1. Neuroscience
Download icon

Insights into the dynamic control of breathing revealed through cell-type-specific responses to substance P

  1. Nathan A Baertsch  Is a corresponding author
  2. Jan-Marino Ramirez
  1. Seattle Children’s Research Institute, United States
  2. University of Washington School of Medicine, United States
Research Article
  • Cited 0
  • Views 508
  • Annotations
Cite this article as: eLife 2019;8:e51350 doi: 10.7554/eLife.51350

Abstract

The rhythm generating network for breathing must continuously adjust to changing metabolic and behavioral demands. Here, we examined network-based mechanisms in the mouse preBötzinger complex using substance P, a potent excitatory modulator of breathing frequency and stability, as a tool to dissect network properties that underlie dynamic breathing. We find that substance P does not alter the balance of excitation and inhibition during breaths or the duration of the resulting refractory period. Instead, mechanisms of recurrent excitation between breaths are enhanced such that the rate that excitation percolates through the network is increased. We propose a conceptual framework in which three distinct phases of inspiration, the burst phase, refractory phase, and percolation phase, can be differentially modulated to control breathing dynamics and stability. Unraveling mechanisms that support this dynamic control may improve our understanding of nervous system disorders that destabilize breathing, many of which involve changes in brainstem neuromodulatory systems.

Introduction

Rhythmicity is ubiquitous in the brain, important for many high-order functions such as consciousness, attention, perception, and memory (Başar and Düzgün, 2016; Başar and Güntekin, 2008; Colgin, 2016; Hanslmayr et al., 2016; Kiehn, 2016; Neske, 2015; Palva and Palva, 2018; Paton and Buonomano, 2018), as well as vital rhythmic motor behaviors including chewing, locomotion, and breathing (Grillner and El Manira, 2015; Kiehn, 2016; Narayanan and DiLeone, 2017; Ramirez and Baertsch, 2018a; Wyart, 2018; Nakamura et al., 2004). Rhythms generated by the brain are diverse, as are the underlying rhythm generating network- and cellular-level mechanisms (Paton and Buonomano, 2018). Indeed, rhythmicity can occur on time scales ranging from milliseconds to days (Golombek et al., 2014). However, most neuronal rhythms must be flexible, able to increase or decrease the rate of oscillation or the degree of synchronization to match changes in physiological or cognitive demands (Brittain et al., 2014; Ramirez and Baertsch, 2018b). Thus, understanding principles that allow dynamic control of rhythm generating networks may provide important insights into the regulation of diverse brain functions.

For half a century, investigations of the networks and cellular mechanisms that generate breathing have provided valuable, generalizable insights into the origins and control of neural rhythmicity (Del Negro et al., 2018; Feldman and Kam, 2015; Ramirez and Baertsch, 2018b; Wyman, 1977; Cohen, 1981; Ezure, 1990; Long and Duffin, 1986; Milsom, 1991). Inspiration, the dominant phase of breathing in mammals (Jenkin and Milsom, 2014), is generated by a spatially dynamic network located bilaterally along the ventrolateral medulla (Baertsch et al., 2019). A region that is both necessary and sufficient for inspiration, the pre-Bötzinger Complex (preBötC), is autorhythmic and forms the core of this network (Smith et al., 1991; Tan et al., 2008; Vann et al., 2018). Like many rhythmic networks, a critical characteristic of the preBötC is that the frequency of its output is dynamic - the rate of breathing changes during for example sleep/wake states, exercise, environmental challenges, and orofacial behaviors such as feeding and vocalization (Moore et al., 2013; Ramirez et al., 2016). Excitatory and inhibitory inputs from other brain regions, as well as neuromodulation, can potently facilitate or depress the frequency of breathing (Doi and Ramirez, 2008; Zuperku et al., 2017). Yet, elucidating how these influences alter the network- and cellular-level rhythm generating mechanisms within the preBötC remains a challenge (Dick et al., 2018).

Glutamatergic synaptic interactions among preBötC interneurons allow this sparsely connected network (Carroll and Ramirez, 2013; Schwab et al., 2010) to periodically synchronize and are therefore obligatory for rhythmogenesis (Ge and Feldman, 1998). However, if left unrestrained, feed-forward excitation in the network leads to hyper synchronization during inspiratory bursts, which subsequently causes a prolonged period of reduced network excitability (Baertsch et al., 2018; Kottick and Del Negro, 2015). This refractory phase delays the onset of the next inspiratory burst and, as a result, the frequency of breathing becomes very slow. Inhibitory interactions within the preBötC are critical for limiting synchronization during bursts (Harris et al., 2017) and reducing the subsequent refractoriness of the network (Baertsch et al., 2018). Indeed, roughly 40% of inspiratory preBötC neurons are inhibitory (GABAergic and/or glycinergic) (Oke et al., 2018; Winter et al., 2009; Morgado-Valle et al., 2010) and by regulating the excitability of glutamatergic neurons during inspiratory bursts, these neurons play an important role in controlling breathing frequency.

However, controlling the inspiratory burst itself is not the only mechanism that regulates the inspiratory rhythm. During the time between bursts, referred to as the inter-burst interval (IBI), recurrent excitatory synaptic connections (Del Negro et al., 2018) and ectopic bursting of individual preBötC neurons (Ramirez et al., 2004) are thought to give rise to a gradual increase in network excitability that drives the onset of the next burst. In this model, spontaneous spiking activity in a small subset of excitatory preBötC neurons begins to percolate stochastically through the network, gradually recruiting more spiking activity among interconnected excitatory neurons (Kam et al., 2013b). During this percolation phase, activation of membrane voltage- and calcium-dependent conductances in an increasing number of neurons causes the excitation to become exponential, culminating in an inspiratory burst (Del Negro et al., 2010; Ramirez et al., 2016). The gradual increase in spiking activity during this phase, or ‘pre-inspiratory ramp’, is thought to be primarily mediated by a subset of glutamatergic preBötC interneurons that have enhanced excitability. Derived from V0-lineage precursors, these neurons express the transcription factor developing brain homeobox one protein (Dbx1) during development (referred to here as ‘Dbx1 neurons’) (Bouvier et al., 2010; Gray et al., 2010; Picardo et al., 2013; Wu et al., 2017). How this process of recurrent excitation may contribute to the dynamic regulation of inspiratory frequency is not well understood.

Here, we examine network- and cellular-level changes in the inspiratory rhythm generator that underlie dynamic frequency responses to the excitatory neuromodulator substance P (SP). A member of the tachykinin neuropeptide family, SP is a key mediator of many physiological and neurobiological processes (Mantyh, 2002). For breathing, SP regulates the stability of the respiratory rhythm (Ben-Mabrouk and Tryba, 2010; Yeh et al., 2017) as well as respiratory responses to hypoxia (Chen et al., 1990; Ptak et al., 2002). The endogenous receptor for SP, neurokinin one receptor (NK1R), is expressed on only ~5–7% of CNS neurons (Mantyh, 2002), but is enriched in the preBötC (Gray et al., 1999; Schwarzacher et al., 2011). SP binding to NK1R causes excitation of preBötC neurons through coupling with voltage-independent cation channels (Hayes and Del Negro, 2007; Ptak et al., 2009), including sodium leak channel, non-selective (Nalcn). Disruption of this ion channel causes pathological respiratory instability (Yeh et al., 2017). SP also promotes robust facilitation of inspiratory frequency (Gray et al., 1999; Peña and Ramirez, 2004b). Therefore, SP is an ideal tool to explore the changes in network activity that regulate breathing stability and underlie the dynamic control of breathing frequency.

By combining electrophysiological, optogenetic, and pharmacological techniques, we find that SP differentially influences the refractory and recurrent excitation phases of the inspiratory rhythm through cell-type-specific effects. We conclude that phase-specific and differential modulation of excitatory and inhibitory network interactions is a key mechanism that allows the frequency of this vital rhythmogenic network to be dynamically controlled.

Results

SP has differential effects on the refractory and percolation phases of the preBötC rhythm

To explore how the neuromodulator SP increases inspiratory frequency at the network level, integrated preBötC population activity was recorded in horizontal brainstem slices (Anderson et al., 2016; Baertsch et al., 2019). from Dbx1CreERT2;Rosa26ChR2-EYFP neonatal mice during bath application of 0.5–1.0 µM SP (n = 6). Horizontal slices that contain the ventral respiratory column were chosen over transverse slices that attempt to isolate the preBötC because inspiratory neurons rostral of the preBötC may be important for the dynamic regulation of the refractory period and breathing frequency (Baertsch et al., 2019). In response to SP, inspiratory burst frequency increased from 0.23 ± 0.02 Hz to 0.31 ± 0.03 Hz (p=0.003) during steady state SP (>~3 min post bath application) similar to previous observations in transverse slice preparations (Peña and Ramirez, 2004b). To determine whether the refractory period is modulated by SP, brief light pulses (200 ms, 0.5 mW/mm2) were delivered randomly during the inspiratory cycle at baseline and in SP. In each condition, the probability of light-evoking a burst in the contralateral preBötC was quantified as a function of elapsed time from the preceding spontaneous population burst and compared to the cumulative distribution of spontaneous inter-burst intervals (IBIs). A representative experiment is shown in Figure 1A,B and the average data are shown in Figure 1C. Evoked bursts were rare if a light pulse occurred immediately following a spontaneous population burst. However, the probability of a light-evoked burst increased with elapsed time until ~2 s following a spontaneous burst when bursts could be evoked with nearly every light pulse (Figure 1A,B). The end of this ~2 s refractory period coincided with a large increase in the number of spontaneous IBIs (Figure 1B,C), indicating that this period of reduced preBötC excitability precludes both light-evoked and spontaneous preBötC burst generation, thereby preventing very short IBIs and fast inspiratory rhythms (Baertsch et al., 2018). However, despite a frequency increase of 31.9 ± 5.6%, the refractory period was not altered by SP (non-linear regression analysis; p>0.05). In SP, IBIs remained limited by the refractory period, but spontaneous bursts occurred more quickly and more consistently following the end of the refractory period. As a result, the average IBI became shorter (4.5 ± 0.4 s to 3.5 ± 0.5 s; paired t-test; p<0.0002) and less variable (SD of 1.3 ± 0.1 to 0.84 ± 0.1; p<0.0139) in SP (Figure 1C). Together these data suggest that SP increases the frequency and regularity of the inspiratory rhythm through differential modulation of two inspiratory phases: The refractory phase remains unchanged, while the duration of the recurrent excitation or percolation phase, which promotes the onset of the subsequent burst, is reduced.

Differential modulation of the refractory phase (RP) and recurrent excitation or percolation phase (PP) of the inspiratory rhythm by SP.

(A) Representative integrated preBötC population recordings from a Dbx1CreERT2;Rosa26ChR2-EYFP horizontal slice during photostimulation of Dbx1 neurons under baseline conditions (black) and in SP (red). (B) Quantified data from the experiment in A showing time-dependent changes in the probabilities of evoking a burst (top) and of a spontaneous burst occurring (bottom). (C) Group data from n = 6 experiments. Spontaneous and evoked probability curves were compared under baseline conditions and in SP using non-linear regression analysis. Inset shows the standard deviation (SD) of the inter-burst intervals under baseline conditions and in SP (paired, two tailed t-test). Data available in Figure 1—source data 1.

Inspiratory spiking patterns of excitatory and inhibitory neurons in the preBötC

Next, we explored the spiking patterns of individual excitatory and inhibitory preBötC neurons to identify mechanisms that may underlie differential modulation of the refractory and percolation phases in the preBötC network. Inspiratory spiking activity was recorded from n = 29 preBötC neurons in whole-cell (WC) or cell-attached (CA) configuration (Figure 2). Example cell-attached recordings are shown in Figure 2—figure supplement 1 and confirmed our results in whole-cell configuration. Membrane potentials of neurons recorded in whole-cell configuration are shown in Figure 2—figure supplement 2. Horizontal slices from Vglut2Cre;Rosa26ChR2-EYFP and VgatCre;Rosa26ChR2-EYFP mice were used so that recorded neurons could be identified as excitatory or inhibitory based on depolarizing responses to light. In Vglut2Cre;Rosa26ChR2-EYFP slices, light pulses were delivered following blockade of synaptic transmission to determine the intrinsic response of the neuron (Baertsch et al., 2018; Baertsch et al., 2019). Recorded neurons were also fluorescently labeled using patch pipets containing AlexaFluor568 to mark their anatomical locations (Figure 2A). There was considerable variability among inspiratory neurons (maximal spiking activity during inspiration), with respect to spike frequency, burst duration, and burst shape; and for any given neuron there was considerable burst-to-burst stochasticity (Carroll and Ramirez, 2013; Carroll et al., 2013). However, excitatory neurons could be clearly grouped based on the presence or absence of spiking during the IBI that typically increases in frequency, or ‘ramps’, before the subsequent inspiratory burst – often referred to as ‘pre-inspiratory (pre-I)’ activity. In contrast, we did not identify any inspiratory inhibitory neurons in the preBötC with pre-I spiking (Figure 2B). Excitatory neurons with pre-I spiking (n = 9; 7 WC, 2 CA) had inspiratory spike frequencies ranging from 12.0 to 50.8 Hz (mean: 28.0 ± 4.4 Hz) and burst durations ranging from 272 to 763 ms (mean: 467 ± 48 ms). Similarly, excitatory neurons without pre-I spiking (n = 13; 11 WC, 2 CA) had inspiratory spike frequencies ranging from 11.3 to 58.2 Hz (mean: 34.9 ± 4.3 Hz) and burst durations ranging from 242 to 698 ms (mean: 419 ± 34 ms). Inhibitory neurons (n = 7; 5 WC, 2 CA) also had considerable variability in spike frequency (16.4 to 82.9 Hz; mean: 48.6 ± 8.6 Hz) and burst duration (144 to 517 ms; mean: 316 ± 54 ms) (Figure 2C,D).

Figure 2 with 2 supplements see all
Baseline spiking patterns of excitatory and inhibitory neurons in the preBötC.

(A) Anatomical locations of n = 29 recorded preBötC neurons. (B) Example traces of an excitatory neuron with pre-I spiking (orange) and without pre-I spiking (green) and an inhibitory neuron (blue) during the inspiratory burst and inter-burst interval (IBI). (C) Quantified spike frequency as a function of time (normalized to burst duration) from n = 9 pre-I excitatory, n = 13 non pre-I excitatory, and n = 7 inhibitory neurons. (D) Quantified mean spike frequency, duration, and shape of inspiratory bursts generated by each type of neuron (one-way ANOVA with Bonferroni post hoc tests). (E) Example quantification of neuronal burst onset time relative to the preBötC population and Poincaré plots showing onset time variability from n = 9 pre-I excitatory, n = 13 non pre-I excitatory, and n = 7 inhibitory neurons (20 inspiratory bursts/neuron). (F) Burst onset time variability or ‘jitter’ quantified as the natural log of the root mean square of successive differences (one-way ANOVA with Bonferroni post hoc tests). Data available in Figure 2—source data 1.

Inspiratory neuron types also differed in burst shape (Figure 2C,D) and burst onset variability (Figure 2E,F). Inhibitory neurons had a much more pronounced decrementing spiking pattern than excitatory neurons with maximal spike frequencies occurring at 24 ± 7% of the inspiratory burst duration. Excitatory neurons with pre-I spiking exhibited a rounded burst pattern with maximal spike frequencies occurring at 49 ± 5% of the inspiratory burst duration, whereas excitatory neurons without pre-I spiking were slightly more decrementing with maximal spike frequencies occurring at 34 ± 6% of the inspiratory burst duration. We also quantified the variability in burst onset times – that is the time between the beginning of each preBötC population burst and the onset of the corresponding neuronal burst. In neurons without pre-I spiking, burst onset was defined as the time of the first action potential, whereas in pre-I neurons burst onset was defined as the time with the largest change in action potential frequency. Poincaré plots of onset times for each inspiratory neuron type are shown in Figure 2E and burst onset variability (quantified as the natural log of the root mean square of successive differences, ln(RMSSD)) is shown in Figure 2F. Overall, average onset times did not differ among neuron types (p>0.05); however, excitatory neurons with pre-I spiking had more cycle-to-cycle variability in burst onset times than excitatory neurons without pre-I spiking or inhibitory neurons (p<0.001).

Effects of SP on excitatory pre-inspiratory neurons in the preBötC are phase-dependent

Spiking activity of pre-I neurons is expected to contribute to both the refractory and percolation phases of the inspiratory rhythm. During inspiratory bursts, spiking of these neurons contributes to synchronization, which promotes the subsequent refractoriness of the network (Baertsch et al., 2018). Following the refractory phase, it is thought that pre-I spiking of these neurons during the IBI facilitates positive-feedback recurrent excitation in the network, which builds up until another inspiratory burst in generated and the cycle restarts (Del Negro et al., 2018). Thus, changes in spiking during inspiratory bursts are predicted to alter the duration of the subsequent IBI through modulation of the refractory phase, whereas changes in pre-inspiratory spiking are predicted to alter the IBI by changing the rate of feed-forward excitation during the percolation phase. We examined SP-induced changes in the spiking activity of pre-I neurons during inspiratory bursts and during the IBI. A representative recording is shown in Figure 3A. Unexpectedly, SP had little effect on spiking during inspiratory bursts (Figure 3B). Changes in burst spike frequency (28.0 ± 4.4 Hz to 29.6 ± 3.8 Hz, p>0.05) and burst duration (467 ± 78 ms to 488 ± 45 ms; p>0.05) were small and inconsistent (Figure 3B,C). In contrast, during the IBI, SP increased the average spiking frequency of pre-I neurons from 5.1 ± 1.0 Hz to 9.9 ± 1.8 Hz (p<0.01), and in all cases increased the slope of the pre-inspiratory ramp (average of 1.7 ± 0.8 to 3.7 ± 1.9 Hz/sec), although this did not reach statistical significance (p=0.113) due to the large variability among neurons (Figure 3D,E). Changes in IBI spike frequency and slope induced by SP were coincident with a significant decrease in the duration of the IBI (Figure 3E). Thus, these phase-dependent changes in spiking activity at the level of individual pre-I excitatory neurons likely contribute to the differential effects of SP on the refractory and percolation phases observed at the network level (see Figure 1).

Effects of SP on pre-I excitatory neurons in the preBötC.

(A) Example intracellular recording from a pre-I neuron at baseline (black) and in SP (orange) with corresponding integrated preBötC population activity (gray). (B) Quantified spike frequency as a function of time (normalized to inspiratory burst duration) in n = 9 pre-I neurons. (C) Mean spike frequency and burst duration of pre-I neurons (paired, two tailed t-tests). (D) Quantified spike frequency vs. time (normalized to IBI duration) during the inter-burst interval showing changes in pre-inspiratory ramp activity induced by SP. (E) Mean spike frequency, pre-inspiratory ramp slope, and IBI duration (paired, two tailed t-tests). (F) Example spiking of a pre-I neuron during a long (black) and short (gray) inter-burst interval under baseline conditions (top), and quantified pre-inspiratory slope during each IBI (below). (G) Inverse relationship between the slope of pre-inspiratory spiking and the length of the IBI from n = 9 pre-I neurons (20 consecutive IBIs/neuron) at baseline and in SP (parameters normalized to baseline values) (linear regression analysis). Data available in Figure 3—source data 1.

Since SP also reduced the variability of the IBI at the network level, we examined the relationship between the duration of individual IBIs and the slope of pre-inspiratory spiking activity. An example recording of an excitatory pre-inspiratory neuron during a long IBI (black) and a short IBI (gray), and the quantified spike frequency over the duration of each IBI, is shown in Figure 3F. Group data for n = 9 neurons is shown in Figure 3G. Twenty consecutive inspiratory cycles were analyzed for each neuron and the duration of each IBI was compared to the pre-inspiratory slope during that cycle. To highlight effects related to cycle-to-cycle variability, values were normalized to the average baseline IBI duration and pre-inspiratory slope for each neuron, respectively. We found that, under control conditions and in SP, there was a significant inverse relationship between the duration of a given IBI and the slope of the corresponding pre-inspiratory ramp, such that pre-inspiratory spiking activity at the level of individual neurons can predict the duration between inspiratory bursts at the network level.

SP recruits a subpopulation of excitatory preBötC neurons to exhibit pre-inspiratory spiking

Unlike pre-I neurons, excitatory neurons that are silent during the IBI are unable to participate in the feed-forward process of recurrent excitation because they lack pre-inspiratory spiking activity. However, these neurons are expected to contribute to network synchronization during inspiratory bursts, and as a result they have the potential to modulate the refractory period. In response to SP, excitatory neurons that did not spike during the IBI under baseline conditions exhibited two distinct phenotypes. Some (8/13) remained silent during the IBI (teal), whereas others (5/13) developed pre-inspiratory spiking (green) (Figure 4A). Excitatory neurons that were not recruited to spike during the IBI (n = 7 WC, 1 CA) also had no change in spike frequency (38.6 ± 5.0 to 37.1 ± 4.9 Hz; p>0.05) or burst duration (430 ± 53 to 428 ± 51 ms, p>0.05) during inspiratory bursts (Figure 4B,C), despite a shortened IBI duration (p<0.05). Since these neurons had no change in spiking throughout the inspiratory cycle, it is unlikely that they contribute to SP-induced frequency facilitation of the inspiratory rhythm. In neurons that were recruited to spike during the IBI (n = 4 WC, 1 CA), spiking frequency increased from 0 to 6.6 ± 1.4 Hz (p<0.01). In SP, these neurons exhibited a pre-inspiratory ramp (4.1 ± 1.7 Hz/sec), which was coincident with a shorter IBI duration (p<0.05) (Figure 4D,E). During individual inspiratory cycles, the slope of the SP-induced pre-inspiratory ramp had a significant inverse relationship with the duration of the IBI (Figure 4F). During inspiratory bursts, spiking patterns did not change in spike frequency (28.9 ± 7.8 to 31.9 ± 6.6 Hz, p>0.05), burst duration (399 ± 28 to 441 ± 36 ms, p>0.05), or burst shape (Figure 4B,C). Thus, a subpopulation of non pre-I excitatory neurons develops pre-I activity in SP without a change in spiking activity during bursts, suggesting that the number of neurons that can participate in the percolation phase increases in the presence of SP, without significant effects on the amount of excitation during bursts and the resulting refractory period.

SP recruits a subpopulation of excitatory preBötC neurons to participate in the percolation phase.

(A) Example intracellular recordings from two excitatory neurons, one that develops pre-I activity in SP (green) and one that does not (teal). Corresponding integrated preBötC population activity is shown below each trace (gray). (B) Quantified spike frequency as a function of time (normalized to baseline inspiratory burst duration) in n = 5 excitatory neurons that were recruited to pre-I (top) and n = 8 excitatory neurons that were not recruited to pre-I (bottom). (C) Mean spike frequency and burst duration in both neuron groups (paired, two-tailed t-tests). (D) Quantified spike frequency vs. time (normalized to baseline IBI duration) during the inter-burst interval showing the recruitment of pre-inspiratory ramp activity by SP. (E) Mean spike frequency, pre-inspiratory ramp slope, and IBI duration in both neuron groups (paired, two tailed t-tests). (F) Inverse relationship between the slope of pre-inspiratory spiking and the length of the IBI in SP from n = 5 excitatory neurons that were recruited to pre-I (20 consecutive IBIs/neuron) (linear regression analysis). Data available in Figure 4—source data 1.

SP does not change spiking activity of inhibitory inspiratory neurons in the preBötC

In contrast to excitatory neurons, the activity of inhibitory neurons during inspiratory bursts reduces network synchronization and the refractory period (Baertsch et al., 2018). Since the RP of the preBötC network was not changed by SP (see Figure 1), we hypothesized that inhibition during bursts would also be unchanged by SP. To test this, we recorded spiking activity from n = 7 (5 WC, 2 CA) inhibitory preBötC neurons under baseline conditions and following application of SP. A representative recording is shown in Figure 5A. In the presence of SP, spike frequency and burst duration of inspiratory inhibitory neurons did not change during bursts (48.6 ± 8.6 to 43.5 ± 6.7 Hz, p>0.05; and 316 ± 54 to 328 ± 56 ms, p>0.05, respectively), despite a coincident shortening of the IBI (p<0.05) (Figure 5B,C). Spiking activity of inhibitory neurons also did not change during the IBI, since all recorded neurons remained silent between inspiratory bursts. Thus, in response to SP, inhibitory neurons in the preBötC had no change in spiking throughout the inspiratory cycle and are therefore unlikely to play a role in SP-induced facilitation of inspiratory frequency.

SP does not change inhibitory network interactions.

(A) Example intracellular recordings from an inhibitory preBötC neuron under baseline conditions (black) and in SP (blue) with corresponding integrated preBötC population activity is shown below (gray). (B) Quantified spike frequency as a function of time (normalized to baseline inspiratory burst duration) in n = 7 inhibitory neurons. (C) Mean spike frequency, burst duration, and IBI (paired, two-tailed t-tests). Data available in Figure 5—source data 1.

SP increases stochasticity among excitatory preBötC neurons during inspiratory bursts

Next, we sought to unravel potential mechanisms that may prevent SP from causing hyper-synchronization of the preBötC network and increased refractory times. Since synchronization is often reduced with increased stochasticity (Carroll and Ramirez, 2013; Harris et al., 2017; Zerlaut and Destexhe, 2017), we compared the cycle-to-cycle variability in burst onset times (see Figure 2E) under baseline conditions and in the presence of SP (Figure 6). Although average burst onset times were not significantly altered by SP for any neuronal type (p>0.05), SP did have effects on burst onset stochasticity. This is demonstrated as greater dispersions in the Poincaré plots shown in Figure 6A. However, these SP-induced changes in burst onset variability (i.e. ‘onset jitter’) differed across neuronal types. Burst onset jitter of excitatory pre-I neurons, which was generally high under baseline conditions (see Figure 2F), did not change significantly in SP (p>0.05) (Figure 6B). In contrast, excitatory neurons that did not exhibit pre-I spiking and had relatively low onset jitter under baseline conditions exhibited increased onset jitter in the presence of SP (p<0.05). Among this group of excitatory neurons, burst onset jitter was increased by SP regardless of whether or not the neuron was recruited to develop pre-inspiratory spiking. Inhibitory preBötC neurons, on the other hand, had relatively inconsistent changes in burst onset variability induced by SP, with no change in mean onset jitter (p>0.05). These results suggest that SP increases the stochasticity of burst onset in a subgroup of preBötC excitatory neurons. Indeed, this increase in stochasticity at the level of individual neurons was reflected in the activity of the preBötC network as a slight increase in the duration (707 ± 22 to 854 ± 33 ms; p<0.0001) and halfwidth (429 ± 18 to 478 ± 20 ms; p<0.0001) of population bursts in SP (Figure 6C).

SP increases onset variability among non pre-I excitatory neurons during preBötC bursts.

(A) Poincaré plots of burst-to-burst variability in onset times under baseline conditions (gray) and in SP for n = 9 excitatory, pre-I (orange); n = 5 excitatory, recruited to pre-I (green); n = 8 excitatory, not recruited to pre-I (teal); and n = 7 inhibitory (blue) neurons. (B) Burst onset time variability or ‘jitter’, quantified as the natural log of the root mean square of successive differences, at baseline and in SP for each neuron type. (paired, two tailed t-tests). (C) Representative population bursts at baseline (black) and in SP (red) with averaged traces in bold and quantified burst duration and burst half-width from the n = 29 slices in which onset jitter was quantified (paired, two tailed t-tests). Data available in Figure 6—source data 1.

Inspiratory neurons rostral to the preBötC have heterogeneous responses to SP

Substance P increase the frequency and stability of the inspiratory rhythm in transverse slices that isolate the preBötC from the rest of the ventral respiratory column (VRC) (Gray et al., 1999; Peña and Ramirez, 2004b). However, neurons with inspiratory activity are not confined to the preBötC but are distributed along the VRC (Barnes et al., 2007; Zuperku et al., 2019). Indeed, the inspiratory network seems to be spatially dynamic since excitatory neurons located rostral to the preBötC can be conditionally recruited to participate in the inspiratory rhythm (Baertsch et al., 2019). This rostral expansion of the active inspiratory network is associated with an increase in the excitation/inhibition ratio, longer refractory times, and slower inspiratory frequencies (Baertsch et al., 2019). Therefore, we explored whether the opposite may also occur: Could the size of the active inspiratory network shrink, and could this be another mechanism that prevents SP from causing increased excitation during inspiratory bursts? To test this, we recoded spiking activity from n = 16 (13 WC, 3 CA) inspiratory neurons located in the rostral VRC (n = 5 excitatory, n = 7 inhibitory, n = 4 unknown). The anatomical locations of these neurons relative to the preBötC neurons described above are shown in Figure 7A. Overall, spiking activity patterns and responses to SP were less consistent among rostral neurons than preBötC neurons. To convey this heterogeneity, spike rasters for each rostral neuron over 20 consecutive inspiratory bursts are shown in Figure 7B,C,D. Despite this variability, spiking frequency during bursts was reduced by SP in all (5/5) rostral excitatory neurons (−55.6 ± 13.5% change from 11.1 ± 4.7 to 6.7 ± 4.0 Hz; p<0.05) (Figure 8A,C), whereas changes were inconsistent among inhibitory rostral neurons with no change on average (39.5 ± 34.7% change from 5.3 ± 1.5 to 6.1 ± 1.5 Hz; p>0.05) (Figure 8B,C). Thus, the potential contribution of rostral excitatory neurons to synchronization of the inspiratory rhythm was reduced by SP, while inhibitory influences were relatively unchanged. Burst onset variability of both excitatory and inhibitory inspiratory neurons rostral of the preBötC was generally high under baseline conditions, and it was further increased by SP (p<0.05) (Figure 8D,E).

Inspiratory neurons rostral of the preBötC have varied responses to SP.

(A) Anatomical locations of n = 16 rostral inspiratory neurons [n = 5 excitatory (green), n = 7 inhibitory (blue), and n = 4 unknown (black)], relative to preBötC neurons (light gray). (B–D) Spike rasters (each row is one burst cycle; 20 consecutive cycles are stacked) and average spike frequency for each excitatory (B), inhibitory (C), and unknown (D) rostral neuron at baseline and in SP. Time is normalized to the preBötC population burst duration, denoted by the x-axis tick marks.

Quantified effects of SP on rostral excitatory and inhibitory inspiratory neurons.

(A–C) Average spike frequency as a function of time (normalized to preBötC population burst duration) in n = 5 excitatory neurons (A), n = 7 inhibitory neurons (B), and in all rostral neurons (n = 16, red). Mean changes in inspiratory spike frequency at baseline and in SP are plotted to the right of each panel (paired, two tailed t-tests, A and B; one-way ANOVA, (C). (D) Poincaré plots showing burst-to-burst variability in onset times among rostral excitatory (n = 5) and inhibitory (n = 7) neurons (20 consecutive inspiratory bursts/neuron). (E) Mean burst onset time variability or ‘jitter’ at baseline and in SP for each excitatory (left) and inhibitory (right) neurons (paired t-tests). (F) Average spike frequency of rostral inhibitory neurons during the inter-burst interval (normalized to baseline IBI) at baseline and in SP. Mean changes in spike frequency and slope during the IBI are plotted to the right (paired, two tailed t-tests). Data available in Figure 8—source data 1.

During the inter-burst interval, the spiking activity of rostral neurons was considerably different from neurons in the preBötC. Unlike the pre-I spiking described for excitatory neurons in the preBötC (see Figures 3 and 4), excitatory rostral neurons were generally silent during the IBI under baseline conditions, and they remained silent following application of SP (Figure 7B). In contrast, four out of seven inhibitory rostral neurons exhibited spiking during the IBI under baseline conditions, and in three of these neurons spike frequency during the IBI was increased by SP. Among the three inhibitory rostral neurons that were silent during the IBI under baseline conditions, two were recruited to spike during the IBI in response to SP. Overall, six of seven exhibited spiking during the IBI in SP (1.5 ± 0.6 Hz at baseline to 3.5 ± 1.2 Hz in SP; p>0.05) (Figure 8F). The spiking of these neurons did not exhibit a pre-inspiratory ramp under baseline conditions (slope: 0.11 ± 0.14 Hz/second; p>0.05) or in the presence of SP (slope: 0.08 ± 0.15 Hz/second; p>0.05), suggesting changes in the spiking of these inhibitory neurons during the IBI is not be regulated by excitatory pre-I neurons in the preBötC.

Discussion

The rhythm generating network that produces breathing movements must constantly adjust to changing metabolic demands and also adapt to overlapping volitional and reflexive behaviors (Feldman et al., 2013; Ramirez and Baertsch, 2018b). Unravelling mechanisms that support this dynamic control may improve our understanding of disorders of the nervous system that destabilize breathing, such as Parkinson’s disease, Rett syndrome, sudden infant death syndrome, congenital central hypoventilation syndrome, multiple-systems atrophy, and amyotrophic lateral sclerosis (Oliveira et al., 2019; Ramirez et al., 2018; Schwarzacher et al., 2011; Katz et al., 2009; Moreira et al., 2016). Changes in neuromodulatory systems within the brainstem have been linked to many of these and other respiratory control disorders (Doi and Ramirez, 2008; Viemari et al., 2005). Here, we introduce the concept that distinct phases of the rhythmogenic process can be differentially regulated by neuromodulation to dynamically control the frequency and stability of breathing (Figure 9).

Intrinsic phases of the inspiratory rhythm generator and their modulation by substance P.

During inspiration, each burst is assembled stochastically via heterogeneous interactions among a combination of intertwined synaptic and intrinsic properties (Ramirez and Baertsch, 2018b). Although exclusively excitatory synaptic interactions (Kam et al., 2013a) or intrinsic bursting mechanisms (Peña et al., 2004a) may be able to produce rhythm in isolation, this is unlikely to occur under normal conditions since these properties interact strongly. To the contrary, the combination of excitatory and inhibitory synaptic interactions with intrinsic bursting properties, known as the ‘rhythmogenic triangle’ (Ramirez and Baertsch, 2018b), is critical for the flexibility of this dynamic network (Ramirez et al., 2004; Rubin and Smith, 2019; Smith et al., 2000; Richter and Smith, 2014). Neuromodulators play important roles in regulating both synaptic and intrinsic bursting properties, perhaps best demonstrated in invertebrate model systems. In these networks, neuromodulators can inhibit or strengthen synaptic interactions (Harris-Warrick et al., 1998; Marder et al., 2014; Nusbaum et al., 2001), as well as induce or suppress intrinsic bursting properties (Elson and Selverston, 1992; Flamm and Harris-Warrick, 1986; Zhang and Harris-Warrick, 1994). These important principles also apply to the aminergic and peptidergic modulation of the mammalian preBötC network (Doi and Ramirez, 2010). For example, in the absence of synaptic interactions, the activity of both INaP and ICAN-dependent autonomous bursting neurons in the preBötC is modulated by SP (Ben-Mabrouk and Tryba, 2010; Peña and Ramirez, 2004b). Although it is known that NK1R expressing preBötC neurons are important for rhythmogenesis (Gray et al., 2001; Guyenet and Wang, 2001), it is unclear how these modulatory effects regulate the frequency and stability of the inspiratory rhythm. Evidence suggests that NK1R is primarily expressed on excitatory neurons (Gray et al., 1999), including those with pre-I activity (Guyenet and Wang, 2001). However, if SP activates excitatory neurons during inspiratory bursts and facilitates their intrinsic bursting properties, one would expect a prolongation of the refractory phase, which would limit rather than promote a frequency increase. Instead, we found that SP does not affect the spiking frequency of either excitatory or inhibitory preBötC neurons during the inspiratory burst phase (Figures 3, 4 and 5). Because local synaptic interactions among interconnected excitatory and inhibitory neurons is a fundamental characteristic of the preBötC network (Ramirez and Baertsch, 2018b), we infer that the changes in neuronal spiking, or lack thereof, that we observed under current clamp conditions reflect local synaptic interactions within the preBötC. Accordingly, we suggest that the balance of excitatory and inhibitory inputs to preBötC neurons is not changed by SP, although voltage-clamp experiments under visual control would be warranted to more directly assess this balance. Nonetheless, an important question is: How is the activity of inspiratory neurons, and presumably the balance of excitation and inhibition, maintained during bursts despite the excitatory effects of SP?

One possibility is that excitatory inputs originating from outside the preBötC could be reduced to prevent SP from causing hyperexcitation during inspiratory bursts. Inspiratory neurons located rostral to the preBötC contribute to the dynamic regulation of breathing frequency. Indeed, transverse brainstem slice preparations that lack rostral inspiratory neurons generate less robust rhythms with shorter refractory periods and faster frequencies than horizontal slices (Baertsch et al., 2019). On the other hand, recruitment of these neurons is associated with increased excitation during inspiratory bursts, a prolonged refractory phase, and consequently slower breathing frequencies (Baertsch et al., 2019). However, under normal conditions, inhibition restrains the rhythm generating ability of these rostral neurons allowing faster, more dynamic breathing. Interestingly, we found that many inhibitory neurons rostral of the preBötC exhibit tonic spiking activity during the IBI, which was moderately increased by SP. In contrast, rostral excitatory neurons did not spike during the IBI, and spiking was only slightly suppressed during inspiratory bursts (Figure 8). Thus, unlike their role in modulating other dynamic breathing responses such as gasping (Baertsch et al., 2019), the contribution of rostral inspiratory neurons to SP-induced frequency increase seems to be relatively modest. Indeed, NK1R is preferentially expressed in the preBötC compared to other regions of the ventral respiratory column (Gray et al., 1999), which may explain the relatively heterogeneous effects of SP on the activity of rostral neurons (Figure 7).

Another possibility is that synchronization of excitatory neurons is reduced during inspiration, thereby preventing hyperexcitation during the burst phase in response to SP. The assembly of inspiratory bursts is known to be stochastic, which is observed as significant variability in the cycle-to-cycle onset time of inspiratory neuron bursts relative to the preBötC population (Carroll and Ramirez, 2013). This variability in activation order, or onset jitter, is related, at least in part, to the sparse functional connectivity within the preBötC network (Carroll and Ramirez, 2013). We found that SP increases the variability of burst onset times among inspiratory neurons located within and rostral to the preBötC. This increase in timing variability was most pronounced among excitatory neurons, specifically those that lack pre-I activity (Figure 6). Thus, although other mechanisms such as depletion of excitatory synaptic vesicles (Rubin et al., 2009) may also contribute, we propose that increased jitter among excitatory neurons impairs synchronization of the network and helps maintain the balance of excitation and inhibition during inspiratory bursts in the presence of SP. Future computational modeling studies will be important to directly test this hypothesis.

The inspiratory burst phase is followed by a period of reduced excitability in the preBötC network. This refractory period is thought to arise from a combination of presynaptic depression (Kottick and Del Negro, 2015) and activation of slow hyperpolarizing current(s) (Baertsch et al., 2018; Krey et al., 2010) in glutamatergic Dbx1 neurons during inspiratory bursts. Indeed, refractoriness is maximal immediately following the inspiratory burst followed by a gradual recovery of excitability (Figure 1), likely as vesicles are recycled and hyperpolarizing conductances are inactivated. This refractory phase manifests experimentally as a period during which the probability of evoking an ectopic inspiratory burst via optogenetic stimulation of Dbx1 neurons is reduced (Figure 1A). However, the refractory period is not absolute as it can be overcome if the stimulus is of sufficient strength (Vann et al., 2018). Using a stimulus procedure consistent with previous reports (Baertsch et al., 2018; Baertsch et al., 2019; Kottick and Del Negro, 2015), we found that SP does not change the duration of the refractory period. This finding is consistent with our demonstration that excitatory neurons do not show an increased activation during the inspiratory burst phase (Figures 3B and 4B). Importantly, the minimum duration of spontaneous inter-burst intervals continues to be restrained by the refractory period in SP (Figure 1). Thus, refractory mechanisms remain an important determinant of breathing frequency in the presence of this neuromodulator. This may be functionally important to prevent excitatory neuromodulators such as SP from driving the respiratory network out of its physiological frequency range. Indeed, neuronal networks must not only be capable of dynamically regulating their frequency, but they must also be able to maintain stability in spite of heterogeneous, intrinsically variable cellular components that receive converging inputs from numerous excitatory neuromodulators (Marder et al., 2014).

As the inspiratory network transitions out of the refractory phase, some excitatory neurons, presumably with more depolarized basal membrane potentials (Butera et al., 1999; Smith et al., 2000), begin to fire action potentials prior to the onset of the next synchronized population burst (Figure 3). The associated synaptic interactions, which may particularly involve Dbx1 neurons (Wang et al., 2014), are thought to stochastically percolate through a network of recurrently connected excitatory neurons (Del Negro et al., 2018). This network-based process, sometimes referred to as the ‘group pacemaker’ hypothesis (Del Negro and Hayes, 2008), likely occurs simultaneously with changes in intrinsic membrane properties such as INaP (Butera et al., 1999; Ramirez et al., 2016; Smith et al., 2000), that together lead to a gradual build-up of excitability within the network. Consequently, the spiking activity of excitatory pre-I neurons between inspiratory bursts exhibits a ramping pattern. We found that the magnitude and slope of this pre-inspiratory ramp, as well as the number of excitatory neurons that have pre-inspiratory activity, is increased by SP. These recruited excitatory neurons also exhibit a pronounced pre-inspiratory ramp, suggesting that they participate in the recurrent excitation process. Together, these effects likely increase the rate of recurrent excitation during this percolation phase to accelerate the onset of the next burst (Figure 9), underling the increase in breathing frequency induced by SP. Moreover, we found that, on a cycle-to-cycle basis, the slope of pre-inspiratory ramp activity is inversely related to the inter-burst interval. Thus, variations in the stochastic percolation of excitation during this phase seem to predict variability in the duration between inspiratory bursts. Our data suggest that, by increasing the rate of recurrent excitation, this process becomes more consistent and breathing irregularity is reduced. We conclude that the dual effects of SP on breathing frequency and stability are primarily a consequence of its effects on the percolation phase of the inspiratory rhythm.

Based on our collective results, we propose a conceptual framework for inspiratory rhythm generation in which three distinct phases, the burst phase, refractory phase, and percolation phase, can be differentially modulated to influence breathing dynamics and stability (Figure 9). Although these three phases seem to be intrinsic to the inspiratory rhythm generator, we anticipate that non-inspiratory neurons and/or respiratory related brain regions outside the preBötC (e.g. Anderson et al., 2016; Fu et al., 2019; Huckstepp et al., 2018; Richter and Smith, 2014) can provide important regulatory control over each phase of the inspiratory cycle, leading to predictable phase-dependent effects on rhythm dynamics. This concept may provide a foundation for understanding breathing in the context of many other physiological and pathological conditions (Bright et al., 2018; Saito et al., 2001), and it may also serve as a guide for understanding the dynamic control rhythm generating networks in general.

Materials and methods

Animals

All experiments and animal procedures were approved by the Seattle Children’s Research Institute’s Animal Care and Use Committee (approved protocol #15981) and conducted in accordance with the National Institutes of Health guidelines. Experiments were performed on neonatal (p6-p12) male and female C57-Bl6 mice bred at Seattle Children’s Research Institute (SCRI). All mice were group housed with access to food and water ad libitum in a temperature controlled (22 ± 1°C) facility with a 12 hr light/dark cycle. For optogenetic experiments, Vglut2Cre (Slc17a6) and VgatCre (Slc32a1) (Vong et al., 2011) homozygous breeder lines were obtained from Jackson Laboratories (Stock numbers 028863 and 016962, respectively). Heterozygous Dbx1CreERT2 mice were donated by Dr. Del Negro (College of William and Mary, VA) and a homozygous breeder line was generated at SCRI. Dbx1CreERT2 dams were plug checked and injected at E9.5 with tamoxifen (24 mg/kg, i.p.) to target preBötC neurons (Kottick et al., 2017). Cre mice were crossed with homozygous mice containing a floxed STOP channelrhodopsin2 fused to an EYFP (Ai32) reporter sequence (JAX #024109). Male and female offspring were chosen at random based on litter distributions.

In vitro medullary horizontal slice preparations

Request a detailed protocol

Horizontal medullary slices were prepared from postnatal day 6–12 mice as described in detail previously (Anderson et al., 2016; Baertsch et al., 2019). Brainstems were dissected in ice cold artificial cerebrospinal fluid (aCSF; in mM: 118 NaCl, 3.0 KCl, 25 NaHCO3, 1 NaH2PO4, 1.0 MgCl2, 1.5 CaCl2, 30 D-glucose) equilibrated with carbogen (95% O2, 5% CO2). When equilibrated with gas mixtures containing 5% CO2 at ambient pressure, aCSF had an osmolarity of 300–312mOSM and a pH of 7.40–7.45. The dorsal surface of each brainstem was secured with super glue to an agar block cut at a ~ 15° angle (rostral end facing up). Brainstems were first sectioned in the transverse plane (200 µm steps) using a vibratome (Leica 1000S) until the VII nerves were visualized. The agar block was then reoriented to position the ventral surface of the brainstem facing up with the rostral end toward the vibratome blade to section the brainstem in the horizontal plane. The blade was leveled with the ventral edge of the brainstem and a single ~850 µm step was taken to create the horizontal slice.

Slices were placed in a custom recording chamber containing circulating aCSF (~15 ml/min) warmed to 30°C. The [K+] in the aCSF was then gradually raised from 3 mM to 8 mM over ~10 min to boost neuronal excitability. Rhythmic extracellular neuronal population activity was recorded by positioning polished glass pipettes (<1 MΩ tip resistance) filled with aCSF on the surface of the slice. Signals were amplified 10,000X, filtered (low pass, 300 Hz; high pass, 5 kHz), rectified, integrated, and digitized (Digidata 1550A, Axon Instruments). The activity of single neurons was recorded using the blind patch clamp approach. Recording electrodes were pulled from borosilicate glass (4–8 MΩ tip resistance) using a P-97 Flaming/Brown micropipette puller (Sutter Instrument Co., Novato, CA) and filled with intracellular patch electrode solution containing (in mM): 140 potassium gluconate, 1 CaCl2, 10 EGTA, 2 MgCl2, 4 Na2ATP, and 10 Hepes (pH 7.2). To map the location of recorded neurons, patch pipettes were backfilled with intracellular patch solution containing 2 mg/ml Alexa Fluor568 Hyrdrazide (ThermoFisher). Neuronal spiking activity was recorded in whole-cell or cell-attached configuration with a multiclamp 700B amplifier in current clamp mode (Molecular Devices, Sunnyvale, CA). Although recordings were performed in ‘current-clamp’ mode, artificial currents were not applied to control spiking activity or to shift membrane potential from its normal resting value. However, in some cases, a small holding current (0 to −20 pA) was applied for leak compensation. Although, in all experiments, recording settings used to obtain baseline values were maintained constant throughout the experiment, in some recordings electrode potential drift precluded analysis of membrane potentials. Extracellular and intracellular signals were acquired in pCLAMP software (Molecular Devices, Sunnyvale, CA). Immediately following electrophysiology experiments, fresh, unfixed slices were imaged to determine the location(s) of the intracellular recording sites.

Optogenetic and pharmacological manipulations

Request a detailed protocol

We chose the Dbx1CreERT2;Rosa26ChR2-EYFP mouse line to assess the refractory period at the network level because it more specifically targets the neurons that are presumed to be rhythmogeneic in the preBötC (Vann et al., 2018). However, we have previously shown that the refractory period in preBötC slices determined using Dbx1CreERT2;Rosa26ChR2-EYFP is not different than the refractory period determined using Vglut2Cre;Rosa26ChR2-EYFP (Baertsch et al., 2018). A 200 µm diameter glass fiber optic (0.24NA) connected to a blue (470 nm) high-powered LED was positioned above the preBötC contralateral to the extracellular electrode and ipsilateral to the intracellular electrode. Power was set ≤1 mW/mm2. To determine the probability of light-evoking inspiratory bursts, 200 ms light pulses were TTL-triggered every 20 s to stimulate Dbx1 neurons (≥50 trials per experiment). Trials were excluded from the analysis if the light pulse occurred during an ongoing spontaneous inspiratory burst. During most intracellular recordings, neurons were classified as excitatory or inhibitory using an optogenetic approach. Because not all excitatory neurons in the preBötC are derived from Dbx1 progenitors, we used a combination of the Vglut2Cre;Rosa26ChR2-EYFP and VgatCre;Rosa26ChR2-EYFP mouse lines to identify individual neurons as excitatory or inhibitory. In VgatCre;Rosa26ChR2-EYFP slices, neurons that depolarized during photostimulation were classified as inhibitory, while those that hyperpolarized or did not respond were presumed to be excitatory. Because a depolarizing response to stimulation of excitatory neurons could be driven synaptically instead of from channelrhodopsin2 expression directly, in Vglut2Cre;Rosa26ChR2-EYFP and Dbx1CreERT2;Rosa26ChR2-EYFP slices, neurons were classified as excitatory or inhibitory based on the presence or absence of a depolarizing response to light, respectively, following pharmacological blockade of excitatory AMPAR- and NMDAR-dependent synaptic transmission (20 µM CNQX, 20 µM CPP).

Substance P was purchased from Tocris (Cat#: 1156), diluted in water to a concentration of 5 mM, and stored in stock aliquots at −20°C. In all experiments, a ~ 10 min baseline period of stable inspiratory activity was recorded prior to bath application of substance P to 0.5–1.0 µM. Intracellular and extracellular population activity was then recorded for >10 min prior to washout into fresh aCSF.

Microscopy

Request a detailed protocol

2.5X brightfield and epifluorescent images of the dorsal surface of horizontal slices were acquired on a Leica DM 4000 B epifluorescence microscope. Following intracellular recording experiments, the location of each recorded neuron within the horizontal slice was immediately quantified by overlaying the brightfield and an epifluorescent image of Alexa Fluor 568 labeled cell(s). Images were then traced in powerpoint and overlaid with the midline and rostral edge (VII nerve) aligned to show the relative locations of recorded cells (see Figures 2A and 7A).

Statistical analysis

Request a detailed protocol

Data was analyzed using Clampfit software (Molecular Devices). Integrated population bursts and individual action potentials were detected using Clampfit’s peak-detection analysis. Statistical analyses were performed using GraphPad Prism6 software and are detailed for each experiment in the Figure Legends. Groups were compared using appropriate two-tailed t-tests, or one-way ANOVAs with Bonferonni’s multiple comparisons post hoc tests. Welch’s correction was used for unequal variances where appropriate. Non-linear regression analysis was used to determine differences between probability curves (Figure 1). Linear regression analyses were used to determine relationships between inter-burst intervals and pre-inspiratory ramp slope (Figures 3G and 5F). Differences were considered significant at p<0.05 and data are displayed as individual data points with overlaid means ± SE. Significance is denoted in the figures as follows: ****p<0.0001; ***p<0.001; **p<0.01; *p<0.05. Experimenters were not blinded during data collection or analysis.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
    Neurobiology of substance P and the NK1 receptor
    1. PW Mantyh
    (2002)
    The Journal of Clinical Psychiatry 63:6–10.
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
    SIDS Sudden Infant and Early Childhood Death: The Past, the Present and the Future
    1. JM Ramirez
    2. SC Ramirez
    3. TM Anderson
    (2018)
    Sudden Infant Death Syndrome, Sleep, and the Physiology and Pathophysiology of the Respiratory Network, SIDS Sudden Infant and Early Childhood Death: The Past, the Present and the Future, University of Adelaide Press.
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
  81. 81
  82. 82
  83. 83
  84. 84
  85. 85
  86. 86
  87. 87
  88. 88
  89. 89
  90. 90
  91. 91
  92. 92
  93. 93
    Multiple receptors mediate the modulatory effects of serotonergic neurons in a small neural network
    1. B Zhang
    2. RM Harris-Warrick
    (1994)
    The Journal of Experimental Biology 190:55–77.
  94. 94
  95. 95

Decision letter

  1. Ronald L Calabrese
    Senior and Reviewing Editor; Emory University, United States
  2. Muriel Thoby-Brisson
    Reviewer; CNRS Université de Bordeaux, France
  3. Jeffrey C Smith
    Reviewer; National Institute of Neurological Disorders and Stroke, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This is a technically impressive study that effectively combines optogenetic, electrophysiological, and pharmacological techniques to analyze cell type specific neuromodulatory actions of substance P (SP) on excitatory and inhibitory neurons in the inspiratory rhythm generator within the mouse preBötzinger complex in vitro. Important novel results are presented suggesting that the demonstrated effects of SP in augmenting inspiratory frequency and rhythm stability are a consequence of its selective effects on the pre-inspiratory recurrent excitation or "percolation" phase of the inspiratory rhythmic cycle where SP is shown to augment the spiking activity and recruit of excitatory pre-I spiking neurons. The important conceptual advance is that neuromodulation of inspiratory cycle dynamics at least in vitro may be understood by considering the respiratory cycle to consist of three distinct phases (inspiratory burst, refractory, and percolation phases) that can be differentially regulated by physiologically important neuromodulators, with predictable phase-dependent effects on rhythm dynamics. The results presented from this study with SP effectively support and emphasize this concept.

Decision letter after peer review:

Thank you for submitting your article "Insights into the dynamic control of breathing revealed through cell-type-specific responses to substance P" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Ronald Calabrese as the Senior and Reviewing Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Muriel Thoby-Brisson (Reviewer #1); Jeffrey C Smith (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Essential revisions:

All the major concerns of the expert reviewers should be addressed in revision. To help guide this process we offer a few consensus comments.

1) There was general concern that the authors cannot really make definitive statements about how SP influences excitatory/inhibitory balance without voltage clamp experiments that directly assess excitatory and inhibitory synaptic input. In the absence of such experiments, it would be satisfactory if the authors present this concept as an inference/speculation with adequate discussion in the manuscript, including further clear discussion about why they think analyses of modulation of more rostral excitatory and inhibitory neuron activity is critical in terms of understanding circuit interactions relevant to the problem, and also indicate the need to perform more comprehensive analyses of synaptic interactions based on direct measurements of synaptic currents/potentials to resolve this issue. Some measurements in the transverse slice would also help clarify if the more complex distributed circuit interactions that the authors seem to be invoking are important. According to the authors' model, the preBötC should be "over-excited" with substance P and measurements of this should include a change burst rate and amplitude. These changes should be then compared to the same dose of SP in the horizontal preparation. The model predicts that SP will have different magnitude of effects in these two networks.

2) There is general concern that the current clamp data needs clear supporting data. (i) Reviewer #2's requirements to provide much more detail about the recordings is essential. (ii) The authors must demonstrate that the data in whole cell recording mode is reproduced by a reasonable number of cell-attached recordings. (iii) The authors also must be clear about how they are defining burst onset. Is it based on the first action potential?

3) The was concern about the issue of "jitter" of burst onset and how the authors measure jitter and about their interpretation of it as a potential mechanism to prevent hypersynchronization. The authors show data that indicating that SP doesn't change the burst duration or spiking frequency for individual cells, but instead increases the "jitter" of burst onset. The authors claim that this is a mechanism to prevent hypersynchronization (subsection “SP increases stochasticity among excitatory preBӧtC neurons during inspiratory bursts”), which is only one potential model and interpretation. If this is included, it must be clearly stated as a model and be limited to the Discussion in the paper. Furthermore, if the burst onset is more variable, then we should expect to see overall changes in burst duration, yet no data on burst characteristics is documented. This should be described and reported.

Reviewer #1:

The very elegant work proposed by Baertsch and Ramirez aims to decipher the cellular mechanisms by which Substance P (SP, a well-known modulator of the respiratory network activity) dynamically controls respiratory rhythmogenesis through cell-type specific effects. They found that SP modulates breathing frequency and stability through an action during a very specific phase of the respiratory cycle: the percolation phase (recurrent excitation phase), but not by changing the inspiratory phase nor the refractory period. In addition, they provide evidences for a cell-specific action as this modulatory mechanism involves preferentially excitatory respiratory neurons capable of exhibiting pre-inspiratory spiking under SP influence without any significant contribution of inhibitory or pre-inspiratory neurons. Thus, they conclude that phase-specific and cell-type specific action of SP is a key mechanism to dynamically control the dynamic of the respiratory network.

Overall the paper is well written, easy to follow and data are nicely illustrated and convincing. I just have a few comments:

1) As stated in the Discussion it is considered that rhythmogenic mechanisms within the respiratory network, supported by the group-pacemaker hypothesis, involve interactions between synaptic and specific membrane properties (such as pacemaker properties) among respiratory neurons. Recording respiratory pacemaker neurons remain challenging due to their poor representativeness. But I was wondering whether the authors had the opportunity to perform their experiments in such a type of respiratory neurons. This could be particularly interesting as, to my knowledge, pacemaker neurons rarely exhibit pre-inspiratory discharge. I would be curious to know whether SP would change this pattern in pacemaker neurons and consequently whether modulation of this specific cell-type could also account for the overall modulatory effect of SP.

2) Do the authors have any indications on the cellular mechanisms underlying the change in discharge pattern of the excitatory neurons in the presence of SP? Is it an increase in recurrent synaptic inputs, changes in membrane properties associated to specific conductances, or both? In other words, do the authors possess data in voltage-clamp mode to reveal possible modulation of either of these features?

Reviewer #2:

This is a technically impressive study that effectively combines optogenetic, electrophysiological, and pharmacological techniques to analyze cell type specific neuromodulatory actions of substance P (SP) on excitatory and inhibitory neurons in the inspiratory rhythm generator within the mouse preBötzinger complex in vitro. Important novel results are presented suggesting that the demonstrated effects of SP in augmenting inspiratory frequency and rhythm stability are a consequence of its selective effects on the pre-inspiratory recurrent excitation or "percolation" phase of the inspiratory rhythmic cycle where SP is shown to augment the spiking activity and recruit of excitatory pre-I spiking neurons. The important conceptual advance is that neuromodulation of inspiratory cycle dynamics at least in vitro may be understood by considering the respiratory cycle to consist of three distinct phases (inspiratory burst, refractory, and percolation phases) that can be differentially regulated by physiologically important neuromodulators, with predictable phase-dependent effects on rhythm dynamics. The results presented from this study with SP effectively support and emphasize this concept.

1) The authors suggest that the inspiratory phase is regulated by the balance of inhibition, excitation, and intrinsic bursting properties of active neurons, and that in the present experiments with SP, the balance of excitation-inhibition is maintained. While the former is certainly the case in general, excitatory and inhibitory inputs to any neuron in the present experiments have not been directly measured (e.g., by voltage clamp measurements) to know for sure about this balance. Similarly, the authors state, but have not directly shown, that SP shifts the balance of excitation and inhibition among rostral inspiratory neurons, which implies that they have performed some analyses of synaptic interactions among neurons in the network, including in preBötC circuits. This needs to be qualified further in the text. I agree that it is important that the data presented show that SP does not change spiking frequency of excitatory inspiratory neurons during the burst phase as well as activity of inhibitory neurons (and by inference probably local inhibitory interactions) during this phase. Perhaps the modulation of leak conductances by SP (noted in the Introduction) can cause depolarizing shifts in neuronal membrane potentials to promote pre-I spiking without altering overall spiking profiles during the inspiratory burst phase.

2) The baseline membrane potentials of neurons before (and therefore) after SP are not indicated in any of the figures (Figure 2, 3, 4, 5) with whole-cell recordings and provided in the text. This information is important and should be provided to understand why the subpopulation of pre-I excitatory neurons is active during the inter-burst interval (presumably they have more depolarized baseline potentials?), the amount of depolarization if any with SP, and why specific neuron types are recruited or not to spike during the IBI interval. It has been postulated for many years that neurons with more depolarized basal membrane potentials exhibit pre-I spiking activity that is important for rhythmogenesis by providing synaptic excitation during the pre-I spiking phase (e.g., Butera et al., 1999).

Reviewer #3:

The authors confirm that SP decreases the interval between respiratory bursts and show this shortening is due to shrinking of a later portion of the interval that they dub the "percolation" phase. This point is cleanly and clearly demonstrated in Figure 1, but this contribution alone does not substantially deepen our understanding of respiratory rhythm generation and modulation. Remaining experiments are purely descriptive, flawed with technical issues, and do not provide the proposed mechanistic understanding. Given Yeh et al., 2017, defined the molecular mechanism for SP modulation of breathing, this work does not provide any advancement necessitating publication to a general audience as is.

1) One primary focus of the paper is to record from previously defined populations of pre-inspiratory and inspiratory excitatory neurons and inspiratory inhibitory neurons to determine if SP has cell-type specific effects. However, a major technical flaw throughout the manuscript is the use of current clamp recordings instead of cell attached recordings. Although the authors claim to have done both in the Materials and methods, no cell-attached recordings are shown. In current clamp, the experimenter has complete control of the cells resting membrane potential and can therefore easily make any cell look pre-inspiratory or inspiratory. With this flaw in mind, the authors cannot then claim to have identified cell-type specific modulation by substance P.

2) In Figure 2E/F and Figure 6 the authors focus attention to the effect of substance P on the onset of neural bursting activity versus preBötC population bursting. They claim that SP increases variability in this timing for pre-inspiratory neurons and that this is key to ensure the preBötC burst size does not increase. As shown in Figure 2E, the onset of the burst appears arbitrarily defined and as shown, would be nearly impossible to definitively decide for a pre-inspiratory neuron. Furthermore, this point is influenced by what membrane potential the experimenter is holding the neuron at in current clamp.

3) Many years of papers have shown that SP increases the preBötC rhythm in a coronal slice preparation. This result shows that brainstem regions rostral to the preBötC are not required for SP modulation of preBötC activity. However, in Figures 7/8 the authors record from neurons rostral to the preBötC and observe changes in activity after SP. They claim these changes cause a shift to increased inhibitory neuron activity that helps to maintain preBötC excitation. This is never directly measured and there are too many logical leaps here, such as:

i) Do these rostral neurons project to the preBötC?

ii) Are there decreased excitatory inputs into the preBötC?

iii) Does bursting in the preBötC become "over-excited" in a coronal slice?

iv) If SP acts through NALCN to depolarize cells, how is it decreasing excitatory neural activity? Given the lack of understanding, these figures add confusion.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for submitting your article "Insights into the dynamic control of breathing revealed through cell-type-specific responses to substance P" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Ronald Calabrese as the Senior and Reviewing Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Muriel Thoby-Brisson (Reviewer #1); Jeffrey C Smith (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

The authors have done an admirable job responding to the previous review. One reviewer asks for further clarifications, which should be seriously considered. When the authors re-submit no external further review will be necessary.

Essential revisions:

The revised manuscript has nicely incorporated the feedback from the three essential revisions. The inclusion of the membrane potential in the featured recordings as well as the number of cell-attached recordings performed addresses essential revision comment #2. Although the N is low for the number of cell-attached recordings, this data is important to confirm that a subset of excitatory-inspiratory neurons display pre-I activity only after SP application. Additionally, as the authors point out, the described membrane potentials suggest that the recruited neurons are likely those with higher resting membrane potentials before SP application. With this in mind, the authors should consider providing the membrane potential before and after SP for all of the recorded neurons of the various neural types. While the depolarization and recruitment of cells into pre-I firing is interesting, it will be essential in the future to determine if the recruitment is required for the SP induced shortening of the percolation phase. Alternative models still remain, like the increased activity of pre-I neurons alone drives the shortening of the percolation phase. The additional description of the methods will enhance reproducibility by others and the modified discussion nicely handles how the various observations made in the manuscript may result in the observed selective effect of SP to shorten the percolation phase and addresses essential revision comments #1 and #3.

https://doi.org/10.7554/eLife.51350.sa1

Author response

Essential revisions:

All the major concerns of the expert reviewers should be addressed in revision. To help guide this process we offer a few consensus comments.

1) There was general concern that the authors cannot really make definitive statements about how SP influences excitatory/inhibitory balance without voltage clamp experiments that directly assess excitatory and inhibitory synaptic input. In the absence of such experiments, it would be satisfactory if the authors present this concept as an inference/speculation with adequate discussion in the manuscript, including further clear discussion about why they think analyses of modulation of more rostral excitatory and inhibitory neuron activity is critical in terms of understanding circuit interactions relevant to the problem, and also indicate the need to perform more comprehensive analyses of synaptic interactions based on direct measurements of synaptic currents/potentials to resolve this issue. Some measurements in the transverse slice would also help clarify if the more complex distributed circuit interactions that the authors seem to be invoking are important. According to the authors' model, the preBötC should be "over-excited" with substance P and measurements of this should include a change burst rate and amplitude. These changes should be then compared to the same dose of SP in the horizontal preparation. The model predicts that SP will have different magnitude of effects in these two networks.

We have added discussion of the limitation of current clamp and that we are inferring synaptic interactions from spiking activity. Follow up voltage clamp experiments are also suggested in the text. See responses to reviewers 1 and 2 for further discussion.

We have not included additional experiments in transverse slices because (i) based on the variable effects of SP on rostral neurons we observed in the horizontal slice, we do not anticipate that changes in the activity of the rostral inspiratory column are a large contributor to SP-induced frequency facilitation. This is consistent with the known distribution of NK1 receptors, the expression of which is less abundant outside the preBötC. The Discussion has been revised to better emphasize this. And (ii) the horizontal slice and transverse slice have different baseline properties including the duration of the refractory period, burst amplitude, burst duration and burst frequency (Baertsch et al., 2019), which may make interpretation of the suggested experiments challenging. Moreover, adding any meaningful experiment in transverse slices would mean repeating all the reported experiments in this preparation. See response to reviewer 3 for further discussion.

2) There is general concern that the current clamp data needs clear supporting data. (i) Reviewer #2's requirements to provide much more detail about the recordings is essential. (ii) The authors must demonstrate that the data in whole cell recording mode is reproduced by a reasonable number of cell-attached recordings. (iii) The authors also must be clear about how they are defining burst onset. Is it based on the first action potential?

We have addressed reviewer 2’s concerns, and now include Vm information for representative whole-cell recordings. We have added the numbers of whole-cell vs. cell-attached recordings in the text and now include example cell-attached recordings in Figure 2—figure supplement 1. We now clearly define how burst onset was determined in both pre-I and non pre-I neurons in the Results section.

3) The was concern about the issue of "jitter" of burst onset and how the authors measure jitter and about their interpretation of it as a potential mechanism to prevent hypersynchronization. The authors show data that indicating that SP doesn't change the burst duration or spiking frequency for individual cells, but instead increases the "jitter" of burst onset. The authors claim that this is a mechanism to prevent hypersynchronization (subsection “SP increases stochasticity among excitatory preBӧtC neurons during inspiratory bursts”), which is only one potential model and interpretation. If this is included, it must be clearly stated as a model and be limited to the Discussion in the paper. Furthermore, if the burst onset is more variable, then we should expect to see overall changes in burst duration, yet no data on burst characteristics is documented. This should be described and reported.

We now present this explicitly as a hypothesis in the Discussion. We also analyzed changes in the duration of population bursts in SP and found that there is a small but significant increase in burst duration as well as burst half-width. This information is now included in the results and Figure 6C.

Reviewer #1:

[…] Overall the paper is well written, easy to follow and data are nicely illustrated and convincing. I just have a few comments:

1) As stated in the Discussion it is considered that rhythmogenic mechanisms within the respiratory network, supported by the group-pacemaker hypothesis, involve interactions between synaptic and specific membrane properties (such as pacemaker properties) among respiratory neurons. Recording respiratory pacemaker neurons remain challenging due to their poor representativeness. But I was wondering whether the authors had the opportunity to perform their experiments in such a type of respiratory neurons. This could be particularly interesting as, to my knowledge, pacemaker neurons rarely exhibit pre-inspiratory discharge. I would be curious to know whether SP would change this pattern in pacemaker neurons and consequently whether modulation of this specific cell-type could also account for the overall modulatory effect of SP.

Based on our previous observations (Carroll and Ramirez, 2013), we assume that many, but not all, pre-I neurons are in fact bursting pacemaker neurons. In the study by Carroll and Ramirez, we characterized the cycle-to-cycle activation of more than 600 neurons that were recorded extracellularly (i.e. without manipulating their membrane potential). We demonstrate that the probability of discharging before the population burst (i.e. pre-I) in the network, was strongly related (<0.0001) to overall spike rate and interspike interval metric of burstiness in synaptic block (<0.0001). We also found that mean spike rate was significantly correlated with mean spike rate after synaptic block – which meant that bursting neurons that have high spike rates after synaptic block have also higher spike rates in the network and were more likely to have pre-inspiratory activity. Of course, in this extracellular and pre-optogenetic study we did not discriminate between inhibitory and excitatory neurons. In the Carroll and Ramirez study we also demonstrated that not all bursting pacemaker neurons had pre-inspiratory activity, and in fact many bursting neurons also discharged later in the burst. Based on the present study, which showed that inhibitory neurons have no pre-inspiratory activity, it is tempting to speculate that bursting neurons that discharged at or after burst onset included inhibitory neurons, while those with pre-I activity were more likely excitatory neurons, but as stated above, this was not determined in this study. Assuming that bursting pacemaker neurons that discharged before the population bursts were excitatory neurons, one could also speculate that ectopic bursting in pacemakers may contribute to mechanisms of recurrent excitation and the percolation phase. Indeed, previous work from our lab has shown that SP can facilitate bursting in both INaP and ICAN pacemaker neurons, which we include in the Discussion (second paragraph). But, since in the present study we did not specifically test for pacemaker properties under synaptically isolated conditions, we can only infer and not make specific conclusions about how intrinsic bursting contributes to the pre-I percolation phase. What, our data indicate however, is that despite the known modulation of intrinsic bursting, substance P does not change spiking activity of excitatory neurons during bursts when respiratory neurons are integrated within the network.

2) Do the authors have any indications on the cellular mechanisms underlying the change in discharge pattern of the excitatory neurons in the presence of SP. Is it an increase in recurrent synaptic inputs, changes in membrane properties associated to specific conductances, or both? In other words, do the authors possess data in voltage-clamp mode to reveal possible modulation of either of these features?

This is an important question. Since it has been previously shown that some preBӧtC neurons depolarize in response to SP (e.g. Yeh et al., 2017; Pena and Ramirez, 2004) including pacemakers (Pena and Ramirez, 2002), we anticipate that intrinsic depolarization of some of these neurons can facilitate spiking activity, and thereby promote mechanisms of recurrent excitation But, in addition to INaP and ICAN, it is also likely that the Nalcn current (Yeh et al., 2017) as well as NMDA mechanisms (Pena and Ramirez, 2004) can contribute to this depolarization. While unraveling which of all the contributing currents is most important is an interesting question, we feel this is beyond the scope of the present study. Thus, we did not apply substance P following blockade of synaptic transmission to assess intrinsic membrane responses. Instead, our aim was to understand the consequences of such a depolarization for the activity of the network as a whole. Hence, we focused on the spiking activity of individual neurons that were identified as excitatory or inhibitory. Our rationale was that a depolarization that does not result in a change in spiking activity will have negligible consequences for the function of the network, whereas a neuron that changes its spiking activity will have consequences for network activity, irrespective of which underlying current caused the increased spiking activity, and irrespective of whether the neuron is intrinsically modulated by Substance P or indirectly via synaptic inputs.

Moreover, we expect bath application of SP to have a tonic influence on NK1R expressing neurons – e.g. causing a persistent shift in Vm under steady state conditions. Thus, although intrinsic changes in membrane potential and increased recurrent synaptic activity are likely linked as described above, we anticipate that the time-dependent changes in spiking or pre-I ramp during the IBI is generated by a build-up of recurrent synaptic activity. Indeed, we found that many rostral inhibitory/inspiratory neurons have spiking during the IBI, but they do not exhibit a time-dependent pre-I ramp, suggesting they are not involved in the recurrent excitation mechanism.

We did not perform voltage clamp experiments to directly assess synaptic inputs. This was intentional because we felt that using the VC approach would not significantly contribute to our understanding of cell-type-specific inputs, since we assume that EPSCs in any given neuron will likely originate from many different neurons. Hence, it would be difficult to quantify which inputs would come from pre-inspiratory neurons vs. from inspiratory neurons without a pre-inspiratory component. Also, just as the post-synaptic targets are unknown when monitoring spiking activity, the locations of pre-synaptic neurons are unknown when monitoring EPSCs in voltage clamp. Instead, we assume that the cell-type specific changes in spiking we observed in current clamp will involve changes in synaptic inputs in other preBӧtC neurons since this network is known to contain prevalent recurrent connections. Indeed, this is a requirement for the network to synchronize. To what extent the changes in synaptic input are driven by changes in leak currents, potassium currents, persistent sodium currents and TRPM, TRPC currents is an important question, but again, we do not expect to obtain a simple answer as all of those currents will somehow contribute. Thus, since we don’t expect a simple answer, we feel that this issue is beyond the scope of the present study.

Reviewer #2:

[…]

1) The authors suggest that the inspiratory phase is regulated by the balance of inhibition, excitation, and intrinsic bursting properties of active neurons, and that in the present experiments with SP, the balance of excitation-inhibition is maintained. While the former is certainly the case in general, excitatory and inhibitory inputs to any neuron in the present experiments have not been directly measured (e.g., by voltage clamp measurements) to know for sure about this balance. Similarly, the authors state, but have not directly shown, that SP shifts the balance of excitation and inhibition among rostral inspiratory neurons, which implies that they have performed some analyses of synaptic interactions among neurons in the network, including in preBötC circuits. This needs to be qualified further in the text. I agree that it is important that the data presented show that SP does not change spiking frequency of excitatory inspiratory neurons during the burst phase as well as activity of inhibitory neurons (and by inference probably local inhibitory interactions) during this phase. Perhaps the modulation of leak conductances by SP (noted in the Introduction) can cause depolarizing shifts in neuronal membrane potentials to promote pre-I spiking without altering overall spiking profiles during the inspiratory burst phase.

It is true that we have not directly assessed the balance of excitatory and inhibitory synaptic input during inspiratory bursts. Indeed, as the reviewer points out, we infer that changes in synaptic interactions reflect the changes in spiking activity we observed under current-clamp conditions. We now consider this limitation throughout the text, and specifically address it in the second paragraph of the Discussion.

However, as the reviewer also points out, it is generally accepted that excitatory and inhibitory preBӧtC neurons are interconnected and, through local synaptic interactions, synchronize together during inspiratory bursts. Indeed, the ability of the isolated preBӧtC network to synchronize depends on local synaptic interactions among excitatory neurons; this is true regardless of which hypothesis of rhythm generation one prescribes to (e.g. “intrinsic bursting”, “group pacemaker”, “burstlet”, or hybrid). Furthermore, in preBӧtC slices (i) inhibitory neurons have inspiratory activity that is dependent on excitatory synaptic transmission (thus, inhibitory neurons receive excitatory inputs during bursts), (ii) the activity of excitatory neurons during bursts is enhanced following blockade of inhibitory synaptic interactions (thus, excitatory neurons receive inhibitory inputs during bursts), and (iii) inspiratory neurons exhibit EPSCs and IPSCs during bursts.

The reviewer’s suggestion that the modulation of leak conductances by SP causes depolarization and promotes an increase spiking during the IBI is exactly what we presume to be true. However, in this study we did not apply substance P following blockade of synaptic transmission to directly assess intrinsic membrane responses to SP and, therefore, we do not want to make any specific claims to correlate membrane depolarization with changes in spiking activity. For more on this, please see our responses to reviewer 1.

2) The baseline membrane potentials of neurons before (and therefore) after SP are not indicated in any of the figures (Figure 2, 3, 4, 5) with whole-cell recordings and provided in the text. This information is important and should be provided to understand why the subpopulation of pre-I excitatory neurons is active during the inter-burst interval (presumably they have more depolarized baseline potentials?), the amount of depolarization if any with SP, and why specific neuron types are recruited or not to spike during the IBI interval. It has been postulated for many years that neurons with more depolarized basal membrane potentials exhibit pre-I spiking activity that is important for rhythmogenesis by providing synaptic excitation during the pre-I spiking phase (e.g., Butera et al., 1999).

We now include Vm information for all representative whole-cell recordings. However, a detailed analysis of Vm information and discussion regarding SP-induced changes in Vm were omitted in the original submission because: (i) Some of the data were obtained in cell-attached configuration; (ii) SP was not applied under conditions with synaptic transmission blocked. Therefore, it cannot be determined whether changes in Vm are due to an intrinsic response of the neuron to SP and/or by changes in the synaptic inputs to that neuron; (iii) Vm changes in the cell body may not reflect changes in the dendrites. Thus, the presence or absence of Vm changes are difficult to interpret, a challenge that we faced when characterizing the role of the Ih current e.g. (iv) SP-induced changes in Vm of preBӧtC neurons have been reported previously using a more appropriate experimental design (e.g. Pena and Ramirez, 2004; Gray et al., 1999), which we have cited in our paper.

We agree with the reviewer that pre-I excitatory neurons likely have a more depolarized basal Vm, as suggested by previous studies (Smith et al., 2000; Butera et al., 1999), and now include this important point in the Discussion (sixth paragraph).

Reviewer #3:

The authors confirm that SP decreases the interval between respiratory bursts and show this shortening is due to shrinking of a later portion of the interval that they dub the "percolation" phase. This point is cleanly and clearly demonstrated in Figure 1, but this contribution alone does not substantially deepen our understanding of respiratory rhythm generation and modulation. Remaining experiments are purely descriptive, flawed with technical issues, and do not provide the proposed mechanistic understanding. Given Yeh et al., 2017, defined the molecular mechanism for SP modulation of breathing, this work does not provide any advancement necessitating publication to a general audience as is.

It is unfortunate that the reviewer did not see much conceptual advance in our study. Although we agree with the reviewer that the study by Yeh et al. was an important contribution to our understanding of SP modulation, we are surprised that the reviewer claims that our intracellular recording experiments are “purely descriptive”, “flawed”, and “do not provide mechanistic understanding”.

Yeh et al., 2017, demonstrated that breathing is destabilized in Nalcn KO mice and that they do not respond to SP. It is a stretch to claim that by describing the consequences of the Nalcn knock-out they defined “the” mechanism for SP modulation, as this description does not explain how activation of Nalcn causes changes to breathing. Similarly, our previously published finding that SP modulates INaP and ICAN is an interesting description, but also does not explain how SP increases frequency. An NK1R KO mouse won’t respond to SP either, but should NK1R activation be considered “the” mechanism for SP modulation of breathing frequency? While molecular components such as NK1R or Nalcn may be required for SP modulation, this does not explain the network changes that actually underlie modulation of breathing frequency and stability. Here, we provide a detailed network mechanism, complete with the essential cellular level recordings required to understand network function. We hope the reviewer will agree, that defining a molecular mechanism is only one aspect in unraveling neuromodulation, and that it is essential to integrate all levels in order to fully understand how neural networks change their activity. Otherwise, understanding breathing would have been solved decades ago: as the reviewer knows, TTX blocks breathing which indicates that TTX-sensitive sodium channels are required for breathing generation.

Our cellular and network level analysis strongly support a novel framework for understanding the dynamic regulation of breathing: That there are three inspiratory phases that can be independently modulated to control the frequency and stability of the rhythm. As reviewer 2 pointed out correctly, this will raise many important questions, including follow-up studies to explore how post inspiratory or late expiratory activity influences these three phases in the intact animal. Or more specifically: does the three-phase activity of inspiratory rhythmogenesis serve as a scaffold for the three-phase breathing rhythm? Addressing these questions is conceptually interesting, but beyond the scope of this study. What we know so far is that inhibitory preBӧtC neurons regulate frequency by controlling activity during the burst phase. Here we show that SP specifically regulates frequency by affecting the activity of excitatory neurons during the percolation phase of the rhythm (and we go on to identify the underling network mechanisms that allow the tonic modulatory effects of SP to have phase specific effects on network function). In doing so we provide an important conceptual advance that we feel does in fact “substantially deepen our understanding of respiratory rhythm generation and modulation”.

1) One primary focus of the paper is to record from previously defined populations of pre-inspiratory and inspiratory excitatory neurons and inspiratory inhibitory neurons to determine if SP has cell-type specific effects. However, a major technical flaw throughout the manuscript is the use of current clamp recordings instead of cell attached recordings. Although the authors claim to have done both in the Materials and methods, no cell-attached recordings are shown. In current clamp, the experimenter has complete control of the cells resting membrane potential and can therefore easily make any cell look pre-inspiratory or inspiratory. With this flaw in mind, the authors cannot then claim to have identified cell-type specific modulation by substance P.

Unfortunately, the reviewer has misunderstood our experiments. Thus, we have added more information to the Materials and methods section to clarify.

Current-clamp allows the experimenter to manipulate current to observe changes in voltage. But, this does not mean that it is necessary to apply current to a cell while recording in current clamp mode. In fact, the cell’s true “resting” membrane potential would most certainly not be determined while applying an artificial current. Although many of our recordings were performed in “current-clamp” mode, artificial currents were not applied to control spiking activity or to shift membrane potential from its normal resting value. Moreover, in all experiments, recording settings used to obtain baseline values were maintained constant throughout the experiment – i.e. for the duration of SP application. Thus, changes in spiking induced by SP cannot be attributed to us making “any cell look pre-inspiratory or inspiratory”.

Further, we assume that this reviewer is aware that we are not the first to report that many inspiratory neurons have pre-inspiratory spiking activity (including studies using sharp electrode recording and computational modelling approaches). This has been known for decades, and we assume that the reviewer doesn’t want to imply that all the published electrophysiological studies that characterized the different respiratory discharge patterns using current clamp, have manipulated these patterns at will. In any case, to address the reviewer’s question we now include the requested cell attached recordings, shown in Figure 2—figure supplement 1.

2) In Figure 2E/F and Figure 6 the authors focus attention to the effect of substance P on the onset of neural bursting activity versus preBötC population bursting. They claim that SP increases variability in this timing for pre-inspiratory neurons and that this is key to ensure the preBötC burst size does not increase. As shown in Figure 2E, the onset of the burst appears arbitrarily defined and as shown, would be nearly impossible to definitively decide for a pre-inspiratory neuron. Furthermore, this point is influenced by what membrane potential the experimenter is holding the neuron at in current clamp.

It seems that the reviewer misunderstood our claims. Yes, we observed highly significant changes in the timing variability between the onset of individual neuronal bursts and the onset of integrated activity of the population. This is an issue that we are very interested in, as much can be learned from quantitatively characterizing the striking onset variability of the respiratory network (Carroll and Ramirez, 2013; Carroll et al., 2013). In the Results, we now more specifically define how onset times were determined (subsection “Inspiratory spiking patterns of excitatory and inhibitory neurons in the preBӧtC”, last paragraph). We found that SP increases onset jitter specifically in neurons without pre-I activity and does not change onset jitter in pre-I neurons, which is large under baseline conditions. The large onset jitter in pre-I neurons confirms our extracellular recordings as published in Carroll and Ramirez, 2013 – i.e. in the absence of any potential current clamp manipulation. Onset times in pre-I neurons were not “arbitrarily defined”, and regardless, our data indicate, and we suggest, that this subgroup of neurons does not contribute to the increased stochasticity we observed. We propose as a hypothesis that this increase in stochasticity among excitatory preBӧtC neurons may be one mechanism that prevents these neurons from becoming hyperexcited during inspiratory bursts. However, the observation of burst onset stochasticity is not new and has been previously described not only by us, but also others (e.g. Carroll and Ramirez, 2013; Smith et al., 2000) without using intracellular recordings.

3) Many years of papers have shown that SP increases the preBötC rhythm in a coronal slice preparation. This result shows that brainstem regions rostral to the preBötC are not required for SP modulation of preBötC activity. However, in Figures 7/8 the authors record from neurons rostral to the preBötC and observe changes in activity after SP. They claim these changes cause a shift to increased inhibitory neuron activity that helps to maintain preBötC excitation. This is never directly measured and there are too many logical leaps here, such as:

i) Do these rostral neurons project to the preBötC?

ii) Are there decreased excitatory inputs into the preBötC?

iii) Does bursting in the preBötC become "over-excited" in a coronal slice?

iv) If SP acts through Nalcn to depolarize cells, how is it decreasing excitatory neural activity? Given the lack of understanding, these figures add confusion.

Yes, many papers have shown that SP increases frequency in the transverse slice, including from our lab (e.g. Pena and Ramirez, 2002; Telgkamp et al., 2002). We fully agree with the reviewer that this indicates that the rostral area is not required. We are puzzled by the reviewer’s comment, since nowhere in the manuscript do we suggest that rostral areas are required for SP modulation of the inspiratory rhythm. However, one cannot automatically assume that if an area is not required, that it does not contribute to a certain phenomenon. To emphasize this logic, here is one example: locomotor and breathing activity do not require the neocortex, but certainly, the neocortex plays an important role in these behaviors. In fact, our recordings from rostral inspiratory neurons showed relatively heterogenous responses to SP, which was explicitly stated in our original submission. This is an important result, however, because it suggests not only that the rostral area is not required, but also that any contributions from rostral inspiratory neurons to SP-induced frequency facilitation are relatively modest. This new insight could not be revealed using transverse slice preparations. In the revised text, we propose that the modest contribution of the rostral area may be due to the enriched expression of NK1R in the preBӧtC relative to the rest of the VRC. Indeed, NK1R is a classical “marker” of the preBӧtC (Gray et al., 1999).

Since we do not suggest that changes rostral are a significant contributor to SP-induced modulation of breathing, we feel that it is unnecessary to address the “logical leaps” implied by the reviewer.

[Editors’ note: the author responses to the re-review follow.]

The authors have done an admirable job responding to the previous review. One reviewer ask for further clarifications, which should be seriously considered. When the authors re-submit no external further review will be necessary.

Essential revisions:

The revised manuscript has nicely incorporated the feedback from the three essential revisions. The inclusion of the membrane potential in the featured recordings as well as the number of cell-attached recordings performed addresses essential revision comment #2. Although the N is low for the number of cell-attached recordings, this data is important to confirm that a subset of excitatory-inspiratory neurons display pre-I activity only after SP application. Additionally, as the authors point out, the described membrane potentials suggest that the recruited neurons are likely those with higher resting membrane potentials before SP application. With this in mind, the authors should consider providing the membrane potential before and after SP for all of the recorded neurons of the various neural types. While the depolarization and recruitment of cells into pre-I firing is interesting, it will be essential in the future to determine if the recruitment is required for the SP induced shortening of the percolation phase. Alternative models still remain, like the increased activity of pre-I neurons alone drives the shortening of the percolation phase. The additional description of the methods will enhance reproducibility by others and the modified Discussion nicely handles how the various observations made in the manuscript may result in the observed selective effect of SP to shorten the percolation phase and addresses essential revision comments #1 and #3.

We appreciate the reviewers for taking the time to carefully assess our study and thank them for their helpful critiques.

We now include quantified Vm information for whole-cell recordings. We did see a small increase in Vm of pre-I neurons in response to SP, and there was also a trend (p=0.07) towards an increase in Vm among excitatory neurons that were recruited to pre-I. However, we refrain from detailed discussion of this in the text since the cell attached recordings cannot be included and because we did not test whether changes in Vm are due to an intrinsic response of the neuron to SP and/or by changes in the synaptic inputs to that neuron.

We agree with the reviewer that it will be important for future studies to more directly test whether the recruitment of excitatory neurons into pre-I spiking is necessary for SP induced facilitation. However, we did find that the slope of pre-I spiking in both pre-I neurons and neurons recruited to pre-I can predict the duration between bursts, strongly suggesting that these processes are linked.

https://doi.org/10.7554/eLife.51350.sa2

Article and author information

Author details

  1. Nathan A Baertsch

    Center for Integrative Brain Research, Seattle Children’s Research Institute, Seattle, United States
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology
    For correspondence
    nathan.baertsch@seattlechildrens.org
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1589-5575
  2. Jan-Marino Ramirez

    1. Center for Integrative Brain Research, Seattle Children’s Research Institute, Seattle, United States
    2. Department of Neurological Surgery, University of Washington School of Medicine, Seattle, United States
    Contribution
    Supervision, Funding acquisition
    Competing interests
    Reviewing editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-5626-3999

Funding

National Heart, Lung, and Blood Institute (R01 HL126523)

  • Jan-Marino Ramirez

National Heart, Lung, and Blood Institute (R01 HL144801)

  • Jan-Marino Ramirez

National Heart, Lung, and Blood Institute (K99 HL145004)

  • Nathan A Baertsch

National Heart, Lung, and Blood Institute (F32 HL134207)

  • Nathan A Baertsch

National Heart, Lung, and Blood Institute (P01 HL090554)

  • Jan-Marino Ramirez

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank NIH grants K99 HL145004 (Awarded to NAB), F32 HL134207 (Awarded to NAB), R01 HL126523 (Awarded to JMR), R01 HL144801 (Awarded to JMR), and P01 HL 090554 (Awarded to JMR) for funding this project.

Ethics

Animal experimentation: All experiments and animal procedures were approved by the Seattle Children's Research Institute's Animal Care and Use Committee and conducted in accordance with the National Institutes of Health guidelines (approved protocol #15981).

Senior and Reviewing Editor

  1. Ronald L Calabrese, Emory University, United States

Reviewers

  1. Muriel Thoby-Brisson, CNRS Université de Bordeaux, France
  2. Jeffrey C Smith, National Institute of Neurological Disorders and Stroke, United States

Publication history

  1. Received: August 26, 2019
  2. Accepted: December 4, 2019
  3. Accepted Manuscript published: December 5, 2019 (version 1)
  4. Version of Record published: January 13, 2020 (version 2)

Copyright

© 2019, Baertsch and Ramirez

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.

Metrics

  • 508
    Page views
  • 97
    Downloads
  • 0
    Citations

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)