Motor Output Patterns Research Articles

Distinct neurogenetic mechanisms establish the same chemosensory valence state at different life stages in Caenorhabditis elegans.

An animal's preference for many chemosensory cues remains constant despite dramatic changes in the animal's internal state. The mechanisms that maintain chemosensory preference across different physiological contexts remain poorly understood. We previously showed that distinct patterns of neural activity and motor output are evoked by carbon dioxide (CO2) in starved adults vs dauers of Caenorhabditis elegans, despite the two life stages displaying the same preference (attraction) for CO2. However, how the distinct CO2-evoked neural dynamics and motor patterns contribute to CO2 attraction at the two life stages remained unclear. Here, using a CO2 chemotaxis assay, we show that different interneurons are employed to drive CO2 attraction at the two life stages. We also investigate the molecular mechanisms that mediate CO2 attraction in dauers vs adults. We show that insulin signaling promotes CO2 attraction in dauers but not starved adults and that different combinations of neurotransmitters and neuropeptides are used for CO2 attraction at the two life stages. Our findings provide new insight into the distinct molecular and cellular mechanisms used by C. elegans at two different life stages to generate attractive behavioral responses to CO2.

A change in behavioral state switches the pattern of motor output that underlies rhythmic head and orofacial movements

The breathing rhythm serves as a reference that paces orofacial motor actions and orchestrates active sensing. Past work has reported that pacing occurs solely at a fixed phase relative to sniffing. We re-evaluated this constraint as a function of exploratory behavior. Allocentric and egocentric rotations of the head and the electromyogenic activity of the motoneurons for head and orofacial movements were recorded in free-ranging rats as they searched for food. We found that a change in state from foraging to rearing is accompanied by a large phase shift in muscular activation relative to sniffing, and a concurrent change in the frequency of sniffing, so that pacing now occurs at one of the two phases. Further, head turning is biased such that an animal gathers a novel sample of its environment upon inhalation. In total, the coordination of active sensing has a previously unrealized computational complexity. This can emerge from hindbrain circuits with fixed architecture and credible synaptic time delays.

Patch‐Seq Analysis of Dbx1‐derived Respiratory Neurons of the preBötzinger Complex in Neonatal Mice

Inspiratory breathing movements depend on neurons in the preBötzinger complex (preBötC) whose progenitors express the transcription factor Dbx1 (henceforth Dbx1 neurons). These Dbx1 preBötC neurons generate both the rhythm and the rudimentary motor output pattern. Electrophysiological properties may differentiate rhythmogenic and pattern-related neurons, but whether two discrete functional classes of Dbx1 preBötC neurons exist remains to be demonstrated. Here, we use the Patch-Seq technique to measure the transcriptomes of individual Dbx1 preBötC neurons. We aim to develop a comprehensive understanding of the molecular components of the inspiratory core oscillator and to investigate whether discrete neuron classes exist therein. We recorded Dbx1 inspiratory neurons in whole-cell patch-clamp mode to assess their electrophysiological properties and then aspirated their cellular contents into the patch electrode for single-cell RNA-sequencing. We sequenced a total of 18 samples: 7 putative rhythm generators, 9 putative output neurons, and 1 that we could not definitely characterize physiologically. We obtained an average of 11,371,215 uniquely aligned paired-end reads per sample with an average length of 151 bp. More than a dozen differentially expressed genes distinguish putatively rhythmogenic and pattern-related Dbx1 preBötC neurons. These data make testable predictions about the composition and function of the mammalian preBötC that advance understanding at the molecular genetic level.

The motor cortex uses active suppression to sculpt movement.

Even the simplest movements are generated by a remarkably complex pattern of muscle activity. Fast, accurate movements at a single joint are produced by a stereotyped pattern that includes a decrease in any preexisting activity in antagonist muscles. This premovement suppression is necessary to prevent the antagonist muscle from opposing movement generated by the agonist muscle. Here, we provide evidence that the primary motor cortex (M1) sends a command signal that generates this premovement suppression. Thus, output neurons in M1 sculpt complex spatiotemporal patterns of motor output not only by actively turning on muscles but also by actively turning them off.

Role of Propriospinal Neurons in Control of Respiratory Muscles and Recovery of Breathing Following Injury.

Respiratory motor failure is the leading cause of death in spinal cord injury (SCI). Cervical injuries disrupt connections between brainstem neurons that are the primary source of excitatory drive to respiratory motor neurons in the spinal cord and their targets. In addition to direct connections from bulbospinal neurons, respiratory motor neurons also receive excitatory and inhibitory inputs from propriospinal neurons, yet their role in the control of breathing is often overlooked. In this review, we will present evidence that propriospinal neurons play important roles in patterning muscle activity for breathing. These roles likely include shaping the pattern of respiratory motor output, processing and transmitting sensory afferent information, coordinating ventilation with motor activity, and regulating accessory and respiratory muscle activity. In addition, we discuss recent studies that have highlighted the importance of propriospinal neurons for recovery of respiratory muscle function following SCI. We propose that molecular genetic approaches to target specific developmental neuron classes in the spinal cord would help investigators resolve the many roles of propriospinal neurons in the control of breathing. A better understanding of how spinal circuits pattern breathing could lead to new treatments to improve breathing following injury or disease.

Phrenic-specific transcriptional programs shape respiratory motor output.

The precise pattern of motor neuron (MN) activation is essential for the execution of motor actions; however, the molecular mechanisms that give rise to specific patterns of MN activity are largely unknown. Phrenic MNs integrate multiple inputs to mediate inspiratory activity during breathing and are constrained to fire in a pattern that drives efficient diaphragm contraction. We show that Hox5 transcription factors shape phrenic MN output by connecting phrenic MNs to inhibitory premotor neurons. Hox5 genes establish phrenic MN organization and dendritic topography through the regulation of phrenic-specific cell adhesion programs. In the absence of Hox5 genes, phrenic MN firing becomes asynchronous and erratic due to loss of phrenic MN inhibition. Strikingly, mice lacking Hox5 genes in MNs exhibit abnormal respiratory behavior throughout their lifetime. Our findings support a model where MN-intrinsic transcriptional programs shape the pattern of motor output by orchestrating distinct aspects of MN connectivity.

Graded Arrays of Spinal and Supraspinal V2a Interneuron Subtypes Underlie Forelimb and Hindlimb Motor Control

The spinal cord contains neural networks that enable regionally distinct motor outputs along the body axis. Nevertheless, it remains unclear how segment-specific motor computations are processed because the cardinal interneuron classes that control motor neurons appear uniform at each level of the spinal cord. V2a interneurons are essential to both forelimb and hindlimb movements, and here we identify two major types that emerge during development: type I neurons marked by high Chx10 form recurrent networks with neighboring spinal neurons and type II neurons that downregulate Chx10 and project to supraspinal structures. Types I and II V2a interneurons are arrayed in counter-gradients, and this network activates different patterns of motor output at cervical and lumbar levels. Single-cell RNA sequencing (RNA-seq) revealed type I and II V2a neurons are each comprised of multiple subtypes. Our findings uncover a molecular and anatomical organization of V2a interneurons reminiscent of the orderly way motor neurons are divided into columns and pools.

Different contributions of primary motor cortex, reticular formation, and spinal cord to fractionated muscle activation

Coordinated movement requires patterned activation of muscles. In this study, we examined differences in selective activation of primate upper limb muscles by cortical and subcortical regions. Five macaque monkeys were trained to perform a reach and grasp task, and electromyogram (EMG) was recorded from 10 to 24 muscles while weak single-pulse stimuli were delivered through microelectrodes inserted in the motor cortex (M1), reticular formation (RF), or cervical spinal cord (SC). Stimulus intensity was adjusted to a level just above threshold. Stimulus-evoked effects were assessed from averages of rectified EMG. M1, RF, and SC activated 1.5 ± 0.9, 1.9 ± 0.8, and 2.5 ± 1.6 muscles per site (means ± SD); only M1 and SC differed significantly. In between recording sessions, natural muscle activity in the home cage was recorded using a miniature data logger. A novel analysis assessed how well natural activity could be reconstructed by stimulus-evoked responses. This provided two measures: normalized vector length L, reflecting how closely aligned natural and stimulus-evoked activity were, and normalized residual R, measuring the fraction of natural activity not reachable using stimulus-evoked patterns. Average values for M1, RF, and SC were L = 119.1 ± 9.6, 105.9 ± 6.2, and 109.3 ± 8.4% and R = 50.3 ± 4.9, 56.4 ± 3.5, and 51.5 ± 4.8%, respectively. RF was significantly different from M1 and SC on both measurements. RF is thus able to generate an approximation to the motor output with less activation than required by M1 and SC, but M1 and SC are more precise in reaching the exact activation pattern required. Cortical, brainstem, and spinal centers likely play distinct roles, as they cooperate to generate voluntary movements.NEW & NOTEWORTHY Brainstem reticular formation, primary motor cortex, and cervical spinal cord intermediate zone can all activate primate upper limb muscles. However, brainstem output is more efficient but less precise in producing natural patterns of motor output than motor cortex or spinal cord. We suggest that gross muscle synergies from the reticular formation are sculpted and refined by motor cortex and spinal circuits to reach the finely fractionated output characteristic of dexterous primate upper limb movements.

Central Pattern Generators: Mechanisms of Operation and Their Role in Controlling Automatic Movements

Central pattern generators consist of sets of interconnected neurons able to generate a basic motor output pattern underlying automatic movements (respiration, locomotion, chewing, swallowing, etc.) without any afferent signals from the executive motor apparatus. They are divided into constitutive pattern generators, which are active throughout life (the respiratory generator), and conditional pattern generators, which control episodic movements (locomotion, chewing, swallowing, etc.). As the motor output of a pattern generator is defined by its internal organization, the activity of conditional pattern generators is initiated by a simple command arriving from the higher centers. The structural-functional organization of the locomotor pattern generators in the marine mollusk Clione, the lamprey, the frog embryo, and laboratory mammals (cats, mice, and rats) are described, along with pattern generators controlling respiratory and swallowing movements in mammals and pattern generators for discharges of the electric organs of gymnotiform fish. Generation of the rhythmic motor output is shown in all cases to be based on the endogenous (pacemaker) activity of specific groups of interneurons and interneural interactions. These two interacting mechanisms supplement each other, ensuring the reliable operation of pattern generators. The question of whether experience gained from studies of central pattern generators can be used for understanding the mechanisms of more complex brain functions, including cognitive functions, is discussed.

Corticospinal and Reticulospinal Contacts on Cervical Commissural and Long Descending Propriospinal Neurons in the Adult Rat Spinal Cord; Evidence for Powerful Reticulospinal Connections.

Descending systems have a crucial role in the selection of motor output patterns by influencing the activity of interneuronal networks in the spinal cord. Commissural interneurons that project to the contralateral grey matter are key components of such networks as they coordinate left-right motor activity of fore and hind-limbs. The aim of this study was to determine if corticospinal (CST) and reticulospinal (RST) neurons make significant numbers of axonal contacts with cervical commissural interneurons. Two classes of commissural neurons were analysed: 1) local commissural interneurons (LCINs) in segments C4-5; 2) long descending propriospinal neurons (LDPNs) projecting from C4 to the rostral lumbar cord. Commissural interneurons were labelled with Fluorogold and CST and RST axons were labelled by injecting the b subunit of cholera toxin in the forelimb area of the primary somatosensory cortex or the medial longitudinal fasciculus respectively. The results show that LCINs and LDPNs receive few contacts from CST terminals but large numbers of contacts are formed by RST terminals. Use of vesicular glutamate and vesicular GABA transporters revealed that both types of cell received about 80% excitatory and 20% inhibitory RST contacts. Therefore the CST appears to have a minimal influence on LCINs and LDPNs but the RST has a powerful influence. This suggests that left-right activity in the rat spinal cord is not influenced directly via CST systems but is strongly controlled by the RST pathway. Many RST neurons have monosynaptic input from corticobulbar pathways therefore this pathway may provide an indirect route from the cortex to commissural systems. The cortico-reticulospinal-commissural system may also contribute to functional recovery following damage to the CST as it has the capacity to deliver information from the cortex to the spinal cord in the absence of direct CST input.

Evidence for a role of the reticulospinal system in recovery of skilled reaching after cortical stroke: initial results from a model of ischemic cortical injury

The purposes of this pilot study were to create a model of focal cortical ischemia in Macaca fascicularis and to explore contributions of the reticulospinal system in recovery of reaching. Endothelin-1 was used to create a focal lesion in the shoulder/elbow representation of left primary motor cortex (M1) of two adult female macaques. Repetitive microstimulation was used to map upper limb motor outputs from right and left cortical motor areas and from the pontomedullary reticular formation (PMRF). In subject 1 with a small lesion and spontaneous recovery, reaching was mildly impaired. Changes were evident in the shoulder/elbow representations of both the lesioned and contralesional M1, and there appeared to be fewer than expected upper limb responses from the left (ipsilesional) PMRF. In subject 2 with a substantial lesion, reaching was severely impaired immediately after the lesion. After 12 weeks of intensive rehabilitative training, reach performance recovered to near-baseline levels, but movement times remained about 50% slower. Surprisingly, the shoulder/elbow representation in the lesioned M1 remained completely absent after recovery, and there was a little change in the contralesional M1. There was a definite difference in motor output patterns for left versus right PMRF for this subject, with an increase in right arm responses from right PMRF and a paucity of left arm responses from left PMRF. The results are consistent with increased reliance on PMRF motor outputs for recovery of voluntary upper limb motor control after significant cortical ischemic injury.

Role of the retrotrapezoid nucleus/parafacial respiratory group in coughing and swallowing in guinea pigs.

The retrotrapezoid/parafacial respiratory group (RTN/pFRG) located ventral to the facial nucleus plays a key role in regulating breathing, especially enhanced expiratory activity during hypercapnic conditions. To clarify the roles of the RTN/pFRG region in evoking coughing, during which reflexive enhanced expiration is produced, and in swallowing, during which the expiratory activity is consistently halted, we recorded extracellular activity from RTN/pFRG neurons during these fictive behaviors in decerebrate, paralyzed, and artificially ventilated guinea pigs. The activity of the majority of recorded respiratory neurons was changed in synchrony with coughing and swallowing. To further evaluate the contribution of RTN/pFRG neurons to these nonrespiratory behaviors, the motor output patterns during breathing, coughing, and swallowing were compared before and after brain stem transection at the caudal margin of RTN/pFRG region. In addition, the effects of transection at its rostral margin were also investigated to evaluate pontine contribution to these behaviors. During respiration, transection at the rostral margin attenuated the postinspiratory activity of the recurrent laryngeal nerve. Meanwhile, the late expiratory activity of the abdominal nerve was abolished after caudal transection. The caudal transection also decreased the amplitude of the coughing-related abdominal nerve discharge but did not abolish the activity. Swallowing could be elicited even after the caudal end transection. These findings raise the prospect that the RTN/pFRG contributes to expiratory regulation during normal respiration, although this region is not an essential element of the neuronal networks involved in coughing and swallowing.

Recruitment of muscle synergies is associated with endpoint force fluctuations during multi-directional isometric contractions.

It has long been assumed that the human central nervous system uses flexible combinations of several muscle synergies to effortlessly and efficiently control redundant movements. However, whether muscle synergies exist in the neural circuit remains controversial, and it is critical to examine the association between the recruitment pattern of synergies and motor output. In this study, we examined the relationship between the activation of muscle synergies and endpoint force fluctuations in the presence of signal-dependent noise. Subjects performed multi-directional isometric force generations around the right ankle on the sagittal plane. We then extracted muscle synergies from measured electromyogram (EMG) data using nonnegative matrix factorization. As a result, the sum of the activation of muscle synergies was correlated with the endpoint force variability from the desired directions. Furthermore, we determined that the activation trace of each synergy reflected the endpoint force fluctuations using cross-correlation analysis. Therefore, these results suggest that muscle synergies statistically calculated from EMG data should be related to the motor output.

Related neuropeptides use different balances of unitary mechanisms to modulate the cardiac neuromuscular system in the American lobster, Homarus americanus.

To produce flexible outputs, neural networks controlling rhythmic motor behaviors can be modulated at multiple levels, including the pattern generator itself, sensory feedback, and the response of the muscle to a given pattern of motor output. We examined the role of two related neuropeptides, GYSDRNYLRFamide (GYS) and SGRNFLRFamide (SGRN), in modulating the neurogenic lobster heartbeat, which is controlled by the cardiac ganglion (CG). When perfused though an isolated whole heart at low concentrations, both peptides elicited increases in contraction amplitude and frequency. At higher concentrations, both peptides continued to elicit increases in contraction amplitude, but GYS caused a decrease in contraction frequency, while SGRN did not alter frequency. To determine the sites at which these peptides induce their effects, we examined the effects of the peptides on the periphery and on the isolated CG. When we removed the CG and stimulated the motor nerve with constant bursts of stimuli, both GYS and SGRN increased contraction amplitude, indicating that each peptide modulates the muscle or the neuromuscular junction. When applied to the isolated CG, neither peptide altered burst frequency at low peptide concentrations; at higher concentrations, SGRN decreased burst frequency, whereas GYS continued to have no effect on frequency. Together, these data suggest that the two peptides elicit some of their effects using different mechanisms; in particular, given the known feedback pathways within this system, the importance of the negative (nitric oxide) relative to the positive (stretch) feedback pathways may differ in the presence of the two peptides.

Assessment of multivariate classification analysis methods on 3‐dimensional anatomical distribution of functionally identified cat pre‐Bötzinger complex microinjection sites (712.15)

Previous work from our laboratory has demonstrated that multiple patterns of phrenic motor output can be elicited by chemical stimulation of anatomically distinct sub‐regions of the in vivo adult cat pre‐Bötzinger complex. To gain further insight into the anatomical distribution of the microinjection sites yielding these different discharge patterns, we implemented two multivariate classification analysis methods: Fisher Linear Discriminant (FLD) and probabilistic neural network (PNN). For FLD analysis, the multivariate data were transformed into univariate observations by computing the inverse of a pooled variance‐covariance matrix, which was then translated back to the histological coordinates to predict physical sub‐region boundaries. For PNN analysis, a 3‐layered radial basis neural network was created with (1) the input vector for each discharge pattern yielding an output value according to how close the input vector was to each pattern weight vector, (2) the contributions for each class of inputs was summed to produce a net output vector of probabilities; and (3) a transfer function was applied to the outputs to identify the maximum probabilities and produce a classification decision. While each analysis method has its advantages (and disadvantages), both were able to predicte all of the observed phrenic discharge patterns based solely on the 3‐D coordinates of the microinjection sites. We suggest, however, that FLD analysis may be better suited for these data.Grant Funding Source: Supported by HL63175

Optogenetic Perturbation of Neural Activity with Laser Illumination in Semi-intact <em>Drosophila</em> Larvae in Motion

Drosophila larval locomotion is a splendid model system in developmental and physiological neuroscience, by virtue of the genetic accessibility of the underlying neuronal components in the circuits1-6. Application of optogenetics7,8 in the larval neural circuit allows us to manipulate neuronal activity in spatially and temporally patterned ways9-13. Typically, specimens are broadly illuminated with a mercury lamp or LED, so specificity of the target neurons is controlled by binary gene expression systems such as the Gal4-UAS system14,15. In this work, to improve the spatial resolution to "sub-genetic resolution", we locally illuminated a subset of neurons in the ventral nerve cord using lasers implemented in a conventional confocal microscope. While monitoring the motion of the body wall of the semi-intact larvae, we interactively activated or inhibited neural activity with channelrhodopsin16,17 or halorhodopsin18-20, respectively. By spatially and temporally restricted illumination of the neural tissue, we can manipulate the activity of specific neurons in the circuit at a specific phase of behavior. This method is useful for studying the relationship between the activities of a local neural assembly in the ventral nerve cord and the spatiotemporal pattern of motor output.

Optogenetic Perturbation of Neural Activity with Laser Illumination in Semi-intact <em>Drosophila</em> Larvae in Motion

Drosophila larval locomotion is a splendid model system in developmental and physiological neuroscience, by virtue of the genetic accessibility of the underlying neuronal components in the circuits1-6. Application of optogenetics7,8 in the larval neural circuit allows us to manipulate neuronal activity in spatially and temporally patterned ways9-13. Typically, specimens are broadly illuminated with a mercury lamp or LED, so specificity of the target neurons is controlled by binary gene expression systems such as the Gal4-UAS system14,15. In this work, to improve the spatial resolution to "sub-genetic resolution", we locally illuminated a subset of neurons in the ventral nerve cord using lasers implemented in a conventional confocal microscope. While monitoring the motion of the body wall of the semi-intact larvae, we interactively activated or inhibited neural activity with channelrhodopsin16,17 or halorhodopsin18-20, respectively. By spatially and temporally restricted illumination of the neural tissue, we can manipulate the activity of specific neurons in the circuit at a specific phase of behavior. This method is useful for studying the relationship between the activities of a local neural assembly in the ventral nerve cord and the spatiotemporal pattern of motor output.

Prediction of inverse relationship between compression phase duration and expulsive airflow during voluntary cough in humans by a joint neural network biomechanical computational model

Previous computational modeling efforts of the cough motor control system have focused on neural networks and predicting motor output patterns. We speculated that a recent joint neuromechanical model of the cough central motor control system that incorporated airway and chest wall mechanics would be useful in predicting mechanical aspects of cough in humans. Simulations of cough with this model produced large airflows and pressures with mechanical and temporal features very similar to those recorded during cough in humans. In particular, compression phase durations and airflows during simulated coughing varied by approximately 60%. Compression phase durations were longer during simulated coughs with low airflows. In human studies, healthy subjects performed a magnitude production protocol by producing weak, moderate, or strong voluntary coughs. Peak expiratory airflows increased by approximately 40% and abdominal electromyogram amplitudes approximately doubled from weak to strong voluntary coughs. Compression phase durations shortened significantly from weak to strong coughs (weak‐351±33 ms, moderate‐253±86 ms, strong‐217±39 ms, p<0.05). The simulation results represent an emergent property and support the predictive value of a joint neuromechanical computational model for human airway protective behaviors. Supported by NIH HL103415, HL89104, HL89071, NS19814.

Role of Inhibition in Respiratory Pattern Generation

Postsynaptic inhibition is a key element of neural circuits underlying behavior, with 20-50% of all mammalian (nongranule) neurons considered inhibitory. For rhythmic movements in mammals, e.g., walking, swimming, suckling, chewing, and breathing, inhibition is often hypothesized to play an essential rhythmogenic role. Here we study the role of fast synaptic inhibitory neurotransmission in the generation of breathing pattern by blocking GABA(A) and glycine receptors in the preBötzinger complex (preBötC), a site essential for generation of normal breathing pattern, and in the neighboring Bötzinger complex (BötC). The breathing rhythm continued following this blockade, but the lung inflation-induced Breuer-Hering inspiratory inhibitory reflex was suppressed. The antagonists were efficacious, as this blockade abolished the profound effects of the exogenously applied GABA(A) receptor agonist muscimol or glycine, either of which under control conditions stopped breathing in vagus-intact or vagotomized, anesthetized, spontaneously breathing adult rats. In vagotomized rats, GABA(A)ergic and glycinergic antagonists had little, if any, effect on rhythm. The effect in vagus-intact rats was to slow the rhythm to a pace equivalent to that seen after suppression of the aforementioned Breuer-Hering inflation reflex. We conclude that postsynaptic inhibition within the preBötC and BötC is not essential for generation of normal respiratory rhythm in intact mammals. We suggest the primary role of inhibition is in shaping the pattern of respiratory motor output, assuring its stability, and in mediating reflex or volitional apnea, but not in the generation of rhythm per se.

Nonlinear variability in a computational model of respiratory rhythm generation

19 Nonlinear variability in a computational model of respiratory rhythm generation Renato De Oliveira , Chris Fietkiewicz b Universidade Federal do Ceara, UFC, Fortaleza, Brazil Department of Electrical Engineering and Computer Science, Case Western Reserve University, Cleveland, OH, USA Excessive variability in neurally generated rhythms may indicate an instability, whereas insufficient variability could result from an inability to adapt to environmental stressors. Previous studies have addressed the presence of variability of the neural control of respiration in both the clinical context and the experimental context. It remains unknown whether variability is a cause or merely a symptom of respiratory dysfunctions such as apneas, period breathing in infants, and Cheyne-Stokes breathing in adults. A fundamental source of respiratory pattern generation is the pre– Botzinger complex (preBotC), which is part of the ventral respiratory column located in the brainstem. Here we use a computational model of the preBotC to analyze the variability of motor output patterns. In particular, we look at the relationship between sensory input and the resulting variability in the motor output pattern. Our model uses Hodgkin-Huxley style neurons that are divided into 3 cell types: hypoglossal motoneurons, premotoneurons, and preBotC cells that are configured for bursting by means of a persistent sodium current. The architecture consists of a feedforward hierarchy such that motoneurons are driven by premotoneurons and premotoneurons are driven by preBotC cells. The preBotC cells receive inputs from each other and also receive random tonic drive in the form of excitatory and inhibitory postsynaptic currents. This tonic drive represents chemosensory and mechanosensory feedback. Hierarchical levels are randomly connected with a fixed connection probability or density. We analyzed the spike histogram of the hypoglossal motoneuron population that represents the whole-nerve activity recorded from cranial nerve XII in experiments. Two nonlinear variability measures were used: mutual information and sample entropy. We studied the relationship between properties of the tonic drive to preBotC cells and the resulting variability in the motor output pattern. We varied the quantity of tonic inputs and connection densities. We found that increases in the quantity of inputs and connection densities to preBotC cells both produced an increase in overall variability. This study proposes a computational model that describes how variability in respiratory patterns may be affected by physiological properties of the respiratory neural control system, including chemosensory and mechanosensory feedback. A nonlinear effect is determined by multiple properties of the control network. Knowledge of this relationship at the kernel level is important in understanding how variability is produced and controlled within the larger closed-loop system that both controls and reacts to changes in respiratory patterns. http://dx.doi.org/10.1016/j.jcrc.2012.10.035