The Multisegmental Motor Supply to Transverse Muscles Differs in a Cricket and a Bushcricket

The transverse muscles of the cricket abdomen are serially homologous muscle bundles that span bilaterally between the left and right edges of the sternites in abdominal segments 2–8 (DuPorte, 1920; Ford, 1923; Richards, 1963; Huber, 1960). They run dorsally across the ventral nerve cord and are among the few skeletal muscles that bridge the midline of the body without a median insertion on a cuticular structure (Voss, 1905).

The tension and contraction of these compact muscles control the width of each sternite and its degree of U-shaped bending. The transverse muscles also keep the abdominal organs from pressing ventrally onto the nerve cord, and their activation pattern is usually synchronized with expiratory dorsoventral compression during abdominal pumping (Hustert and Schneider, 1986). At the location of the transverse muscles of crickets, other orthopteran insects have similar (bushcrickets) or different hyperneural muscle systems in the sternal region of their abdomen: locusts and mole crickets have a ventral diaphragm consisting of an almost continuous network of delicate muscle fibres (Albrecht, 1953), whereas cockroaches have a longitudinal system of hyperneural muscles (Shankland, 1965). The inter-species difference in hyperneural muscle systems is more extreme than in other homologous muscle systems described for orthopterans; for example, the rostral, tergal, sternal or appendage muscles. Comparisons of the nervous and motoneurone supply have often shown a very similar Bauplan for homologous motor systems in all orthopteran insects. We expected a similar situation for hyperneural muscle systems, but the innervation of the locust diaphragm (Peters, 1977; Yang and Burrows, 1983) and the cockroach system (Hertel and Penzlin, 1986) differed from each other and from that of crickets (Hustert and Schneider, 1986).

A preliminary study (Hustert and Schneider, 1986) has demonstrated an extraordinary peripheral branching pattern for single motoneurones in the transverse muscles of crickets: branches diverge to transverse muscles in several segments, and several motoneurones converge from different ganglia onto single transverse muscles. This design may either be a primitive form of intersegmental integration based on the convergence of efferents or a specific system of peripheral integration based on efferent redundancy that cooperates with central control of motoneurones in the central nervous system (CNS).

In bushcrickets, the innervation pattern of transverse muscles is quite different because, typically, they are innervated by four bilateral pairs of neurones originating from different neighbouring ganglia. One pair are inhibitory neurones whereas the branches of another pair of motoneurones diverge, supplying transverse muscles of two adjacent segments (Consoulas, 1990). The developmental significance of the peripheral and intersegmental collaterals has not been assessed here.

In this paper, as a first step in a general comparison of transverse muscle systems, we studied the motor supply of the apparently similar muscular systems of the cricket Gryllus bimaculatus and of the bushcricket Decticus albifrons.

Crickets, Gryllus bimaculatus (DeGeer) (Grylloidea, Orthoptera), were taken from a laboratory culture (Zoologisches Institut der Universität Göttingen). Bushcrickets Decticus albifrons (Fabricius) (Tettigonioidea, Orthoptera) were collected from fields in the Peninsula of Halkidiki and from fields in Vergina (the capital of ancient Macedonia) during the summer months. Dissections were started after chilling the animals to 2–4°C for anaesthesia. In experiments requiring the suppression of ventilatory activity, the neck connectives were severed. After the wings and legs had been removed, the animal was pinned down ventrally and its abdomen was dissected by dorsal longitudinal incisions. Reproductive organs and the gut were dissected from the interior and the tergites were gently pinned out laterally so that the abdomen could be filled with cricket saline without leakage. In experiments using D. albifrons, the body cavity was filled with physiological saline (in mmol l⁻¹: NaCl, 140; KCl, 5; CaCl₂.2H₂O, 5; NaHCO₃, 4; MgCl₂, 1; saccharose, 140; and Tes buffer, 5; pH7.0) Transverse muscles (TM), nerves and ganglia were then accessible for stimulation, for extracellular recording with hook and suction electrodes, for intracellular recording with glass microelectrodes, and also for neuroanatomical tracing by conventional cobalt backfilling from nerve stumps, followed by silver intensification (Bacon and Altman, 1977; Hustert, 1978).

The neural connections extending from each segmental TM to TMs in other segments were studied by stimulating antidromically from the motoneurone branches using surface suction electrodes placed on several TM fibres or hook electrodes placed on the appropriate peripheral nerves. Suprathreshold stimuli produced twitching in every TM showing a common motor innervation. These results were reaffirmed by centrifugal cobalt filling of the TM motoneurones selectively via the dorsal nerve close to its ganglionic nerve root. In crickets, the intersegmental motoneurone collaterals through the paramedian nerves to other segments were then traced in silver-intensified wholemounts of the abdominal sternal regions. In bushcrickets, the ganglionic sheath was treated with 1 % protease (Sigma XIV) for 1min before intracellular recording and staining of neurones. Glass microelectrodes were filled with cobalt hexammine chloride (resistance 50–80 MΩ). To monitor the muscle tension generated by a spontaneously active TM, one of its insertions, including a small area of the membrane, was cut off, denervated ipsilaterally and attached to the probe of an isometric tension transducer. The other end remained intact.

Ganglia were numbered according to the number of their corresponding abdominal segment (AG6, ganglion of the sixth abdominal segment). In G. bimaculatus, the bilateral dorsal and ventral nerves are termed d and v respectively (Edwards and Palka, 1974). TM7 is the transverse muscle and 7d is the dorsal nerve of the seventh segment. In D. albifrons, the nomenclature of Cambell (1961) was applied with some modifications to the nerve branches of the abdominal system.

The transverse muscles bridge the sternites in the anterior half of abdominal segments 2–7 in G. bimaculatus (Fig. 1A) and 2–8 in D. albifrons (Fig. 1B). Their muscle fibres connect directly between left and right sternal insertions across the midline. In male crickets, the TMs of each segment remain separate in segments 2–8. The only link between transverse muscles by intersegmental muscle strands is present between the seventh and eighth abdominal segments of females (Fig. 1A). From their very fine TM8, a pair of lateral oblique muscle bundles branches anteriorly and converges on the middle of TM7 (Fig. 1A). This connection resembles the pattern of oblique intersegmental muscle strands in the ventral diaphragm of locusts and, in some respects, the hyperneural muscle of cockroaches (Richards, 1963; Shankland, 1965).

Transverse muscle contractions in the cricket and the bushcricket narrow the sternites, which bulge ventrally at the same time (Fig. 2A,C). Sternite bending in the cricket is eased by two flexible sutures some distance from the segmental longitudinal midline (Gnatzy and Hustert, 1989). When the TMs relax, the sternites extend again as a result of elastic recoil in these sutures. In the bushcricket, with its less sclerotized sternite, this is achieved by lateral recoil of the previously compressed tergites. TMs contribute to the compression phase of expiratory dorsoventral pumping (Fig. 2B,D) by decreasing the width and volume of the segments. They also prevent the sternites from being pulled too far laterally by the lateral tergosternal muscles (Fig. 2A,C). During strong expiration, the serial arrangement of left dorsoventral, transverse sternal and right dorsoventral muscles can stabilize or even narrow the width of each segment. At the same time, sternal bending protects the CNS from being compressed by a filled gut and by genital organs lying dorsally in the abdomen. One major transverse muscle function, therefore, is to support expiration (Hustert and Schneider, 1986).

Visual inspection under the light microscope, and serial longitudinal sections, reveals that transverse muscle fibres bridge the segmental midline without a morphological discontinuity in the median region. The nervous supply, in contrast, never crosses the segmental midline. This strictly hemisegmental extent of the motoneurones of each muscle is seen both in peripheral cobalt fills (Fig. 3B) and in electrophysiological recordings (Fig. 3A,C). Only attenuated excitatory postsynaptic potentials (EPSPs) can be recorded from the contralateral side of any muscle fibre in the vicinity of the midline (Fig. 3A). In D. albifrons, where inhibitory innervation is present, TM6 muscle fibres receive converging inputs from both left and right inhibitory neurones in only a small area around the midline (Fig. 3C).

In G. bimaculatus, there is only one bilateral pair of excitatory motoneurones from each ganglion to its corresponding segmental TM. They can be backfilled from the axon in the nerve branch connecting to the dorsal nerve (Fig. 4A). One of the motoneurones innervating the TMs in D. albifrons is homologous to the cricket transverse motoneurones (Fig. 4B). In both cases, the motoneurone somata lie ventrally in the anterior region of the contralateral hemiganglion. The primary neurite crosses the midline through the anterior dorsal commissure and gives rise to extensive dorsal branches which extend far into the more posterior neuropiles of both hemiganglia (Fig. 4A,B).

Exceptions to this strictly segmental pattern occur in the cricket, where the two fused ganglia contain abdominal neuromeres (Fig. 4A). (i) In the metathoracic ganglion, which contains the neuromeres for the first and second abdominal segments, the TM2 motoneurones develop additional anterior dorsal branches that extend into the first abdominal neuromere. (ii) In the terminal ganglion, the TM7 motoneurones (supplying both TM7 and the rudimentary TM8) extend their posterior dorsal branches into the neuromere of the eighth abdominal segment.

The transverse muscles from the third to the seventh segments are linked by the left and right paramedian nerves (Fig. 1A). This is the only nervous supply other than the segmental supply via the dorsal nerve. The intersegmental paramedian nerves only contain intersegmental collaterals of excitatory motoneurone axons that branch to the segmental transverse muscle and send collaterals anteriorly, posteriorly or in both directions to one or more successive transverse muscles (Fig. 5). The intersegmental collaterals connect to most, if not all, fibres of every muscle they reach, but the density of contacts seems to decrease at increasing distance from the segment of motoneurone origin.

The intersegmental divergence of the TM motoneurones through the paramedian nerves in the middle abdominal segments circumvents the CNS functionally and connects the TMs of the third to eighth segments by a separate pathway. As a consequence, several intersegmental motoneurone branches, in addition to the segmental ones, converge on each transverse muscle (TM3–TM7) (Fig. 5). Conduction speeds in these axons are 0.35–0.4 ms⁻¹.

Stimulation of the terminal branches of motoneurones from a single muscle demonstrates that muscles innervated by the same motoneurone contract in synchrony. Neuroanatomy and electrical stimulation showed that the innervation patterns of the transverse muscles include (i) exclusively heterosegmental motoneurones to TM8 from the neighbouring segmental neuromere (seventh abdominal neuromere) in female crickets; (ii) a bilateral homosegmental supply to TM2 from its second abdominal neuromere in the metathoracic ganglion; and (iii) supply by the homosegmental and three additional heterosegmentally converging pairs of motoneurones to TM6, totalling eight different motoneurones for this single transverse muscle. This complex intersegmental innervation pattern raises the question of how efferents to these muscles are coordinated in the abdomen of crickets.

The innervation pattern in the bushcricket D. albifrons differs from that in the cricket. Each of the five adjacent mid-abdominal transverse muscles TM3–TM7 is innervated bilaterally by both an anterior and a posterior pair of nerve branches. Transverse muscle 6 (TM6), for example, is supplied by the pair of N1B1 branches of the sixth abdominal ganglion (AG6) and by the pair of N1B3 branches arising from the next anterior ganglion (AG5) (Fig. 1B). TM2, the most anterior transverse muscle, is supplied only by N1B1 branches of the second abdominal ganglion, whereas the most posterior muscle (TM8) receives only the N1B3 branches from the seventh abdominal ganglion (Fig. 1B). The sixth transverse muscle was selected as typical of a mid-abdominal segment. Backfilling of posterior nerve 1B1 of TM6 (see inset in Fig. 6) reveals median and contralateral neurones in the sixth abdominal ganglion. The contralateral neurone is homologous to one of the cricket neurones, as described above. Backfilling of the anterior N1B3 branch reveals a contralateral motoneurone in each of the next two anterior ganglia (AG5 and AG4, Fig. 6). One motoneurone, which is homologous to the cricket transverse muscle motoneurones, has an axon that bifurcates in different directions via N1B1 to the segmental TM and via N1B3 to the posterior TM on the same side (see inset of Fig. 6).

The axons of contralateral motoneurones located in the fifth and fourth abdominal ganglia, innervating the right half of TM6 (see inset in Figs 6, 7), pass through N1B3. This makes it possible to record their action potentials extracellularly (Fig. 7A), to identify the motoneurones physiologically (Fig. 7) and to stain them intracellularly. Paired recordings reveal that the tonically activated neurone is located in the fourth abdominal ganglion in a medio-posterior position (Fig. 7B). Its cell body lies ventrally, its main dendritic field dorsally, while its axon descends through the connective to the next posterior ganglion (AG5) from where it branches to the segmental muscle (TM6) via N1B3. This tonic motoneurone is rarely active during normal or strong ventilation, but it is active during shallow ventilation or after the connectives have been severed anterior to the fourth abdominal ganglion. By homology with cricket transverse motoneurones, the bushcricket’s anterior contralateral motoneurone innervates TM5 and TM6, and its large-amplitude action potentials may be recorded from N1B3 (Fig. 7A). It fires in a typical bursting pattern during expiration (Fig. 7C).

Cobalt fills and stimulation of peripheral nerves 1B1 and 1B3 of all of the segments revealed the complete innervation plan of all transverse muscles in D. albifrons (Fig. 5B).

Myograms from a transverse muscle and a dorsoventral ventilatory muscle in the same abdominal segment showed that the transverse muscles are active mainly during expiratory pumping movements and are relatively inactive in the intervening periods (Fig. 2B). The muscles usually relax during inspiration and during the ventilatory pause.

During natural ventilation, the expiratory bursts are well synchronized in all abdominal ganglia. The convergent bursts of motor output therefore superimpose at each TM of this area in reasonable synchrony (Fig. 8), so there is no effective time lag between the efferent bursts transmitted from the metathoracic centres through the segmental ganglion and those signals that have travelled intersegmentally for a longer distance through the paramedian nerves. The overall frequency of incoming motor impulses is highest in TM6 because it has the greatest convergence of motoneurones. The intersegmental convergence on muscles causes an increased frequency of incoming motor spikes. Additionally, during expiratory bursts, temporal and spatial summation, and possibly mutual facilitation of PSPs by different motoneurones, occur on multiply innervated transverse muscles.

Only EPSPs were recorded from any cricket TM. Their amplitudes at different frequencies of motoneurone discharge (both from a single or from several converging motoneurones) often showed signs of facilitation (Fig. 9A,D), but occasionally a slight decrease in amplitude was seen (Fig. 9C). In simultaneous extracellular recordings from the paramedian nerve and intracellular recordings from transverse muscles innervated by just two ipsilateral efferents, large differences in the amplitude of EPSPs caused by two motoneurones could be demonstrated (Fig. 9A).

In D. albifrons, the entire set of transverse muscles contracts rhythmically to cause expiration. Most of the muscle fibres are controlled by bursts of EPSPs during expiration and IPSPs during inspiration (Fig. 10A). During regular ventilation many EPSPs summate in TM5 fibres (Fig. 10A), but during shallow ventilation the EPSPs may be separated into three amplitudes (Fig. 10B). In neighbouring TMs, subsets of EPSPs correspond 1:1 because they are generated by the same bifurcating neurone, i.e. the anterior contralateral neurone of the fifth abdominal ganglion. Nerve 1B3 contains only the axons of the two excitatory contralateral motoneurones (Fig. 7). Paired recordings from nerve branch 1B1 and the ipsilateral TM5 fibres reveal (i) that the large action potentials recorded from N1B1 generate one of the two group of EPSPs (Fig. 10D); (ii) that EPSPs without corresponding action potentials are generated by the motoneurone that enters via nerve branch 1B3 (Fig. 10D); and (iii) that the smaller-amplitude action potentials generate IPSPs (Fig. 10 C,D). From these recordings and the total number of efferent neurones to TM6, it is obvious that the posterior and anterior contralateral neurones are excitatory and the median ones are inhibitory.

The excitatory bifurcating motoneurones are continuously active, and their firing pattern and coordination may be recorded from their efferent axons in the appropriate N1B3 branches. Recordings from left and right N1B3 branches show that, although the action potentials do not correlate 1:1, they discharge at similar frequencies during ventilatory bursts (Fig. 11A). The effect of a bifurcating motoneurone on the contractions of TM was studied in the sixth transverse muscle. One side of TM6 was attached to a tension transducer and both N1B3 branches and the right N1B1 branch were cut off so that it was possible to study the effect of the only remaining active contralateral bifurcating motoneurone. The rhythmical contractions of the muscle followed the rhythmical activation of this neurone alone and correlated well with the summating EPSPs (Fig. 11B). Single EPSPs did not induce contractions here but may increase the tension of the muscle between the expiratory phases.

Topographical criteria characterize the transverse muscles of G. bimaculatus and D. albifrons as being homologous to other hyperneural muscle systems of orthopteran sternites, such as the ventral diaphragm of the locust (Albrecht, 1953) or the hyperneural muscles of the cockroach (Shankland, 1965). However, the muscular structures of the cricket and bushcricket systems show remarkable differences and similarities in their morphology, insertion sites, dimensions and position in relation to the nerve cord. Differences are also seen in the way these muscles are innervated. Yang and Burrows (1983) have suggested that the ventral diaphragm of the locust, in the first abdominal segment, is innervated by six pairs of neurones whereas the other segments each have two pairs of motoneurones located in two adjacent ganglia (Peters, 1977; M. Ferber, personal communication). The hyperneural muscles of cockroaches are innervated by an unknown number of neurones with axons in the median nerve (Hertel and Penzlin, 1986). Since there is no information available about the origin and development of the above systems, one could claim that, although these muscles occupy similar positions in the abdominal segments, they are not homologous and therefore have different innervation. This would not apply to the differences in the innervation patterns of the tranverse muscle systems of the cricket G. bimaculatus and the bushcricket D. albifrons. In both species, these muscles fulfil the morphological criteria of homology because (i) both occupy the same position in the segment, running across the nerve cord, (ii) both are supplied by nerve branches running through the tergal nerve of each abdominal ganglion; and (iii) both have the same function, participating in ventilatory pumping movements of the abdomen. If these serially homologous muscles have evolved from a common ancestor, one would expect that they should be innervated by homologous neurones. The results presented here only partly confirm this hypothesis. In G. bimaculatus, six pairs of serially homologous excitatory motoneurones innervate the transverse muscles, while their axons diverge to supply different (up to four) transverse muscles. In D. albifrons, each transverse muscle, with the exception of the first and last, is innervated by three pairs of excitatory and one pair of inhibitory neurones. The axon of only one type of excitatory motoneurone divides to innervate two adjacent transverse muscles. This motoneurone appears to be homologous to the cricket motoneurone as far as the location, intraganglionic architecture and type of EPSPs generated in the fibres are concerned. This neurone, therefore, seems to be the only one common to the innervation of both systems. The inhibitory neurones specifically and exclusively innervate the transverse muscles. Four other groups of segmental muscles in D. albifrons (ventral and dorsal intersegmentals and first and second dorsoventrals) are innervated by different inhibitors (Consoulas, 1990; Consoulas and Theophilidis, 1992). Moreover, inhibitory innervation has never been described in other hyperneural muscle systems.

Because of the presumed close common ancestry of the cricket and bushcricket lines, it is not possible to evaluate which system is more advanced. Further clues may be gained from the TM system in ‘primitive’ crickets, such as Stenopelmatidae (Jerusalem crickets), that receives inhibitory input on TM muscle fibres as do bushcrickets (C. Consoulas and R. Hustert, unpublished observation). This animal has morphological features of both bushcrickets (fused second and third abdominal ganglion separate from the metathoracic ganglion and just one lateral nerve root from each ganglion) and crickets (clavate hairs on the cerci) (C. Consoulas and R. Hustert, unpublished observations).

The differences in the TM structure and innervation in crickets and bushcrickets may be required by differences in the functional design of their anatomy: G. bimaculatus has a completely sclerotized sternite, D. albifrons has only muscular attachments sclerotized in this region of the abdomen. In bushcricket, active ventilation is achieved mainly by narrowing the abdominal segments in the pleural region and, to a lesser extent, by lifting the sternite region. Both types of ventilatory movements require strong involvement of the TMs during expiration to pull the tergal regions towards each other. This is favoured by the less sclerotized sternites of D. albifrons.

We have analysed the basic functions of the transverse muscles and of their peculiar nervous supply in the orthopterans G. bimaculatus and D. albifrons. Their role in shaping and stabilizing the sternite during ventilation is obvious, for they usually contract in phase with the expiratory dorsoventral muscles. During expiration, the dorsoventral expiratory muscles in a segment and the interconnecting transverse muscle make up a muscular sling that, in addition to bending and lifting the sternite, stabilizes or narrows the tergite.

In crickets, the expiratory discharge rates in ventilatory motoneurones that arise from each abdominal ganglion are not strictly proportional to each other, indicating that, for the expiratory and transverse muscles, there is no common final output in a segment. Similarly, transverse muscle discharge can differ to some extent from one segment to another. The rate of contraction in a segment may also depend on internal conditions, such as the volume of haemolymph or the presence of more solid material (gut, reproductive organs) that requires adaptability in the compression rate of a segment during ventilation. These specific needs in each abdominal segment may vary according to the variable volumes of the gut contents and genital products. The gut and gonads do not end at the segmental borders and so proprioceptors should regulate each segment and its vicinity. Correspondingly, in most insects, many primary mechanosensory afferents in the abdomen extend into several ganglia (Zawarzin, 1924; Hustert, 1978). In contrast, most afferents from the legs reach the ipsilateral hemisegment only (Pflüger et al. 1989).

In bushcrickets, which have a very soft sternal region, pressure caused by ventilation and the weight of gut and reproductive organs spreads directly to most of the other segments. This may require a high degree of synchrony of discharges at the common final output to all transverse muscles. The morphological (body shape) and functional (the way the abdomen pumps during ventilation) differences observed between the two species could reflect the differences found in the innervation pattern of the transverse muscles.

Stress on the sternites is monitored by campaniform sensilla located in the cuticle and by the chordotonal organ that spans its flexible sutures (Gnatzy and Hustert, 1989). Additionally, pleural chordotonal organs of cricket segments, homologous to those of locusts (Hustert, 1974), can record the strain on the extending pleural folds when the segment expands dorsoventrally (R. Hustert, unpublished results). Since most abdominal afferent axons extend into the segmental and neighbouring ganglia, direct proprioceptive information about the volume of a segment is available for local control and also diverges for the coordination of transverse and other muscles in neighbouring segments. This may result in a lower ‘drive’ for a transverse muscle contraction in a segment that contains bulky tissue than for a neighbouring normal segment, even if high motor output converges from other ganglia. A comparable mechanosensory control function is achieved by the multipolar sensilla of the locust ventral diaphragm, which respond more strongly when the diaphragm is loaded (Hustert, 1975). In that system, the transverse sternal muscles (ventral diaphragm) contract during inspiration and are not synchronous with expiration as are the TMs in crickets (Fig. 2).

Direct motoneurone supply to different muscles of insects has so far only been described for common inhibitory neurones (Hale and Burrows, 1985). Several examples are known of individual insect wall muscles being innervated by several excitatory motoneurones converging from two adjacent ganglia or neuromeres (Lewis et al. 1973) or for visceral musculature (Klemm et al. 1986) and genital muscles (Hustert and Topel, 1986; Heckmann and Hustert, 1989; Hustert, 1980). However, convergence from several segmental ganglia is unusual. Could there be a specific functional reason for this? An explanation is required for the high convergence of the segmental plus intersegmental motoneurone supplies onto the transverse muscles of most abdominal segments, especially in crickets. The rate and strength of muscle contraction are increased by this convergence: several motoneurones discharging in the same expiratory bursting period produce a higher frequency of EPSPs in a TM than if only the segmental efferents are active in bilateral synchrony. In bushcrickets, single EPSPs in a TM are not able to elicit contractions. In crickets, the large gut and genital contents present in the middle abdominal segments require a larger variability in segmental volume and a larger range of motor control during expiratory compression of these segments. The high convergence of motoneurones onto the transverse muscles in the fourth, fifth and sixth abdominal segments supports this view.

Why should specific peripheral pathways bridging several segments be required for transverse muscle motoneurones? Why do these motoneurone collaterals not use the CNS as a pathway? Only developmental reasons come to mind. Collateral branching of the motoneurones within the CNS and through the connectives may not be possible during development. Developmental conditions and/or functional requirements apparently favour collateral sprouting in the periphery. As a result, the TM motoneurones that diverge to different segments in the periphery can receive synaptic inputs only in their segmental ganglion of origin.

We thank Henning Schneider for his contribution to the early stages of this study and M. Ferber for their helpful comments. In particular, C. Consoulas wishes to thank R. Hustert for his support and help in the study of the transverse muscle systems. This research project was supported partly by the ERASMUS program to C.C. and by the Greek Ministry of Industry, Energy and Technology (Branch Research and Technology). We thank Deutsche Forschungsgemeinschaft for some of the equipment we used (Hu223).