THE JOURNAL OF COMPARATIVE NEUROLOGY 283143-152 (1989)

Interganglionic Dendrites Constitute an Output Pathway From the Procerebrum of

the Snail Achatimjidica

RONALD CHASE AND BARBARA TOLLOCZKO Department of Biology, McGill University, Montreal, Quebec H3A 1B1 Canada

ABSTRACT The procerebrum is an olfactory processing region that occupies approxi-

mately one-third of the total brain area in pulmonate gastropod molluscs. It has many unusual features, including a development separate from the rest of the brain and the absence of axons belonging to its intrinsic neurons. We have investigated the input and output pathways of the procerebrum in the terres- trial snail Achatina fulica by using hexamminecobalt chloride as a selective label. Both the tentacle nerve and the cerebropedal connective nerve contrib- ute to a fine neural plexus that is distributed throughout the neuropile region of the procerebrum. The fibers from the tentacle nerve are predominantly pre- synaptic, whereas those from the cerebropedal connective are predominantly postsynaptic. The postsynaptic fibers (dendrites) were traced to two groups of nerve cells (total number, 20-25) near the ventral surface of the ipsilateral pedal ganglion. No evidence was obtained for any other numerically signifi- cant output pathway from the procerebrum. Since locomotion is known to be controlled by the pedal ganglion, these results provide an anatomical sub- strate for the strong influence of olfaction on locomotor behavior in snails. The pathway is unusual in that the dendrites are interganglionic and can be as long as 5 mm.

Key words: mollusea, brain, olfactory pathways, anatomy, ganglia

The procerebrum constitutes a major portion of the cere- bral ganglion of pulmonate molluscs. I t differs strikingly from the rest of the brain, as well as from the other ganglia (Bullock and Horridge, '65). Because of its distinctiveness, its universal presence in all species of the subclass Pulmon- ata, and its absence from the brains of other gastropod sub- classes, the procerebrum has been described as a taxonomic criterion (Van Mol, '67). In Achatina fulica, it is estimated that each one of the two bilaterally symmetrical procere- brums contains 2 x lo4 neurons, which is more than is found in the entire remainder of the central nervous system (Chase, '86). These neurons differ from those elsewhere in the central nervous system in the following ways: 1) They constitute a coherent, densely packed population that is uniform in size and appearance. 2) They are the smallest neurons, with diameters 5-7 pm. 3 ) They have abundant chromatin and very little cytoplasm. 4) They are diploid, not polyploid (Chase and Tolloczko, '87). The neuropile of the procerebrum is more finely textured than any other and, uniquely, it lies beside, not interior to, the cell groups that contribute processes to it. An additional unusual feature of the procerebrum is that it develops separately from the rest

of the brain (Haller, '13; Van Mol'67). It shares an embryo- logical origin with the tentacular ganglion and only joins with the rest of the brain late in development.

The function of the procerebrum is probably to process olfactory information (Gelperin et al., '89). This can be inferred from its embryological association with the tentacle ganglion and from the fact that in its fully developed state it remains connected to the tentacle ganglion via the tentacle nerve. Abundant evidence points to olfaction as the princi- pal function of the tentacles (Chase, '86). Experimental studies have demonstrated that stimulation of the tentacles with odors increases metabolic activity in the procerebrum (Chase, '85). Because the tentacles and the procerebrum are well developed in the order Stylommatophora, which is ter- restrial, and poorly developed in the order Basommato- phora, which is aquatic, Van Mol ('67) has argued that the evolution of these coupled structures has been important in the successful adaptation of the pulmonates to the terres- trial environment.

Accepted December 26,1988.

0 1989 ALAN R. LISS, INC.

144

On the basis of its apparent role in olfactory integration, its dense population of small neurons, and its finely tex- tured neuropile, many authors have recognized an analogy between the procerebrum of pulmonate molluscs and the corpus pedunculatum of arthropods and annelids (discussed in Bullock and Horridge, '65). Moreover, with the foregoing criteria, as well as others, a case can be made for an analogy between the procerebrum and the vertebrate olfactory bulb (Chase, '86). Although the actual physiological contribu- tions of the procerebrum to olfactory function are unknown, the structure is obviously of functional significance owing to the general importance of olfaction in snail behavior (Chase, '86) and given the relatively large endowment to the procer- ebrum of neurons and neuropile.

The inputloutput organization of the procerebrum was the subject of considerable investigation by anatomists early in this century (reviewed in Bullock and Horridge, '65). Hanstrom's work ('25), based on Golgi staining, is the most informative. He found that fibers travelling in the tentacle nerve terminate in the neuropile of the procerebrum. We have confirmed this observation in the present study. It can be assumed, therefore, that the tentacle nerve is a major source of afferent input. More interesting, Hanstrom never saw any of the neurites of the intrinsic procerebrum cells projecting to any other part of the brain. Haller ('13) had made similar observations, and our results offer no contra- diction. Therefore, the only way for olfactory information to pass out of the procerebrum is via recipient processes (den- drites) that come into the procerebrum from cell bodies located elsewhere.

In this investigation, we have discovered an output path- way from the procerebrum. The anatomical arrangement is unusual in that fibers that are functionally dendritic arrive in the procerebrum from a distant ganglion.

R. CHASE AND B. TOLLOCZKO

MATERIALS AND METHODS Specimens of Achatina fulica weighing 2-8 g were ob-

tained from a laboratory culture (Pawson and Chase, '84). To label neural pathways, a 0.2 M solution of hexammineco- balt chloride in distilled water was used in all experiments. The experiments utilized freshly dissected preparations.

Nerve backfills were accomplished by sucking the cut nerve end into a fitted glass pipette containing hexammine- cobalt. The time that the nerve was kept in the pipette, and the length of the nerve stump, were carefully chosen to avoid extracellular label and to limit the possibility of tran- synaptic transport. The optimum times were 2 hours for the tentacle nerve and 4 hours for the cerebropedal connective. The optimum lengths were about 3 mm for the tentacle nerve and about 2 mm for the cerebropedal connective. The preparations were processed immediately after the nerves

were removed from the pipettes (see below). Using these procedures we observed no background (extracellular) la- belling, and the labelling of neuronal structures was both consistent from preparation to preparation and in agree- ment with known anatomy. Regardless, the fiber pathways reported in this paper were in every case scrutinized for structural continuity by observations at the highest practi- cal optical power. The same filling protocols were used for electron microscopy as for light microscopy. The integrity of the fills was confirmed in the electron microscope by the observation that plasma membranes were intact and no label was seen in the extracellular space. Also, doubly labelled synapses, which could result from transynaptic mi- gration of the label, were uncommon (Table l).

The cellular sources of fibers projecting to the procere- brum were investigated by iontophoretic injections of hex- amminecobalt into the procerebrum. A glass micropipette was successively inserted into the neuropile at two to four separate locations. At each site, 0.1-0.5 pA positive current was passed through the pipette for 10 minutes in 500 msec pulses delivered a t 1 Hz. The preparations were kept in snail saline at 4°C for 16 hours before further processing.

For light microscopy, hexamminecobalt was precipitated by using H,S dissolved in snail saline. After thorough wash- ing in saline, tissues were fixed in Carnoy's solution for 5-7 days. Intensification was achieved either by the Timm's method (Bacon and Altman, '77) or by the method of Davis ('82). With the latter method, to aid penetration, 4% Triton X-100 was added to the sodium tungstate preincubation solution. Development was done in darkness, with solution replaced at 15 minute (Davis) or 30 minute (Timm's) inter- vals, and a total duration of up to 60 minutes. The tissues were dehydrated in an ethanol series and either cleared in methyl salicylate for viewing as whole mounts or embedded in soft resin (Spurr, '69) and sectioned a t 24-30 pm.

For electron microscopy, hexamminecobalt was precipi- tated by H,S dissolved in 0.1 M PO, buffer (pH 7.6) for 2 minutes. After washing in buffer solution, the tissues were fixed on ice for 1 hour in 1% paraformaldehyde, 1% gluta- raldehyde in 0.05 M PO, buffer (pH 7.6). Intensification was by the method of Croll ('86). The tissues were then washed again in PO, buffer and prepared for microscopy as previ- ously described (McCarragher and Chase, '85). Individual sections were scanned in their entirety in a Philips 410 elec- tron microscope. All candidate labelled synapses were pho- tographed for subsequent evaluation.

RESULTS In histological section (Fig. 1) the cells of the procerebrum

appear small (diameters, 5-7 pm) and numerous in relation to those in other regions of the brain. The neuropile of the

TABLE I. Polarity of Synapses Involving Labelled Fiber Pathways'

Both pre- and No. of Presynaptic Postsynaptic Symmetrical postsynaptic

Labelled

Labelled pathway brains profiles profiles synapses processes labelled

Tentacle

Cerehropedal

'The label was silver-intensified hexamminecobalt. The criteria for counting a synapse were 1) vesicular accumulation at the membrane and 2) membrane specialization an defined in McCarragher and Chase ('85). Symmetrical synapses satisfy both criteria at both sides of the cleft (McCarragher and Chase, '85). Numbers in parentheses include putative synapses in which only one of the two criteria was satisfied.

nerve 2 11 (17) 0 (3) 1(5) 1

connective 3 2 (3) 12 (30) 1 2

INTERGANGLIONIC DENDRITES IN SNAIL PROCEREBRUM 145

Fig. 1. Histological section of the left half of the cerebral ganglion of Achatina fulica. Toluidine blue staining. TN, tentacle nerve; CPC, cere- bropedal connective; CC, cerebral commissure; mesocbm, mesocere- brum. Scale bar = 300 um.

procerebrum is finely textured and is separated from the area of the cell bodies, at relatively dorsal planes of section, by a fiber tract that is continuous with the tentacle nerve.

The neurites of the procerebrum cells were visualized by injecting hexamminecobalt into the neuropile. Each cell sends a single, apparently unbranched, slender process into the neuropile. Because the neurites follow straight and par- allel paths from cell bodies to neuropile, a small iontophor- etic injection of hexamminecobalt into the neuropile at any location results in the labelling of a discrete fiber bundle and a contiguous group of procerebral cells (Fig. 2C). No procer- ebral neurites were seen projecting to any area other than the procerebral neuropile.

Given that the intrinsic neurons of the procerebrum do not project outside the region, it was assumed that informa- tion transfer is effected via dendrites coming into the pro- cerebrum from elsewhere. Two complementary experimen- tal strategies were employed to identify these dendrites: 1) anterograde labelling by applying hexamminecobalt to the proximal ends of cut cerebral nerves, and 2) retrograde labelling by iontophoretic injections of hexamminecobalt into the procerebral neuropile.

In different experiments, all the major nerves of the cere- bral ganglion were filled with hexamminecobalt, including the tentacle nerve, the three lip nerves, the cerebrobuccal connective, the cerebropedal connective and the cerebro- pleural connective. Only the medial lip nerve, the tentacle nerve, and the cerebropedal connective contain fibers that

terminate in the procerebrum. The projection of the medial lip nerve, also described by Gelperin et al. (’89) in Lirnax maximus, is sparse in comparison to that of the other two projections. I t presumably represents the olfactory input to the procerebrum from the inferior tentacles, and in this respect it may be considered as functionally parallel to the tentacle nerve projection. Because the tentacle nerve pro- jection is more substantial, it alone was examined in detail.

The cerebral projections of the tentacle nerve were exam- ined in 19 preparations. As shown histologically in Figure 2A and schematically in Figure 3A, the nerve arborizes extensively in the neuropile region of the ipsilateral procere- brum. Other fiber bundles deriving from the tentacle nerve can be traced to the neuropile of the postcerebrum, the cere- brobuccal connective, the cerebropedal connective, and the cerebral commissure (Fig. 3A). Contrary to Hanstrom’s (’25) description of procerebral cells with processes in the tenta- cle nerve, we saw no backfilled procerebrum cells. However, we did backfill four to ten neurons in the ipsilateral postcer- ebrum, consistent with Hanstrom’s descriptions. In addi- tion, about 25 neurons in the ipsilateral posterior mesocere- brum were backfilled (Fig. 3A). These latter cells were not identified by Hanstrom as having axons in the tentacle nerve.

The cerebropedal connective nerve also projects to the procerebrum (Fig. 3B), both ipsilaterally (Fig. 2B) and con- tralaterally (Fig. 2D). The contralateral projection appears to be sparser than the ipsilateral projection, and this obser- vation is not likely due to the greater distance to the contra- lateral terminus because the contralateral projection is sparse along its entire length (Fig. 3B). The ipsilateral pro- jection to the procerebrum was seen in 17 fills of the cere- bropedal connective nerve. In histological sections (Fig. 2B), it appears as a widely distributed fibrous plexus that is coex- tensive with the projection of the tentacle nerve (compare Fig. 2A,B). One difference in the projections of the two nerves is that fills of the cerebropedal connective, but not the tentacle nerve, produce labelled fibers penetrating into the region of procerebral cell bodies (Fig. 3B). The labelled fibers of the procerebral plexus have the same appearance regardless of their nerve of origin. They are very fine (diam- eters, 0.2-1.0 Wm in the electron microscope), multiply branched, and commonly marked by small beads.

When viewed a t high magnification, it can be seen that the fibers of the cerebropedal connective reach the procere- bra1 neuropile by means of a tortuous path through the post- cerebral neuropile (Figs. 3B, 4). After entering the ganglion as a ribbon of mostly parallel fibers, they become densely intermingled with other fibers, mostly oriented in crossing directions, before emerging again as a fairly coherent bundle that enters the procerebrum. Careful tracing of single fibers has established the continuity of the projection from nerve to procerebrum. In contrast, the passing fibers joining the cerebropedal connective with the tentacle nerve and the peritentacular nerves (Fig. 3B) are relatively larger and less convoluted than those going to the procerebrum.

In addition to the fiber pathways described above, numer- ous neuronal cell bodies are backfilled from the cerebrope- dal connective. These are found primarily in the postcere- brum and the posterior mesocerebrum, both ipsilaterally and contralaterally. The major cell groups are shown in Fig- ure 3B.

To locate the neurons whose fibers travel in the cerebro- pedal connective and arborize as a plexus in the procere- brum, hexamminecobalt was injected into the procerebral

146 R. CHASE AND B. TOLLOCZKO

Fig. 2. Fiber projections to the procerebral neuropile. A: Labelling resulting from a hexamminecobalt fill of the tentacle nerve. Section, 24 prn thick. The fibers of the nerve enter the ganglion a t the site marked TN, between the cell bodies and the neuropile (Np) of the procerebrum. B: Labelling resulting from a fill of the ipsilateral cerebropedal connec- tive nerve. Note fibers of passage entering the tentacle nerve and fibers in neuropile coextensive with those shown in panel A. C: Labelling

neuropile for retrograde labelling. These injections produce a uniformly dense precipitation throughout the procerebral neuropile (Fig. 5A). Some fibers travelling to the contralat- era1 cerebrum from the ipsilateral tentacle nerve (see Fig. 3A) are labelled, but none was ever traced to a labelled cell body. Of special interest is a fiber bundle that can be identi- fied (Fig. 5A, arrowhead) and traced from the procerebrum to the cerebropedal connective. These fibers are further observed along the length of the connective nerve (approxi- mately 3 mm in an animal weighing 6 8). They enter the pedal ganglion as a coherent bundle and can be traced directly to two clusters of neuronal somata (Fig. 5B,C). The

resulting from iontopboretic injections of hexamminecobalt at two sites (stars) in procerebral neuropile. Section, 30 pm thick. A group of procer- ebral cells and their neurites (arrowhead) have been labelled retro- gradely. D: Labelling resulting from a fill of the contralateral cerebrope- dal connective. Whole mount preparation. Arrowhead points to incom- ing fiber bundle prior to its arborization in procerebral neuropile. Scale bars = 100 um.

same groups of pedal ganglion cells were labelled in 16 of 20 preparations in which hexamminecobalt was injected into the procerebrum; the remaining four preparations produced no labelling in the pedal ganglion.

The pedal ganglion cells that are labelled by procerebral injections are situated near the ventral surface of the gan- glion. They lie beneath a more superficial layer of cell bod- ies. The lateral group of cells (Fig. 5B) is situated approxi- mately 60 pm from the surface of the ganglion, whereas the medial group is about 90 pm deep. The lateral group con- tains eight to ten cells and the medial group 13-15 cells. The individual cells are oval in shape and measure 10-13 pm by

INTERGANGLIONIC DENDRITES IN SNAIL PROCEREBRUM 147

CPlC

Fig. 3. Schematic drawings showing extent of labelling with nerve fills. A Fills of tentacle nerve (TN). B Fills of cerebropedal connective (CPC). Only the fiber tracts that impinge on the procerebrum are shown. Somata, but not axons, of backfilled cellular groups are indicated as solid markings. The number of drawn somata in each group indicates the relative numbers, but not the actual numbers, of observed cells. The

fiber thicknesses have been drawn to highlight the projections to the procerebrum (outlined cells) and do not reflect actual relative diame- ters. Peri TN, peritentacular nerve; CBC, cerebrobuccal connective; CPlC, cerebropleural connective; ant, med, and post lip N, anterior, medial, and posterior lip nerves.

A n t

Med

P o s t

Fig. 4. Projection of the cerebropedal connective to the procerebral neuropile, The nerve was injected with hexamminecobalt via a micropi- pette. This procedure limited the number of fibers that were labelled, and thus rendered more visible the pathway of interest. Fibers were

traced with the aid of a drawing tube. Outlines of the ganglion and the procerebral cells are schematic. Arrowheads delimit a region where the labelled fibers are intricately interwoven with other tissues. Nerves labelled as in Figure 3. Scale bar = 200 pm.

148 R. CHASE AND B. TOLLOCZKO

Fig. 5. Identification of pedal ganglion cells projecting to the procer- ehrum. All photographs are of whole mount preparations. A: Dorsal view of left half of cerebral ganglion showing distribution of hexammine- cobalt in procerebral neuropile (star) following multiple focal injections. Arrowhead points to fiher bundle directed toward cerebropedal connec- tive (CPC). T N , tentacle nerve; CPlC, cerehropleural connective. B:

16-19 pm (Fig. 5C). Other than the long neurite linking the cell body with the procerehrum, no processes attributable to these cells were seen, either branching from the main neu- rite or joining the cell soma.

In addition to the pedal ganglion cells, procerebral neu- rons were also consistently labelled by injections into the neuropile, as noted earlier (Fig. 2C). Far less consistently labelled were neurons in the postcerebrum. In six of 20 prep-

Ventral view of paired pedal ganglia from same preparation as shown in A. Arrowheads point to clusters of labelled neurons. C: High-magnifica- tion photograph of labelled pedal ganglion cells from a different prepa- ration. 11, right pedal ganglion; I,, left pedal ganglion. Scale bars ~ 200 fim in A, B; 50 gm in C.

arations, five to ten such cells were labelled. In only three preparations were there any neurons labelled in the same region of the postcerebrum in which are found neurons backfilled from the cerebropedal connective (Fig. 3B). Neu- rons in the mesocerebrum were labelled in only a single preparation, one in which an injection of hexamminecobalt was probably misplaced from the neuropile to the fiber tract that contains efferent axons headed for the tentacle nerve

INTERGANGLIONIC DENDRITES IN SNAIL PROCEREBRUM 149

Fig. 6. Synaptic arrangements of labelled processes. The stars mark processes containing silver grains from Timm’s intensification of hex- amminecobalt. Arrowheads delimit synapses. A: Labelled presynaptic profile from a snail in which the tentacle nerve was filled with hexammi- necobalt. R: Labelled postsynpatic profile, f i l l of cerebropedal connec-

(Fig. 3A). These data argue against the existence of neurons in the cerebral ganglion that could be the cellular source of the procerebral plexus revealed by filling the cerebropedal connective.

Electron microscopy confirmed that the fiber plexus of the procerebrum that was seen in the light microscope is neither artifact nor composed only of passing fibers. More

tive. C: Symmetrical synapse, fill of tentacle nerve. D: Serial/parallel synaptic arrangement, fill of cerebropedal connective; unlabelled pro- cess 1 synapses on labelled process 2 and on labelled process 3; process 2 synapses on process 3. Scale bars = 200 nm.

importantly, the synaptic arrangements entered into by ten- tacle nerve fibers and cerebropedal connective fibers were clearly distinguished (Table 1, Fig. 6). Labelled tentacle nerve processes are predominantly presynaptic, whereas la- belled cerebropedal connective fibers are predominantly postsynaptic. The labelled postsynaptic profiles often have the appearance of an enlarged termination (Fig. 6B). How-

160

ever, the anatomical source of any given process (i.e., tenta- cle nerve, cerebropedal connective, or intrinsic neuron) can- not be reliably ascertained by reference to vesicular content or any other ultrastructural criteria. The synaptic profiles contain either clear vesicles or mixed clear and dense core vesicles in both labelled and unlabelled processes. Some synapses have an ultrastructural symmetry such that it is impossible to distinguish pre- and postsynaptic processes (Fig. 6C). In other cases, there are complex synaptic arrangements involving serial and parallel connections (Fig. 6D).

R. CHASE AND B. TOLLOCZKO

neurons into the tentacle nerve (Fig. 3 A Hanstrom, '25) and by the entry into the neuropile of fibers from the cerebrope- dal connective (Figs. 2B, 3B, 4).

The direct pathway from pedal ganglion neurons to the neuropile of the procerebrum has been established by con- sistent results with three complementary methods: light mi- croscopy of both anterograde and retrograde labelling, and electron microscopy. The ultrastructural evidence allows us to conclude that the cerebropedal fibers in the procerebrum are recipient in function, not transmitting, because a large majority of labelled cerebropedal processes are postsynap- tic, whereas processes of the tentacle nerve are predomi- nantly presynaptic (Table 1). This observation is important for understanding the functional organization of the procer- ebrum because it provides for an output pathway from the region (Fig. 7).

It is significant that the output of the procerebrum goes directly to the pedal ganglion. In several species of gastro- pod molluscs, it has been shown that the pedal ganglion con- tains motor neurons and motor programs essential for loco- motion (Audesirk, '78; Hening et al., '79; Jahan-Parwar and Fredman, '79). Olfactory information, transmitted to the procerebrum from the specialized epithelium at the tip of the tentacle, is a major determinant of locomotor behavior

DISCUSSION The absence of procerebral cell processes projecting to

other brain regions (Haller, '13; Hanstrom, '25; our observa- tions) may be accounted for by the separate early develop- ment of the procerebrum and the rest of the brain (Haller, '13; Van Mol '67). When the procerebrum eventually joins the rest of the brain at a late embryonic stage there may be physical barriers, or other restrictions, preventing the out- growth of procerebral fibers. There is obviously no prohibi- tion on the growth of fibers into the procerebrum, however, as shown by the projection of postcerebral and mesocerebral

Fig. 7. Schematic summary of results. Arrowheads indicate direction of information transfer as revealed by electron microscopy (Table 1). TN, tentacle nerve; CPC, cerebropedal connective; V, visceral ganglion; Pa, parietal ganglion; P1, pleural ganglion; Ped, pedal ganglion.

INTERGANGLIONIC DENDRITES IN SNAIL PROCEREBRUM 151

in Achatina and other terrestrial molluscs (Chase, '82, '86). In Aplysia, it has been demonstrated that the cerebral gan- glion is a source of commands for the initiation of the pedal wave motor program (Jahan-Pawar and Fredman, '79; Fred- man and Jahan-Parwar, '83).

Unfortunately, were were not able to visualize the axon, nor any branched fiber, belonging to any labelled pedal gan- glion cell. This precludes definitive interpretation, at this time, of the ultimate destination of signals leaving the pro- cerebrum via the cerebropedal connective. It should be noted, however, that the appearance of the labelled pedal ganglion cells was typical of all labelled cells in our mate- rial-that is, branched or multiple processes were rarely observed. Only in a few large, exceptionally well-filled neu- rons were branches visible. By contrast, when some of the same cells (e.g., the metacerebral giant) are filled by intra- cellular injection, the branching is invariably abundant. Other workers who have used cobalt to backfill molluscan nerves have had similar results (Jahan-Parwar and Fred- man, '76; Bicker et al., '82, Sonetti et al., '82; Croll, '86). In an effort to overcome this problem, we attempted to use either horseradish peroxidase or carboxyfluorescein as a label of the pedal ganglion cells. However, neither substance produced satisfactory results when injected into the procer- ebrum. Likewise, we were unsuccessful in our attempts to identify the appropriate pedal ganglion cells by electrophy- siological tests, or other means, so that we could inject a label intrasomatically.

Kunze ('21) had noted a fiber bundle communicating between the external mass of the procerebrum and a region near the origin of the cerebropedal connective, but her methods were evidently inadequate to reveal either the ter- minal plexus in the procerebrum or the direct link to the connective nerve. Our material has also revealed the intri- cate and tortuous appearance of the pathway a t the edge of the procerebrum (Fig. 4). This appears to be due to its pas- sage through a dense network of fibers passing in other directions and, conceivably, it is a consequence of the fibers' delayed entry into the procerebrum.

Our interpretation of information transfer from the pro- cerebrum to the pedal ganglion (Fig. 7) necessarily implies that action potentials are initiated in the processes of the pedal cells at or near the procerebrum and conducted cen- trifugally to the pedal ganglion. Since the processes in the procerebrum are very fine ( < l . O wm) and the distance to the pedal ganglion is very long (about 5 mm in a fully grown snail), passively conducted signals would not suffice. In other animals, dendritic spikes are unusual, but not un- known. Their occurrence in Purkinje cells and other neu- rons of the vertebrate brain is well documented (Llinas and Sugimori, '80). They also occur in certain visual interneu- rons of the leech (Peterson, '84). In both of these examples, the dendritic spikes are propagated noncontinuously. In crustacea, however, several examples have been described in which action potentials are generated and propagated at

multiple sites on processes distant from the soma (Calabrese and Kennedy, '74; Moulins and Nagy, '81). The same phe- nomena have also been described recently in the neuron R2 in the mollusc, Aplysia (Ambron et al., '88). The neuronal structures discussed in these latter examples combine the functional properties of dendrites and axons, as conven- tionally conceived. Especially relevant to the scheme shown in Figure 7 are the molluscan mechanoreceptor neurons which, like vertebrate dorsal root ganglion neurons, com- monly have somata in the central ganglia, and peripheral dendrites that conduct action potentials centrally (Byrne et al., '74; Cobbs and Pinsker, '78). Although the pedal gan- glion cells of Achatina are interneurons, not primary sen- sory neurons, their dendrites could be considered to have a peripheral locus, given that the procerebrum has a struc- tural and developmental affinity with the tentacular ex- tremity. Finally, in regard to this aspect of the results, it is noteworthy that Bullock and Horridge ('65), in their in- sightful review, also stated the necessity for "cellulipedal" conduction out of the procerebrum, although they had in mind only neurons of the postcerebrum.

The pathway to the pedal ganglion may not be the only one carrying information from the procerebrum, but it is almost surely the most significant. Hanstrom ('25) de- scribed neurons in the postcerebrum (his Fig. 8) and in the mesocerebrum (his Fig. 6) having processes that enter the procerebral neuropile. We found no unequivocal evidence of mesocerebral cells having dendrites in the procerebrum, as opposed to having axons in the tentacle nerve. Rarely, a few postcerebral cells were retrogradely labelled by injections into the procerebrum. However, the majority of these neu- rons, if not all of them, probably possess neuropile processes only as branches of efferent axons travelling in the tentacle nerve, as noted by Hanstrom ('25). They can be backfilled from the tentacle nerve (Fig. 3A). Thus, these postcerebral cells are likely involved in tentacular reflexes and are not significant in the control of locomotion or other behaviours requiring integration by the central nervous system. There- fore, while the possibility of species differences cannot be ruled out (Hanstrom used two species of Helix) and neither can the possibility that the retrograde uptake of hexam- minecobalt is somehow selective, our results indicate that output pathways from the procerebrum to regions other than the pedal ganglion are minor at best.

These results imply a high degree of neuroanatomical convergence in the snail's olfactory system, from roughly lo5 receptor cells (Chase, '86) to only about 25 pedal ganglion cells. Part of the convergence occurs peripherally, in the ten- tacle (Chase, '86). A comparison between convergence in the snail procerebrum and the mammalian olfactory bulb re- veals (Table 2) that while the absolute number of olfactory receptors in the rabbit and the snail differs by a factor of about 500, the comparable convergence ratios differ by a factor of only four. The similar, and large, convergences in these two distantly related animals emphasizes the seem-

TABLE 2. Comparison of Neuroanatomical Convergences in the Olfactory Systems of the Rabbit and the Snail'

Anaxonal Receptors interneurons Outputs to CNS Convergence ratios

1,Mx):ZMI:l

Snail 1 105 z 10' 25 4,000:8M):1

'The data for the rabbit are from Allison and Warwick ('49). The snail data refer t o Achatina fcrlica and are from Chase ('86) and the present study.

Rabbit 5 107 1 107 5 x 10' (granule cells) (tufted and mitral cells)

(procerehral cells) (pedal ganglion cells)

152

ingly general requirement for impressively large numbers of neurons to process olfactory information, for reasons un- known (see Chase, '86).

The results obtained with the electron microscope, while establishing the pre-and postsynaptic character of the ten- tacle nerve fibers and the cerebropedal connective fibers, respectively, have not provided an overall understanding of the functional organization of the procerebrum. This struc- ture is not likely a simple relay between tentacular neurons and pedal ganglion neurons. The presence of symmetrical synapses (Fig. 6C; see also, Zs.-Nagy and Sakharov, '70), serial/parallel arrangements (Fig. 6D), and instances of pre- synaptic profiles among labelled cerebropedal connective fibers (Table 1) is one argument against such a view. Addi- tionally, a major contributor to the procerebral neuropile, the intrinsic procerebral neuron, was not identified in our electron microscope investigation. We know, however, from our light microscope observations (Fig. 2) as well as those of Hanstrom ('25), that the processes of the intrinsic cells over- lap extensively with those of the afferent and efferent fibers in the plexus of the procerebral neuropile. Thus, the princi- pal mystery about the procerebrum is the role played by the intrinsic neurons. Any insight into the functional organiza- tion of this tissue depends upon an understanding of how the procerebral neurons connect with the input and output pathways. Such knowledge, in turn, must await the develop- ment of a method for the reliable identification of procere- bra1 processes in the electron microscope.

R. CHASE AND B. TOLLOCZKO

Bullock, T.H., and G.A. Horridge (1965) Structure and Function in the Ner-

Calabrese, R.L., and D. Kennedy (1974) Multiple sites of spike initiation in a

Chase, R. (1982) The olfactory sensitivity of snails, Achatina fulica. J. Comp.

Chase, R. (1985) Responses to odors mapped in snail tentacles and brain by

Chase, R. (1986) Lessons from snail tentacles. Chem. Senses 11:411-426. Chase, R., and B. Tolloczko (1987) Evidence for differential DNA endorepli-

cation during the development of a molluscan brain. J. Neurobiol. 18:39& 406.

Cobbs, J.. and H. Pinsker (1978) In vivo responses of paired giant mechanore- ceptor neurons in ApZysia abdominal ganglion. J. Neurobiol. 9:121-141.

Croll, R.P. (1986) Modified cobalt staining and silver intensification tech- niques for use with wholemount gastropod ganglion preparations. J. Neu- robiol. 17:569-576.

Davis, N.T. (1982) Improved methods for cobalt filling and silver intensifica- tion of insect motor neurons. Stain Technol. 57239-244.

Fredman, S.M., and B. Jahan-Parwar (1983) Command neurons for locomo- tion in Aplysia. J. Neurophysiol. 49:1092-1117.

Gelperin, A,, D.W. Tank, and G. Tesauro (1989) Olfactory processing and associative memory: Cellular and modeling studies. In J.W. Bryne and W.O. Berry (eds): Neural Models of Plasticity: Theoretical and Empirical Approaches. New York Elsevier, pp. 133-159.

Haller, B. (1913) Die Intelligenzsphiiren des Molluskengehirns. Arch. Mi- krosk. Anat. 81:233-322.

Hanstrom, B. (1925) Uber die sogennannten Intelligenzsphiiren des Mollus- kengehirns und die Innervation des Tentakels von Helix. Acta Zool. (Stockh.) 6:183-215.

Hening, W.A., E.T. Walters, T.J. Carew, and E.R. Kandel(1979) Motor neu- ronal control of locomotion in Aplysia. Brain Res. 279:231-253.

Jahan-Parwar, B., and S.M. Fredman (1976) Cerebral ganglion of Aplysia: Cellular organization and origin of nerves. Comp. Biochem. Physiol. [A] 54r347-357.

Jahan-Panvar, B., and S.M. Fredman (1979) Neural control of locomotion in Aplysia: Role of the central ganglia. Behav. Neural Biol. 27:39-58.

Kunze, H. (1921) Zur Topographie und Histologie des Centralnervensystems von Helix pomatia L. Z. Wiss. Zool. 118.25-203.

Llinas, R., and M. Sugimori (1980) Electrophysiological properties of in vitro Purkinje cell dendrites in mammalian cerebellar slices. J. Physiol. (Lond.) 305:197-213.

McCarragher, G., and R. Chase (1985) Quantification of ultrastructural sym- metry at molluscan chemical synapses. J. Neurobiol. 16:69-74.

Moulins, M., and F. Nagy (1981) Participation of an unpaired motor neuron in the bilaterally organized oesophageal rhythm in the lobsters Jasus lal- andii and Palinurus oulgaris. J. Exp. Biol. 90r205-230.

Pawson, P.A., and R. Chase (1984) The life-cycle and reproductive activity of Achatina fulica (Bowditch) in laboratory culture. J. Molluscan Stud. 50%-91.

Peterson, E.L. (1984) Two stages of integration in a leech visual interneuron. J. Comp. Physiol. 155.543-557,

Sonetti, D., M.A. Rassu, and F. Lomardo (1982) Mapping of neurons by retro- grade cobalt filling of the tentacular nerves of Planorbis corneus. Comp. Biochem. Physiol. [Aj 73~47-56.

Spurr, A.R. (1969) A low-viscosity epoxy resin embedding medium for elec- tron microscopy. J. Ultrastruct. Res. 26:31-43.

Van Mol, J.-J. (1967) Etude morphologique et phyloghetique du ganglion cerebroide des GastLropodes Pulmones (Mollusques). MBm. Acad. R. Belg. (Classe Sci.) 37:l-168.

2s.-Nagy, I., and D.A. Sakharov (1970) The fine structure of the procerebrum of pulmonate molluscs, Heliz and Limax. Tissue Cell 2:399411.

vous System of Invertebrates, Vol. 2. San Francisco: Freeman.

single dendritic system. Brain Res. 82:316-321.

Physiol. 148:225-235.

"C-2-deoxyglucose autoradiography. J. Neurosci. 5:2930-2939.

ACKNOWLEDGMENTS We thank Shelley Adamo and Gerald Pollack for their

comments on an earlier version of the manuscript. The research described here was supported by a grant from the Natural Sciences and Engineering Research Council of Ca- nada.

LITERATURE CITED Allison, A.C., and R.T. Warwick (1949) Quantitative observations of the

olfactory system of the rabbit. Brain 72:18&197. Ambron, R.T., S.G. Rayport, and J. Babiarz (1988) Giant Aplysia neuron R2

has distal dendrites: Evidence for protein sorting and a second spike ini- tiation site. J. Neurosci. 8:722-731.

Audesirk, G. (1978) Properties of central motor neurons exciting locomotory cilia in Tritonia diomedia. J. Comp. Physiol. 128259-267.

Bacon, J.P., and J.S. Altman (1977) A silver intensification method for cobalt-filled neurons in wholemount preparations. Brain Res. 138t359- 363.

Bicker, G., W.J. Davis, and E.M. Matera (1982) Chemoreception and mecha- noreception in the gastropod mollusc Pleurobranchaea californica. 11. Neuroanatomical and intracellular analysis of afferent pathways. J. Comp. Physiol. 149:235-250.

Byme, J.H., V. Castellucci, and E.R. Kandel (1974) Receptive fields and response properties of mechanoreceptor neurons innervating siphon skin and mantle shelf in Aplysia. J. Neurophysiol. 37:1041-1064.