GABA and Glutamate-LikeImmunoreactivity at Synapses on
Depressor Motorneurones of the Leg ofthe Crayfish, Procambarus clarkii
ALAN H.D. WATSON,1* MICHELLE BEVENGUT,2 EDOUARD PEARLSTEIN,2
AND DANIEL CATTAERT,2
1School of Biosciences, University of Wales Cardiff, Cardiff, CF10 3US, United Kingdom2Laboratoire de Neurobiologie et Mouvements, UPR 9011, CNRS, 13402 Marseille, France
ABSTRACTTo investigate their synaptic relationships, depressor motorneurones of the crayfish leg
In the crustacea, motorneurones supplying muscles ofthe locomotory or stomatogastric systems make extensivecentral connections and may play a significant role incentral pattern-generating circuitry (Heitler, 1978, 1981;Harris-Warrick et al., 1992). Crustacean excitatory motor-neurones in the same motor pool are often extensivelylinked by electrotonic synapses, whereas those in antago-nistic motor pools may be linked by graded inhibitorychemical synapses (Graubard et al., 1980; Marder andEisen, 1984; Chrachri and Clarac, 1989). The motorneu-rones in the crayfish locomotory system are glutamatergic,but although release of glutamate at the neuromuscularjunction causes depolarisation, its release centrally, forexample onto antagonistic motorneurones, activates apicrotoxin-sensitive chloride conductance and produces aninhibitory response (Pearlstein et al., 1994, 1998). Thisappears to be the same receptor channel complex acti-vated by g-aminobutyric acid (GABA) (Franke et al.,1986a,b; Pearlstein et al., 1994). The long latency of theconnections between antagonistic motorneurones hasstimulated some debate as to whether they are mono- orpolysynaptic (Sherff and Mulloney, 1996; Pearlstein et al.,1998); however, small numbers of close appositions be-tween third- and fourth-order neurites of antagonistic mo-
torneurones that might constitute synaptic sites havebeen observed using confocal microscopy (Pearlstein et al.,1998).
GABA is also a major inhibitory transmitter in thecrayfish nerve cord. Immunocytochemical studies showthat it is present in many neurones within each ganglion(Mulloney and Hall, 1990), and its effects on motorneu-rones are well documented (Sherff and Mulloney, 1996;Pearlstein et al., 1994, 1998). One hypothesised source ofGABAergic input onto excitatory motorneurones is frominhibitory motorneurones, although there is as yet littledirect evidence for this (Sherff and Mulloney, 1996).
Although some models for locomotor pattern generationhave concentrated almost entirely on the motorneurones
Grant sponsor: Centre Nationale de la Recherche Scientifique; Grantsponsor: European Science Foundation; Grant number: 177.
Daniel Cattaert’s current address is: UMR 5816, Neurobiologie des Re-seaux, CNRS, Universite Bordeaux I, Bat 2 - Biologie Animale, Avenue desFacultes, 33405, Talence cedex, France.
*Correspondence to: Alan H.D. Watson, School of Biosciences, Biomedi-cal Building, University of Wales Cardiff, Museum Avenue, PO Box 911,Cardiff, CF10 3US, United Kingdom. E-mail: [email protected]
Received 8 July 1999; Revised 8 March 2000; Accepted 8 March 2000
THE JOURNAL OF COMPARATIVE NEUROLOGY 422:510–520 (2000)
alone (Sherff and Mulloney, 1996), several populations ofspiking and nonspiking local interneurones have beenidentified in the thoracic ganglia that are likely to play animportant role in such circuits through their orchestrationand modulation of interactions between the motorneu-rones innervating the muscles of one or more leg joints(Heitler and Pearson, 1980; Paul and Mulloney, 1985a,b;Chrachri and Clarac, 1989). Indeed, it is probable that therhythm-resetting properties described for some swim-meret motorneurones are mediated via electrotonic syn-apses made with local interneurones (Heitler, 1978; Pauland Mulloney, 1985a; Sherff and Mulloney, 1996).
In the present study we have used ultrastructural im-munocytochemistry to investigate the nature of the inputand output synapses made by one class of motorneuroneinnervating the crayfish leg. As GABA and glutamateclearly play such an important role in the control of mo-torneurone activity, we have used antibodies againstthese transmitters to analyse their patterns of synapticinput. Taken together, synapses immunoreactive for thesetransmitters accounted for over 80% of the input receivedby the motorneurones. The significance of this is discussedin the context of the known connectivity of the motorneu-rones.
MATERIALS AND METHODS
Both male and female crayfish (Procambarus clarkii),about 10 cm long from rostrum to telson, were obtainedfrom a commercial supplier (Chateau Garreau, Landes,France), maintained in aerated fresh water at room tem-perature (18–21°C), and fed once a week. An in vitropreparation of the thoracic ventral nerve cord (Sillar andSkorupski, 1986; Chrachri and Clarac, 1989), consisting ofthe three last thoracic ganglia together with the motornerves from the fifth thoracic ganglion to the proximal legmuscles, was used. The thoracic nerve cord was pinneddorsal side up to a Sylgard-lined Petri dish, and the fourthand fifth thoracic ganglia were desheathed to improveperfusion of central neurones and to allow intracellularrecordings from motorneurons. The composition of thebathing solution, which was maintained at 9°C using aPeltier cooler, was (in m mol l21) 195 NaCl, 5.5 KCl, 13.5CaCl2, 2.5 MgCl2, and 10 Tris at pH 7.6.
For intracellular staining, depressor motorneuroneswere impaled with electrodes containing 5% horseradishperoxidase (HRP) in 0.2 mol l21 Tris buffer containing 0.5mol l21 KCl and identified physiologically. The HRP wasinjected with 500-ms pulses of 5 nA of depolarising cur-
Fig. 1. A: Camera lucida drawing of an intracellularly labelleddepressor motorneurone from the fifth thoracic ganglion. The dottedoutline indicates the extent of the neuropile. B: Drawing of a sectionof the ventral nerve cord showing the position of the neuropile areasand the nerve roots arising from the ganglion. These are labelled
according to the muscle groups they supply. PRO, promotor; A LEV,anterior levator; P LEV, posterior levator; REM, remotor; DEP, de-pressor; ADR, anterior distal root; PDR, posterior distal root; T3–5,third to fifth thoracic ganglia; A1, first abdominal ganglion. Scalebars 5 100 mm in A; 500 mm in B.
511SYNAPSES ON CRAYFISH MOTORNEURONES
rent at 1 Hz for 1–1.5 hours. The ganglia were then fixedin 2.5% glutaraldehyde in 0.05 mol l21 phosphate buffer,pH 7.4, containing 0.15 mol l21 sucrose for 1–2 hours.After washing in phosphate buffer, the ganglia were incu-bated for 1 hour in the same buffer containing diamino-benzidine in which hydrogen peroxide was generated bythe action of glucose oxidase on b-d glucose (Watson andBurrows 1981). The ganglia were then sliced into 2–3pieces and treated with 1% osmium tetroxide in distilledwater for 30 minutes and then 2% aqueous uranyl acetatefor a further 20 minutes before rapid dehydration andembedding in Durcupan ACM (Fluka).
Immunocytochemistry
Four depressor motorneurones were examined usingimmunocytochemical methods, of which three were sam-pled extensively. For labelling with a single antibody,ultrathin sections were mounted on PB100C barred gridscoated with Pioloform (Agar Scientific, Stansted, UK).Sections were etched at room temperature using 2% peri-odic acid (5 minutes) and sodium metaperiodate saturatedat 50°C (5 minutes) and then washed in Tris buffer, pH7.2. After 30 minutes in 5% normal goat serum in thesame Tris buffer, the grids were transferred for 2 hours torabbit anti-GABA at 1:2000 dilution or rabbit anti-glutamate antiserum at 1:7000 (both from Sigma, Poole,UK) in Tris buffer. After further washing they were trans-ferred for 1 hour to 15-nm gold-labelled goat anti-rabbitantibody (British Biocell International, Cardiff, UK) di-luted 1:30 in Tris buffer, pH 8.2 for 1 hour. The grids werefinally stained in uranyl acetate and lead citrate beforeexamination in the electron microscope.
For labelling with the two antibodies, ultrathin sectionswere mounted on uncoated 200-mesh hexagonal grids thathad been covered with poly-L-lysine. The sections werethen etched on one surface in the usual way, and immu-nostaining for glutamate was carried out as above using a15-nm gold-conjugated goat anti-rabbit antibody. Thegrids were then coated with Pioloform on the labelled face.The opposite face was then etched, and immunostainingfor GABA was carried out using a 10-nm gold-conjugatedgoat anti-rabbit antibody. The sections were then driedand stained with uranyl acetate and lead citrate.
The levels of gold particle labelling over immunoreac-tive terminals were compared quantitatively with levels oflabelling over nonimmunoreactive terminals for bothGABA and glutamate. As described previously (Pearlsteinet al,. 1998), the ratio of specific antibody labelling tobackground was determined using image analysis. ForGABA, the average particle density over immunolabelledprocesses was 64.4 particles/mm2 compared with 4.9particles/mm2 for unlabelled terminals, corresponding toan enrichment factor of 13.2. For glutamate, the density ofparticles for labelled terminals was 49.7 particles/mm2
compared with 7.5 particles/mm2 for unlabelled terminals,corresponding to an enrichment factor of 6.6. The presenceof low levels of glutamate-like immunoreactivity in pro-cesses showing strong immunoreactivity for GABA isprobably a reflection of the role of glutamate as the imme-diate precursor for GABA.
To determine the dimensions of synaptic vesicles, elec-tron micrograph negatives taken at a magnification of316,000 were scanned at 600 dpi using an Epson GT-7000Photo scanner. The measurements were made using Im-agePro Plus v4.0 software (Media Cybernetics) after a
further magnification on screen to 3128,000. In each of sixterminals immunoreactive for GABA, or glutamate, orneither antibody, the maximum and minimum diametersof 30 agranular vesicles were determined. For each classof terminal, the mean of the two diameters was expressedas a ratio to provide an index of the degree of flattening.The dimensions of the different populations were com-pared with an unpaired t-test using the Microsoft Excelstatistics package.
One neurone was selected for a detailed quantitativesurvey of immunolabelling in the terminals presynaptic toit. From this neuron, a total of 122 input synapses wereassessed, 55 from sections immunostained for GABA and67 from sections immunostained for glutamate. Figures2–4 come from this neurone, in which no unequivocalexamples of output synapses were seen. The micrographsin Figure 5 are taken from another neurone, which madenumerous output synapses.
RESULTS
The major movements of the leg of the crayfish duringwalking are brought about by the action of muscles mov-ing the proximal leg joints between the thorax, coxopodite,and basipodite. The depressor muscle acts on thecoxopodite/basipodite joint and is supplied by excitatorymotorneurones with somata in the posterior cortex of theganglion (Fig. 1). The neurones studied were from the fifththoracic ganglion.
The depressor motorneurones received synapses both ontheir fine processes (e.g., Figs. 2A, 3, 4) and on majorneurites up to 2–4 mm in diameter, either directly or ontoshort spines arising from them (Fig. 2B). Most of thesesynapses were dyadic, with the labelled motorneuronecontributing one of the two postsynaptic processes. Thesecond postsynaptic neurite in the dyad was not generallyimmunoreactive for glutamate, suggesting that it did notusually belong to another motorneurone. Approximatelyone-third of inputs (31%, n 5 55 synapses) were immuno-reactive for GABA (Figs. 2A,C), and one-half, (51%, n 5 67synapses) were immunoreactive for glutamate (Figs. 2B,4B,D). The neuropile also contained many processes thatwere labelled by neither antibody (e.g., Figs. 2D, 3). Ob-servations made from adjacent pairs of sections labelledseparately with the two antibodies or single sections la-belled with both antibodies demonstrated that a signifi-cant proportion of input synapses received by the coxal
Fig. 2. A: A small-diameter motorneurone neurite receives input(arrowhead) from a GABA-immunoreactive process. B: Two shortspines from a large-diameter neurite receive synaptic contacts (arrow-heads) from a glutamate-immunoreactive process. C: A section immu-nolabelled for GABA (10 nm gold) and glutamate (15 nm gold) clearlydistinguishes, on the basis of their immunoreactivity, between twoneuropilar process that lie close to a motorneurone neurite. TheGABA-immunoreactive process makes a synapse (arrowhead) ontothe motorneurone. The inset shows the gold labelling of the GABA-immunoreactive process at higher magnification. Most of the goldparticles in the process are 10 nm in diameter. A single 15-nm particleis indicated by the arrow. D: A section labelled with GABA andglutamate antibodies demonstrates three categories of neuropilar pro-cess in the vicinity of a motorneurone neurite. One (GABA) is labelledmainly with 10-nm gold particles, one (glut) only with 15-nm parti-cles, and a third (asterisk) is not immunoreactive, showing only back-ground levels of gold particles. Scale bars 5 0.25 mm; 0.1 mm in inset.
512 A.H.D. WATSON ET AL.
Figure 2
Fig. 3. A,B: Two nearby sections are labelled with antibodiesagainst GABA (A) and glutamate (B). Five processes (a–d) are markedto aid comparison. The dashed line indicates the boundary between band c where the membranes are sectioned obliquely. Process a isimmunoreactive for glutamate, b and d for GABA, and c for neither.Process c makes two inputs (arrowheads) onto a motorneurone neu-rite. C–E: Higher magnification images of processes immunoreactive
for GABA (C), glutamate (E), or neither of these antibodies (D) toallow comparison of their synaptic vesicles. F,G: Adjacent sections ofa synapse (arrowhead) onto a fine motorneurone neurite from a pro-cess that contains many large granular vesicles as well as smallagranular ones and is immunoreactive for neither GABA (F) norglutamate (G). Scale bars 5 0.25 mm in B (applies to A,B), F (appliesto F,G); 0.1 mm in C–E.
514 A.H.D. WATSON ET AL.
depressor neurones were made by processes of this type(Figs. 3A,B,F,G, 4A).
The three immunocytochemically distinct classes of ter-minal presynaptic to the motor neurone contained mainlyagranular vesicles. The vesicle population in each of thesewas statistically significantly different from the other two(P . 0.01). Agranular vesicles in processes immunoreac-tive for glutamate were nearly spherical (Fig. 3E), with amean maximum diameter (6SEM) of 44.8 6 0.47 nm andmean minimum diameter of 39.5 6 0.37 nm (n 5 180), aratio of 1.13. Small numbers of round granular vesicleswith diameters of 75–85 nm were also sometimes seen.Agranular vesicles in GABA-immunoreactive processeswere more flattened (Fig. 3C) than those in the glutamate-immunoreactive processes, with a mean maximum diam-eter of 34.9 6 0.55 nm and mean minimum diameter of23.4 6 0.43 (n 5 180), giving a ratio of 1.49.
Some of these processes also contained a few granularvesicles. In many cases these had similar dimensions tothe ones in glutamatergic processes, but occasionally theywere larger (100–110 nm). The agranular vesicles in theprocesses labelled by neither antibody (Fig. 3D) had amean maximum diameter of 33.0 6 0.49 nm and meanminimum diameter of 21.6 6 0.37 nm (n 5 180), giving aratio of 1.52. Often, a scattering of 75–85-nm-diametergranular vesicles were seen in the nonimmunoreactiveprocesses, but occasionally many, much larger granularvesicles were present (Fig. 3F,G). These were 100–125 nmin diameter with a dark core, which had a paracrystallineappearance. This suggests that the nonimmunoreactivepresynaptic processes may represent more than one cate-gory of neurone, which may use different transmitters ortransmitter combinations. Even though, within the neu-ropile, there were several distinct populations of processesthat contained mainly granular vesicles of various dimen-sions, these were not seen to make synaptic contact withthe motorneurones.
It is clear that different categories of input synapse areintermingled on the neurites of coxal motorneurones. Fig-ure 4B shows that synapses immunoreactive for gluta-mate and those that are not can lie within 1 mm of eachother. Serial synapses involving different classes of pre-synaptic process are also seen. In Figure 4C, a GABA-immunoreactive process synapses on a process that is notimmunoreactive for GABA, which contacts a motorneu-rone. In Figure 4D, a similar situation is seen, but herethe process presynaptic to the motorneurone that is con-tacted by the GABA-immunoreactive one is seen to belabelled by the glutamate antibody.
In physiological studies of depressor motorneurones,their central output connections have been demonstrated(Pearlstein et al,. 1998). These were seen in abundance insome preparations of labelled neurones (Fig. 5A,B). Clus-ters of synaptic vesicles surrounded single or multipleoutput synapses, often from quite large diameter neurites(1.5–3.5 mm). These were frequently made onto small neu-ropilar processes that rarely appeared immunoreactive forGABA or glutamate; however, with the methods used itcan be difficult to determine unequivocally whether suchsmall processes are immunoreactive. Intensive analysis ofother labelled depressor motorneurones revealed no out-put synapses, however, suggesting that neurones within asingle motor pool may differ in their central connections.
DISCUSSION
Vesicle morphology and transmitter content
The four classes of neuronal terminal that we observedin presynaptic contact with the motorneurones were con-sistent with classes already described in crustacea on thebasis of vesicle morphology and hypothesised transmittercontent.
Early studies of the neuropile of the stomatogastricganglion of the spiny lobster Panulirus interruptus iden-tified three classes (types A–C) of agranular vesicle-containing terminals (King, 1976a,b). The vesicles in typeA were predominantly pleomorphic, whereas those in typeB were round and rather larger. In type C, small pleomor-phic agranular vesicles were accompanied by numerouslarge granular vesicles. Neurones thought to be cholin-ergic gave rise to type A terminals and glutamatergicneurones to type B. The glutamate-immunoreactive pro-cesses we observed were similar to type B terminals invesicle content.
Two further categories of terminal were identified inmore recent studies of crab stomatogastric ganglion: typeD, in which dense-core and agranular vesicles were inter-mingled rather than separate as in type C, and type E,which contained larger (round) agranular vesicles thanthose in type B (Kilman and Marder, 1996). Types C andD were stained by GABA antibodies, but it was suggestedthat type A included both GABAergic and cholinergic ter-minals (Kilman and Marder, 1996). Although the agranu-lar vesicles we observed in the GABA-immunoreactiveand the nonimmunoreactive terminals were similar insize and shape, they were nonetheless statistically distin-guishable. Given that both populations of processes maycontain more than one physiological class whose vesiclescould exhibit subtle differences in morphology, this is notnecessarily inconsistent with the conclusions of Kilmanand Marder (1996). We could not, however, confirm theirsuggestion that GABA immunoreactivity was sometimesspecifically associated with vesicles, although this wascertainly true for glutamate. The pattern of immunolabel-ling we observed in Procambarus is consistent with obser-vations from Pacifasticus (Leitch et al., 1990; Falconer etal., in preparation), with the interpretation of the classicalstudies of Uchizono (1967) and Atwood et al. (1982), andwith immunocytochemical studies of the insect nervoussystem (Watson, 1988; Watson and Pfluger, 1994).
One model for the pattern generator for the swimmeretmotor program (Sherff and Mulloney 1996) proposes thatinhibitory motorneurones, which are immunoreactivefor GABA (Bevengut and Cournil, 1990), make directsynapses onto homonymous excitatory motorneurones.These could be candidates for some of the GABA-immunoreactive synapses seen in the present study. Theonly GABAergic arthropod inhibitory motorneurone so farstudied ultrastructurally appears to make no central out-put synapses (Watson et al., 1985); however, given thedifferences between the properties of excitatory motorneu-rones in the crustacea and insects, this may not be areliable indication of synaptic relationships of crayfishinhibitory motorneurones. The abundance of GABA-immunoreactive synapses suggests that even if some doarise from inhibitory motorneurones, many other neu-rones may contribute to this input.
Agranular vesicle-containing processes unlabelled by ei-ther GABA or glutamate antibodies appear to correspond
515SYNAPSES ON CRAYFISH MOTORNEURONES
Figure 4
to King’s (1976a,b) type A and so may contain acetylcho-line. This is a neurotransmitter of some stomatogastricmotorneurones (Marder and Eisen, 1984); however, cho-linergic motorneurones have not been described in crus-tacean thoracic ganglia. Acetylcholine is present in crus-tacean sensory neurones (Florey, 1973) and has beendemonstrated pharmacologically at a synapse between ab-dominal mechanoreceptors and first-order interneuronesin the crayfish lateral giant neurone escape circuit (Milleret al., 1992). In the thoracic nervous system, several pop-ulations of motorneurones supplying leg muscles, includ-ing the depressors, receive monosynaptic inputs from sen-sory neurones of the coxo-basipodite chordotonal organ(Le Ray et al., 1997a,b). Cholinergic agonists can inducerhythmic activity in the crayfish locomotor network (Chra-chri and Clarac, 1987, 1990; Cattaert et al., 1994), andthis may mimic the activity of descending cholinergic in-terneurones, which may make at least some of their con-nections directly onto motorneurones (Braun and Mul-loney, 1993).
Although small agranular vesicles were the dominantcategory in the processes presynaptic to the motorneu-rones examined, small numbers of large granular vesicleswere also present. Co-localization of putative transmitters
is well known in the crustacean nervous system. Acetyl-choline may co-localise with serotonin in sensory neurones(Katz and Harris-Warrick, 1989; Katz et al., 1989) such asgastropyloric receptors in the crab, which have a rapidnicotinic excitatory effect on the stomatogastric pattern-generating network as well as a more prolonged modula-tory effect that is mimicked by serotonin. Peptides such asproctolin may be colocalized with dopamine, serotonin, oracetylcholine in central interneurones (Siwicki andBishop, 1986; Siwicki et al., 1987) and may be present insome motorneurones that are thought to be glutamatergic(Siwicki et al., 1987). Proctolin is also found in someGABAergic interneurones in the stomatogastric systemthat excite the gastric mill rhythm (Coleman and Nus-baum, 1994) or elicit a particular pyloric rhythm (Blitz etal., 1999)
Variability in the abundance of outputsynapses among the depressor
motorneurone population
Many groups of homonymous motorneurones in thecrayfish are connected to antagonistic motor pools viadirect, nonspiking inhibitory glutamatergic synapses(Chrachri and Clarac, 1989; Pearlstein et al., 1994, 1998).In a previous paper we reported that we were unable tofind output synapses on depressor motorneurones despiteunequivocal physiological evidence for their existence(Pearlstein et al., 1998). However, from the results of thefuller investigations presented here, it is clear that indi-vidual members of the depressor motorneurone pool differin the presence or abundance of central output synapses.Intensive sampling of some neurones revealed no evidenceof outputs, whereas in others these were numerous onboth small and large neurites. Contacts between antago-nistic motorneurones do not involve large-diameter neu-rites (Pearlstein et al., 1998) and this, together with theabundance of the output synapses made by certain depres-sor motorneurones, suggests that many are made onto
Fig. 5. A: A 3-mm-diameter neurite from a depressor motorneu-rone makes an output synapse (arrowhead) onto two neuropilar pro-cesses. The synaptic site is indicated by a presynaptic bar. B: A seconddyadic output synapse from a depressor motorneurone is here seen at
higher magnification, which clearly reveals the structure of the pre-synaptic bar (arrowhead) and the increase in the density of thepostsynaptic membrane (arrows) in each of the two postsynaptic pro-cesses. Scale bars 5 0.25 mm.
Fig. 4. A: A section labelled with antibodies against GABA (10 nmgold) and glutamate (15 nm gold) shows two input synapses (arrow-heads) from a process labelled with neither antibody (asterisk). B: Inthis section immunostained only for glutamate, synapses (arrow-heads) from immunoreactive and nonimmunoreactive processes arefound close together on the motorneurone. C: In this section labelledonly for GABA, an immunoreactive process synapses (arrowhead)onto an unlabelled one that in turn makes two synapses with amotorneurone neurite. D: Serial synapses (arrowheads) in a sectionlabelled with GABA (10 nm gold) and glutamate (15 nm gold) anti-bodies. A GABA-labelled process synapses onto a glutamate-labelledone that in turn contacts a motorneurone neurite. Scale bars 5 0.25mm.
517SYNAPSES ON CRAYFISH MOTORNEURONES
other classes of neurone. This supports the suggestionthat motorneurones are more than just passive effectors ofthe motor programme.
The neurites postsynaptic to the depressor motorneu-rones were generally of small diameter, did not containsynaptic vesicles, and rarely appeared to be immunoreac-tive for glutamate, the putative transmitter of the motor-neurones. Glutamate immunoreactivity, however, is en-riched above background levels only over the synapticvesicle pool and mitochondria, so this does not precludethe possibility that some postsynaptic processes do belongto motorneurones.
The possibility that the properties and synaptic connec-tions of individual members of a single motorneurone poolmay differ has received considerable support from physi-ological studies. Only 60% of the depressor motorneuronepool studied here could be demonstrated electrophysiologi-cally to make output connections onto antagonistic motor-neurones (Pearlstein et al., 1998; Chrachri and Clarac,1989). In the swimmeret motor system, less than half ofantagonistic motorneurone pairs tested were connected(Sherff and Mulloney, 1996). Although the appropriatepartners might simply not have been found, the highsuccess rate in identifying connections between most or allmembers of other antagonistic motorneurone pools sug-gests that difference in connectivity is at least a plausibleexplanation. The fact that injection of current into somepowerstroke motorneurones in the swimmeret system canreset the locomotory rhythm, whereas current injectioninto others does not (Heitler, 1978), reinforces this idea.Subpopulations of promotor and remotor motorneuronesof the crayfish leg also differ in the input they receive frommuscle receptors and chordotonal organs (Skorupski etal., 1992).
Similar arguments have been invoked to explain thelack of output synapses seen in ultrastructural studies ofcrayfish uropodite closer motorneurones (Kondoh et al.,1987) despite physiological evidence for their existence(Nagayama et al., 1983). In the locust, ultrastructuralstudies have also revealed differences in synaptic interac-tions between motorneurones innervating the same mus-cle. The extensor muscle of the tibia receives input fromtwo excitatory motorneurones, one of which makes manycentral output synapses and the other apparently none(Watson and Burrows, 1982; Burrows et al., 1989). Thismay not be a strict analogy to the situation in the crayfishleg motorneurone pools, however, as the two locust motor-neurones are active in different behaviours.
Despite the important role of electrotonic synapses inconnecting motorneurones of the same motor pool (Chra-chri and Clarac, 1989), and in linking motorneurones withsome interneurones (e.g., Paul and Mulloney, 1985a),none was seen in the present study. These may besparsely distributed and hard to find, as even in the smallneuropile of the much studied stomatogastric ganglion, itis only recently that ultrastructural evidence for them hasbeen obtained (Kilman and Marder, 1996).
Abundance of GABA- and glutamate-immunoreactive synapses
From quantitative studies of inputs onto the centralbranches of several types of insect neurone, it is clear thatthe distribution of different classes of synapse may varyover a neurone’s arborisation. For example, interneuronebranches that are predominantly input sites may receive
quite different proportions of GABA-immunoreactive syn-apses than branches, which are predominantly sites ofoutput (Leitch and Laurent, 1993; Watson and Hardt,1996). If we compare input branches of several differenttypes of neurone, however, a relatively consistent patternemerges. In one class of locust local spiking sensory inter-neurones, these branches receive 43% of their input fromGABA-immunoreactive processes (Leitch and Laurent,1993), while for the input branches of dorsal unpairedmedian neurones (that modulate muscle activity), the fig-ure is 39% (Pfluger and Watson, 1995). In the cricketprothoracic ganglion, ascending auditory interneuronesAN1 and AN2 receive, respectively. 27% and 19% of theirinput from GABA-immunoreactive processes (Hardt andWatson, 1994), and local auditory interneurones ON1 andON2 receive 29% and 31%, respectively (Watson andHardt, 1996). The figure of 31% for the crayfish motorneu-rones observed here is very similar even though the otherneurones are quite different in function.
There is little quantitative information available forglutamate-immunoreactive synapses on arthropod neu-rones. Locust dorsal unpaired median neurones receive21% of their input from such synapses (Pfluger andWatson, 1995) compared with 51% for the crayfish depres-sor motorneurones. The picture is complicated by the factthat glutamate may have more than one role in the ar-thropod central nervous system. At the neuromuscularjunction, it is an excitatory neurotransmitter in insectsand crustacea (Takeuchi and Takeuchi, 1965; Usherwoodet al., 1968) and at an insect central synapse between twomotorneurones (Emson et al., 1974; Sombati and Hoyle,1984; Watson, 1988; Bicker et al., 1988) it is also excita-tory (Hoyle and Burrows, 1973; Heitler and Burrows,1977; Burrows et al., 1989). The central connections be-tween antagonistic crayfish motorneurones are inhibitoryand are mediated by a glutamate-evoked increase in chlo-ride conductance (Pearlstein et al., 1994, 1998). In theinsect nervous system, bath application of glutamate canalso evoke hyperpolarising responses (Usherwood et al.,1980; Dubas, 1990, 1991), although neither the ionicmechanism, nor the role this may play in neuronal cir-cuitry is known. Thus the glutamate-immunoreactive syn-apses seen on the crayfish motorneurone dendrites couldconceivably represent a mixture of excitatory and inhibi-tory connections. The distribution and identity of gluta-matergic neurones within the crayfish nervous system islargely unexplored, but in the locust these include inter-segmental interneurones as well as motorneurones(Watson and Seymour-Laurent, 1993). The effect of gluta-mate release from these interneurones, which appear toplay a role in flight initiation (Pearson et al., 1985), isagain unknown.
A number of spiking and nonspiking local interneuronesin the crayfish thoracic nervous system appear to have arole in walking activity (Chrachri and Clarac, 1989).These can have inhibitory or excitatory influences on mo-torneurone activity, but their transmitters are unknown.At least one population of locust spiking local sensoryinterneurones that receive sensory input from the leg areGABA immunoreactive (Watson and Burrows, 1987). Thetransmitters of most other spiking local interneurone pop-ulations in the locust have yet to be determined, but theydo not appear to contain GABA, glutamate (Watson andSeymour-Laurent 1993), or acetylcholine (Lutz and Tyrer,1987).
518 A.H.D. WATSON ET AL.
LITERATURE CITED
Atwood HL, Lang F, Morin WA. 1982. Synaptic vesicles: selective depletionin crayfish excitatory and inhibitory axons. Science 176:1353–1355.
Bevengut M, Cournil I. 1990. Gamma-amino butyric acid-inhibitory motorinnervation of leg muscles of the shore crab. Eur J Neurosci 2:132–139.
Bicker G, Schafer S, Ottersen OP, Storm-Mathisen, J. 1988. Glutamate-like immunoreactivity in identified neuronal populations of insect ner-vous systems. J Neurosci 8:2108–2122.
Blitz DM, Christie AE, Coleman MJ, Norris BJ, Marder E, Nusbaum MP.1999. Different proctolin neurons elicit distinct motor patterns from amultifunctional neuronal network. J Neurosci 19:5449–5463.
Braun G Mulloney B. 1993. Cholinergic modulation of the swimmeretmotor system in crayfish. J Neurophysiol 70:2391–2398.
Burrows M, Watson AHD, Brunn DE. 1989. Physiological and ultrastruc-tural characterization of a central synaptic connection between identi-fied motor neurones in the locust. Eur J Neurosci 1:111–126.
Cattaert D, Araque A, Washington B, Clarac F. 1994. Nicotinic and mus-carinic activation of motoneurones in the crayfish locomotor network.J Neurophysiol 72:1622–1633.
Chrachri A, Clarac F. 1987. Induction of rhythmic activity in motorneu-rones of crayfish thoracic ganglia by cholinergic agonists. Neurosci Lett77:49–54.
Chrachri A, Clarac, F. 1989. Synaptic connections between motor neuronesand interneurons in the fourth thoracic ganglion of the crayfish. J Neu-rophysiol 62:1237–1249.
Chrachri A, Clarac F. 1990. Fictive locomotion in the fourth thoracicganglion of the crayfish. J Neurosci 10:707–710.
Coleman MJ, Nusbaum MP. 1994. Functional consequences of compart-mentalization of synaptic input. J Neurosci 14:6544–6552
Dubas F. 1990. Inhibitory effect of L-glutamate on the neuropile arboriza-tions of flight motoneurones in locusts. J Exp Biol 148:501–508.
Dubas F. 1991. Actions of putative amino-acid transmitters on the neuro-pile arborisations of locust flight motoneurones. J Exp Biol 155:337–356.
Emson PC, Burrows M, Fonnum F. 1974. Levels of glutamate decarboxyl-ase, choline acetyl transferase and acetylcholine esterase in identifiedmotor neurones of the locust. J Neurobiol 5:33–42.
Florey E. 1973. Acetylcholine as sensory transmitter in crustacea. J CompPhysiol 83:1–16.
Franke C, Hatt H, Dudel J. 1986a. The excitatory glutamate activatedchannel recorded in cell-attached and excised patches from the mem-branes of tail, leg and stomach muscles of crayfish. J Comp Physiol A159:579–589.
Franke C, Hatt H, Dudel J. 1986b. The inhibitory chloride channel acti-vated by glutamate as well as GABA. J Comp Physiol A 159:591–609.
Hardt M, Watson AHD. 1994. Distribution of synapses on two ascendinginterneurones carrying frequency-specific information in the auditorysystem of the cricket: evidence for GABAergic inputs. J Comp Neurol345:481–495.
Hardt M, Watson, AHD. 1999. Distribution of input and output synapseson the central branches of bushcricket and cricket auditory afferentneurones: immunocytochemical evidence for GABA and glutamate indifferent populations of presynaptic boutons. J Comp Neurol 403:281–294.
Harris-Warrick RM, Marder E, Selverston AI, Moulins M, editors. 1992.Dynamic biological networks. The stomatogastric system. Cambridge:MIT Press.
Heitler WJ. 1978. Coupled motoneurones are part of the crayfish swim-meret central oscillator. Nature 275:231–234.
Heitler WJ. 1981. Neural mechanisms of central pattern generation in thecrayfish swimmeret system. In: Salanki G, editor. Neurobiology ofinvertebrates, vol 23, Advances in physiological science. Oxford: Per-gamon Press, Oxford. p 369–383.
Heitler WJ, Burrows M. 1977. The locust jump. II. Neural circuits of themotor programme. J Exp Biol 66:221–241.
Heitler WJ, Pearson KG. 1980. Non-spiking interactions and local inter-neurons in the central pattern generator of the crayfish swimmeretsystem. Brain Res 187:206–211.
Hoyle G, Burrows M. 1973. Neural mechanisms underlying behavior in thelocust Schistocerca gregaria. I. Physiology of identified motorneuronsin the metathoracic ganglion. J Neurobiol 4:3–41.
Katz PS, Harris-Warrick RM. 1989. Serotonergic/cholinergic muscle recep-tor cells in the crab stomatogastric nervous system. II. Rapid nicotinicand prolonged modulatory effects in neurons in the stomatogastricganglion. J Neurophysiol 62:571–581.
Katz PS, Eigg MH, Harris-Warrick RM. 1989. Serotonergic/cholinergicmuscle receptor cells in the crab stomatogastric nervous system. I.Identification and characterization of gasteropyloric receptor cells.J Neurophysiol 62:558–570.
Kilman VL, Marder E. 1996. Ultrastructure of the stomatogastric ganglionneuropil of the crab, Cancer borealis. J Comp Neurol 374:362–375.
King DG. 1976a. Organisation of crustacean neuropile. I. Patterns ofsynaptic connections in lobster stomatogastric ganglion. J Neurocytol5:207–237.
King DG. 1976b. Organisation of crustacean neuropile. II. Distribution ofsynaptic contacts on identified motor neurons in the lobster stomato-gastric ganglion. J Neurocytol 5:239–266.
Kondoh Y, Sato M, Hisada M. 1987. Neuronal structure and synapticdistribution of a uropod closer motor neuron in the crayfish terminalganglion. J Neurocytol 16:39–54.
Le Ray D, Clarac F, Cattaert D. 1997a. Functional analysis of the sensorymotor pathway of resistance reflex in crayfish. I. Multisensory codingand motor neuron monosynaptic responses. J Neurophysiol 78:3133–3143.
Le Ray, D, Clarac, F Cattaert, D. 1997b. Functional analysis of the sensorymotor pathway of resistance reflex in crayfish. II. Integration of sen-sory inputs in motor neurons. J Neurophysiol 78:3144–3153.
Leitch B, Laurent G. 1993. Distribution of GABAergic synaptic terminalson the dendrites of locust spiking local interneurones. J Comp Neurol337:461–470.
Leitch B, Heitler WJ, Cobb JLS, Pitman, RM. 1990. Anti-GABA antibodieslabel a subpopulation of chemical synapses which modulate an electri-cal synapse in crayfish. J Neurocytol 19:929–936.
Lutz EM, Tyrer NM. 1987. Immunohistochemical localisation of cholineacetyl transferase in the central nervous system of the locust. BrainRes 407:173–179.
Marder E, Eisen JS. 1984. Transmitter identification of pyloric neurons:electrically coupled neurons use different transmitters. J Neurophysiol51:1345–1361.
Miller MW, Vu ET, Krasne FB. 1992. Cholinergic transmission at the firstsynapse of the circuit mediating the crayfish lateral escape reaction.J Neurophysiol 68:2174–2184.
Mulloney B, Hall WM. 1990. GABA-ergic neurons in crayfish nervoussystem: an immunocytochemical census of the segmental ganglia andthe stomatogastric system. J Comp Neurol 291:383–394.
Nagayama T, Takahata M, Hisada M. 1983. Local spikeless interaction ofmotoneuron dendrites in the crayfish Procamburus clarkii. J CompPhysiol 152:335–345.
Paul DH, Mulloney B. 1985a. Local interneurons in the swimmeret systemof the crayfish. J Comp Physiol A 156:489–502.
Paul DH, Mulloney B. 1985b. Nonspiking local interneuron in the motorpattern generator for the crayfish swimmeret. J Neurophysiol 54:28–39.
Pearlstein E, Marchand AR, Clarac F. 1994. Inhibitory effects ofL-glutamate on central processes of crustacean leg motoneurons. EurJ Neurosci 6:1445–1452.
Pearlstein E, Watson AHD, Bevengut M, Catteart D. 1998. Inhibitoryconnections between antagonistic motor neurons of the crayfish walk-ing legs. J Comp Neurol 399:241–254.
Pearson KG, Reye DN, Parsons DW, Bicker G. 1985. Flight-initiatinginterneurones in the locust. J Neurophysiol 53:910–925.
Pfluger HJ, Watson AHD. 1995. GABA and glutamate-like immunoreac-tivity at synapses received by dorsal unpaired median neurones in theabdominal nerve cord of the locust. Cell Tissue Res 280:325–333.
Sherff CM, Mulloney B. 1996. Tests of the motor neuron model of the localpattern-generating circuits in the swimmeret system. J Neurosci 16:2839–2859.
Sillar KT, Skorupski P. 1986. Central input to primary afferent neurons incrayfish, Pascifastacus leniusculus, is correlated with rhythmic motoroutput to the thoracic ganglia. J Neurophysiol 55:678–688.
Siwicki KK, Bishop CA. 1986. Mapping of proctolinlike immunoreactivityin the nervous systems of lobster and crayfish. J Comp Neurol 243:435–453.
519SYNAPSES ON CRAYFISH MOTORNEURONES
Siwicki KK, Beltz BS, Kravitz EA. 1987. Proctolin in identified serotoner-gic, dopaminergic, and cholinergic neurons in the lobster, Homarusamericanus. J Neurosci 1987:522–532.
Skorupski P, Rawat BM, Bush BMH. 1992. Heterogeneity and centralmodulation of feedback reflexes in crayfish motor pool. J. Neurophysiol67:648–688
Sombati S, Hoyle G. 1984. Glutamatergic central nervous transmission inlocusts. J Neurobiol 15:507–516.
Takeuchi A, Takeuchi N. 1965. Localised action of gamma aminobutyricacid in the crayfish muscle. J Physiol 177:225–238.
Uchizono K. 1967. Inhibitory synapses on the stretch receptor neuron ofthe crayfish. Nature 214:833–834
Usherwood PNR, Giles D, Suter C. 1980. Studies of the pharmacology ofinsect neurones in vitro. In: Insect neurobiology and pesticide action(Neurotox ’79). London: Society of Chemical Industry. p 115–128.
Usherwood PNR, Machili P, Leaf G. 1968. L-glutamate at insect excitatorynerve-muscle synapses. Nature 219:1169–1172.
Watson AHD. 1988. Antibodies against GABA and glutamate label neu-rones with morphologically distinct synaptic vesicles in locust centralnervous system. Neuroscience 26:33–44.
Watson AHD, Burrows, M. 1981. Input and output synapses on identified
motor neurones of a locust revealed by the intracellular injection ofhorseradish peroxidase. Cell Tissue Res 215:325–332.
Watson AHD, Burrows, M. 1982. The ultrastructure of identified locust motorneurones and their synaptic relationships. J Comp Neurol 205:383–397.
Watson AHD, Burrows, M. 1987. Immunocytochemical and pharmacolog-ical evidence for GABAergic spiking local interneurones in the locust.J Neurosci 7:1741–1751.
Watson AHD, Hardt M. 1996. Distribution of synapses on two local audi-tory interneurones, ON1 and ON2, in the prothoracic ganglion of thecricket: relationships with GABA-immunoreactive neurones. Cell Tis-sue Res 283:231–246.
Watson AHD, Pfluger H-J. 1994. Distribution of input synapses fromprocesses exhibiting GABA- and glutamate-like immunoreactivity ontoterminals of prosternal filiform afferents in the locust. J Comp Neurol343:617–629.
Watson AHD, Seymour-Laurent KJ. 1993. The distribution of glutamate-like immunoreactivity in the thoracic and abdominal ganglia of thelocust (Schistocerca gregaria). Cell Tissue Res 273:557–570.
Watson AHD, Burrows M, Hale JP. 1985. The morphology and ultrastruc-ture of common inhibitory motor neurones in the thorax of the locust.J Comp Neurol 239:341–359.
