1. Computational and Systems Biology
  2. Neuroscience
Download icon

Brain Networks: Modeling breathing rhythms

  1. Jan-Marino Ramirez  Is a corresponding author
  2. Nathan A Baertsch
  1. Seattle Children's Research Institute, United States
  2. University of Washington, United States
  • Cited 0
  • Views 1,040
  • Annotations
Cite this article as: eLife 2019;8:e46033 doi: 10.7554/eLife.46033


Computational models are helping researchers to understand how certain properties of neurons contribute to respiratory rhythms.

Main text

Orchestral music is beautiful, rich and complex, yet analyzing the contribution of any one instrument can be difficult. In many ways, neuroscientists face similar challenges when they try to understand how the networks in the brain work. These circuits are made of many types of neurons, each of which has different properties. Moreover, the properties of a neuron can change because of its interactions with other cells in the network.

Researchers have started to unravel this complexity by building computational models of both neurons and networks of neurons, and by focusing on pairs of interacting properties. This is similar to how one might study how the left and right hand of a violinist work together to produce melody and rhythm, and then use this knowledge to better understand other stringed instruments and their roles in the orchestra.

Within the ‘rhythmogenic’ networks of the brain, different types of neurons work together to create the body rhythms that are essential for life. For example, a complex network generates breathing rhythms, and it is often used to understand rhythmogenic circuits in general. The balance between excitatory and inhibitory connections between neurons is critical to shape network activity (Ramirez and Baertsch, 2018). Equally important, but less well understood, are the interactions between intrinsic properties that are built into individual neurons.

In particular, two intrinsic properties are thought to play a role in controlling breathing rhythms and network activity in general. The first is a persistent sodium current (INaP), which is slowly activated and inactivated by changes in the voltage across the neuronal membrane. The second does not depend on voltage: rather, the calcium-activated non-selective cation current (ICAN) is triggered when the level of calcium ions inside the cell increases. The source of these calcium ions is unknown, but there is experimental evidence that they could be provided by mechanisms at the synapses between neurons (Del Negro et al., 2010; Del Negro et al., 2011). Both INaP and ICAN allow some neurons to generate rhythms on their own, without being stimulated by their neighbors. Yet, in the actual network, these properties also enhance the signals transmitted by excitatory synapses (Ramirez et al., 2004). Because INaP and ICAN interact with synaptic properties, as well as with each other, it is difficult to isolate their relative contributions.

Pharmacological and genetic manipulations have shed light on how INaP and ICAN work in the respiratory network (see, for example, Koizumi et al., 2018; Picardo et al., 2019), but there is still no consensus on how they contribute to rhythmogenesis (Feldman and Del Negro, 2006; Ramirez et al., 2004). Now, in eLife, Jeffrey Smith of the National Institute Neurological Disorders Stroke (NINDS) and colleagues – including Ryan Phillips as first author – report results from a computational modeling approach that re-examines how INaP and ICAN control breathing (Phillips et al., 2019).

The researchers, who are based at NINDS, the University of New Hampshire and Georgia State University, established a simplified model of the breathing network, which only takes into account excitatory synaptic interactions. Then, they tuned the contributions of three variables, INaP, ICAN, and the source of intracellular calcium ions, and explored how this affected the frequency and amplitude of breathing. First, they examined how changes in the source of calcium ions influenced the contribution of ICAN to the network. If calcium came from within neurons, ICAN controlled the frequency, but not the strength, of breathing. On the other hand, if calcium depended on synaptic activity, ICAN acted as a synaptic amplifier to control the strength of the rhythm, but it had little effect on its frequency. This scenario best matched experimental data and prior conclusions (Del Negro et al., 2011; Koizumi et al., 2018), prompting Phillips et al. to conclude that ICAN is activated by calcium ions that are primarily of synaptic origin.

The group then went on to demonstrate that ICAN does not establish the rhythm in their computational model, but that it increases the strength of breathing by recruiting more neurons to participate in the network. Indeed, when ICAN was eliminated from the model network, a weak breathing rhythm remained, which was generated by a small number of neurons with high levels of INaP activity (Figure 1). Eliminating this ‘rhythmogenic kernel’ from the model stopped the rhythm altogether, in agreement with some (Peña et al., 2004), but not all (Pace et al., 2007) experimental data.

How the intrinsic properties of neurons contribute to the breathing rhythm.

Phillips et al. established a model of the brain network that controls breathing, and used it to deduce how two built-in properties of neurons, the INaP sodium current (blue) and the ICAN current (green), control the breathing rhythm. Without INaP, the network is silent, with the neurons exhibiting non-rhythmic activity (grey). However, a small number of neurons with high levels of INaP activity can produce a weak rhythm, even in the absence of ICAN. In turn, the model suggests that ICAN is activated by calcium ions coming from synapses; the current would then amplify excitatory interactions between neurons. This amplification leads to additional neurons participating in the rhythm (red), producing a robust functional network.

Unlike the model built by Phillips et al., the actual network that controls breathing is not exclusively excitatory, but is subjected to important inhibitory and neuromodulatory control. Further, alternative mechanisms of rhythmogenesis that do not depend on INaP have also been proposed (Del Negro et al., 2010; Rubin et al., 2009). Could different rhythmogenic mechanisms therefore contribute to breathing, depending on the demands of the network? In particular, could the contributions of INaP and ICAN change based on the neuromodulatory state of the network? Addressing such questions will require further back-and-forth between experiments and increasingly complex models.

Overall, the results of Phillips et al. allow us to understand how intrinsic neuronal properties independently control the strength and frequency of the breathing rhythm. Their model is also a useful framework in which to explore how changes in the way INaP and ICAN interact can dynamically impact rhythmogenic properties. For example, it could shed light on the way how the network reconfigures when the body lacks oxygen. Ultimately, describing the duet between INaP and ICAN in the respiratory network may help dissect how rhythmic activity is controlled in other regions of the brain (Penn et al., 2016; Riquelme et al., 2018).


Article and author information

Author details

  1. Jan-Marino Ramirez

    Jan-Marino Ramirez is in the Center for Integrative Brain Research, Seattle Children's Research Institute, and the departments of Pediatrics and Neurological Surgery, University of Washington, Seattle, United States

    For correspondence
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5626-3999
  2. Nathan A Baertsch

    Nathan A Baertsch is in the Center for Integrative Brain Research, Seattle Children's Research Institute, Seattle, United States

    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1589-5575

Publication history

  1. Version of Record published: March 25, 2019 (version 1)


© 2019, Ramirez and Baertsch

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.


  • 1,040
    Page views
  • 162
  • 0

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Computational and Systems Biology
    2. Neuroscience
    Sean R Bittner et al.
    Research Article

    A cornerstone of theoretical neuroscience is the circuit model: a system of equations that captures a hypothesized neural mechanism. Such models are valuable when they give rise to an experimentally observed phenomenon -- whether behavioral or a pattern of neural activity -- and thus can offer insights into neural computation. The operation of these circuits, like all models, critically depends on the choice of model parameters. A key step is then to identify the model parameters consistent with observed phenomena: to solve the inverse problem. In this work, we present a novel technique, emergent property inference (EPI), that brings the modern probabilistic modeling toolkit to theoretical neuroscience. When theorizing circuit models, theoreticians predominantly focus on reproducing computational properties rather than a particular dataset. Our method uses deep neural networks to learn parameter distributions with these computational properties. This methodology is introduced through a motivational example of parameter inference in the stomatogastric ganglion. EPI is then shown to allow precise control over the behavior of inferred parameters and to scale in parameter dimension better than alternative techniques. In the remainder of this work, we present novel theoretical findings in models of primary visual cortex and superior colliculus, which were gained through the examination of complex parametric structure captured by EPI. Beyond its scientific contribution, this work illustrates the variety of analyses possible once deep learning is harnessed towards solving theoretical inverse problems.

    1. Cancer Biology
    2. Computational and Systems Biology
    Ioannis Galdadas et al.
    Research Article

    Mutations within the kinase domain of the epidermal growth factor receptor (EGFR) are common oncogenic driver events in non-small cell lung cancer. Although the activation of EGFR in normal cells is primarily driven by growth-factor-binding-induced dimerization, mutations on different exons of the kinase domain of the receptor have been found to affect the equilibrium between its active and inactive conformations giving rise to growth-factor-independent kinase activation. Using molecular dynamics simulations combined with enhanced sampling techniques, we compare here the conformational landscape of the monomers and homodimers of the wild-type and mutated forms of EGFR ΔELREA and L858R, as well as of two exon 20 insertions, D770-N771insNPG, and A763-Y764insFQEA. The differences in the conformational energy landscapes are consistent with multiple mechanisms of action including the regulation of the hinge motion, the stabilization of the dimeric interface, and local unfolding transitions. Overall, a combination of different effects is caused by the mutations and leads to the observed aberrant signaling.