Netherlands

Research Reports

THE MORPHOLOGY OF I D E N T I F I E D N E U R O N S IN T H E A B D O M I N A L G A N G L I O N OF A P L Y S I A C A L I F O R N I C A

W. WlNLOW* and E. R. KANDEL Department of Neurobiology and Behavior, The Public Health Research Institute of the City of New York and The Division of Neurobiology and Behavior, Departments of Physiology and Psychiatry, College of Physicians and Surgeons of Columbia University, New York, N.Y. 10032 (U.S.A.)

(Accepted February 18th, 1976)

SUMMARY The morphology of identified neurons and of one multiaction interneuron (L 10) of the abdominal ganglion of Aplysia has been studied using cobalt chloride, injected intracellularly. Cells with little synaptic input, R3-R14, had a relatively poorly developed dendritic tree, whereas the dendritic tree of cells L7 and L10, with extensive synaptic input, was highly complex. Cells L1-L6 and the RB cell cluster were found to have intermediate complexity of synaptic inputs and dendritic morphology. Within cell clusters, individual cells were often morphologically distinct. Identified cells have both invariant and variant axonal branches. Variant axons often project down other than their customary nerve trunks or are supernumerary. Three features of neuropil architecture were encountered. (1) When cells from the same cluster send their axons down the same nerve the axons often run in fascicles. (2) Although an identified cell's dendritic geometry varies from preparation to preparation, its dendrites always occupy approximately the same position in the neuropil. (3) The postsynaptic follower cells of L10 send their main axons through the axonal arborization of L10.

INTRODUCTION The central ganglia of invertebrates are separated into a peripheral region of cell bodies that is free of synapses and a central neuropil region where synaptic inter* Present address: Ethology and Neurophysiology Group, School of Biological Sciences, University of Sussex, Palmer, Brighton, Sussex BN1 9QG, Great Britain.

222 actions occur. The peripheral location of neuronal perikarya has facilitated the identification of individual nerve cells on the basis of position and electrophysiological properties of their cell bodies, and has allowed these identified neurons to be related to specific behaviors. In contrast, the complexity of the centrally located neuropil has been a consistent deterrent to the analysis of the synaptic regions. The analysis of neuropil architecture will ultimately require combined studies using light and electron microscopy (see for example refs. 21, 28, 29). But a significant beginning has recently been made using neuronal markers that are detectable with the light microscope; for instance, the Procion dyes 4,~7, cobalt chloride 2a and horseradish peroxidase 21. Injection of these substances into nerve cell bodies leads to the filling of their major processes and at least partial filling of their minor processes. Several studies using these techniques have now been carried out that have related the physiological properties of neurons to their morphology. For example, Kennedy et al. 15 have found that the electrical responsiveness of a given crayfish neuron soma is a function of the length and diameter of the axonal process connecting it to its main axon and dendritic tree. Selverston and Remler ~6 have correlated anatomical connectivity patterns with the physiologically determined connections between crayfish fast flexor motor neurons and lateral and medial giant fibers (see also ref. 20). Intracellular marking techniques have also proven useful for analyzing dendritic branching patterns. Thus, Stretton and Kravitz in the lobster 2v and Burrows 1,z and Goodman 9 in the locust found that physiologically identified neurons have the same general morphological outline from preparation to preparation, but the fine detail of their dendritic branching patterns is quite variable. Tweedle et al. a° have demonstrated that, unlike the situation in vertebrates, deafferentation of a cockroach motor neuron does not alter its dendritic branching pattern (for other studies, see also ref. 32; for review see ref. 14). This study represents an initial attempt to analyze the architecture of the neuropil within the abdominal ganglion of Aplysia. We have studied a multi-action cholinergic neuron (LI0) and a population of follower cells which it innervatesla,17,ak We first examined the morphology and dendritic field of the pre- and postsynaptic cells individually. Then, by injecting several cells at a time, we have examined, at a light microscopic level, the pattern of interrelationships between them. We were specifically interested in seeking answers to the following questions. (1) How consistent is the main branching pattern of the pre- and postsynaptic neurons? (2) Do the various postsynaptic cells of a given presynaptic neuron vary in the complexity of their dendritic arborization? (3) Is there a relation between the complexity of the dendritic input as inferred morphologically and the complexity of the somatic input as inferred from physiological studies? (4) How do the processes of the postsynaptic cell contact those of the presynaptic neuron? We have found (confirming others) that despite considerable variability in the branching pattern of individual cells there is often a consistent difference between different postsynaptic cells. We also noted a close correlation between the intricacy of a cell's dendritic arborization and the complexity of its synaptic input. Finally, we found that neuron LI0 does not extend its presynaptic branches over a great distance

223 to reach the postsynaptic cells. Rather, the processes of the postsynaptic cells traverse the neuropil and converge upon the synaptic region of the presynaptic cell. A preliminary abstract of some of these results has been published elsewhere 33. In the following paper Thompson et al ~9. examine the fine structure of L10 and its processes on the ultrastructural level. They also compare L10 to cell R2, a cholinergic cell which is thought not to be presynaptic to other cells in the ganglion. MATERIALS AND METHODS Approximately 200 small Aplysia (40-130 g) were used in the course of these experiments. Abdominal ganglia were excised from the animals and pinned to the Sylgard (Dow Corning) base of a lucite chamber as previously described7. The chamber contained 6 pairs of Ag-AgC1 stimulating electrodes and was slowly perfused with artificial seawater (Instant Ocean) at room temperature (20-22 °C). Individual nerve cell bodies were identified according to topography and physiological properties 7. The cells were penetrated with double-barrelled microelectrodes consisting of a conventional recording barrel, filled with 2 M potassium citrate, and an injection barrel. The injection barrel was prepared from clean, siliconized electrode glass and attached to a piece of polyethylene tubing through which pressure or a vacuum could be applied 18. Electrode tips were broken to a diameter of 2-5/~m giving a tip resistance of 1-5 MfL Cobalt chloride solution (1 M) was then drawn into the empty injection barrel under vacuum. The electrodes were mounted on a Prior micromanipulator and neuronal activity was monitored by means of a cathode follower and a Tektronix D12 dual beam oscilloscope. Stimuli were delivered from an Ortec 4710 dual-channel stimulator via a Grass SIU5 stimulus isolation unit. After physiological identification cobaltous chloride was pressure-injected into neurons. Following the injection the ganglion was incubated in seawater at room temperature for 1.5-2 h. Cobalt sulfide was then precipitated intraneuronally by soaking the ganglion in 3 ~ ammonium sulfide in seawater for 10-30 min ~5. The ganglion was then washed in seawater for 5 min, fixed for at least 2 h in Carnoy's fixative, dehydrated in 95 ~ and 100K ethanol and finally cleared and stored in methyl salicylate. Successfully injected neurons were photographed using a Leitz-Wetzlar 35 mm microscope camera attached to the trinocular head of a Wild M5 binocular microscope. Photographs were taken from dorsal, lateral and ventral aspects of the ganglion. Tracings of the major anatomical features of a given neuron were made from the photographs and the finer branches traced free-hand, while viewing the preparation through the Wild microscope. Drawings obtained in this manner were more accurate than those obtained using a camera lucida attached to the microscope. RESULTS

White cells of the right hemiganglion In common with the bag cells which are established neuroendocrine cells, the

224 A

L.C.

B

L.C.

R.C

R.C.

DORSAL

R.LATERAL R.C.

R.C.

R6

B.

S.

C

B

DORSAL

B

2

R,LATERAL

Fig. 1. Drawings showing some of the white cells of the right rostral quadrant of the abdominal ganglion of Aplysia cali[ornica, in situ. The cells were filled with cobalt chloride by pressure injection and the drawings prepared from whole mounts of the ganglion immersed in methyl salicylate. A: cell R3; B: R6; C: R7. A1, B1 and C1 are dorsal views, and A2, B2 and C2 are right lateral views of the abdominal ganglion. Cells R1-R8 usually lie on the dorsal surface of the ganglion, whereas R9-R13 lie ventrally. The main axons of the white cells always exit from the ganglion via the branchial nerve and are almost devoid of dendrites. Their axons run through the neuropil in a fascicle. Fine branches to the ganglionic sheath may arise from the axons or somata of the white cells. In this and the subsequent figures the following letters have been assigned to the nerves arising from the ganglion: L.C., left connective; R.C., right connective; S., siphon nerve; G., genital nerve; P., pericardial nerve (G. and P. often fused close to the ganglion but divide more distally); SP., sperrnathecal nerve (not shown unless containing an axon); B., branchial nerve.

225

A

I R.C.

L.C. __,_-

2

R.C.

_

RII

B. VENTRAL

B

R.C.

PoG.

B.

S.

R. LATERAL R.C.

L.C.

2

RIZ'~~:= /B°~ ~.~

1

5001J

S. VENTRAL

PG.

B. R.I.ATERAL

Fig. 2. Drawings of white cells R l l and R12 in situ. A1 and BI, ventral views. A2 and B2, right lateral views.

B

D

Fig. 3. P h o t o m i c r o g r a p h s of cobalt chloride injected neurons from whole m o u n t s of the a b d o m i n a l ganglion. A: dorsal view of R7. B: dorsal view of R14. C: dorsal view of L7 (rostral end of the ganglion is on the right of the picture). D : ventral view o f LI 0. T h e n a k e d axons of R7 a n d R 14 contrast m a r k e d l y with the complex dendritic trees of L7 and L10. R7 and R I 4 receive little or no synaptic input, while the input to L7 and L I 0 is extremely rich a n d varied. Scale, 250 # m .

227

t"4

¢'q

t"q

0

~1 ,-~

ee~

i

228 rostral white cells (R3-RI 3) are known to project into the connective tissue sheath of the ganglion v and are probably also neuroendocrine in character. The white cells have an endogenous pacing rhythm and fire at a rate of 0.5-1 Hz. Several of the rostral white cells were examined, and the earlier findings that they send processes into the ganglionic sheath was confirmed. These cells receive little or no evoked or spontaneous synaptic input 7. Consistent with this we found that all seem to have very few fine branches or arborizations coming off the main axons (Figs. 1, 2 and 3A). Although we cannot distinguish presynaptic from postsynaptic process in the arborization of any of these cells, we will call the total arborization the dendritic field or arbor. The efferent axons of these cells run in a discrete tract in the ventral neuropil (Figs. 1 and 2, Table I; see also ref. 7). Although the caudal white cell, R 14, exhibits a more extensive axonal branching pattern than the rostral white cells it shares many features in common with them. It occasionally displays an endogenous rhythm, contains large granules similar to those of the rostral white cells and sends processes into the ganglionic sheath v. Its axonal branching pattern is quite varied (Figs. 3B and 4, and Table I). It may have branches in the branchial, spermathecal, genital and pericardial nerves. As with the rostral white cells it has limited synaptic input 7 and a restricted dendritic arborization. In both RI4 and the rostral white cells the processes into the sheath often arise directly from the cell body and not as branches of the main axon (Fig. 4B and C). The diameters of the somata of these cells average 250-430 #m.

Cell LIO L10 is located in the left caudal quadrant of the ganglion. Its soma lies on the ventral surface medial to the insertions of the siphon nerve and genital-pericardial nerves. It is often spontaneously active, firing in an irregular pattern, and may be moderated by spontaneous IPSPs. Both its spontaneous and evoked synaptic inputs are complex v. It is most easily identified by its synaptic actions on L5, L3 or L7, which can usually be impaled from the ventral surface. Its axons invariably exits from the ganglion via the pericardial nerve (Figs. 6, 9, ll, 13, 14 and 16) and has never been seen to branch (Table II). The soma diameter of L10 varies between 250 and 325 #m. L10's dendritic tree is both complex and dense and extends dorsally and rostrally. In many cases a cluster of dendrites arises from a small branch on the soma (Figs. 6A, 9A-D, 13A and B, 14, 16A and B). In other cases most of the dendritic mass arises from the axon hillock. In addition to the very rich main dendritic mass many fine fibers arise from the main axon as it passes out of the ganglion. Some of these fibers actually arise within the pericardial nerve itself (e.g. Figs. 11 and 14). The same general area of neuropil is always occupied by the dendrites of LI0. The dendrites filled with cobalt have never been seen to enter areas of neuropil more than 1.5 soma diameters away from its soma and only rarely does a process cross the commissure. Although L10 is structurally limited to the left caudal quadrant (see ref. 8), its synaptic effects are distributed to cells dispersed throughout the ganglion.

Inhibitory follower cells" of the left rostral quadrant The identified cells L1-L6 all receive inhibitory synaptic input from L10. They

229 A

L.C.

R.C.

R.C.

I

S. ~=:'/ ~

.

B.

R. LATERAL

DORSAL

B

L.C.

R.C.

R.C. 2

--

B.

DORSAL C

L.C.

R.LATERAL R.C. 2

S.

G.P.

R.C.

B.

DORSAL

R.LATERAL

Fig. 4. Drawings of R14 in situ. A 1 - C I : dorsal views; A2-C2" views drom the right lateral aspect. Like the rostral white ceils, R14 possesses few dendrites, but has a much more complex axonal branching pattern. It also sends fine branches into the ganglionic sheath.

also receive excitatory inputs from the maj or ganglionic nerve trunks following stimulation 7 and cell L2 (and to a slightly lesser degree L4) have considerable spontaneous synaptic activity. L2-L4 and L6 are usually bursting cells, while L1 and L5 are usually silent. The diameters of their somata range from 250 to 400/~m. Their branching patterns are summarized in Table II.

230

I~

r~ i

¢-q

0

k~

2 °~

,.d Q ~

231 Cell L1. The soma of L1 is almost always found in the left dorso-lateral corner of the ganglion to the left of the left connective. Unlike cells L2-L6 it does not usually have a branch in the pericardial nerve. It invariably sends its main axon out of the left connective and has a fairly extensive dendritic tree arising from the proximal region of its axon (Figs. 5A and 6A). Its dendrites extend ventromedially in the dorsal neuropil, but do not appear to contact those of L10 (Fig. 6A). Its known inhibitory connection to L10 is not explained by our data. We presume that some fine branch of either cell could not be filled with COC12 or that an intermediate neuron, electrotonically coupled to L10, exists. CellL2. Cell L2 is structurally the most complex cell of the identified cells of the left rostral quadrant. Its soma usually lies medial to that of L1 on the dorsal surface of the ganglion in line with the left connective (Figs. 5B1, 5B3 and 6B), but has been found close to the midline of the ganglion (Fig. 5B2). Its axon is always multibranched, invariably sending a process down the pericardial nerve and usually branching into the spermathecal nerve. Its other branches are rather more variable, but they often pass out of the siphon and genital nerves and more rarely out of thebranchial nerve and left connective. Cell L2 possesses an extensive proximal dendritic tree and a second sparser set of dendrites more distally in the neuropil of the left hemiganglion usually within L10's dendritic field. All its branches pass dorsally through L10's dendritic field (Fig. 6B) so that one or more synaptic contacts from LI0 to L2 could easily be made in this region, while other contacts could occur in the pericardial nerve. Cells L3-L6. These cells form a fairly homogenous group in both morphological and functional terms (Figs. 7 and 8, Table II). The soma of L3 lies medial to that of L2 and may occasionally occur on the ventral surface of the ganglion (Fig. 7). It is always very close to the midline. L4's cell body usually occurs posterior and slightly medial to that of L2 (Fig. 8B), but may lie more laterally (Fig. 8A). The soma of L5 is the only one of this group always found on the ventral surface of the ganglion. It lies posterior to L3 very close to the midline of the ganglion (Fig. 8C and D). Cell L6 lies on the dorsal surface close to the midline, posterior to L3 and medial to L4 (Fig. 8E and F). In common with the main axon of L2, the axons of L3-L6 run in a fascicle (Fig. 14) along the ventral surface of the neuropil (Fig. 7E) and exit from the ganglion into the pericardial nerve together with the axon of L10. In only one case (one L4 out of 9 ceils examined) has one of these cells sent a branch into another nerve (the genital) besides the pericardial nerve. The cells are relatively simple in form, but do exhibit a limited variability in that some bifurcate in the neuropil and others in the pericardial nerve (Figs. 7C and E, 8F and Table III). The variability is, however, restricted since the two branches appear to follow a parallel course out of the ganglion. All the ceils examined were found to have an extensive dendritic field on the proximal region of the axon and in most cases this dendritic field extended distally to the posterior border of the neuropil. The axons of L3-L6 all pass through the network of dendrites emanating from L10 and other possible points of contact occur in the pericardial nerve (Fig. 9). Cell LT. Cell L7 is located in the left caudal quadrant of the ganglion. It is spontaneously active, receives both excitatory and inhibitory input from L10 and

232 ¢q

..J d

a5

~

/ i

(

o

a~

.u 0 r~

o5

ro n"

t'kl

t5 tY

m ¢q

t'~

.-J

~

o5

._1

O5

t6

m

_

_

Fig. 5. Dorsal views o f L1 and L2 in situ. A1 and A2: cell LI ; B1-B3: cell L2. Unlike the other cells o f the left rostral quadrant L1 projects out o f the left connective and not the pericardial nerve. L2's multiple axonal branching pattern is also quite distinctive.

233

L.C.

_.C.R.C.

A

R.C°

2

LI

LIo

~B. S.

G.P. DORSAL

L. LATERAL L.C. R.C. B

I

L.C.

R.C.



Lz

LIO

S. G. P.B. L. LATERAL

S. G. P.

B. DORSAL

Fig. 6. Drawings of L10 and its follower cells L1 (A) and L2 (B) in situ. A1 and B1 : left lateral views of the ganglion. A2 and B2: dorsal views. Although L1 and L10 are known to be in physiological contact, their injected dendritic fields do not overlap and no marked branches appear to pass from one cell to the other. The axons of L2 pass close to L10 except for the variant supernumerary axon in the left connective.

234

r~

at _.1

n.

8

0 ...I

~6 _J

o.

t6

at

_J

09

..J

W

_7

O

-I

~o

cn

_a

(.9

v~

t~ m

of) at O Q •

Fig. 7. D r a w i n g s o f several e x a m p l e s o f L3, s h o w i n g variability f r o m one preparation to the next. A D a n d E (right): dorsal views. E (left): left lateral view.

235

A

R.C.

L.C.

B

B. C

L.C.

R.C.

LC,

RC.

S . p ~

L4

G.

t B.

D

L.C.

R.C.

L5

L5

S. G.P. DORSAL E

L_.c

L6

B.

S.

G.R DORSAL

R.C

F

L.C.

B. R.C.

.

P.

DORSAL

B.

S. G.P.

DORSAL

B.

Fig. 8. Drawings of IA (A, B), L5 (C, D) and L6 (E, F) from the dorsal aspect.

236 TABLE llI Branching patterns oI L3--L6 Cell

Number injected

B~furcation Neuropil

No bifurcation Pericardial nerve

L3 L4 L5 L6

17 7 11 13

5 2 5 7

4 1

8 5 6 5

Total

48

19

5

24

extensive spontaneous inhibitory and excitatory inputs from other sources. It is a motor neuron for the gill and mantle shelf and is involved in the defensive withdrawal reflex of these mantle organs 10. It receives both direct input and indirect (via interneurons) from the mechanoreceptor sensory neurons of the siphon skin 3. The diameter of the soma of L7 varies between 250 and 400 #m. It lies on the left lateral surface of the ganglon posterior to LI and can be viewed from either surface of the ganglion. Its main axon usually passes posteriorly and medially through the neuropil until it reaches the level of the commissure at which point it divides into several branches. The largest axonal branch usually traverses the commissure and exits via the branchial nerve, while smaller branches pass out of the siphon, pericardial and genital nerves (Fig. 10). An axonal branch of L7 is occasionally (2/12) found in the right connective (Fig. 10B). As might be expected from a cell with a highly complex neuronal input, L7 usually possesses a rich dendritic field (Figs. 3C and 10). The densest feltwork of dendrites is found proximally and tends to lie horizontally in the neuropil. Smaller clusters of dendrites on the main axonal branches are also found in both the ipsilateral and contralateral neuropil. Like those of L10, the dendrites of L7 occupy a relatively constant volume of neuropil from preparation to preparation. As can be seen from Fig. 11, the morphological relationships between L10 and L7 are complex. Their dendritic fields tend to overlap and mingle with one another. Fine lateral branches are found on the axons of both cells as they enter the pericardial nerve, thus increasing the possible sites of synaptic interaction. Multiple connections from L10 to L7 are thus a strong possibility (ref. 22). Cells o f the right caudal quadrant

Both R 15 and the RB cell cluster are located in the right caudal quadrant of the ganglion. R15 is a whitish cell and lies dorsally and medially at the caudal pole of the ganglion. Its soma diameter varies between 375 and 500 #m. The RB cells studied here lie close to R15, but more anteriorly. Other RB cells lie on the ventral surface of the ganglion. Those studied ranged between 140 and 220 #m in diameter. R15 is a spontaneously active neuron possessing an endogenous bursting rhythm which may be modulated by spontaneous or evoked PSPs 7,11,24. It receives excitatory input from L10.

237

u.

-

J

!

M n.

--

ro

j

o

M

_

0

o

J~

o6

J

Fig. 9. Drawings o f L3-L6 paired with L10. A 1 - D I : left lateral views. A 2 - D 2 : dorsal views. The

axons of these left rostrai quadrant cells run through the ventral neuropil, as does the main axon of L2. L10's dendritic field is variable in its pattern, but occupies approximately the same volume of neuropil in each case.

The morphology of R15 is rather variable (Fig. 12). It always sends an axonal branch out of the pericardial nerve and other branches are usually found in the siphon nerve, genital nerve, spermathecal nerve and left connective. This latter branch has not previously been detected electrophysiologically TM. An antidromic spike from the branch in the left connective was only discernible using a high magnesium saline to block synaptic transmission. R15 very occasionally sends a branch down the branchial nerve. The number of axonal branches projecting from R15 down a particular nerve

238 =k O O tO

.d

\ t,1)

m

r~ •

a,

o.

m

-~ r~