JOURNAL OF NEUROBIOLOGY, VOL. 6, NO. 5, PP. 463-474
Ultrastructural Correlates of Interneuronal Function in the Abdominal Ganglion of Aplysia californica RHANOR GILLETTE"
and BRUCE POMERANZ
Department of Zoology, University of Toronto, Toronto, Canada
SUMMARY
The synaptic inputs and outputs of the major interneuron L10 of the abdominal ganglion of Aplysia were studied using an intracellular staining technique for the electron microscope. The sites of both the chemical synaptic input and output of L10 are localized to the dendritic arborizations that arise from the axon in the ganglion neuropil. Thus, the interneuronal functions are mediated at the dendritic processes and could occur in the absence of spiking in the axon and cell body. The sites of L10 synaptic output are presumed to be at aggregations of vesicles and mitochondria in the dendrites. The synaptic vesicle content of L10, a cholinergic neuron, with many large dense vesicles resembles that described for serotonergic cells in Aplysia, making distinction of synaptic pharmacology by ultrastructure difficult. Focal membrane specializations with a clear synaptic cleft were not observed between L10 and its large population of postsynaptic cells. In contrast, clear focal input sites were frequently found on L10. Gap junctions, sites of probable electrical coupling between L10 and other neurons, were also found. These observations are discussed as evidence that many synapses do not have focal specializations. INTRODUCTION
Correlating morphology with the physiological characters of a neuron helps to define the functional role of that neuron as a circuit element in a network. For example, knowledge of the sites of synaptic inputs and outputs within the geometry of the neuron is important because these relations affect how synaptic signals are integrated and coded into trains of action potentials (Rall, 1959; Purpura, 1972); the structure of synaptic terminals can often be correlated with pharmacology and physiological actions (Bodian, 1966; Atwood, Lang, and Morin, 1972); and anatomical mapping of axon pathways and terminations certifies and extends physiological data. This report documents * Present address: Thimann Laboratories, Division of Natural Sciences, University of Natural Sciences, University of California, Santa Cruz, California 95060. 463 01975 by John Wiley & Sons, Inc.
464
GILLETTE AND POMERANZ
the fine structure of a cholinergic molluscan interneuron and attempts to relate structure to the known physiology of the neuron. The abdominal ganglion of Aplysia provides an identifiable major interneuron, known as L10, whose physiology and pharmacology has been well described. The synaptic transmitter of L10 is acetylcholine (Giller and Schwartz, 1971). L10 is presynaptic to an estimated 40 neurons in the ganglion; some of these neurons receive excitatory postsynaptic potentials from L10, others receive inhibitory postsynaptic potentials from L10 (Kandel, Frazier, Waziri, and Coggeshall, 1967), and one neuron receives a biphasic excitatory/inhibitory postsynaptic potential (Blankenship, Wachtel, and Kandel, 1971). Two neurons have been found to be directly electrically coupled to L10 (Waziri, 1969, 1971). L10 also receives synaptic excitation and inhibition from other neurons of the nervous system. Behaviorally, L10 has a significant role as a cardiac “command” interneuron, modulating heartbeat (Koester, Mayeri, Liebeswar, and Kandel, 1974).
METHODS The ultrastructure of L10 and its synaptic relations with other neurons were studied via an intracellular staining technique for the electron microscope; neurons are selectively stained by an osmium-binding polymer which is catalytically generated by intracellularly injected cobalt ions (Gillette and Pomeranz, 1973a). For this study, both L10 and a postsynaptic neuron were stained. The abdominal ganglia were dissected from two healthy Aplysia californica weighing approximately 200 g. Ganglia were mounted under Aplysia saline (Sato, Austin, Yai, and Maruhashi, 1968) a t room temperature. Conventional electrophysiological techniques were used for intracellular recording; a Brush chart recorder was used for making permanent records of neuronal electrical activity. In one ganglion, the neuron L10 was penetrated with a glass microelectrode filled with 1 M cobaltous chloride, with a tip diameter of 2-3 pm. The identity of LlO was confirmed by penetrating an adjacent neuron with a 3 M KC1-filled electrode and recording postysynaptic potentials simultaneous with action potentials in L10 (Kandel e t al., 1967). L10 was then pressure-injected with cobalt chloride until all spontaneous electrical activity ceased, input resistance dropped about 5096, and the cell appeared greyish as seen in the dissecting microscope. In another ganglion, both L10 and a rostrally adjacent follower neuron were injected with cobalt. This neuron was found to receive hyperpolarizing postsynaptic potentials of 1-5 mV amplitude from L10 (Fig. 1). The physiology and position of this neuron identify it as the cell L12 by the criteria of Frazier, Kandel, Kupfermann, Waziri, and Coggeshall (1967). L12 was injected with somewhat less cobalt than L10. After injection, ganglia were left a t 15°C for 45 min to allow diffusion of cobalt throughout the neurons. Ganglia were then incubated for 25-35 min in a solution of 1 mg/ml diaminohenzidine tetrahydrochloride (DAB) in Aplysia saline adjusted to pH 7.5 with NaOH. NaOH was used because previously Tris, phosphate, and bicarbonate buffers were found to inhibit cobalt-catalyzed polymerization of DAB. During incubation the injected cells turned dark blue, indicating formation of the osmiophilic polymer. Ganglia were rinsed 20 min in saline alone, then fixed for 15 min in a cold mixture of 2% formaldehyde, 1% glutaraldehyde, 4% sucrose, and 3% NaCl in 0.1 M sodium phosphate buffer pH 7.4, according to the method of Sherman and Atwood (1972) for marine tissue. The left caudal quarter ganglion containing the injected cells was cut out with a razor blade and left in the fixative for 12-36 hr a t 4OC. This ganglion region encompasses all, or nearly all, of the dendritic field of L10 (Gillette and Pomeranz, 1973b, and further unpublished observations). After a 1 hr rinse in buffer containing 8% sucrose, the tissues were postfixed in
INTERNEURON STRUCTURE IN APLYSIA
465
Fig. 1. Discharge of L10 (lower trace) causes a hyperpolarizing postsynaptic potential in L12 (upper trace). The fast biphasic inflection (arrow) on top of the synaptic potential in L12 is an artifact of the L10 spike due to some anomalous rectified coupling in the amplifier circuits. 1%OsOd in a buffer containing 4% sucrose. The tissues were dehydrated, embedded in Spurr's resin, and mounted for sectioning. Along the entire course of L10 through the neuropil and out a nerve serial thick sections (1and 2 fim) were cut and stained with 1% methylene blue. At 10 or 20 fim sectioning intervals 25-30 ultrathin silver sections were taken and mounted on formvarcoated 50-mesh or celloidin-coated 75-mesh copper grids. Serial order of the ultrathin sections was maintained from ahout one-fourth of the sampling regions. In this way, 16 and 18 sectioning depths past the cell body layer into the neuropil were sampled from the two ganglia for electron microscopy. Grids were stained with uranyl acetate and lead citrate and then examined in a Phillips 200 electron microscope equipped with a 35 mm camera.
RESULTS
Gross neuron morphology In the light microscope the cell body and axon of L10 in both ganglia studied were easy to find and trace in the light microscope because the DAB polymer outlined the membrane and the axoplasm stained somewhat more lightly than that of other cells. The finer processes of L10 could be easily traced down to a size of 4-5 pm. The axon of L12 and its fine processes did not contrast as well and tracing of finer processes required closer examination. The general morphology of L10 has been previously described (Frazier et al., 1967; Gillette and Pomeranz, 1973b). In the two ganglia studied here a large furrowed and invaginated axon of irregular diameter (30-50 pm) arises from the cell body to enter the neuropil core of the ganglion and course out the pericardial nerve trunk, Fine processes (diam 1-5 pm) arise from the axon along its entire course in the neuropil, however, the large majority of the many fine processes of L10 are ramifications of a few large branches or dendrites (diam 10-15 pm) arising from the axon near the base of the soma (Fig. 2). The fine processes of L10 may extend to several hundred micrometers. In the ganglion where both L10 and L12 were stained, the axon of L12 is smaller and more irregular in cross section than that of L10, and sprouts fewer, thinner, and generally shorter processes. The L12 axon trunk does not contact the L10 axon trunk as it courses to the pericardial nerve. A
466
GILLETTE AND POMERANZ
Fig. 2. Freehand reconstruction of the neurons L10 and L12 from light microscopy of serial sections. Only the larger processes are drawn while the extensive finer ramifications are generalized or omitted. Most of the fine processes of L10 arise from a few large branches near the cell body. A single large process of L12 ramifies in the same area as does a process of L10 (rectangle); an electron micrograph from this area appears as Fig. 4.
branch of 5 pm diameter arises fiom the axon of L12, and courses about 200 pm to ramify and interdigitate with the twigs of a similar branch of L10, as illustrated in Figure 2.
Ultrastructural studies The characteristics of the staining technique have already been described (Gillette and Pomeranz, 1973a). Generally, the processes of the injected neuron are identified by the darkly staining polymer of DAB which appears as amorphous clumps in the axoplasm, or lines the inner surface of the neuron membrane. The axoplasm is lighter than that of unstained cells due to a slight condensation of the fibrillaryt matrix. Stained neurons also show mitochondrial distortion. The amorphous polyrrer contrasts best a t low magnifications (2,000-8,000X), there a well-stained process of an injected neuron is unmistakable. The ultrastructure of unstained neurons shows no ill effects from incubation in DAB and the character of synaptic'vesicles appears no different from control tissues fixed immediately upon dissection (Gillette, 1974). Synaptic input to L10. Small vesicle-filled processes contacting L10 were found much more often at the fine processes of L10 rather than a t the L10 axon. However, synapses with well-defined synaptic membrane specializations consisting of a widened, even synaptic cleft and moderate membrane and intermembrane densities by a polarized cluster of vesicles were only found synapsing onto fine processes and were never found on the main axon of L10 (Figs. 3a,b). Usually four or five of these well-defined synaptic inputs to L10 could be located in any given ultrathin section through the quarterganglion neuropil. Serial sections through such synapses showed that they are discrete en passant junctions with an average width of about 1 pm. Well-defined synaptic input to L12 was found to be qualitatively similar, except that synapses were found occasionally on the axon trunk.
INTERNEURON STRUCTURE IN APLYSIA
467
A variety of vesicle types occur in the processes presynaptic to L10; this variety is common to many other processes of the neuropil. Arbitrary classification of the vesicle types is not realistic, since the vesicles were found to lie on a continuum of morphology. However, for convenience, we denote four types in the continuum: dense, granular, round, and ellipsoid, as described in Table 1. Vesicles intermediate in size, shape, and content to the dense and the round types, the dense and the granular types, the round and the granular types, and the round and the ellipsoid types are common. Many combinations of different vesicle types and their intermediates could be found to occur when scans were made of many fine neuropilar processes. Attempts to classify synaptic junctions by vesicle content of the presynaptic process were only partially successful because ( I ) several types of vesicle are often found in any process, and (2) analysis of serial sections showed that the distribution of vesicles within a process is inhomogeneous (Figs. 3a,b) so that sectioning plane may determine apparent vesicle content. Synapses onto L10, and also synapses found elsewhere in the neuropil, mostly show the dense and the round types of vesicles together; less commonly, presynaptic processes contain predominantly the granular and the ellipsoid types. Correlates of L10 synaptic output. Fine processes of L10 are often found to contain conglomerates of vesicles and mitochondria. The conglomerates are most striking in irregular swellings of the dendrites up to 5 or 10 pm in diameter and can be recognized in the light microscope by their high
Fig. 3 (continued)
GILLETTE AND POMERANZ
468
Fig. 3. Nearby sections from a complete series through a synapse onto a fine process of L10. The synapse shows a well-defined synaptic cleft (cl). The apparent vesicle content of the presynaptic process (P) varies with the section plane. (a) Vesicles are mainly of the round (r) type (see text). (b) Vesicles are of the round (r), dense (d), and granular (gr) types. Calibration bar is 0.5 um. TABLE 1 Vesicle Types Found in the Fine Processes of the Neuropil of the Abdominal Ganglion Vesicle type Round Ellipsoid
Granular Dense
Description Round, diam 500-700 8,contents uniform and moderately electron-dense. Ellipsoid or flattened, common dimensions 290 x 650-360 x 480 contents uniform and moderately electron-dense. Round, diam 500-725 8, containin a diam. darkly staining granule 200-300 Round to slightly ellipsoid, diam 700-1200 8,contents uniform and very electrondense.
a,
f
basophilia. The DAB polymer is more dense in the cytoplasm among conglomerates than in cytoplasm elsewhere, suggesting that there are more divalent cation-binding sites which bind cobalt in these regions. The presumed synaptic vesicles are mainly of the dense and the round types (Fig. 4);granular vesicles are rarely present. No focal synaptic membrane specializations were ever found for the known cholinergic output of L10, although sections
INTERNEURON STRUCTURE IN APLYSIA
469
Fig. 4. A fine process of L10 is filled with vesicles of the dense (d) and round (r) types, with a few granular (g) vesicles. Mitochondria1 profiles (m) are frequent. The profuse DAB polymer in this process of L10 should not be confused with glycogen (gly) in other neurons. Calibration bar is 1 Fm.
were searched intensively in each sampling region. Scattered dense vesicles were also found in the cell body and main axon of L10. Large accumulations of vesicles and mitochondria could also be found occasionally in processes of
470
GILLETTE AND POMERANZ
Fig. 5. A process of L10 is found in direct physical contact with a process of its postsynaptic neuron, L12. Several such contacts were found within the area indicated in Fig. 1. Calibration bar is 2 l m .
unstained neurons, indicating this clustering of organelles is not an artifact of cobalt injection. The juncture of L10 and L12. Fine branches of the long process arising from the axon of L12 were found to come into direct apposition with twigs of a branch of L10 (Figs. 2 and 5). Figure 5 shows that the axon and fine processes of L12 can be distinguished from those of L10 in the electron microscope, presumably because L12 was injected with less cobalt than L10. However, the distinction between L10 and L12 is a subtle one based mostly on axoplasmic texture. Therefore, as the branches of L12 could not be traced in the serial sections in the light microscope far beyond the region shown in Figures 2 and 5, certain identification of distal sites of synaptic contact was not made. Probable electrotonic connections of L10. Gap junctions were found between L10 and the fine processes of other neurons in both ganglia studied
INTERNEURON STRUCTURE IN APLYSIA
47 1
Fig. 6. (a) A gap junction (arrow) indicates electrotonic coupling between fine processes of L10 and another neuron. Cobalt has crossed the junction and caused DAB polymer formation in both processes. Analysis of adjacent sections allowed L10 identification. Calibration bar is 0.5 wm. (b) The symmetrical junction of (a) shows a typical 7-layered structure and a 20 A “gap.” Calibration bar is 0.1 um.
(Figs. 6a,b). Two gap junctions were found a t the axon trunk of L10 in one ganglion and one gap junction was found a t a fine process in the second ganglion. Cobalt ions from L10 appear to have crossed the junctions and cata-
472
GILLETTE AND POMERANZ
lyzed polymer formation in the probable electrical followers of L10. Analysis of adjacent ultrathin sections tracing the processes away from the gap junctions showed that intensity of the DAB stain decreased rapidly over a distance of 10-15 pm on one side of the junction, indicating rapid dilution of cobalt ions crossing the junction from L10. DISCUSSION
These morphological findings indicate that L10 must mediate its cholinergic synaptic output via the fine processes of its dendritic tree. This follows from the findings of the rarity of axo-axonal appositions, the observed apposition between L10 and its postsynaptic neuron L12 only at their fine processes, and the presence of the masses of presumed synaptic vesicles in tlhe fine processes of L10. However, the sites of L10 output could not be correlated with the focal membrane specializations which have been used as the morphological criterion for identification of synapses in this and in previous studies of molluscan neuropil (Gerschenfeld, 1963; Coggeshall, 1967). There are three immediate explanations. First, intracellular cobalt on the presynaptic side may have destroyed them; second, L10 output sites could be so rare or highly localized that they were missed by the sampling procedure; and third, L10 output sites are not characterized by focal membrane specializations. The first possibility has not been answered by direct test, but is unlikely because such specializations have bec,ii found (in another mollusc) to include rigid, mechanical intercellular junctions (Jones, 1970) not apparently obliterated by high divalent cation concentration (Hillman and Llinfis, 19743 and not exposed to the intracellular milieu. The second possibility is unlikely on morphological grounds, since large masses of synaptoid vesicles were found in different processes of L10 in widely separate neuropil regions and in any given section through the neuropil. It is also improbable on physiological grounds. L10 is a command interneuron with a large postsynaptic population (Frazier et al., 1967) endogenously capable of patterning the cardiac rhythm independent of synaptic input (Koester et al., 1974). It is doubtful that so many well-defined input sites could have been found to the exclusion of even a single outpdt site. Therefore, the likely alternative is that L10 output sites lack clear membrane specializations. Thompson, Schwartz, and Kandel (1973), in another study of L10 fine structure also did not report such specializations. Synaptic membrane specializations serve in vertebrate and invertebrate neurons as foci of vesicle release (Pappas and Waxman, 1972). The apparent lack of clear focal output sites could reflect that L10 synaptic output is a nonfocal occurrence over large areas of the fine processes. Finally, these findings support previous conjecture that synapses in molluscan neuropil are more frequent than is indicated by the sparseness of synapses that are structurally akin to well-studied vertebrate synapses (Coggeshall, 1967; Cobb and Mulilins, 1973).
INTERNEURON STRUCTURE IN APLYSIA
473
As the synaptic inputs to L10 are also found at the fine processes of the dendritic tree, the interneuronal transactions of L10 must be mediated there. The proximity of synaptic input and output sites is reminiscent of the granule cells of the vertebrate olfactory bulb, where action potential generation is not necessary to synaptic transmission (Shepherd, 1970). The large constants of electrotonic decay (A) of Aplysia neurons further assure the electrical proximity of input and output sites (Graubard, 1973). Thus, although it is not known if the fine processes are capable of supporting action potentials, it is likely that the interneuronal function of L10 is not necessarily dependent on spike initiation or invasion in the axon trunk or cell body. Depolarizing excitatory dendritic input could cause synaptic output either by direct electronic spread, or by local initiation of an active membrane response. Discovery of gap junctions on L10 provides the probable morphological substrate for the physiological finding that L10 is directly electrically coupled to other neurons of the ganglion (Waziri, 1969, 1971). Three characteristics of electrotonic junctions listed by Pappas and Waxman in their review (1972) are evident: close apposition of opposing cell membranes, symmetry, and the ability to pass small ions, in this case, cobalt ions. No strong morphological evidence could be drawn from this study for the known existence of different kinds of synapses on L10 (Kandel et al., 1967) based on vesicle appearance. The morphological continuum of vesicle types found in the processes of the neuropil could represent a functional continuum for transmitter release and vesicle recycling. Thus, the frequency of different types in any neuron process could reflect the state of physiological activit y at the time of fixation. It is important to note that the vesicle content of L10, a cholinergic neuron, shows marked similarity to that described for serotonergic neurons in Aplysia (Taxi and Gautron, 1969; Weinreich, McCaman, McCaman, and Vaughn, 1973). We conclude that there is presently little basis for inferring transmitter pharmacology or synaptic action in molluscan brain solely from ultrastructural observation. A possible future exception may come from pharmacological identification of those synapses which show focal membrane specializations. The authors thank Dr. Harold L. Atwood for his expert advice, encouragement, and generous loan of equipment during the course of this study. Much of the data presented here was the subject of a talk presented at the Society for Neuroscience Third Annual Meeting.
REFERENCES ATWOOD, H. L., LANG,F. and MORIN,W. A. (1972). Synaptic vesicles: Selective depletion in crayfish excitatory and inhibitory axons. Science 176: 1353-1355. J. E., WACHTEL,H. and KANDEL,E. R. (1971). Ionic mechanisms of excitatoBLANKENSHIP, ry, inhibitory, and dual synaptic actions mediated by an identified interneuron in abdominal ganglion of Aplysia. J. Neurophysiol. 34: 76-92. BODIAN,D. (1966). Electron microscopy: Two major synaptic types on spinal motoneurons. Science 151: 1093-1095. CORB, J. L. S., and MULLINS,P. A. (1973). Synaptic structure in the visceral ganglion of the lamellibranch mollusc, Spisula solida. 2. Zellforsch. 138: 75-83.
474
GILLETTE AND POMERANZ
COGGESHALL, R. (1967). A light and electron microscope study of the abdominal ganglion of Aplysia californica. J. Neurophysiol. 3 0 1263-1287. FRAZIER,W. T., KANDEL,E. R., KUPFERMANN,I., WAZIRI,R. and COGGESHALL, R. E. (1967). Morphological and functional properties of identified neurons in the abdominal ganglion of Aplysia californica. J. Neurophysiol. 3 0 1288-1351. GERSCHENFELD,H. M. (1963). Observations on the ultrastructure of synapses in some pulmonate molluscs. Z. Zellforsch. 6 0 258-275. J. H. (1971). Choline acetyltransferase in identified neurons of the GILLER,E. and SCHWARTZ, abdominal ganglion of Aplysia californica. J. Neurophysiol. 34: 94-108. GILLETTE, R. (1974). Microstructural and ultrastructural studies on identified neurons of the abdominal ganglion of Aplysia californica. Thesis, University of Toronto. GILLETTE,R. and POMERANZ, B. (1973a). Neuron geometry and circuitry via the electron microscope: Intracellular staining with osmiophilic polymer. Science 182 1256-1258. GILLETTE,R. and POMERANZ, B. (197313). A study of neuron morphology in Aplysia califorrtica using Procion Yellow dye. Comp. Biochem. Physiol. 4 4 A 1257-1259. GRAUBARD, K. (1973). Electrotonic decrements within Aplysia neurons. Abstracts of the Society for Neuroscience, Third Annual Meeting. HILLMAN,D. E. and LLINAS, R. (1974). Calcium-containing electron-dense structures in the axons of the squid giant synapse. J . Cell Biol. 61: 146-155. JONES, D. G. (1970). A further contribution to the study of the contact region of Octopus synaptosomes. Z. Zellforsch. 103: 48-60. R. E. (1967). Direct and cornKANDEL,E. R., FRAZIER,W. T., WAZIRI,R. and COGGESHALL, mon connections among identified neurones in Aplysia. J . Neurophysiol. 3 0 1352-1376. KOESTER,J., MAYERI,E., LIEBESWAR,G. and KANDEL,E. R. (1974). Neural control of circulation in Aplysia. 11. Interneurons. J. Neurophysiol. 37: 476-496. PAPPAS,G. D. and WAXMAN,S. G. (1972). Synaptic fine structure: Morphological correlates of chemical and electrotonic transmission. In: Structure and Function of Synapses. G. D. Pappas and D. P. Purpura, Eds., Raven Press, New York, pp. 1-44. PURPURA,D. P. (1972). Intracellular studies of synaptic organizations in the mammalian brain. Zbid., pp. 257-302. RALL, W. (1959). Branching dendritic trees and motoneuron resistivity. Exptl. Neurol. 1: 491-527. SATO,M., AUSTIN,G. M., YAI, H. and MARUHASHI,J. (1968). The ionic permeability changes during acetylcholine-induced responses of Aplysia ganglion cell. J . Gen. Physiol. 51: 321-345. SHEPHERD,G. M. (1970). The olfactory bulb as a simple cortical system. In: The Neurosciences, Second Study Program. F. 0. Schmitt, Ed., Rockefeller University Press, New York, pp. 539-551. SHERMAN,R. G. and ATWOOD,H. L. (1972). Correlated electrophysiological and u1trastructu:ra1 studies of a crustacean motor unit. J. Gen. Physiol. 5 9 586-615. TAXI,J. and GAUTRON,J. (1969). Donnkes cytochimiques in faveur de l’existence de fibres nerveuses serotoninergiques dans le coeur de I’hplysie, Aplysia californica. J . Microscop. 8: 627-636. J. H. and KANDEL,E. R. (1973). Autoradiographic analysis THOMPSON,E. B., SCHWARTZ, with the light and electron microscope of Aplysia identified neurons, their processes, and synapses after intrasomatic injection of 3H-L-fucose. Abstracts of the Society for Neuroscience, Third Annual Meeting. WAZIRI,R. (1969). Electrical transmission mediated by an identified cholinergic neuron of Aplysia. Life Sci. 8: 469-476. WAZIRI,R. (1971). Electrotonically coupled interneurones produce two types of inhibition in Aplysia neurones. Nature New B i d . 232: 286-288. R. E. and VAUGN,J. E. (1973). Chemical, enWEINREICH,D., MCCAMAN,M. W., MCCAMAN, neurons from zymatic and ultrastructural characterization of 5-hydroxytryptamine-containing the ganglia of Aplysia californica and Tritonia diomedia. J. Neurochem. 20: 969-976. Accepted for publication January 21,1975
Ultrastructural Correlates of Interneuronal Function in the Abdominal Ganglion of Aplysia californica RHANOR GILLETTE"
and BRUCE POMERANZ
Department of Zoology, University of Toronto, Toronto, Canada
SUMMARY
The synaptic inputs and outputs of the major interneuron L10 of the abdominal ganglion of Aplysia were studied using an intracellular staining technique for the electron microscope. The sites of both the chemical synaptic input and output of L10 are localized to the dendritic arborizations that arise from the axon in the ganglion neuropil. Thus, the interneuronal functions are mediated at the dendritic processes and could occur in the absence of spiking in the axon and cell body. The sites of L10 synaptic output are presumed to be at aggregations of vesicles and mitochondria in the dendrites. The synaptic vesicle content of L10, a cholinergic neuron, with many large dense vesicles resembles that described for serotonergic cells in Aplysia, making distinction of synaptic pharmacology by ultrastructure difficult. Focal membrane specializations with a clear synaptic cleft were not observed between L10 and its large population of postsynaptic cells. In contrast, clear focal input sites were frequently found on L10. Gap junctions, sites of probable electrical coupling between L10 and other neurons, were also found. These observations are discussed as evidence that many synapses do not have focal specializations. INTRODUCTION
Correlating morphology with the physiological characters of a neuron helps to define the functional role of that neuron as a circuit element in a network. For example, knowledge of the sites of synaptic inputs and outputs within the geometry of the neuron is important because these relations affect how synaptic signals are integrated and coded into trains of action potentials (Rall, 1959; Purpura, 1972); the structure of synaptic terminals can often be correlated with pharmacology and physiological actions (Bodian, 1966; Atwood, Lang, and Morin, 1972); and anatomical mapping of axon pathways and terminations certifies and extends physiological data. This report documents * Present address: Thimann Laboratories, Division of Natural Sciences, University of Natural Sciences, University of California, Santa Cruz, California 95060. 463 01975 by John Wiley & Sons, Inc.
464
GILLETTE AND POMERANZ
the fine structure of a cholinergic molluscan interneuron and attempts to relate structure to the known physiology of the neuron. The abdominal ganglion of Aplysia provides an identifiable major interneuron, known as L10, whose physiology and pharmacology has been well described. The synaptic transmitter of L10 is acetylcholine (Giller and Schwartz, 1971). L10 is presynaptic to an estimated 40 neurons in the ganglion; some of these neurons receive excitatory postsynaptic potentials from L10, others receive inhibitory postsynaptic potentials from L10 (Kandel, Frazier, Waziri, and Coggeshall, 1967), and one neuron receives a biphasic excitatory/inhibitory postsynaptic potential (Blankenship, Wachtel, and Kandel, 1971). Two neurons have been found to be directly electrically coupled to L10 (Waziri, 1969, 1971). L10 also receives synaptic excitation and inhibition from other neurons of the nervous system. Behaviorally, L10 has a significant role as a cardiac “command” interneuron, modulating heartbeat (Koester, Mayeri, Liebeswar, and Kandel, 1974).
METHODS The ultrastructure of L10 and its synaptic relations with other neurons were studied via an intracellular staining technique for the electron microscope; neurons are selectively stained by an osmium-binding polymer which is catalytically generated by intracellularly injected cobalt ions (Gillette and Pomeranz, 1973a). For this study, both L10 and a postsynaptic neuron were stained. The abdominal ganglia were dissected from two healthy Aplysia californica weighing approximately 200 g. Ganglia were mounted under Aplysia saline (Sato, Austin, Yai, and Maruhashi, 1968) a t room temperature. Conventional electrophysiological techniques were used for intracellular recording; a Brush chart recorder was used for making permanent records of neuronal electrical activity. In one ganglion, the neuron L10 was penetrated with a glass microelectrode filled with 1 M cobaltous chloride, with a tip diameter of 2-3 pm. The identity of LlO was confirmed by penetrating an adjacent neuron with a 3 M KC1-filled electrode and recording postysynaptic potentials simultaneous with action potentials in L10 (Kandel e t al., 1967). L10 was then pressure-injected with cobalt chloride until all spontaneous electrical activity ceased, input resistance dropped about 5096, and the cell appeared greyish as seen in the dissecting microscope. In another ganglion, both L10 and a rostrally adjacent follower neuron were injected with cobalt. This neuron was found to receive hyperpolarizing postsynaptic potentials of 1-5 mV amplitude from L10 (Fig. 1). The physiology and position of this neuron identify it as the cell L12 by the criteria of Frazier, Kandel, Kupfermann, Waziri, and Coggeshall (1967). L12 was injected with somewhat less cobalt than L10. After injection, ganglia were left a t 15°C for 45 min to allow diffusion of cobalt throughout the neurons. Ganglia were then incubated for 25-35 min in a solution of 1 mg/ml diaminohenzidine tetrahydrochloride (DAB) in Aplysia saline adjusted to pH 7.5 with NaOH. NaOH was used because previously Tris, phosphate, and bicarbonate buffers were found to inhibit cobalt-catalyzed polymerization of DAB. During incubation the injected cells turned dark blue, indicating formation of the osmiophilic polymer. Ganglia were rinsed 20 min in saline alone, then fixed for 15 min in a cold mixture of 2% formaldehyde, 1% glutaraldehyde, 4% sucrose, and 3% NaCl in 0.1 M sodium phosphate buffer pH 7.4, according to the method of Sherman and Atwood (1972) for marine tissue. The left caudal quarter ganglion containing the injected cells was cut out with a razor blade and left in the fixative for 12-36 hr a t 4OC. This ganglion region encompasses all, or nearly all, of the dendritic field of L10 (Gillette and Pomeranz, 1973b, and further unpublished observations). After a 1 hr rinse in buffer containing 8% sucrose, the tissues were postfixed in
INTERNEURON STRUCTURE IN APLYSIA
465
Fig. 1. Discharge of L10 (lower trace) causes a hyperpolarizing postsynaptic potential in L12 (upper trace). The fast biphasic inflection (arrow) on top of the synaptic potential in L12 is an artifact of the L10 spike due to some anomalous rectified coupling in the amplifier circuits. 1%OsOd in a buffer containing 4% sucrose. The tissues were dehydrated, embedded in Spurr's resin, and mounted for sectioning. Along the entire course of L10 through the neuropil and out a nerve serial thick sections (1and 2 fim) were cut and stained with 1% methylene blue. At 10 or 20 fim sectioning intervals 25-30 ultrathin silver sections were taken and mounted on formvarcoated 50-mesh or celloidin-coated 75-mesh copper grids. Serial order of the ultrathin sections was maintained from ahout one-fourth of the sampling regions. In this way, 16 and 18 sectioning depths past the cell body layer into the neuropil were sampled from the two ganglia for electron microscopy. Grids were stained with uranyl acetate and lead citrate and then examined in a Phillips 200 electron microscope equipped with a 35 mm camera.
RESULTS
Gross neuron morphology In the light microscope the cell body and axon of L10 in both ganglia studied were easy to find and trace in the light microscope because the DAB polymer outlined the membrane and the axoplasm stained somewhat more lightly than that of other cells. The finer processes of L10 could be easily traced down to a size of 4-5 pm. The axon of L12 and its fine processes did not contrast as well and tracing of finer processes required closer examination. The general morphology of L10 has been previously described (Frazier et al., 1967; Gillette and Pomeranz, 1973b). In the two ganglia studied here a large furrowed and invaginated axon of irregular diameter (30-50 pm) arises from the cell body to enter the neuropil core of the ganglion and course out the pericardial nerve trunk, Fine processes (diam 1-5 pm) arise from the axon along its entire course in the neuropil, however, the large majority of the many fine processes of L10 are ramifications of a few large branches or dendrites (diam 10-15 pm) arising from the axon near the base of the soma (Fig. 2). The fine processes of L10 may extend to several hundred micrometers. In the ganglion where both L10 and L12 were stained, the axon of L12 is smaller and more irregular in cross section than that of L10, and sprouts fewer, thinner, and generally shorter processes. The L12 axon trunk does not contact the L10 axon trunk as it courses to the pericardial nerve. A
466
GILLETTE AND POMERANZ
Fig. 2. Freehand reconstruction of the neurons L10 and L12 from light microscopy of serial sections. Only the larger processes are drawn while the extensive finer ramifications are generalized or omitted. Most of the fine processes of L10 arise from a few large branches near the cell body. A single large process of L12 ramifies in the same area as does a process of L10 (rectangle); an electron micrograph from this area appears as Fig. 4.
branch of 5 pm diameter arises fiom the axon of L12, and courses about 200 pm to ramify and interdigitate with the twigs of a similar branch of L10, as illustrated in Figure 2.
Ultrastructural studies The characteristics of the staining technique have already been described (Gillette and Pomeranz, 1973a). Generally, the processes of the injected neuron are identified by the darkly staining polymer of DAB which appears as amorphous clumps in the axoplasm, or lines the inner surface of the neuron membrane. The axoplasm is lighter than that of unstained cells due to a slight condensation of the fibrillaryt matrix. Stained neurons also show mitochondrial distortion. The amorphous polyrrer contrasts best a t low magnifications (2,000-8,000X), there a well-stained process of an injected neuron is unmistakable. The ultrastructure of unstained neurons shows no ill effects from incubation in DAB and the character of synaptic'vesicles appears no different from control tissues fixed immediately upon dissection (Gillette, 1974). Synaptic input to L10. Small vesicle-filled processes contacting L10 were found much more often at the fine processes of L10 rather than a t the L10 axon. However, synapses with well-defined synaptic membrane specializations consisting of a widened, even synaptic cleft and moderate membrane and intermembrane densities by a polarized cluster of vesicles were only found synapsing onto fine processes and were never found on the main axon of L10 (Figs. 3a,b). Usually four or five of these well-defined synaptic inputs to L10 could be located in any given ultrathin section through the quarterganglion neuropil. Serial sections through such synapses showed that they are discrete en passant junctions with an average width of about 1 pm. Well-defined synaptic input to L12 was found to be qualitatively similar, except that synapses were found occasionally on the axon trunk.
INTERNEURON STRUCTURE IN APLYSIA
467
A variety of vesicle types occur in the processes presynaptic to L10; this variety is common to many other processes of the neuropil. Arbitrary classification of the vesicle types is not realistic, since the vesicles were found to lie on a continuum of morphology. However, for convenience, we denote four types in the continuum: dense, granular, round, and ellipsoid, as described in Table 1. Vesicles intermediate in size, shape, and content to the dense and the round types, the dense and the granular types, the round and the granular types, and the round and the ellipsoid types are common. Many combinations of different vesicle types and their intermediates could be found to occur when scans were made of many fine neuropilar processes. Attempts to classify synaptic junctions by vesicle content of the presynaptic process were only partially successful because ( I ) several types of vesicle are often found in any process, and (2) analysis of serial sections showed that the distribution of vesicles within a process is inhomogeneous (Figs. 3a,b) so that sectioning plane may determine apparent vesicle content. Synapses onto L10, and also synapses found elsewhere in the neuropil, mostly show the dense and the round types of vesicles together; less commonly, presynaptic processes contain predominantly the granular and the ellipsoid types. Correlates of L10 synaptic output. Fine processes of L10 are often found to contain conglomerates of vesicles and mitochondria. The conglomerates are most striking in irregular swellings of the dendrites up to 5 or 10 pm in diameter and can be recognized in the light microscope by their high
Fig. 3 (continued)
GILLETTE AND POMERANZ
468
Fig. 3. Nearby sections from a complete series through a synapse onto a fine process of L10. The synapse shows a well-defined synaptic cleft (cl). The apparent vesicle content of the presynaptic process (P) varies with the section plane. (a) Vesicles are mainly of the round (r) type (see text). (b) Vesicles are of the round (r), dense (d), and granular (gr) types. Calibration bar is 0.5 um. TABLE 1 Vesicle Types Found in the Fine Processes of the Neuropil of the Abdominal Ganglion Vesicle type Round Ellipsoid
Granular Dense
Description Round, diam 500-700 8,contents uniform and moderately electron-dense. Ellipsoid or flattened, common dimensions 290 x 650-360 x 480 contents uniform and moderately electron-dense. Round, diam 500-725 8, containin a diam. darkly staining granule 200-300 Round to slightly ellipsoid, diam 700-1200 8,contents uniform and very electrondense.
a,
f
basophilia. The DAB polymer is more dense in the cytoplasm among conglomerates than in cytoplasm elsewhere, suggesting that there are more divalent cation-binding sites which bind cobalt in these regions. The presumed synaptic vesicles are mainly of the dense and the round types (Fig. 4);granular vesicles are rarely present. No focal synaptic membrane specializations were ever found for the known cholinergic output of L10, although sections
INTERNEURON STRUCTURE IN APLYSIA
469
Fig. 4. A fine process of L10 is filled with vesicles of the dense (d) and round (r) types, with a few granular (g) vesicles. Mitochondria1 profiles (m) are frequent. The profuse DAB polymer in this process of L10 should not be confused with glycogen (gly) in other neurons. Calibration bar is 1 Fm.
were searched intensively in each sampling region. Scattered dense vesicles were also found in the cell body and main axon of L10. Large accumulations of vesicles and mitochondria could also be found occasionally in processes of
470
GILLETTE AND POMERANZ
Fig. 5. A process of L10 is found in direct physical contact with a process of its postsynaptic neuron, L12. Several such contacts were found within the area indicated in Fig. 1. Calibration bar is 2 l m .
unstained neurons, indicating this clustering of organelles is not an artifact of cobalt injection. The juncture of L10 and L12. Fine branches of the long process arising from the axon of L12 were found to come into direct apposition with twigs of a branch of L10 (Figs. 2 and 5). Figure 5 shows that the axon and fine processes of L12 can be distinguished from those of L10 in the electron microscope, presumably because L12 was injected with less cobalt than L10. However, the distinction between L10 and L12 is a subtle one based mostly on axoplasmic texture. Therefore, as the branches of L12 could not be traced in the serial sections in the light microscope far beyond the region shown in Figures 2 and 5, certain identification of distal sites of synaptic contact was not made. Probable electrotonic connections of L10. Gap junctions were found between L10 and the fine processes of other neurons in both ganglia studied
INTERNEURON STRUCTURE IN APLYSIA
47 1
Fig. 6. (a) A gap junction (arrow) indicates electrotonic coupling between fine processes of L10 and another neuron. Cobalt has crossed the junction and caused DAB polymer formation in both processes. Analysis of adjacent sections allowed L10 identification. Calibration bar is 0.5 wm. (b) The symmetrical junction of (a) shows a typical 7-layered structure and a 20 A “gap.” Calibration bar is 0.1 um.
(Figs. 6a,b). Two gap junctions were found a t the axon trunk of L10 in one ganglion and one gap junction was found a t a fine process in the second ganglion. Cobalt ions from L10 appear to have crossed the junctions and cata-
472
GILLETTE AND POMERANZ
lyzed polymer formation in the probable electrical followers of L10. Analysis of adjacent ultrathin sections tracing the processes away from the gap junctions showed that intensity of the DAB stain decreased rapidly over a distance of 10-15 pm on one side of the junction, indicating rapid dilution of cobalt ions crossing the junction from L10. DISCUSSION
These morphological findings indicate that L10 must mediate its cholinergic synaptic output via the fine processes of its dendritic tree. This follows from the findings of the rarity of axo-axonal appositions, the observed apposition between L10 and its postsynaptic neuron L12 only at their fine processes, and the presence of the masses of presumed synaptic vesicles in tlhe fine processes of L10. However, the sites of L10 output could not be correlated with the focal membrane specializations which have been used as the morphological criterion for identification of synapses in this and in previous studies of molluscan neuropil (Gerschenfeld, 1963; Coggeshall, 1967). There are three immediate explanations. First, intracellular cobalt on the presynaptic side may have destroyed them; second, L10 output sites could be so rare or highly localized that they were missed by the sampling procedure; and third, L10 output sites are not characterized by focal membrane specializations. The first possibility has not been answered by direct test, but is unlikely because such specializations have bec,ii found (in another mollusc) to include rigid, mechanical intercellular junctions (Jones, 1970) not apparently obliterated by high divalent cation concentration (Hillman and Llinfis, 19743 and not exposed to the intracellular milieu. The second possibility is unlikely on morphological grounds, since large masses of synaptoid vesicles were found in different processes of L10 in widely separate neuropil regions and in any given section through the neuropil. It is also improbable on physiological grounds. L10 is a command interneuron with a large postsynaptic population (Frazier et al., 1967) endogenously capable of patterning the cardiac rhythm independent of synaptic input (Koester et al., 1974). It is doubtful that so many well-defined input sites could have been found to the exclusion of even a single outpdt site. Therefore, the likely alternative is that L10 output sites lack clear membrane specializations. Thompson, Schwartz, and Kandel (1973), in another study of L10 fine structure also did not report such specializations. Synaptic membrane specializations serve in vertebrate and invertebrate neurons as foci of vesicle release (Pappas and Waxman, 1972). The apparent lack of clear focal output sites could reflect that L10 synaptic output is a nonfocal occurrence over large areas of the fine processes. Finally, these findings support previous conjecture that synapses in molluscan neuropil are more frequent than is indicated by the sparseness of synapses that are structurally akin to well-studied vertebrate synapses (Coggeshall, 1967; Cobb and Mulilins, 1973).
INTERNEURON STRUCTURE IN APLYSIA
473
As the synaptic inputs to L10 are also found at the fine processes of the dendritic tree, the interneuronal transactions of L10 must be mediated there. The proximity of synaptic input and output sites is reminiscent of the granule cells of the vertebrate olfactory bulb, where action potential generation is not necessary to synaptic transmission (Shepherd, 1970). The large constants of electrotonic decay (A) of Aplysia neurons further assure the electrical proximity of input and output sites (Graubard, 1973). Thus, although it is not known if the fine processes are capable of supporting action potentials, it is likely that the interneuronal function of L10 is not necessarily dependent on spike initiation or invasion in the axon trunk or cell body. Depolarizing excitatory dendritic input could cause synaptic output either by direct electronic spread, or by local initiation of an active membrane response. Discovery of gap junctions on L10 provides the probable morphological substrate for the physiological finding that L10 is directly electrically coupled to other neurons of the ganglion (Waziri, 1969, 1971). Three characteristics of electrotonic junctions listed by Pappas and Waxman in their review (1972) are evident: close apposition of opposing cell membranes, symmetry, and the ability to pass small ions, in this case, cobalt ions. No strong morphological evidence could be drawn from this study for the known existence of different kinds of synapses on L10 (Kandel et al., 1967) based on vesicle appearance. The morphological continuum of vesicle types found in the processes of the neuropil could represent a functional continuum for transmitter release and vesicle recycling. Thus, the frequency of different types in any neuron process could reflect the state of physiological activit y at the time of fixation. It is important to note that the vesicle content of L10, a cholinergic neuron, shows marked similarity to that described for serotonergic neurons in Aplysia (Taxi and Gautron, 1969; Weinreich, McCaman, McCaman, and Vaughn, 1973). We conclude that there is presently little basis for inferring transmitter pharmacology or synaptic action in molluscan brain solely from ultrastructural observation. A possible future exception may come from pharmacological identification of those synapses which show focal membrane specializations. The authors thank Dr. Harold L. Atwood for his expert advice, encouragement, and generous loan of equipment during the course of this study. Much of the data presented here was the subject of a talk presented at the Society for Neuroscience Third Annual Meeting.
REFERENCES ATWOOD, H. L., LANG,F. and MORIN,W. A. (1972). Synaptic vesicles: Selective depletion in crayfish excitatory and inhibitory axons. Science 176: 1353-1355. J. E., WACHTEL,H. and KANDEL,E. R. (1971). Ionic mechanisms of excitatoBLANKENSHIP, ry, inhibitory, and dual synaptic actions mediated by an identified interneuron in abdominal ganglion of Aplysia. J. Neurophysiol. 34: 76-92. BODIAN,D. (1966). Electron microscopy: Two major synaptic types on spinal motoneurons. Science 151: 1093-1095. CORB, J. L. S., and MULLINS,P. A. (1973). Synaptic structure in the visceral ganglion of the lamellibranch mollusc, Spisula solida. 2. Zellforsch. 138: 75-83.
474
GILLETTE AND POMERANZ
COGGESHALL, R. (1967). A light and electron microscope study of the abdominal ganglion of Aplysia californica. J. Neurophysiol. 3 0 1263-1287. FRAZIER,W. T., KANDEL,E. R., KUPFERMANN,I., WAZIRI,R. and COGGESHALL, R. E. (1967). Morphological and functional properties of identified neurons in the abdominal ganglion of Aplysia californica. J. Neurophysiol. 3 0 1288-1351. GERSCHENFELD,H. M. (1963). Observations on the ultrastructure of synapses in some pulmonate molluscs. Z. Zellforsch. 6 0 258-275. J. H. (1971). Choline acetyltransferase in identified neurons of the GILLER,E. and SCHWARTZ, abdominal ganglion of Aplysia californica. J. Neurophysiol. 34: 94-108. GILLETTE, R. (1974). Microstructural and ultrastructural studies on identified neurons of the abdominal ganglion of Aplysia californica. Thesis, University of Toronto. GILLETTE,R. and POMERANZ, B. (1973a). Neuron geometry and circuitry via the electron microscope: Intracellular staining with osmiophilic polymer. Science 182 1256-1258. GILLETTE,R. and POMERANZ, B. (197313). A study of neuron morphology in Aplysia califorrtica using Procion Yellow dye. Comp. Biochem. Physiol. 4 4 A 1257-1259. GRAUBARD, K. (1973). Electrotonic decrements within Aplysia neurons. Abstracts of the Society for Neuroscience, Third Annual Meeting. HILLMAN,D. E. and LLINAS, R. (1974). Calcium-containing electron-dense structures in the axons of the squid giant synapse. J . Cell Biol. 61: 146-155. JONES, D. G. (1970). A further contribution to the study of the contact region of Octopus synaptosomes. Z. Zellforsch. 103: 48-60. R. E. (1967). Direct and cornKANDEL,E. R., FRAZIER,W. T., WAZIRI,R. and COGGESHALL, mon connections among identified neurones in Aplysia. J . Neurophysiol. 3 0 1352-1376. KOESTER,J., MAYERI,E., LIEBESWAR,G. and KANDEL,E. R. (1974). Neural control of circulation in Aplysia. 11. Interneurons. J. Neurophysiol. 37: 476-496. PAPPAS,G. D. and WAXMAN,S. G. (1972). Synaptic fine structure: Morphological correlates of chemical and electrotonic transmission. In: Structure and Function of Synapses. G. D. Pappas and D. P. Purpura, Eds., Raven Press, New York, pp. 1-44. PURPURA,D. P. (1972). Intracellular studies of synaptic organizations in the mammalian brain. Zbid., pp. 257-302. RALL, W. (1959). Branching dendritic trees and motoneuron resistivity. Exptl. Neurol. 1: 491-527. SATO,M., AUSTIN,G. M., YAI, H. and MARUHASHI,J. (1968). The ionic permeability changes during acetylcholine-induced responses of Aplysia ganglion cell. J . Gen. Physiol. 51: 321-345. SHEPHERD,G. M. (1970). The olfactory bulb as a simple cortical system. In: The Neurosciences, Second Study Program. F. 0. Schmitt, Ed., Rockefeller University Press, New York, pp. 539-551. SHERMAN,R. G. and ATWOOD,H. L. (1972). Correlated electrophysiological and u1trastructu:ra1 studies of a crustacean motor unit. J. Gen. Physiol. 5 9 586-615. TAXI,J. and GAUTRON,J. (1969). Donnkes cytochimiques in faveur de l’existence de fibres nerveuses serotoninergiques dans le coeur de I’hplysie, Aplysia californica. J . Microscop. 8: 627-636. J. H. and KANDEL,E. R. (1973). Autoradiographic analysis THOMPSON,E. B., SCHWARTZ, with the light and electron microscope of Aplysia identified neurons, their processes, and synapses after intrasomatic injection of 3H-L-fucose. Abstracts of the Society for Neuroscience, Third Annual Meeting. WAZIRI,R. (1969). Electrical transmission mediated by an identified cholinergic neuron of Aplysia. Life Sci. 8: 469-476. WAZIRI,R. (1971). Electrotonically coupled interneurones produce two types of inhibition in Aplysia neurones. Nature New B i d . 232: 286-288. R. E. and VAUGN,J. E. (1973). Chemical, enWEINREICH,D., MCCAMAN,M. W., MCCAMAN, neurons from zymatic and ultrastructural characterization of 5-hydroxytryptamine-containing the ganglia of Aplysia californica and Tritonia diomedia. J. Neurochem. 20: 969-976. Accepted for publication January 21,1975
