SSB, an Antigen that Selectively Labels Morphologically Distinct Synaptic Boutons at the Drosophila Larval Neuromuscular Junction Vivian Budnik* and Michael Gorczyca Department of Biology, Morrill Science Center, University of Massachusetts, Amherst, Massachusetts 01 003

SUMMARY In this report we describe the expression of Small Synaptic Bouton (SSB), an antigen that is selectively expressed in a specific subset of neuromuscular junction terminals in the body wall of Drosophila larva. The expression of SSB was studied with a polyclonal antibody raised against the C A M Pphosphodiesterase of the Drosphilu learning mutant dunce (Nighorn et al., 1991, Neuron 6:455-467); however, immunctreactivity was not abolished by the dunce (dnc) alleles dncMI4and dncM" or deficiencies of the dnc gene, indicating that the antigen labelled could not be the dnc gene product, but another antigen that we termed SSB. Immunoreactivity was localized in the body wall muscles to a specific subset of neuromuscular junction terminals that have been implicated in activity-dependent plasticity. This demonstrates that

these morphologically distinct terminals can be immunocytochemicall) distinguished and that the) probablq represent innervation b j a distinct neuronal population. Confocal and electron microscopic examination demonstrated that staining was restricted to the synaptic boutons themselves, not to neurites or motor axons. Ultrastructural analysis showed label close to synaptic Fesicles in the presynaptic terminal and in the surrounding subsynaptic reticulum. Central nerbous system ( C N S ) staining was restricted to a segmentally repeated pattern of cell bodies in the ventral ganglion and to a few small groups of cells in the brain lobes. o 1992 John NIIC)&

INTRODUCTION

strength of synaplic contacts that we observe at a given developmental state is the result o f a number of regulatory sigrials that can change over time ( Herrera and Werle, 1989). One of the regulatory signals that has been documented in numerous systems is the electrical activity of neurons (Brown and Ironton, 1977; Hubel, Wiesel, and LeVay, 1977; Harris, 1980; Meyer, 1982; Sanes and Constantine-Paton, 1983; Lnenicka, Atwood, and Marin, 1986; Schmidt and Tieman, 1989; Lnenicka, Hong, Combatti, and LePage, 199 1 ). In these systems, changes in the electrical activity of neurons lead to changes in connectivity during development, changes in the number of synaptic contacts between the presynaptic neuron and the postsynaptic target, and morphological changes in the presynaptic terminals. We have been studying the regulation of syn-

How the number of synaptic contacts at a particular target is regulated remains one of the central questions of developmental neurobiology. This becomes particularly important in light of many studies indicating that synaptic contacts are not static structures, but may undergo a continuous process of remodelling ( Anzil, Bieser, and Wernig, 1984; Murphey and Lemere, 1984; Wernig and Fisher, 1986; Bailey and Chen, 1988a,b; Greenough and Bailey, 1988). Under this view, the number and Received May 15, 1992; accepted July 23, 1992 Journal ofNcurobiology, Vol. 23, No. 8, pp. 1054-1066 (1992) 0 1992 John Wiley & Sons, Inc. CCC 0022-3034/92/09 1054-13 * To whom correspondence should be addressed

1054

Sons, Inc

Keywords: synaptic bouton, motor terminal, neuromuscular junction, dunce, phosphodiesterase.

Selecfive Labelling of S-vnuptic Boutons

apses at the Drosophilu larval neuromuscular junction. This system is particularly suitable because it is composed of large, identifiable muscle cells on which motorneurons synapse in a specific pattern (Johansen, Halpern, Johansen, and Keshishian, 1989a; Budnik, Zhong, and Wu, 1990). In addition to its accessibility for electrophysiologicaltechniques (Jan and Jan, 1976; Wu and Haugland, 1985) as well as its increasingly well-characterized development (Johansen, Halpern, and Keshishian, 1989b; Sink and Whitington, 199 la,b; Halpern, Chiba, Johansen, and Keshishian, 199l ), the possibility of molecular and genetic approaches makes this system especially attractive. We have previously studied the morphology of motor axon terminals in mutant combinations that increase neuronal activity (Budnik et al., 1990) by either disrupting potassium currents (Ganetzky and Wu, 1983) or increasing the number of sodium channels ( Wu and Ganetzky, 1980). In these mutants there is an increase in the ramification of terminals and in the number of synaptic boutons (Budnik et a]., 1990). These alterations were found to be more pronounced in a morphologically distinct subset of terminals, and could be rescued by a mutation that reduces neuronal activity by decreasing the expression of sodium channels. Involvement of CAMP metabolism on these morphological changes was suggested by the finding that the mutant dunce ( d n c ) , which eliminates a CAMP-dependent phosphodiesterase ( PDE ) (Byers, Davis, and &ger, 1981), can mimic the morphological alterations caused by increased activity. In addition, it interacts in a synergistic fashion with the potassium channel mutants to enhance further the phenotype at the neuromuscular junction (Zhong, Budnik, and Wu, 1992). At the same terminals, dnc disrupts synaptic facilitation and potentiation (Zhong and Wu, 199 1 ), and affects a CAMP-dependent K + channel (Delgado, Hidalgo, Diaz, Latorre, and Labarca, 199 1 ) . Because of the effects of dnc on the structure of motor terminals and the physiology of the neuromuscular junction, we decided to study the expression of the dnc gene product at the body wall muscles, using polyclonal antibodies against the dnc protein (Nighorn, Healy, and Davis, 199 1 ). Confocal and electron microscopy, as well as double-labelling techniques showed that immunoreactivity was exclusively expressed in a subset of synaptic terminals, the same class of terminals that show activity-dependent plasticity (Budnik et al., 1990). In the central nervous system (CNS) the expression was restricted to a small group of cell bodies.

I055

Surprisingly, however, animals genetically deficient for the dnc region did not eliminate the dnclike immunoreactivity. This result indicates that the antigen expressed in terminals is not the dnc product, but an antigen that shares an epitope with the CAMP-phosphodiesterase. We have termed this antigen Small Synaptic Bouton antigen (SSB). This is the first demonstration in Drosophilu that morphologically distinct terminals at the body wall muscles can be immunocytochemically distinguished, and, therefore, they probably represent innervation by a separate motorneuron. Whereas the significance of this antigen to structural plasticity or to synaptic function is unclear, its exclusive ultrastructural localization to synaptic boutons suggests that it may be important for these processes.

MATERIALS AND METHODS Flies For these studies flies were reared at 25°C in a standard Drosophila medium. The wild-type strain was Canton-S (CS). Mutantlinesused were N 6 4 " 6 / Y / X X yw;Dp(l:2) w + " ~ ~ / + ,Df(1) N64i15/w+Y/XX y w,Df(1) dm7Se19/ FM7a, D f f I ) N 7 l h / X Xy 2; Dp(1:2) w+'lb7/+, y w cv v f dncMt4/XXy J y w cv v,f dncM1'/XXyf; Df(2L) 130 rdo pr cho/CyO, Ddcnz7pr/CyO, shitsl (for chromosomal markers and balancers see Lindsley and Zimm, 1992; for description of dnc alleles and deficiencies see Kiger and Salz, 1985; and Davis and Davidson, 1984). The dnc mutant stocks were obtained from Drs. R. Davis, T. Tully, and the Bloomington Stock Center. The Ddc stocks were from Dr. K. White. Animals deficient for the dnc gene (Ofdnc) were created as follows: Df ( I ) N6"16/ X Iw+Y/XXy w (50% of male larvae Y; Dp W + ~ ' ~ ' / S M from this cross should be D,f dnc) and N64'16/Y; Dp(l:2) w"lb7/+ X D f ( l ) d m 7 S e 1 9 / F M (25% 7 ~ of the progeny should be D f d n c ) . To obtain larvae deficient in the Ddc gene ( D f D d c ) D f (2L) 130 rdo pr cho/CyO were crossed to D d ~ " ~ ~ p r / C DfDdc y O . larvae were obtained according to Vallts and White ( 1986). For developmental studies, wild-type embryos were staged according to Budnik, Wu, and White ( 1989). A

lmmunocytochemistry Unless otherwise specified all incubations were done at room temperature. Body wall muscles of larvae were dissected in Ca*+-freeDrosuphilri saline (concentrations in mM: 2 KCI; 128 NaCI; 4 MgCI,; 2 EGTA; 35.5 sucrose; 5 Hepes, buffered at pH 7.2 with NaOH) and fixed for 2.5 h in freshly made 4% paraformaldehyde, 0.1 Mphosphate buffer, pH 7.2. Samples were incubated in rabbit

I056

Biidnik and Gorczyca

anti-horseradish peroxidase ( H R P ) ( 1:200; Sigma), rabbit anti-dunce ( 1 5 0 ; Nighorn et al., 1991), or rabbit anti-serotonin (5-HT) ( 1:200; INCTAR or 1:50 Accurate Chemicals) overnight or for 48 h at 4°C. For 5-HT immunocytochemistry dissections, fixation, and rinses were done at 4°C. As secondary antibodies, H R P o r fluorescein isothiocyanate- ( FlTC) conjugated goat anti-rabbit immunoglobulin ( 1:20 and 1:100, respectively; Cappel) were used ( 3-4 h incubation). Rinses and dilution of the antibodies were done in 0.1 M phosphate buffer, pH 7.2, containing 0.2% Triton X-100. Visualization of immunoreactivity when using H R P conjugated secondary antibody was done using diaminobenzidine (DAB) and 0.06% H,O, as substrate in 0.5 M Tris buffer, pH 7.6. For cryostat sections of adult brains, the method of Nighorn et al. ( 1991) was used. The SSB antibody was an affinity-purified antibody (gift from Dr. R. Davis), directed against a fusion protein composed of LacZ protein and a sequence of the dnc molecule corresponding to the phosphodiesterase-conserved region ( Nighorn et al., 1991). Samples were observed using a Bio RAD M600 confocal unit attached to a Nikon microscope or using epifluorescence on a Zeiss Axioskop. Images in Figures 1 and 5 were enhanced using the NIH program Image (version 1.4).

Double-Labelling Experiments For double-labelling experiments, the primary antibodies were goat anti-HRP (Sigma), goat anti-5-HT (INCTAR), and rabbit anti-SSB. For these experiments, simultaneous incubations with a mixture of primaries (either anti-HRP and anti-SSB or anti-5-HT and anti-SSB) were followed by incubation in a mixture of rhodamineconjugated anti-goat and FITC-conjugated anti-rabbit ( 1:lOO: Chemicon). Both secondaries were made in donkey.

lmmunoelectron Microscopy For SSB immunoelectron microscopy, body wall muscles were dissected in Drosophila saline containing 0.1 m M CaCI, ( n o EGTA), fixed in 4% paraformaldehyde, and processed as above using the HRP-conjugated secondary antibody. Triton-X 100 (0.2%)was used only for the primary incubation. All rinses were done in 0.1 M phosphate buffer, pH 7.2. After the H R P reaction, the DAB reaction product was silver-gold intensified according to Liposits, Sherman, Phelix, and Paul ( 1986) as follows. Samples were rinsed twice in 2% sodium acetate solution (SA) ( 15 min), incubated 4 h in 10% thioglycolic acid at 4°C to suppress nerve tissue argyrophilia, rinsed 4 times ( I 5 min) in SA, and transferred to freshly prepared physical developer (solution A: 2.5% sodium carbonate; 0.1% ammonium nitrate, 0.1% silver nitrate, 0.5% tungstosilic acid, 0.7% paraformaldehyde) for about 6-8 min. The time ofthis incubation was found to

be one of the most critical steps for successful specific staining. Incubations over 8 min in physical developer were found to increase unspecific staining significantly. Development was stopped by incubating the samples in 1% acetic acid for 5 min. Samples were then rinsed in SA for 5 min, gold toned in 0.05% gold chloride to replace silver grains by gold particles, rinsed in SA for 5 min, and rinsed twice (10 min) in 3% sodium thiosulfate to remove unbound silver. Samples were rinsed twice in SA ( 5 inin), transferred to 0.1 M cacodylate buffer, 35’70 sucrose, pH 7.0, for 15 min, postfixed in 1% Os,O, and embedded in Spurr’s by conventional procedures. Immunoreactive boutons in ventral longitudinal muscles 12 and 13 were initially localized in thick sections, and then ultrathin ( 100 nm) sections were obtained.

RESULTS Expression of SSB lmmunoreactivity at the Body Wall Muscles

The body wall muscles of Drosuphilu larvae consist of a segmentally repeated pattern of identifiable muscle fibers (Crossley, 1 9 7 8 ) . Each of these muscle fibers is innervated in a stereotypic muscle-specific fashion (Johansen et al., 1 9 8 9 a ) by one or more motorneurons. As in most invertebrate muscles, each motor axon terminal makes multiple synaptic contacts along each muscle fiber. These contacts are believed to be localized at synaptic boutons or varicosities along the motor axon terminals (Lnenicka et al., 1986; Budnik et al., 1 9 9 0 ) . At least two types of terminals can be distinguished according to their length and size of synaptic boutons (Johansen et ,d., 1 9 8 9 a ) : short terminals bearing large (3-8 pmi synaptic boutons (type-I terminals) and long terminals with small ( 1-2 pm) boutons (type-I1 terminals). Several physiological and anatomical studies have indicated that body wall muscles are polyinnervated (Jan and Jan, 1976; Halpern et al., 1991; Sink and Whitington, 1991a). Therefore, it is possible that these two types of terminals belong to two different populations of motorneurons that may express different physiological properties; however, the evidence for this has been so far lacking. Antibodies against the dnc gene product (Nighorn et al., 199 I ) were used to ascertain whether this CAMP-dependent phosphodiesterase is expressed at the body wall motor terminals. Immunoreactivity was found to be selectively localized to the same terminals, type 11, that show activity-dependent morphological plasticity (Fig. 1 ) . Because immunoreactivity was not eliminated by dnc alleles that

Figure I Expression ofthe SSB antigen at motoraxon terminals. (A)Anti-HRPimmunoreactivity in muscles 12 and 13 stains all terminals as visualized with the confocal microscope. Notice the presence ofboth type-I (arrow) and -11 (arrowhead) terminals. ( B ) Anti-SSB immunoreactivity in muscles 12 and 13 in a different sample. Notice that only type-11-terminals are labelled by this antibody. (C) High-magnification view of two stretches of type-I1 terminals at a sample labelled with the anti-HRP antibody and ( D ) at a sample stained with the SSB antibody. Notice that in contrast to H R P immunoreactivity. SSB immunoreactivity is concentrated in the boutons. Scale bar = 40 Nm in ( A , B ) , and 15 pm in (C,D).

eliminate the dnc gene product (see below) the antigen was termed Small Synaptic Bouton antigen. Confocal microscopy and double-labelling techniques were employed to identify unambiguously the terminals labelled by anti-SSB at type-I1 terminals. Anti-HRP antibody was used to label all nerve terminals, because in fruit flies virtually all nervous tissue, including motor nerve terminals, is fortuitously recognized by this antibody (Jan and Jan, 1982). Comparison of anti-SSB with antiHRP immunoreactivity revealed that the localization of SSB was exclusive to synaptic boutons of type-I1 terminals in all body wall muscles that are innervated by this type of terminal. In SSB- and HRP-double-labelled preparations we compared stretches of terminals containing from approximately three to 80 synaptic boutons. Virtually all boutons systematically examined (more than 10,000) showed complete correspondence be-

tween type-I1 HRP- and SSB-labelled boutons; however, no SSB immunoreactivity was found in type-I terminals (Fig. 2). For example, muscles 6 and 7 contain only type-I terminals as revealed by anti-HRP labeling [Fig. 2( A)]. We found that these muscles were completely devoid of SSB immunoreactivity [Fig. 2( B)]. Muscles 12 and 13, on the other hand, show both types-I and -11 terminals [Fig. 2( C)] ; however, only type-I1 terminals are labelled by the SSB antibody [Fig. 2( D)]. Figure 3 is a diagrammatic representation of the musculature of a larval body segment showing the expression of SSB immunoreactivity at different muscles (analysis of seven samples, approximately eight hemisegments per sample). We found that most exterior muscles (about one-third of all body wall muscles) expressed SSB immunoreactivity with a high frequency (70% of hemisegments or more). In addition, a few internal and superior

Figure 2 Anti-HRP (A$) and anti-SSB (B,D) double labelling of muscles 6 and 7 ( A,B) and 12 (C,D). Notice that in muscle 6 and 7, which are exclusively innervated by type-I terminals, no SSB immunoreactivity is observed. Muscle 12 contains both type-I and -11 terminals (C), but only type-I1 terminals are labelled by the SSB antibody (D). Scale bar = 20 pm.

4

=:

a

Selcctivr Labelling ($Synaptic Bozitons

__

Ventral Midline

1059

_ .

US-69%

0 2 5 - 49 %

00-241

Figure 3 Diagrammatic representation of SSB immunoreactivity at the body wall muscles. Muscles containing SSB-immunoreactive terminals in more than 70% of cases are shown in black. Muscles found to contain SSB-positive boutons in 50%-69% of cases are stippled in a black background, and muscles showing immunoreactivity in 25%-49% of cases are shown in gray. Muscle fibers that were not identified routinely from preparation to preparation are indicated by "?" The muscles shown are the fibers of a single abdominal segment that has been divided into three layers [external ( E X ) , internal (INT), and superior (SUP)] for clarity. This drawing was based on a diagram by Anderson et al., 1988. Quantification was based on an analysis of seven samples using approximately eight hemisegments per sample.

muscles also showed SSB immunoreactivity at a moderate frequency (50%-69%). We did not observe any segmental differences in this pattern (examination of abdominal segments 2-7). Analysis of SSB immunolabelling using confocal microscopy revealed that SSB immunoreactivity was concentrated in the synaptic boutons themselves [ Fig. 1 (C,D)]. No detectable levels ofimmunoreactivity were found in either neurites between synaptic boutons or the axons before they contacted the muscles. The postsynaptic muscle fibers appeared not to have any significant level of staining above background. Expression of SSB during Development

SSB immunoreactivity is first observed at about 18 h (stage 17 ) of embryonic development in type-I1 terminals. At this stage, the neuromuscular junction has been recently established (Johansen et al., 1989b). No other structure appeared to be labelled during the embryonic period.

Examination of SSB immunoreactivity at different times of larval and early pupal development demonstrated that immunoreactivity was present at type41 terminals throughout the larval period and as far as 8 h after puparium formation. After this period the larval muscles degenerated and SSB-immunoreactive boutons could not be seen. Ultrastructural Localization of SSB

The subcellular localization of SSB at the electron microscopical level was studied using HRP-conjugated secondary antibodies with diaminobenzidine as substrate. In some preparations the intensity of the signal was amplified by using a silver-gold intensification technique (Liposits et al., 1986). We concentrated on muscles 12 and 13 of the second and third abdominal segment (Crossley, 1978). Because glutaraldehyde fixation precluded SBS-I1 immunoreactivity, only paraformaldehyde was used, sacrificing some fine ultrastructural details. However, with this fixation protocol, the main fea-

tures of the synaptic terminals such as vesicles, mitochondria, synaptic densities, and subsynaptic reticulum, were conserved (Budnik et al., 1990). Two types of stained synaptic terminals, according to synaptic vesicle type and ultrastructural anatomy, could be distinguished. The most frequently observed and most densely stained type of terminal contained 30-nm clear vesicles, and was surrounded by a well-developed subsynaptic reticulum at the postsynaptic muscle [Fig. 4( A,B)]. The vesicles at these synaptic terminals are presumed to contain glutamate, the primary excitatory transmitter at this neuromuscular junction (Jan and Jan, 1976; Johansen et al., 1989a). ,$nother, much less frequently observed type of terminal contains 120to 150-nm dense vesicles, and has a greatly reduced subsynaptic reticulum [Fig. 4 ( C ) ]. The transmitter/ neuromodulator at these vesicles is unknown, but proctolin, octopamine, and insulin-like immunoreactivity has been reported in these muscles (Anderson et al., 1988; Halpern et al., 1988; Budnik and Gorczyca, submitted). In stained terminals containing clear vesicles, SSB was concentrated in the presynaptic terminal and the postsynaptic subsynaptic reticulum (Fig. 4 ) . Within the presynaptic terminal, label was distributed in close proximity to synaptic vesicles. No staining appeared in mitochondria or was associated with other structures. In addition, label was not observed at regions of the synaptic terminals that do not contain vesicles, such as the axonal segment between terminals. Label in the subsynaptic reticulum was not confined to any particular area or structure. Terminals containing dense vesicles also showed some degree of immunoreactivity above background, but to a much lesser extent than terminals containing clear vesicles [Fig. 4(C)]. Occasionally some label also appeared in the muscle fiber [Fig. 4 ( A ) J , but this staining was very sparse. We do not know whether this staining represents background staining amplified by the silver intensification protocol, or whether the SSB antigen is expressed at very low levels in the muscle.

Distribution of SSB lrnmunoreactivity in the Ventral Ganglion SSB immunoreactivity in the larval CNS was localized to a small, segmentally repeated subset of neurons and fibers in the ventral ganglion (Fig. 5 ) and a number of cell clusters and neuropil in the brain (not shown). The intensity of the immunoreactivity was strongest in CNS of young larvae (first and

second instar) and became very dim in the third instar. This age-dependent difference was not detected at neuromuscular junction nerve terminals at which immunoreactivity was strong at all stages observed. Because of the regularity and consistency of the pattern of immunoreactive cell bodies observed in the ventral ganglion as well as its relevance to the potential staining at the neuromuscular junction, we will limit our description to this region of the CNS. SSB immunoreactivity was found as a pair of neurons per hemisegment in most abdominal segments, as a single neuron in the last abdominal hemisegment [Fig. 5 ( A ) ] , and as midline neurons in the subesophageal segments [Fig. 5 (C)]. Staining was also observed in the neuropil. The neurons at each segment sent projections to the contralatera1 side of the same segment, with terminals surrounding the contralateral SSB-positive neurons, but in a plane several microns away from the cell bodies [Fig. 5(A,C)]. The staining was most intense at the seventh segment, where SSB-immunoreactive cells were more dorsally located [Fig. 5 ( B ) ] .The pattern ofdistribution ofSSB-inimunoreactive neurons was highly reminiscent of the pattern of monoamine-containing neurons previously described by VallCs and White ( 1988) and Budnik and White ( 1988 ). Therefore, we double-labelled larval CNS using serotonin and SSB antibodies. We found exact correspondence between 5-HTand SSB-containing neurons in the abdominal segments: however, there was no or little correspondence of 5-HT- and SSB-containing neurons in the thoracic segments and in the brain. The observation that SSB-positive neurons in the ventral ganglion also contain serotonin is surprising, because no serotonin immunoreactivity was found in typeI1 terminals, or any terminals, at the body wall muscles. Because of incompatibility in the techniques, we did not attempt to double label dopamine-containing and SSB-immunoreactive cells; however, we found that some SSB-positive cells in the subesophageal region were similar, with regard to pattern and localization, to the catecholamine-containing neurons at this region [Fig. 5 ( C ) ] (Budnik and White, 1988). The pattern of immunoreactivity in the larval CNS was in contrast with the pattern described by Nighorn et al. ( 199 1 ) using the same antisera. In that study it was reported that immunoreactivity was concentrated in the mushroom bodies of larvae and adults. We found no staining at larval

Figure 4 Ultrastructural localization of SSB to synaptic terminals. ( A ) Synaptic terminal on muscle 12 stained with anti-SSB primary and HRP-conjugated secondary, followed by silvergold intensification of the diaminobenzidine reaction product. This synaptic terminal contained clear vesicles. ( B ) High-magnification view of the synaptic terminal shown in ( A ) . Notice that the label is most concentrated at regions of the terminal that contain synaptic vesicles. ( C ) Synaptic terminal on muscle 13 contains dense vesicles and has a less developed subsynaptic reticulum, characteristic of this type of terminal. Also notice that SSB immunoreactivity at this terminal is less concentrated than in terminals containing clear vesicles. a = axon terminal; M = muscle; ssr = subsynaptic reticulum; arrow = vesicles. Scale bar = 1.5 fim in ( A ) and 0.4 pm in (B,C).

1062

Biidnik und G'OYCZ~CU

Figure 5 SSB immunoreactivity in the ventral ganglion visualized with the confocal microscope. ( A ) SSB-immunoreactivecells and fibers in the abdominal segments. ( B ) High-magnification view of SSB-positive cells and fibers in the 7th abdominal segment. Cells shown in ( A ) and (€3) are also serotonin immunoreactive (see text). (C) SSB-immunoreactivecells in the subesophageal region. Scale bar = 32 pm in ( A ) and 25 pm in (B,C).

mushroom bodies with our immunocytochemical technique. We are not certain as to the source of this difference, but it may be related to the fixation procedure (see Discussion). Besides the expression in the nervous system and at type-I1 terminals at the neuromuscular junction, we also observed SSB immunoreactivity in the nuclei of testis cells in male larvae. This nuclear staining was also observed in ovaries, but it appeared much dimmer. No other organs in the larvae showed SSB immunoreactivity.

Studies in Mutants Deficiency of dnc. If the antigen localized with the anti-SSB antibody were the dnc gene product, animals deficient for the dnc gene product should be devoid of staining at type-I1 terminals. To investigate this, we examined SSB immunoreactivity in larvae deficient for the dnc gene product as well as in two dnc alleles that completely eliminate CAMP-phosphodiesterase activity. We processed ca. 30 body wall muscles of male larvae progeny

from each of the two deficiency crosses (see Methods) for SSB immunocytochemistry. All samples showed SSB immunoreactivity at type-I1 terminals similar to wild type. Similarly, other alleles of dnc, dncM14,and dncM'I did not suppress staining at the body wall muscles. All these alleles of dnc were reported to eliminate CAMP-phosphodiesteraselike immunoreactivity at the mushroom bodies (Nighorn et al., 199 1 ). This demonstrates that the antigen recognized by the SSB antibody is not the dnc gene product, but an antigen that shares an epitope with it. Deficiency of Ddc. Because the pattern of SSB-immunoreactive cells in the abdominal segments of the ventral ganglion correspond with the pattern of serotonin-containing neurons, we also analyzed mutant larvae deficient in the gene Ddc ( Q f Ddc mutants). Ddc encodes for the enzyme dopa decarboxylase, which is required for the synthesis of both dopamine and serotonin (reviewed in Wright, 1987). Previous studies in D f Ddc mutants demonstrated that serotonin-containing neurons develop in normal number and distribution in these mutants (Vallks and White, 1986). Similarly, we found that D f Ddc larvae showed normal expression of SSB both in neuronal somata and in type-I1 terminals at the body wall muscles. The pattern of serotonin-containing neurons in dncMl4 and dncM" as well as in D f dnc was also normal.

shibiue'". The temperature-sensitive mutant shi"' causes a very specific temperature-dependent block of endocytosis (Poodry and Edgar, 1979; Kosaka and Ikeda, 1983).This results in the blockade of synaptic vesicle reformation, and therefore, when transmitter release is induced at restrictive temperature, there is a depletion of synaptic vesicles (Koenig, Saito, and Ikeda, 1983; Koenig, Kosaka, and Ikeda, 1989; Koenig and Ikeda, 1989). No other defects such as nerve impulse conduction, neurotransmitter release mechanisms, or muscle excitability, have been detected in this mutant (Salkoff and Kelly, 1978). Because SSB was found in regions of the terminal rich in synaptic vesicles we studied the possibility of its release in conjunction with synaptic vesicle content. We incubated third-instar shi larvae at 30°C to produce synaptic vesicle depletion. At this temperature shi larvae become rapidly paralyzed. After 2 h body wall muscles were dissected and processed for SSB immunoreactivity. We found that SSB immunore-

activity was not affected at the light microscopical level by this procedure.

DISCUSSION

In this report we describe the expression of SSB, an antigen restricted to a subset of Drosophila larval motor axon terminals as well as a small number of cell bodies and neuropil in the CNS. The muscle nerve terminals correspond to the previously described type-I1 terminals (Johansen et al., 1989a) which in certain muscles have been shown to be affected by neuronal activity levels (Budnik et al., 1990). The pattern of staining is highly specific to these boutons-no other terminals or structures are stained at the body wall muscles with this antibody. The restricted labelling of the boutons reinforces the idea that boutons are molecularly distinct, not simply random width fluctuations in the neurite. In animals deficient in the dnc gene product, this staining is not abolished. Therefore, we believe that the SSB antigen is not the dnc phosphodiesterase but an antigen that shares an epitope with it. Because this antibody was generated against a phosphodiesterase-conserved region (Nighorn et al., 1991), it is possible that the SSB antigen is as yet another phosphodiesterase or another unrelated antigen that shares an epitope with it. The fact that the SSB antigen is so specifically expressed in a small subset of plastic synaptic terminals and in only a few cells in the CNS, makes this antigen very interesting; however, it probably is not a neurotransmitter because of its ultrastructural localization outside most synaptic vesicles, as well as the fact that terminal staining in shi larvae, heat pulsed to restrictive temperature, is not eliminated. It is interesting to note that the expression of the SSB antigen is remarkably similar to the expression of synapsins at some vertebrate presynaptic terminals (Hirokawa, Sobue, Kanda, Harada, and Yorifuji, 1989; Mandell, Czernik, De Camilli, Greengard, and Townes-Anderson, 1992). Our results are in contrast with those of Nighorn et al. ( 199 1 ) in two respects. They found intense staining at the mushroom bodies of larvae yet failed to observe immunoreactivity at cell bodies in the larval ventral ganglion. We found a number of cell bodies in the ventral ganglion, but no mushroom body staining. We attribute this difference to our different fixation and staining protocols. In fact, we were able to replicate their staining at the

mushroom bodies of adult wild-type flies only if the brains were lightly fixed (undissected adult heads without proboscis were immersed in fixative). If brains were dissected away from the cuticle and air sacs before fixation, then no immunoreactivity was observed at the mushroom bodies. Thorough fixation, therefore, probably destroys the antigenicity of the dnc gene product. whereas it increases SSB immunoreactivity. Another methodological difference that might explain the different results is the fact that the study of Nighorn et al. used cryostat sections, whereas our larval studies were done with whole mounts. It is possible that the dnc gene product is expressed at an intracellular location that is inaccessible to antibodies, and that sectioning is required to increase accessibility. We observed the localization of SSB immunoreactivity at the body wall muscles from embryonic development to early metamorphosis. At all stages, SSB immunoreactivity was restricted to type-I1 terminals at the body wall muscles. The earliest expression of SSB that we could observe was at stage 17, after neuromuscular junctions are formed (Johansen et al., 1989b). Two types of terminals at the Drosophla body wall muscles have been previously described (Johansen et al., 1989a; Budnik et al., 1990) based on their characteristic morphology. Physiological analysis and anatomical studies in embryos and larvae demonstrate that the larval body wall muscles are polyinnervated (Jan and Jan, 1976; Halpern et al., 1991; Sink and Whitington, 1991a); however, our study is the first demonstration that these morphologically different terminals can be immunocytochemically separated and. therefore, that they probably correspond to a distinct motorneuron. Sink and Whitington ( 199 l a ) have demonstrated that body wall muscles in a single hemisegment are innervated by about 30 motorneurons. Because we observe that more than one-third of the muscles contain SSB-positive terminals, it is likely that more than one motorneuron provides type-I1 innervation. Remarkably, the expression of SSB is restricted to terminals that can undergo niorphological plasticity. We have previously demonstrated, for example, that the number and density of type-I1 terminals is increased in mutant animals with increased neural activity (Budnik et al., 1990) or in mutations with increased CAMP levels (Zhong et al., 1992). These terminals are also the same terminals that undergo the most dramatic growth (increase in length of the neurite and increase in number of boutons) during the growth of larval muscles from

hatching to the third larval instar (V. Budnik, unpublished observations). At present we do not know whether the SSB antigen is related to these plastic phenomena; however, the fact that SSB might be related to the dnc gene product, a protein that in Drosophila has been implicated in learning and memory, and in morphological and physiological plasticity at the neuromuscularjunction, makes our observation intriguing. We hope that a genetic approach will allow us in the future to ascertain the exact molecular nature of SSB. This study also indicates that SSB immunoreactivity is expressed in a small subset of neurons in the larval CNS. Surprisingly, the pattern of distribution of these neurons in the ventral ganglion is remarkably similar to the distribution of monoamine-containing neurons (Budnik and White, 1988; VallCs and White, 1988). Serotonin and SSB immunoreactivity was identical in the ventral ganglion. The equivalence of dopamine-containing neurons with the SSB-containing neurons was not investigated because of the lack of compatible immunological or histochemical methods. However, our impression is that at least some ofthe SSB containing neurons are located in similar positions to the dopamine-containing neurons. We believe that these SSB immunoreactive neurons are not type-I1 motorneurons because no amine-containing fiber is seen to exit the ventral ganglion, and because no serotonin immunoreactivity is observed at type-I1 terminals (or at any terminal) in the body wall muscles. T o address this problem conclusively, we note that some method of backfilling type-I1 motor axons in conjunction with serotonin immunocytochemistry would have to be performed. The authors thank Dr. R. Davis for his generous gift of anti-dnc antiscra and for providing us with some of the dnc stocks, as well as Ms. Lucy Yin for technical help, and Drs. E. Connor and R. Murphey for helpful comments on the manuscript. This work was supported by NIH grant NS30072-01 and an Alfred P. Sloan Fellowship to V.B.

REFERENCES ANDERSON,M. S., HALPERN,M. E., and KESHISHIAN, H. ( 1988). Identification ofthe ncuropeptide transmitter proctolin in Duosophilu larvae: characterization of fiber-specific neuromuscular endings. J. Nezirosci. 8:242-255. ANZIL,A. P., BIESER,A., and WERNIG,A. ( 1984). Light and electron microscopic identification of nerve termi-

Selective Labelling of Synaptic Boutons nal sprouting and retraction in normal adult frog muscle. J. Physiol. (Lond.) 350:393-399. BAILEY,C. H . , ~ ~ ~ C HM. E (N1988a). , Long-term memory in Aplysiu modulates the total number of varicosities of single identified sensory neurons. Proc. Nutl. Acud. Sci. USA 85:2373-2377. BAILEY,C. H., and CHEN,M. ( 1988b). Long-term sensitization in A p l ~ s i uincreases the number of presynaptic contacts onto the identified gill motor neuron L7. Proc. Nutl. Acud. Sci. USA 859356-9359. BROWN,M. C., and IRONTON, R. ( 1977). Motoneurone sprouting induced by prolonged tetrodotoxin block of nerve action potential. Nature 265:459-46 1. BUDNIK,V., and WHITE,K. (1988). Catecholaminecontaining neurons in Drosophilu tnelunoguster. distribution and development. J. Comp. Neurol. 268~400-413. BUDNIK,V., Wu, C. -F., and WHITE,K. ( 1989). Altered branching of serotonin-containing neurons in Drosophilu mutants unable to synthesize serotonin and dopamine. J. Nmrosci. 9:2866-2877. BUDNIK,V., ZHONG,Y., and Wu, C. -F. (1990). Morphological plasticity o f motor axon terminals in Drosophila mutant with altered excitability. J. Neurosci. 10:3754-3768. BYERS,D., DAVIS,R. L., and KIGER,J. A. ( 198 1 ). Defect in CAMPphosphodiesterase due to the dunce mutation of learning in Drosophilu melunogustrr. Nature 289~79-81. CROSSLEY, C. A. ( 1978). The morphology and development of the Drosophilu muscular system. In: The Genetics and Biology of Drosophila, Vol. 2b. M. Ashburner and T. R. F. Wright, Eds. Academic Press, New York, pp. 499-560. DAVIS,R. L., and DAVIDSON, N. ( 1984). Isolation o f the Drosophilu mclunoguster dunce chromosomal region and recombinational mapping of dunce sequences with restriction site polymorphisms as genetic markers. Mol. Cell Biol. 4:358-367. DELGADO,R., HIDALGO,P., DIAZ,F., LATORRE.R., and LABARCA,P. ( 1991). A cyclic AMP-activated K + channel in Drosophilu larval muscle is persistently activated in dunce. PNAS 88:557-560. GANETZKY,B., and Wu, C. -F. (1983). Neurogenetic analysis of potassium currents in Drosophilu; synergistic effects on neuromuscular transmission in double mutants. J . Neurogenet. 1: 17-28. GREENOUGH, W. T., and BAILEY,C. ( 1988). The anatomy of a memory: convergence of results across a diversity of tests. T I N S 11:142-147. HALPERN,M. E., ANDERSON,M. S., JOHANSEN, J., and KESHISHIAN, H. ( 1988). Octopamine-immunoreactive nerve terminals are found on a single identified muscle fiber o f the Drosophilu larval body wall. Soc. Neurosci. A hstr. 14:38 3. HALPERN,M. E., CHIBA,A., JOHANSEN, J., and KESHISHIAN, K. ( 199 1 ). Growth cone behavior underlying

1065

the development of stereotypic synaptic connections in Drosophilu embryos. J. Neurosci. 11:3227-3228. HARRIS,W. A. (1980). The effects of eliminating impulse activity on the development of the retinotectal projection in salamander. J. Comp. Neurol. 194:303317. HERRERA,A. A. and WERLE,M. (1989). Mechanisms of elimination, remodeling and competition at frog neuromuscular junctions. J . Neurobiol. 21:73-98. HIROKAWA, N., SOBUE,K., KANDA,K., HARADA,A., and YORIFUJI,H. (1989). The cytoskeletal architecture of the presynaptic terminal and molecular structure of synapsin 1. .I. Cell Biol. 108:111-126. HUBEL,D. H., WIESEL,T. N., and LEVAY,S. (1977). Plasticity of ocular dominance columns in monkey striate cortex. Phil. Truns. R.Soc. Lond. 278:377-409. JAN, L. Y., and JAN,Y. N. (1976). L-Glutamate as an excitatory transmitter at the Drosophilu larval neuromuscular junction. J. Physiol. (Lond.) 262:2 15-236. JAN,L. Y., and JAN,Y. N. ( 1982). Antibodies to horseradish peroxidase as specific neuronal markers in Drosophilu and grasshopper embryos. Proc. Null. Acud. Sci. USA 72:2700-2704. JOHANSEN, J., HALPERN,M. E., JOHANSEN, K. M., and KESHISHIAN, H. ( 1989a). Stereotypic morphology of glutamatergic synapses on identified muscle cells of Drosophilu larvae. J. Neurosci. 9:7 10-725. JOHANSEN, J., HALPERN,M. E., and KESHISHIAN,H. ( 1989b). Axonal guidance and the development of muscle fiber specific innervation in Dro.sophilu embryos. J. Neurosci. 9:43 18-4332. KIGER, J. A,, and SALZ,H. K. ( 1985). Cyclic nucleotide metabolism and physiology of the fruit fly Drosophilu melunoguster. Adv. Insect Phjniol. 18: 14 I - 179. KOENIG,J. H., SAITO,K., and IKEDA,K. ( 1983). Reversible control of synaptic transmission in a single gen mutant of Drosophilu me1unogu.ster. J . Cell Biol. 96: 15 17-1 522. KOENIG,J. H., KOSAKA,T., and IKEDA, K. ( 1989). The relationship between the number o f synaptic vesicles and the amount of transmitter released ( 1989). J. Neurosci. 9: 1937- 1942. KOENIG,J. H. and IKEDA,K. (1989). Disappearance and reformation of synaptic vesicle membrane upon transmitter release observed under reversible blockade o f membrane retrieval. J. Neurosci. 9:3844-3860. KOSAKA,T., and IKEDA,K. ( 1983). Possible temperature-dependent blockage of synaptic vesicle recycling induced by a single gene mutation in Drosophilu. J. Neurobiol. 14:207-225. LINDSLEY,D. L., and ZIMM,G. G. ( 1992). The Genome of Drosophila melanogaster. Harcourt Brace Jovanovich, Publishers, Academic Press, Inc., San Diego, CA. LNENICKA, G. A., HONG, S. J., COMBATTI, M., and LEPAGE,S. ( 199 1 ). Activity-dependent development of synaptic varicosities at crayfish motor terminals. J. Neurosci. 11:1040-1048. LNENICKA,G. A., ATWOOD,H. L., and MARIN, L.

1066

Biidnik and Gorczyca

( 1986). Morphological transformation of synaptic terminals of a phasic motorneuron by long-term tonic stimulation. J . Neurosci. 6:2252-2258. LIPOSITS,Z., SHERMAN,D., PHELIX, C., and PAUL, W. K. (1986). A combined light and electron microscopic immunocytochemical method for the simultaneous localization of multiple tissue antigens. Histochemistry 85:95-106. MANDELL,J. W., CZERNIK,A. J., DE CAMILLI,P., GREENGARD,and TOWNES-ANDERSON, E. ( 1992). Differential expression of synapsins I and I1 among rat retinal synapses. .I. Neurosci. 12:1736- 1749. MEYER,R. L. (1982). Tetrodotoxin blocks the formation of ocular dominance columns in goldfish. Science 218589-59 1. MURPHEY,R. K., and LEMERE,C. P. ( 1984). Competition controls the growth of an identified axonal arborization. Science 2241 352-1355. NIGHORN,A., HEALY,M. J., and DAVIS,R. ( 199 1 ). The cyclic AMP phosphodiesterase encoded by the diince gene is concentrated in the mushroom body neuropil. Neuron 6:455-467. POODRY,C. A,, and EDGAR,L. ( 1979). Reversible alterations in the neuromuscular junctions of Drosophilu melunoguster bearing a temperature sensitive mutation shihire. J. Cell. Biol. 81520--527. SALKOFF,L., and KELLY,L. ( 1978). Temperature induced seizure and frequency dependent neuromuscular block in a ts mutant of Drosophilu. NuttirLj 273: 156- 1 58. SANES,D. H., and CONSTANTINE-PATON, M. (1983). Altered activity patterns during development reduce neural tuning. Science 221:l 183-1 185. SCHMIDT,J. T., and TIEMAN, S. B. (1989). Activity, growth cones and the selectivity of visual connections. Comments Dev. Neurohiol. 1: I 1-28. SINK, H., and WHITINGTON,P. M. ( 1991a). Location

and connectivity of abdominal motorneurons in the embryo and larva of Drosophilu rnrlunoguster. J. Neurobiol. 22:298-3 1 1. SINK,H., and WHITINGTON, P. M. ( 199 1 b ) . Pathfinding in the central nervous system and periphery by identified embryonic Drosophilu motor axons. Development 112:307-3 16. VALL~S, A. M., and WHITE, K. (1986). Development of serotonin-containing neurons in Drosophilu mutants unable to synthesize serotonin. .J. Ncwrosci. 6:1482-149 I . VALL~S, A. M., and WHITE,K. (1988). Serotonin-containing neurons in Drosophilu mdunogarler. development and distribution. J. c'omp. Nciirol. 268:4 I4428. WERNIG,A., and FISHER,M. ( 1986). The nerve-muscle junction: a remodeling contact. E.xp. Bruin Rex series 14. In: Calcium Electrogenesis and Neuronal Functioning. U. Heidelberg, Ed. Heidelberg. Springer-Verlag, Berlin, pp. 245-255. WRIGHT,T. R. F. (1987). The genetics of biogenic amine metabolism, sclerotization and melanization in Drosophilu tnelunoguster. Adv. Genef. 24: I 27-222. Wu, C. -F., and GANETZKY, B. ( 1980). Genetic alteration of nerve membrane excitability in temperaturesensitive para1yt i c mu tan t s of Drosoph ilu indunoguster. Nutiire 2862314-816. Wu, C. -F., and MAUGLAND. F. N. (1985). Voltage clamp analysis of membrane currents in larval muscle fibers of Drosophilu: alterations of potassium currents in Shuker mutants. J . Nciirosci. 52626-2640. ZHONG,Y., and Wu, C. -F. ( 1991 ). Altered synaptic plasticity in Drosophilu memory mutants with a defective cAMP cascade. Science 251: 198-201. ZHONG,Y., BUDNIK,V., and Wu, C. -F. ( 1992). Synaptic plasticity in Dro.cophilu memory and hyperexcitable mutants: role of cAMP cascade. J . Nciirosci. 12:644-65 I .