Endogenous rhythm and pattern-generating circuit interactions in cockroach motor centres

Abstract

Cockroaches are rapid and stable runners whose gaits emerge from the intricate, and not fully resolved, interplay between endogenous oscillatory pattern-generating networks and sensory feedback that shapes their rhythmic output. Here we studied the endogenous motor output of a brainless, deafferented preparation. We monitored the pilocarpine-induced rhythmic activity of levator and depressor motor neurons in the mesothoracic and metathoracic segments in order to reveal the oscillatory networks' architecture and interactions. Data analyses included phase relations, latencies between and overlaps of rhythmic bursts, spike frequencies, and the dependence of these parameters on cycle frequency. We found that, overall, ipsilateral connections are stronger than contralateral ones. Our findings revealed asymmetries in connectivity among the different ganglia, in which meta-to-mesothoracic ascending coupling is stronger than meso-to-metathoracic descending coupling. Within-ganglion coupling between the metathoracic hemiganglia is stronger than that in the mesothoracic ganglion. We also report differences in the role and mode of operation of homologue network units (manifested by levator and depressor nerve activity). Many observed characteristics are similar to those exhibited by intact animals, suggesting a dominant role for feedforward control in cockroach locomotion. Based on these data we posit a connectivity scheme among components of the locomotion pattern generating system.

Keywords: Central pattern generator; Cockroach; Connectivity model; Extracellular-recording; Locomotion control.

Fig. 1.

Electrode positioning and the recorded rhythmic activity. (A) Schematic presentation of a thoracic hemiganglion and its peripheral nerves. Hooks mark recording sites. Recorded nerves were crushed distal to the recording site in order to prevent the transmission of afferent signals from leg-sensors into the hemiganglionic neuropile. (B) Recording of rhythmic activity of levator and depressor MNs. Nomenclature is presented according to: body side, thoracic segment, and nerve function, e.g. L3Dep, left-metathoracic-depressor. L2Lev and R3Lev burst approximately in-phase with each other, as do R2Lev and L2Dep. Latencies are defined in the text (see ‘Results’). The antagonistic L2Lev and L2Dep alternate in approximate anti-phase, as does the ipsilateral levator pair R2Lev and R3Lev. Average burst duration is greater for depressor in comparison to its antagonistic levator (e.g. L2Dep duration>L2Lev duration).

Fig. 2.

The pilocarpine-induced motor pattern. (A) Depressor MNs burst termination is dependent on levator activity. Asterisks above levator trace denote double-bursts; grey bars: depressor tonic firing in the absence of levator bursts. After levator activity halts, the depressor continues to fire tonically for a highly variable time period (seconds-minutes), with highly variable spike-frequency. In addition, multiple bursting was relatively common in levators and rare in depressors. (B) An example of prolonged pilocarpine-induced rhythmic activity in levator and depressor MNs. Activity of levator MNs 5, 6 and larger units (LevMN-5, LevMN-6 and Lev-L, respectively) is marked. The L3Dep trace comprises bursts of DepMN-Ds. The low-amplitude activity seen between depressor bursts is in--phase with levator bursts and represents the activity of common-inhibitory-neurons (CIN). The two diagonal levators burst in-phase with each other and in anti-phase with the left-metathoracic-depressor, in accordance with predicted activity during the double-tripod gait.

Fig. 3.

Endogenous temporal characteristics are dependent on burst frequency. Grey bands represent confidence interval of the linear regression lines. Correlation coefficients' [Pearson's r (A) and Spearman's ρ (B)] are noted. Data are averaged from left and right hemiganglia. Details of correlations are presented in ‘Results’. (A) Spike frequency is dependent on burst frequency in the absence of proprioceptive or descending inputs. Spike frequency during a burst of activity increases with increasing burst frequency; this dependency is greater for levators than for depressors. Levators of both ganglia exhibit similar spike frequencies and rate of change (slope). Spike frequency significantly differs between the two depressors. (B) L/D ratio positively correlates with burst frequency. The range of recorded frequencies is smaller in the metathoracic samples. L/D ratio is not significantly different between the two ganglia, and in most cases it is greater in the metathorax. The slopes are similar, although the correlation is stronger in the mesothorax.

Fig. 4.

Transition latency between levator-to-depressor is greater than between depressor-to-levator in antagonist mesothoracic and metathoracic pairs. Latencies of transition between levator-to-depressor (solid lines) and depressor-to-levator (dashed lines) are asymmetric. Blue: both transitions are shorter in the mesothorax (A) than in the metathorax (B). Green: in the mesothorax, overlap (i.e. negative latency) is much more common in the depressor-to-levator transition than vice versa.

Fig. 5.

Latencies within hemiganglia and between neighbouring in-phase MNs. Line, box and whiskers represent median, interquartile range and non-outlier range (1.5×interquartile range), respectively; means are marked by +; meta, metathoracic; meso, mesothoracic; *P<0.05. (A) Levator-to-depressor transitions are greater than those between depressor and levator in both hemiganglia. Transition latency between Lev-Dep is similar in both hemiganglia, as is the latency between Dep-Lev. However, the significant difference in mean latency between the two different transitions within each pair Wilcoxon signed-rank test might indicate that different mechanisms negotiate each transition. (B) Dashed line: latency=0. From left to right: latencies between burst onsets of diagonal, ipsilateral, and contralateral in-phase active pairs. The ipsilateral pathway exhibits short latency and low variability, indicating strong ipsilateral connectivity. Onset latencies of contralateral and diagonal pairs are similar (P > 0.1, Mann–Whitney test). Mesothoracic MN activities precede their metathoracic agonistic MNs, as indicated by the negative means, demonstrating the front-to-back activation sequence that characterizes the double-tripod gait.

Fig. 6.

Phase relations and phase-locking. (A) Scheme of phase relations s.d., as a measure of coupling strength. Arrow end, levator; round end, depressor; circular connection, coupling between antagonistic MNs within a single hemiganglion. Solid and dashed lines represent pairs of in-phase and anti-phase MN pairs, respectively. Lower s.d. indicates stronger coupling. Endogenous coupling strength was found to be dependent upon these parameters: (i) direction, ipsilateral coupling is stronger than contralateral and diagonal coupling; (ii) hemiganglia involved, contralateral coupling differs between ganglia; and (iii) function of the coupled MNs, coupling between levator and depressor is stronger than between two levators. (B) Phase-locking strength between mesothoracic and metathoracic MNs is asymmetrical and stronger in the ascending pathway. Data for three pairs are normally distributed (Shapiro-Wilk test, P > 0.05) and are presented. Line, box, and whiskers represent mean, s.e.m. and s.d., respectively. Significance level is marked as *P<0.05. Transition phase-lock is stronger in the ascending pathway for each of the pairs. Asymmetric phase-locking indicates differences in the mechanisms of coordination in different directions.

Fig. 7.

Schemes of the connectivity within and between the coxa-trochanter CPGs. (A) A three-component local hemiganglionic control architecture. Arrow end, excitatory synapse; round end, inhibitory synapse; τ, time constant; 5, LevIN-5; Ds, LevIN-Ds; Kernel, oscillating CPG kernel that excites LevIN-5 and DepIN-Ds. LevIN-5 is first to burst and activate its following MN, while inhibiting DepIN-Ds. Due to the suggested greater τ in the kernel-DepIN-Ds path, once LevIN-5 excitation by the oscillating kernel is terminated, DepIN-Ds escapes its inhibition but still receives excitation from the kernel, resulting in a DepIN-Ds burst that induces its follower MN to fire its plateau potentials. (B) A suggested minimal connectivity model of the CPG network generating the double-tripod gait activity pattern of the coxa-trochanter-joints. K, oscillating CPG kernel. Shaded grey: data obtained or postulated from previous research. The model employs nearest-neighbour architecture with a front-to-back propagation sequence that can generate the motor patterns observed in this work, corresponding to straight walking on a smooth horizontal surface. Round and arrow ends represent inhibitory and excitatory synapses, respectively. Grey and black lines represent weak and strong connections [e.g. the direct inhibitory K-K connection descending from the mesothoracic hemiganglia represents weaker (grey) meso-meta connection in comparison to the ascending, stronger (black), meta-meso pathway]. Lev and Dep represent LevIN-5, and DepIN-Ds, respectively. 1, tonic drive generated by the subesophageal ganglion to activate local oscillatory-kernels; 2, mutual inhibition between neighbouring oscillatory-kernels; 3, oscillatory-kernel simultaneously excites LevIN-5, which is first to burst, and DepIN-Ds; 4, LevIN-5 activity inhibits its antagonistic DepIN-Ds; 5, LevIN-5 excites all DepIN-Ds of its neighbouring hemiganglia. The architecture is minimal and allows the addition of various centrally-generated inputs, as well as head-descending and proprioceptive inputs.