Aplysia cell adhesion molecules and serotonin regulate sensory cell- motor cell interactions during early stages of synapse formation in vitro

Long-term facilitation of sensorimotor synapses of Aplysia in culture by serotonin (5-HT) is accompanied by two changes: an increase in the number of sensory cell branches and varicosities contacting the major axons of the target motor cell L7, and a downregulation of Aplysia cell adhesion molecules (apCAM) from the surface of the presynaptic sensory cell. We tested the hypothesis that the two changes may be linked; the 5-HT-induced decrease of apCAM levels from sensory neurites may defasciculate sensory neurites from each other and make the surface of the motor axons a more attractive substrate for new growth and synapses. We used developing cultures to examine the relationship of neuritic branching, varicosity formation, and efficacy of the connections formed by sensory cells to levels of apCAM expression on the motor cell. We then determined the consequences of 5-HT applied during the early period of interaction between sensory and motor cells (day 1 or 2 in culture) on the pattern of sensory cell growth and synapse formation. We report that the number of sensory cell branches and varicosities, and the ability of sensory growth cones to fasciculate with L7 axons and form chemical connections correlate with the level of apCAM expression on different regions of L7. Early exposure to 5-HT increased the number of sensory cell branches and varicosities contacting newly regenerated distal neurites of L7 to levels that would normally occur when the sensory neurites interact with the major proximal axons of L7. Treatment with 5-HT also modulated the efficacy of the developing synaptic connections. The change in synapse efficacy was accompanied by an increase in the formation of new sensory varicosities and branches with pioneering growth cones extending on the major axons of L7. The results are consistent with the hypothesis that treatment with 5-HT modulates local differences in the expression of cell adhesion molecules on the surface of the interacting cells making motor neurites more attractive for sensory growth cones, thereby affecting new sensory neuritic growth and synapse formation.

Long-term facilitation of sensorimotor synapses of Aplysia in culture by serotonin (5HT) is accompanied by two changes: an increase in the number of sensory cell branches and varicosities contacting the major axons of the target motor cell L7, and a downregulation of Aplysia cell adhesion molecules (apCAM) from the surface of the presynaptic sensory cell. We tested the hypothesis that the two changes may be linked; the 5HT-induced decrease of apCAM levels from sensory neurites may defasciculate sensory neurites from each other and make the surface of the motor axons a more attractive substrate for new growth and synapses.
We used developing cultures to examine the relationship of neuritic branching, varicosity formation, and efficacy of the connections formed by sensory cells to levels of apCAM expression on the motor cell. We then determined the consequences of S-HT applied during the early period of interaction between sensory and motor cells (day 1 or 2 in culture) on the pattern of sensory cell growth and synapse formation.
We report that the number of sensory cell branches and varicosities, and the ability of sensory growth cones to fasciculate with L7 axons and form chemical connections correlate with the level of apCAM expression on different regions of L7. Early exposure to 5-HT increased the number of sensory cell branches and varicosities contacting newly regenerated distal neurites of L7 to levels that would normally occur when the sensory neurites interact with the major proximal axons of L7. Treatment with 5-HT also modulated the efficacy of the developing synaptic connections. The change in synapse efficacy was accompanied by an increase in the formation of new sensory varicosities and branches with pioneering growth cones extending on the major axons of L7. The results are consistent with the hypothesis that treatment with 5-HT modulates local differences in the expression of cell adhesion molecules on the surface of the interacting cells making motor neurites more attractive for sensory growth cones, thereby affecting new sensory neuritic growth and synapse formation.
[Key words: growth cones, varicosities, synapse formation, 5-HT, sensory neurons, Aplysia cell adhesion molecule] The precise wiring of the nervous system during development is established by a multistep process that includes guidance of axonal pathways, recognition of postsynaptic targets, formation of synapses, and activity-dependent stabilization or elimination of synapses. The local cues that influence each of these steps in development include diffusible substances released by neighboring cells that affect direction or motility of the growth cones (Lumsden and Davies, 1983;Haydon et al., 1984;Tessier-Lavigne et al., 1988;Klar et al, 1992;Pini, 1993;Peter et al., 1994). In addition, the expression and distribution of specific extracellular matrix or cell adhesion molecules and their regulation by electrical or biochemical activity may contribute to axon guidance and synapse formation Landmesser et al., 1990;Bixby and Harris, 199 1;Edelman and Cunningham, 199 1;Reichardt and Tomaselli, 199 1;Takeichi, 199 1;Greenningloh and Goodman, 1992). However, the nature of the intercellular and intracellular signaling mechanisms associated with initiating synapse formation between appropriate partners, and of the role of activity in the maturation or stabilization of the newly formed synapses remain poorly understood.
The immunoglobulin-like (Ig) family of cell adhesion molecules present in both invertebrates and vertebrates appear to play critical roles throughout neural development including axonogenesis (Diamond et al., 1993), pathfinding (Edelman, 1983;Rutishauser and Jessell, 1988;Furley et al., 1990), and synapse formation (Fraser, 1980;Chow, 1990;Landmesser et al., 1990;Cunningham and Edelman, 199 1). ApCAM, or Aplysia cell adhesion molecule, is a major cell surface glycoprotein expressed on most neurons in the CNS of Aplysia from the earliest stages of development through adult life (Keller and Schacher, 1990;Schacher et al., 199 1). ApCAM isoforms are homologous to members of the Ig family of cell adhesion molecules represented by vertebrate NCAM and Drosophila fasciclin II (Mayford et al., 1992). The highest levels of expression in both the developing and mature CNS of Aplysia are in the neuropil region of the ganglia or in the synaptosome fraction (Keller and Schacher, 1990;Schacher et al., 1991). Consistent with the in vivo distribution, apCAM is expressed at the highest levels at sites of synaptic interaction between sensory and motor cells in cell culture (Keller and Schacher, 1990). Support for the role of apCAM in synapse formation comes from studies on long-term functional and structural plasticity of the sensorimotor synapse. Long-term facilitation by 5-HT is accompanied by the formation of new sensory branches and varicosities (Montarolo et al., 1986;Glanzman et al., 1990). By contrast, long-term depression of the same synapse by the neuropeptide Phe-Met-Arg-Phe-amide (FMRFamide) is accom-panied by the loss of presynaptic sensory neurites and varicos-periods, each period followed by two washes of 10 ml of perfusion ities (Montarolo et al., 1988;Schacher and Montarolo, 1991). medium at 10 min intervals to remove each control or 5-HT application. Both forms of plasticity are also accompanied by a rapid and cell-specific change in the distribution of apCAM. Treatment with 5-HT evokes a downregulation of apCAM from the surface of the sensory cell via a CAMP-dependent increase in endocytosis of clathrin-coated vesicles (Bailey et al., 1992;Mayford et al., 1992;Hu et al., 1993). On the other hand, FMRFamide causes a downregulation ofapCAM from the surface ofthe target motor cell by a similar CAMP-dependent mechanism (Peter et al., 1994;Wu and Schacher, 1994). The changes in apCAM distribution on either the presynaptic (facilitation) or the postsynaptic (inhibition) cell correlate with the expression of the long-term functional and structural changes (Bailey et al., 1992;Wu and Schacher, 1994). The changes in apCAM distribution also correlate with changes in the behavior of growth cones interacting with neurites of homologous cells (Peter et al., 1994). These results suggest that cell-specific differences in the distribution of apCAM, and extrinsic factors that may regulate the expression of these molecules on presynaptic versus postsynaptic cells may affect growth cone behavior during interactions with target cells both during the early stages of synapse formation and at later stages when mature neural circuits are modified by stimuli evoking behavioral plasticity.
The cultures were then rinsed with standard culture medium (Aplysia hemolymph and L15; Schacher and Proshansky, 1983) and returned to the incubator (18°C). For experiments examining the consequences of anti-apCAM mAb on synapse formation, the motor cell L7 was plated alone and allowed to regenerate for 1 d. Each L7 culture was treated either with anti-apCAM mAb (4E8; 250 j&ml) or control IgG antibodies (250 &ml) for 5 hr using a blind protocoi. The antibodies were washed out with perfusion medium and standard culture medium and a single sensory neuron was added to each dish with its stump 50-100 pm from the proximal motor axon. Control and experimental cultures were examined 24 hr later for the efficacy of the synaptic connection and for the pattern of sensory-motor interaction using intracellular dye injection.
Electrophysiology. The stimulation and recording techniques for measuring the amplitude of the EPSP evoked in L7 in cocultures have been described (Montarolo et al., 1986(Montarolo et al., , 1988Dale et al., 1988;Ghirardi et al., 1992). After 1 or 2 d in culture, the motor cell was impaled with a microelectrode (15-20 MQ) containing 2.0 M KC1 and held at a potential of -30 mV below the resting level to permit accurate measurement of the amplitude of the EPSP. The amplitudes of the EPSPs for the same cultures were retested a day later and for some cultures on day 4 as well. Control solution or 5-HT was added to the cultures on day one either after recording the amplitude of the EPSP or after photographing the structure of the sensory cell (see below). Synaptic potentials were evoked in L7 by stimulating each sensory cell with a brief (0.4-0.6 msec) depolarizing pulse sufficient to fire a single action potential using an extracellular electrode placed near the cell body of the sensory neuron. Cultures were matched bv the amnlitude of their initial EPSP -and divided into control and 5-HT-treated groups.
hesion-molecules on the presynaptic cell and the postsynaptic We report here that the cell-and region-specific level of ex-Dression of a&AM and the cell-specific modulation of that target may influence target selection and synapse formation.
expression by-S-HT may contribute significantly to the overall pattern of neurite outgrowth, the extent of presynaptic celltarget interaction, and synapse formation. The results support Materials and Methods the hypothesis that local differences in expression of cell ad-Dye injection and imaging structural changes. The fluorescent dye 5(6)-carboxyfluorescein (Molecular Probes; 6% in 0.44 M KOH; pH 7.0) or lissamine rhodamine (Gurr; 5% in water) was injected into each sensory cell immediately after recording the amplitude of the EPSP on dav 1 and dav 2 with 0.4-0.6 nA hvnernolarizine current nulses (500 Cell culture. Mechanosensory neurons of Aplysia were isolated from pleural ganglia dissected from adult animals (70-l 00 gm) and cocultured either with homologous sensory cells or with identified motor cell L7 isolated from the abdominal ganglion of juvenile animals (l-3 gm; Howard Hughes Medical Institute Mariculture facility, Miami, FL) and maintained for 24 d as described previously (Schacher and Proshansky, 1983;Schacher, 1985;Rayport and Schacher, 1986;Glanzman et ai., 1990). Each SN-L7 culture contained a sinale L7 cocultured with one or two sensory cells. Cells were isolated with the proximal segments of their original axons (100-200 pm for sensory cells and 250-800 pm for L7). The placement of the sensory cells with respect to the motor cell was dependent on the nature ofthe experiment. To quantify the regional differences in the structure of sensory neurons when interacting with the motor targets, a single sensory cell was plated such that its axon stumn was 100-500 urn from the oriainal proximal axons of L7 (initial segment, axon stump, or between thetwo):To study the effects of5-HT (Sigma, St. Louis, MO) or pretreatment of the motor cell L7 with high levels of anti-apCAM mAb on both synapse formation and sensory cell structure, a single sensory cell was plated with its cut axon stump near (O-100 pm) the proximal axons of the motor cell. To examine the effect of 5-HT on sensory cell growth near the newly regenerated motor or sensory net&es, two sensory cells were plated either with a third sensory cell or with a motor cell such that their axon stumps were more than 500 pm from either the cell body or axon stump of the target sensory or motor cell. To examine regional differences in the distribution of apCAM, individual motor cells and two to four sensory cells were plated in the same culture dish but far apart so that their neurites did not interact. the distal motor neurites, or other sensory neurites were taken on a Nikon Diaphot microscope with an SIT (Dage) video camera. The im-msec at 1 Hzjfor 5-6 min (Glanzman et-al., 1989: Schacher and Monages were processed by a Dell 310 computer with a PC Vision Plus tarolo, 199 1). For experiments on sensory cell interaction with distal frame grabber, and subsequently stored on a Storage Dimension or Panasonic optical disk drive. Alignment of the live view area with the motor neurites or with other sensory neurites, dye was injected on day first recorded image was aided by the computer with fine adjustments made with manual rotation of the culture dish. Illumination used for 2 and day 3. Nomarski or phase contrast and fluorescent images of the obtaining fluorescent images was kept as low as possible to prevent same view areas either along the major axons of the motor cell, along photodamage. To minimize differences in imaged structures that might arise as a result of differences in the extent of dye filling, light intensities used on the second day were adjusted to match the intensity of the stored images taken on the first day. Micrographs of the images were made with a Panasonic or Sony video printer.
Treatments. For experiments examining the effects of 5-HT, each SN-L7 or SN-SN coculture was exposed to either control solution [perfusion medium; 1: 1 by volume of L-15 medium (Sigma) with appropriate salts added to reflect marine environment of Aplysia and seawater (Instant Ocean); Montarolo et al., 19881 or to 2.5 FM 5-HT (Sigma) in perfusion medium. The cultures were exposed to these solutions for two 1 hr QuantiJication of structural change. Counts of the number of varicosities, branches, and growth cones of sensory neurons, and measurements of sensory neurite lengths were obtained from fluorescent images of sensory neurites contacting either the distal neurites or the major axons of L7. Because the cultures were photographed with a relatively low-power (20 x) objective lens, most if not all the neurites and varicosities contacting the motor cell in a given region could be viewed in a single photograph. Since the major axons of L7 are relatively thick structures, it was necessary in some cases to photograph a second focal plane to image all of the labeled neurites in a given view area. The matched fluorescent images of each focal plane for the two time points were compared and the total number of varicosities, branches, and growth cones counted. Net change was measured for all treatments. We did not examine the effect of treatment on the turnover of preexisting varicosities or neurites. Structures that were round or slightly elongated ellipses greater than or equal to 2.5 pm and connected by narrow neuritic necks were counted as varicosities Chen, 1983,1988). Growth cones were identified as elongated structures on the ends of neurites with fine-diameter filopodia. Counts were performed blind; the individual did not know the amplitude ofthe EPSPs before or after treatment or the nature of the treatments.
ApCAM immunojluorescence and distribution. or cultures of isolated motor cell L7 along with isolated sensory cells were rinsed two times at 30 min intervals with perfusion medium and exposed for 30 min to FITC-conjugated anti-ApCAM monoclonal antibody (mAb 4E8; 1 O-20 &ml) dissolved in perfusion medium (Keller and Schacher, 1990;Mayford et al., 1992;Peter et al., 1994;Wu et al., 1994). Unbound mAB was washed out with perfusion medium. Axons and neurites of each cell in a given culture dish were photographed at the same fluorescent light intensity (Mayford et al., 1992;Peter et al., 1994;. Average fluorescent intensities were measured along the entire length of the primary motor axon from cell body to axon stump (200-600 pm), and compared to average intensity measured for the newly regenerated neurites emerging from the axon stumps. These values were compared to the average fluorescent intensities of sensory neurites and growth cones regenerated by cells plated in the same dish as the motor cells. For each culture the intensity values for each region were normalized relative to the value obtained for the sensory cell growth cones. Each culture dish provided a single set of ratios that were then averaged to obtain final values.
Analysis ofdata. All data are represented as the mean rt SEM (standard error of the mean). Student's t tests (two tailed) or ANOVAs (one or two factor) followed by corrected multicomparison t tests (Bonferroni) were used to measure significance of the changes in EPSP or sensory cell structure. Differences were considered significant when p values were less than 0.05.

Results
Branching and formation of varicosities by SNs are associated with high levels of apCAM on L7 extended along the substrate or the regenerated distal motor neurites. With contact along the major motor axons after 3 d in culture, their was a twofold increase in the total number of branches (198 f 29%; N = 9; p < 0.005) and nearly a threefold increase in the number of varicosities (287 * 45%; N = 9; p < 0.005) compared to that found for sensory neurites interacting with regenerated distal motor neurites. Sensory neurite growth following contact with the original motor axons appeared directed so that the individual sensory neurites retain close apposition to the original motor axons. Contact by the large sensory neurites in the middle of the motor axon ( Fig. 1) was followed by bidirectional sensory growth, while contact at either end of the motor axons ( Fig. 2) was followed by unidirectional growth along the original motor axons. One factor affecting the overall pattern and direction of sensory cell growth may be the high level of apCAM on the surface of the original motor axons (Peter et al., 1994;Wu et al., 1994) relative to the apCAM levels on the distal motor neurites or on the sensory neurites themselves. Significant changes in the pattern of sensory cell growth may not be triggered by contact with the distal regenerated neurites of L7, because the relative differences in apCAM expression between the sensory and motor neurites a;e too small. To test this hypothesis, we examined the relative ratio of apCAM expression for different regions of L7 and sensory cells maintained under the same growth conditions to that expressed by the sensory cell growth cones (N = 6 cul-Earlier studies described qualitatively the effects of the original tures). After staining live cells with FITC-conjugated anti-apCAM proximal axons of L7 on the pattern of growth and structure of mAb, we photographed the cells and measured the average flusensory cell neurites and varicosities in culture (Glanzman et orescence intensity for the axons and neurites of the motor and al. , 1989, 1990, 199 1). These proximal axons of L7 served as sensory cells (Fig. 3). Compared to the level of anti-apCAM an excellent substrate for neuritic growth. In addition, the sen-fluorescence expressed on the growth cones of sensory cells (norsory varicosities located on the original proximal axons contain malized to a level equal to l), the level expressed on the original transmitter release sites while sensory varicosities contacting axons of L7 was nearly sixfold greater (5.76 f 0.83). The level the more distal newly regenerated motor neurites did not contain of anti-apCAM fluorescence expressed on the distal motor neumature properties of transmitter release sites (Glanzman et al., rites was twofold higher (2.06 f 0.15) than sensory growth 1989). We first quantified the regional variations in the sensory cones. Sensory neurites expressed levels that were about 40% cell neuritic structure by comparing the structure of sensory higher than the growth cones (1.42 + 0.13). The difference in neurites interacting with distal neurites of L7 with the neurites staining intensity between proximal motor axon and distal moof the same sensory cell interacting with the original proximal tor neurites is not a consequence of more membrane, since the axons of L7. We then determined whether differences in apCAM distal motor neurites contain stacks of fine-diameter processes levels expressed on proximal motor axons and distal motor while the proximal motor axons contain primarily the original neurites correlate with the overall structure of the sensory neu-large-diameter axons (Schacher and Proshansky, 1983). rons and the ability to form chemical connections.
These data are consistent with the hypothesis that one feature Single sensory cells were plated such that their neurites would that makes the original proximal axons of L7 most attractive first encounter newly regenerated distal neurites of L7 before for sensory cell growth cones and the formation of varicosities making contact with different areas of the original proximal may be the relatively high levels of apCAM expression comaxons of L7-intermediate zones (N = 3; Fig. l), cell body and pared to the level expressed on the distal motor neurites or the axon hillock (N = 3; Fig. 2A), or motor axon stump (N = 3; sensory neurites themselves. As growth cones of sensory neurites Fig. 2B). After 3 d in culture, each sensory cell was injected with contact the proximal motor axon they defasciculate from other fluorescent dye and sensory neurites extending toward the orig-sensory neurites to extend as separate processes along the surface inal proximal axons of L7, and those near or in contact with of the proximal motor axon. The relatively small difference the axons were photographed. As previously reported, the orig-between apCAM expression on the distal motor neurites cominal proximal axons of L7 had a significant effect on the number pared to the neighboring sensory neurites may not be large enough of sensory cell branches and varicosities (Figs. 1, 2). Although to alter the pattern of sensory growth while interacting with the basic pattern of sensory neurite growth was not influenced distal L7 neurites. To test these predictions, we examined the by the location of the initial interaction of the sensory neurites changes in the pattern of sensory cell growth and in synapse with the original motor axons, the location of this interaction formation when only the motor cells are treated with levels of influenced the subsequent directions of sensory growth. Rela-anti-apCAM antibodies that evoke defasciculation of homolotively large-diameter sensory neurites, formed primarily by fas-gous neurites (Keller and Schacher, 1990;Peter et al., 1994). ciculated bundles of finer-diameter neurites (Schacher and Pro-The pattern of sensory neurite outgrowth and the formation shansky, 1983; Glanzman et al., 1989) with few varicosities, of chemical connections were altered significantly by pretreat- ment of L7 with anti-apCAM mAb (Fig. 4). Treatment with the proximal motor axons (Fig. 4B2J33). The neurites either excontrol IgG had no effect on the pattern of expression and dis-tended across and beyond the motor axon (Fig. 4B3), stopped tribution of apCAM (Fig. 4AI) or the pattern of sensory neuritic growing upon reaching the motor axon, or extended parallel to growth when interacting with the major axons of L7 (Fig. 4A2,A3).
the motor axon following a turn. This failure to fasciculate with Sensory neurites defasciculated from each other and refascicu-the motor axon was correlated with a failure to form connections lated on the motor axon. The average amplitude of the EPSP (four of six cultures) such that the average EPSP amplitude was was 6.8 mV 2 2.7 (N= 6). Pretreatment ofL7 with anti-apCAM only 0.5 mV -t 0.3 (N= 6;~ < 0.001). Thus, altering the normal mAb 24 hr earlier resulted in marked decrease in the intensity difference in the level of apCAM expression on the surfaces of of anti-apCAM fluorescence from remaining unbound apCAM the sensory and motor cells can have a profound effect on the on the surface of L7 (Fig. 4Bl). Treatment also resulted in the overall pattern of interaction and the formation of synaptic failure of most of the sensory neurites to fasciculate on the connections (Hawver and Schacher, 1993).

5-HT induces branching and varicosity formation by SN neurites with distal L7 neurites
Previous studies indicated that transmitter applications that downregulated apCAM from the surface of neurons also evoke a decline in fasciculation by the cell's growth cones with neurites of homologous cells (Peter et al., 1994). Application of 5-HT downregulates apCAM from the surface of sensory net&es, but not motor cell L7, and reduces fasciculation of sensory growth cones with other sensory net&es, but has no effect on the interactions of L7 growth cones with other neurites of L7 (Peter et al., 1994). The same treatments with 5-HT evoke long-term facilitation of sensorimotor connections that is accompanied by a target-dependent increase in SN branches and varicosities interacting with the motor axons. Thus, the selective downregulation of apCAM from the surface of sensory neurites but not the motor cell may facilitate the formation of new branches and varicosities by first decreasing SN-SN interactions and increasing SN-L7 interactions leading to the formation of new synaptic contacts. These long-term changes might arise as a consequence of the transmitter-induced shift in the relative ratio of apCAM levels on the surfaces of the two cells. To test this hypothesis, we examined the long-term effects of 5-HT on the structure of sensory neurites interacting with the distal neurites of L7 or with the distal neurites of other sensory cells.
The structure of sensory neurites interacting with distal neurites of L7 after control treatment (Fig. 5) is relatively simple with few branches and varicosities compared to the structure of cells contacting the proximal axons (see Figs. 1, 2). By contrast, 24 hr after treatment with 5-HT, sensory cells elaborate many more branches and varicosities in regions of the distal motor neurites (Fig. 6). 5-HT (N = 6 cultures; 12 sensory cells) increased the number of new branches by 326 f 37% compared to a change of 178 f 20% in the number of new branches for controls (N = 6 cultures; 12 sensory cells; p < 0.01). The differential change in the number of varicosities was even greater; 385 -t 31% with 5-HT treatment and 150 + 12% for controls (p < 0.008). These changes in the pattern of growth are target dependent. The overall structure of sensory neurites contacting other sensory neurites was not affected 24 hr after treatment with 5-HT (data not shown). When reexamined 24 hr after treatment, experimental (N = 9) and control (N = 8) cells had comparable changes in the number of new branches (4.4 f 2.0 for 5-HT vs 4.7 f 1.4 for control). The defasciculation of sensory growth cones observed immediately after treatment with 5-HT (Peter et al., 1994) was not maintained 24 hr later when sensory neurites interact with homologous cells. Treatment with 5-HT while sensory neurites are actively growing did increase the rate of neuritic growth. Each branch extended 58 1.7 f 53.2 pm after 5-HT treatment compared to 316.8 + 33.1 pm after control treatment (p < 0.002). These data are consistent with the hypothesis that one of the long-term actions of 5-HT, the selective downregulation of apCAM on the surface of the extending sensory net&es, but not the motor net&es, results in a change in the ratio of apCAM expression between distal motor neurites and the sensory neurites such that sensory-motor int The Journal of Neuroscience, November 1994, 74 (11) 6891 teractions are favored over sensory-sensory interactions. These changes, along with the direct actions on the rate of neurite extension, may affect the degree of synaptic interactions between the presynaptic cell and the target.
5-HT modulates ejicacy of developing sensorimotor synapse and the number of new sensory varicosities and branches with pioneering growth cones contacting original proximal axons ofL7 Sensory cells form stable chemical connections with motor cell L7 after 4 d in culture. The efficacy of these connections and the structure of sensory neurites in contact with the original axons of L7 can be modulated for long durations, however, after exposure to specific neuromodulators (Montarolo et al., 1988;Schacher et al., 1990). During the first 4 d in culture, there is a progressive increase in the efficacy of sensorimotor connections that is correlated with a progressive increase in the number of sensory varicosities in contact with the original proximal axons of L7 (Glanzman et al., 1991;Bank and Schacher, 1992). To explore further the hypothesis that 5-HT enhances SN-L7 interaction, we examined how exposure to 5-HT alters the development of the connection and the structure of sensory neurites extending on the target axon. Treating l-d-old cultures with 5-HT had a significant and prolonged effect on the amplitude of the EPSP evoked in L7 during the first 4 d in culture. After measuring the amplitude of the EPSP and treating with either control solution (N = 6) or 5-HT (N = 6) the same connections were reexamined on day 2 and on day 4. 5-HT increased the efficacy of sensorimotor connections by more than twofold on day 2 (Fig. 7); the change in the amplitude of the evoked EPSP went from 11.0 mV + 2.4 to 24.7 mV ? 3.4 in controls versus a change from 12.0 mV f 2.1 to 40.2 mV f 2.4 with 5-HT. The difference in the efficacy of the connections persisted for 48 hr. The EPSP amplitude evoked in control cultures on day 4 was 42.3 mV -+ 5.5, while the EPSP evoked in the 5-HT treated cultures was 58.7 mV f 1.1. The difference on day 4 is an underestimate of the actual difference since the evoked response in L7 in five of the six cultures treated with 5-HT was an action potential (and the EPSP was scored as 60 mV), despite the fact that L7 was hyperpolarized by 30 mV below the resting potential (range of resting potential was 50-59 mV for the 5-HT group vs a range of 47-60 mV for the control group).
Treatment with 5-HT also had a significant effect on the structure of the sensory neurites extending along the motor axons (Figs. 8,9). In a second set of cultures, each sensory cell was injected with the fluorescent dye and photographed after recording the amplitude of the EPSP both on day 1 and 24 hr later. As was observed for the other groups, the amplitude of the EPSP after treatment with 5-HT (N = 7) increased by more than twofold compared to the increase observed with controls (N = 7), an increase of 13.1 mV f 2.8 for controls compared to an increase of 3 1.9 mV & 4.1 for 5-HT. The increase in EPSP amplitude was accompanied by a twofold increase in the formation of new sensory varicosities (9.7 f 1.6 for control vs 20.1 & 1.8 for 5-HT) and a nearly threefold difference in the or along distal motor neurites then defasciculate upon reaching the cell body/initial segment of L7 (arrowhead in A) or the axon stump of L7 (arrow in B) to generate a greater number of finer branches and varicosities. The direction of growth of the sensory neurites after contact with the proximal axons maintains the high level of interaction with the proximal axons of L7. Scale bar, 50 pm. f blgure 4. Anti-apCAM mAb treatment of L7 alters subsequent pattern of sensory-motor interaction. Distribution of apCAM and pattern of sensory neurite growth along motor axon 24 hr following treatment with mouse IgG (A) or anti-apCAM mAb (B). Sensory cells were added after treatment. Note the nonuniform distribution and relatively high-intensity staining of FITC-anti-apCAM along the motor axon following control treatment (AZ) compared to the relatively uniform and low-intensity staining following treatment with anti-apCAM mAb (Bl). The sensory neurites, visualized by rhodamine epifluorescence after intracellular injection of lissamine rhodamine (Glanzman et al.,199 1), defasciculate from each other (.12) and extend along the main motor axon (,13) with control treatment. Following treatment with anti-apCAM mAb, sensory neurites extend (BZ) and bypass the main motor axon (BS). The dye-filled sensory neurites are depicted in red and superimposed on the Nomarski view of the cocultures (A3 and R3). Scale bar, I5 pm. + Figure 3. Nonuniform distribution of apCAM along axons and neurites of L7 (A) and sensory cell (B) plated in the same dish. FTTC-anti-apCAM was applied to the live culture after 4 d and photographed as described previously (Keller and Schacher, 1990;Mayford et al., 1992;Peter et al., ! 994;Wu and Schacher, 1994). The same phase-contrast views of L7 (Al) and sensory cell (BI) are viewed with epifluorescence in A2 and B2, respectively. The fluorescent intensities are normalized to the same level for both cells and are depicted in pseudocolor with high (red) or low (dark ~/UP) levels of apCAM (range of pixel intensity between 0 and 75 arbitrary units). Note that the initial segment of L7 (arrowhead in A) and two axon stumps (arrows in A) show "hot spots" or patches of high-intensity staining. Stripes of higher intensity are also common along the axon. On average. the regenerated distal neurites of L7 are stained at a relatively low intensity. The growth cones of the sensory cell (small urrow~ in B), as well as the neurites, axon stump (lurp arrow in B) and the initial segment (arrowhead in B) of the sensory cell are stained at a low intensity relative to the major axon of L7. Scale bar, 50 pm. Figure 5. The structure of sensory neurites is relatively simple when interacting with the distal neurites of L7. A, Nomarski contrast view of SN-L7 culture on day 3. The sensory cell was plated about 600 pm from the proximal axons of the target motor neuron L7. The large arrow points to the axon stump of the sensory cell; the open arrow points to location of a sensory growth cone on day 2, small arrowhead added as reference point. B, Epifluorescent photograph of sensory neurites emerging from axon stump (soIid arrow) and extending toward the motor axon on day 2 following injection of 5(6)-carboxyfluorescein. One growth cone (open arrow) has made contact with the most distal motor neurites at this time. C, Epifluorescent photograph of regenerated sensory neurites 24 hr later. Sensory neurites have extended substantially and made contacts with many distal motor neurites, but the growth pattern remains relatively simple; sensory neurites extend in fasciculated thick bundles with few varicosities formed while contacting distal neurites of L7. Scale bars: A, 70 pm; B and C, 30 Wm. number of new sensory growth cones extending along the proximal axons of L7 (4.1 k 1.1 for control vs 11.8 f 2.2 for 5-HT). These results are consistent with the hypothesis that 5-HT enhances synapse formation by simultaneously decreasing sensory-sensory interactions while increasing sensory-motor interactions.

Discussion
Our results suggest that the regional distribution of a single type of cell adhesion molecule on the surfaces of both the potential target and the ingrowing presynaptic processes may influence the direction and pattern of growth by the presynaptic growth Figure 6. 5-HT changes the structure of regenerating sensory neurites interacting with distal neurites of L7. A, Nomarski contrast photograph of SN-L7 culture on day 3. The sensory cell was plated about 500 pm from the stump of one of the proximal axons of the target motor neuron L7. Arrows indicate reference points. B, Epifluorescent photograph of all regenerated sensory neurites contacting the distal neurites of L7 on day 2. Arrow indicates reference point. C, Epifluorescent photograph of regenerated sensory neurites contacting distal neurites of L7 24 hr later and following applications of 5-HT. New sensory neurites branch extensively and form numerous varicosities; a pattern of growth that is more typical of sensory neurites contacting the proximal axons of L7 (see Figs. 1,2). The arrows indicate reference points. Scale bars: A, 70 pm; B and C, 30 pm. Day. cones as well as the formation of structures associated with synaptic contacts. The preference of the sensory growth cone to extend along the surfaces of the proximal axons of L7 (Glanzman et al., 1989(Glanzman et al., , 1990(Glanzman et al., , 1991 correlates with the high level of expression of apCAM. Thus, the overall pattern of growth by a presynaptic cell when contacting potential zones within a target region may be influenced by the same types of reciprocal ligandreceptor interactions between the surfaces of the cells that may mediate growth cone responses for different substrates during axonal guidance (Landmesser, 1980;Goodman et al., 1984;Cox et al., 1990;Raper and Kapthammer, 1990;Kuwada, 1992;Schwabe et al., 1993). When sensory cell growth cones are given a choice between extending along the substrate or other sensory neurites, there is a preference for the sensory neurites (Peter et al., 1994). The distal motor neurites on average express a level of anti-apCAM fluorescence that is about 40% greater than that observed for sensory net&es. This difference correlates with a small increase in branching and in the number of varicosities formed by sensory neurites with distal motor neurites compared to the growth pattern observed with interactions with other sensory neurites (Glanzman et al., 1989). The most profound change in growth is associated with the original proximal motor axons that express nearly a threefold higher level ofanti-apCAM fluorescence compared to the distal motor net&es. Thus, as a sensory growth cone contacts this region it fasciculates with the major motor process and extends along it rather than other neighboring sensory or motor neurites. This correlation suggest that the sensory growth cone may be capable of detecting differences in apCAM levels either at a choice point or along a gradient. Such choices within a target region may contribute to the mechanism by which homogeneous populations of presynaptic axons (e.g., motor neurons or retinal ganglion cells) select specific zones within a homogeneous population of target cells to establish connections (Sperry, 1963;Trisler et al., 198 1;Stryker and Harris, 1986;Shatz, 1990;Baier and Bonhoeffer, 1992;Goodman and Shatz, 1993;Kishishian et al., 1993;Sanes, 1993).
Although we cannot rule out the possibility that other surface molecules are mediating some of the target-dependent structural changes in the sensory cell, the fact that apCAM isoforms make up a large fraction of total membrane protein (Keller and Schacher, 1990) favors the importance of these molecules in contributing to the process of synapse formation between the sensory and motor cells. This role in synapse formation is supported by the profound changes in sensory cell growth and synapse formation when anti-apCAM mAb bound on the motor axon alone interferes with sensory neurite fasciculation with the motor axon and the establishment of effective synaptic connections. We do not know how regional differences in the level of apCAM expression are generated. Perhaps the original motor axons and dendrites attached to the cell body of L7 at the time of plating maintain their high levels while insufficient time has elapsed for the insertion of newly synthesized apCAM molecules on the surfaces of the regenerated distal neurites. The high level of apCAM expression at sites of contact between sensory varicosities and the motor axons suggests the possibility that sites of cell-cell interaction may represent regions of preferred insertion or capping of these cell surface molecules, leading to the uneven distribution (Keller and Schacher, 1990;Bailey et al., 1992;Singer, 1992). In addition, the overall distribution of these molecules on either the presynaptic or postsynaptic cells may have been influenced by the earlier actions of specific neuromodulators (Bailey et al., 1992;Mayford et al., 1992;Peter et al., 1994;Wu and Schacher, 1994).
One consequence of the selective downregulation of apCAM from the surface of the sensory cells by 5-HT is a shift in the relative ratio of apCAM expression to favor sensory cell interaction with the motor cell. A sensory growth cone, after 5-HT treatment, is given a new cellular substrate choice between a sensory neurite and a distal motor neurite that now show about a twofold difference in the intensity of fluorescently tagged apCAM levels on their surfaces. The calculation of this difference is based on the average 5-HT-induced reduction in apCAM on the surface of the neurites of isolated sensory cells and the absence of any change evoked by 5-HT on the distribution of apCAM on L7 neurites (Mayford et al., 1992;Peter et al., 1994). This difference of about 1 OO%, compared to a 40% differential in anti-apCAM fluorescence in control situations, may now be large enough to contribute to the 5-HT-induced change in the pattern of growth when sensory neurites contact the distal neurites of L7.
The mechanism by which a sensory growth cone selects between different cell substrates (even those on the same target cell) which express different levels of apCAM is not known. Second messenger cascades triggered by apCAM interactions at the surface (Atashi et al., 1992;Doherty and Walsh, 1992) may alter cytoskeletal organization affecting growth cone motility Figure 8. Early application of 5-HT alters the pattern of sensory growth on the proximal axons of L7. AI, Epifluorescent micrograph of sensory neurites growing on the major axon of L7 after 1 d in culture. The major axon of L7 extends from left to right in the middle of the photograph. The bright fluorescent signal at far left is the sensory axon and stump. Note the sensory neurites extending on the motor axon in the middle of the field. The EPSP evoked in L7 was 15 mV. A2, Growth pattern of sensory neurites on L7 axon 24 hr after treatment with 5-HT. Note the appearance of additional growth cones of the extending sensory neurites and new varicosities along individual sensory neurites. The EPSP on day 2 was 46 mV. Scale bar, 30 pm. and direction (Forscher et al., 1987;Kater et al., 1988;Lohof et al., 1992). These changes, coupled with the others evoked by 5-HT in the sensory cells-alterations in the expression of a number of proteins including Bip, calmodulin, calriticulin, and the light chain of clathrin (Barzilai et al., 1989;Kennedy et al., 1992;Kuhl et al., 1992;Byrne et al., 1993;Hu et al., 1993), changes in the activity of various protein kinases including PKA and PKC (Kandel and Schwartz, 1982;Occor and Byrne, 1985;Greenberg et al., 1987;Bergold et al., 1990;Saktor and Schwartz, 1990;Byrne et al., 1991;Ghirardi et al., 1992;Sossin and Schwartz, 1992), and the extent of phosphorylation of the substrates of these kinases )-may contribute to the structural changes associated with the formation of new branches and varicosities (Nazif et al., 1991;Schacher et al., 1993). Some of these other changes may contribute to the 5-HT-induced increase in neurite extension by sensory cells plated with other sensory cells when measured 24 hr after treatment. This effect on the rate of neurite extension is not observed for isolated sensory cells immediately after the 5-HT treatment (Peter et al., 1994), or for sensory cells that have been in culture for 5 d and are no longer in the active phase of neurite growth (Glanzman et al., 1990). Thus, 5-HT actions may regulate a number of ongoing processes associated with growth and synapse formation with different kinetics, but cannot initiate those processes in cells that have achieved their final morphology and lack an appropriate synaptic partner.
The exposure to 5-HT during the early stages of SN-L7 interactions appears to accelerate transiently the normal time- dependent changes in synaptic efficacy. This functional change in efficacy is then maintained for an additional 48 hr. The initial functional change in the connection is paralleled by a change in the number of varicosities and branches as was observed for 5-HT treatment of mature connections after 5 d in culture (Glanzman et al., 1990;Bailey et al., 1992). The downregulation of apCAM from the surface of the sensory cells may also contribute to the initial changes in the developing cultures, especially the increase in the number of sensory growth cones extending along the proximal motor axons. The increase in growth cones may arise from the 5-HT-induced defasciculation of sensory neurites from other sensory neurites (Mayford et al., 1992;Peter et al., 1994). Since each neurite is capable of forming varicosities along the motor axon, a change in the number of sensory branches extending along the axon of the motor cell can lead to a change in the degree of interaction between the surfaces of the sensory neurites and motor axons. This change, in turn, may lead to the formation of more synaptic contacts (Hawver and Schacher, 1993). Thus, even a relatively small change in the ratio of apCAM expressed by the sensory neurites and the proximal motor axon at a critical point in the formation of synapses can lead to a relatively large change in the overall structure of the sensory cells and in the efficacy of their synaptic connections with the target cells. The change evoked by early exposure to 5-HT parallels the findings of Greenough et al. (1973Greenough et al. ( , 1985 on the consequence of early experience on the morphological development of various areas of the CNS in vertebrates. Activity-dependent activation of specific pathways during critical periods in the development of neural circuits may enhance the functional and structural features of the activated pathway (Constantine-Paton et al., 1990;Shatz, 1990;Goodman and Shatz, 1993) by modulating the levels of expression or distribution of cell adhesion molecules that continue to be expressed throughout the life of the animal. Additional changes in these circuits via the modulation of NCAM-like molecules with stimuli that evoke behavioral plasticity in the adult animal (Mayford et al., 1992;Cremer et al., 1994) could be mediated in part by the same mechanisms that are first utilized during the formation and stabilization of the connections in development.