Heartbeats entrain breathing via baroreceptor-mediated modulation of expiratory activity

Abstract

New findings: Cardio-ventilatory coupling refers to the onset of inspiration occurring at a preferential latency following the last heartbeat (HB) in expiration. According to the cardiac-trigger hypothesis, the pulse pressure initiates an inspiration via baroreceptor activation. However, the central neural substrate mediating this coupling remains undefined. Using a combination of animal data, human data and mathematical modelling, this study tests the hypothesis that the HB, by way of pulsatile baroreflex activation, controls the initiation of inspiration that occurs through a rapid neural activation loop from the carotid baroreceptors to Bötzinger complex expiratory neurons.

Abstract: Cardio-ventilatory coupling refers to a heartbeat (HB) occurring at a preferred latency prior to the next breath. We hypothesized that the pressure pulse generated by a HB activates baroreceptors that modulate brainstem expiratory neuronal activity and delay the initiation of inspiration. In supine male subjects, we recorded ventilation, electrocardiogram and blood pressure during 20-min epochs of baseline, slow-deep breathing and recovery. In in situ rodent preparations, we recorded brainstem activity in response to pulses of perfusion pressure. We applied a well-established respiratory network model to interpret these data. In humans, the latency between a HB and onset of inspiration was consistent across different breathing patterns. In in situ preparations, a transient pressure pulse during expiration activated a subpopulation of expiratory neurons normally active during post-inspiration, thus delaying the next inspiration. In the model, baroreceptor input to post-inspiratory neurons accounted for the effect. These studies are consistent with baroreflex activation modulating respiration through a pauci-synaptic circuit from baroreceptors to onset of inspiration.

Keywords: CVC; arterial baroreflex; baroreceptors; coupling; mathematical modeling; respiration.

FIGURE 1

Temporal raster plot of the heartbeats relative to the inspiratory onset. Every blue circle represents a single heartbeat (representative data shown from one supine, male subject). The coordinates of the circle are the occurrence times in seconds of a heartbeat relative to the start of the recording (x-coordinate) and of the interval between the heartbeat time and the onset of inspiration of the closest breath (y-coordinate). Negative y-values correspond to heartbeats occurring at the end of expiration before the inspiratory onset and positive y-values correspond to the heartbeats occurring after the inspiratory onset (shown by horizontal black dashed line). Vertical dashed lines show the beginning (green) and the end (red) of recording segments selected from the baseline, SDB and recovery parts of the experiment. We did not analyse the transition periods between baseline and SDB and between SDB and recovery

FIGURE 2

Distribution of heartbeats preceding the inspiratory onset. Each panel shows the cumulative distribution function (CDF, blue lines) as well as the histograms (probability density function, green bars) of the heartbeat occurrence times relative to the onset of the next inspiration (heartbeat latency) for the recording segments corresponding to baseline breathing, slow deep breathing (SDB) and recovery in supine male subjects. The three rows show data for different individuals. Orange lines are the CDFs of the uniform probability distribution. Red bars indicate maximal distance between the actual CDF and the uniform CDF. The distributions for all 10 subjects were statistically significantly different from uniform distributions

FIGURE 3

The measure of CVC strength and last heartbeat latency before inspiration. We used the maximal difference between the heartbeat latency cumulative distribution function and the uniform distribution (see red bars in Figure 2) as a measure of CVC strength in a particular individual. (a) Group data for CVC strength that appeared to be consistent among individuals and did not vary significantly across the three experimental conditions. (b) The characteristic heartbeat latency from inspiration (calculated as x-coordinates of the red bars in Figure 2) also had similar values (approximately 200 ms) across individuals and did not change significantly from baseline to SDB to recovery

FIGURE 4

Experimental set-up of artificially perfused brainstem–spinal cord preparation in a rodent. (a) The preparation is referred to as in situ because the brainstem, spinal cord and connectivity to peripheral mechano-, baro- and chemo-sensory and to homeostatic motor fibres remain intact. Thus, reflex evoked responses can be recorded. (b) Traces of the physiological recordings. A pulse in the perfusion pressure (PP) can be delivered in different phases of the respiratory cycle defined by phrenic nerve activity (PNA, blue trace). Bursts in PNA correspond to the inspiratory phase and interburst intervals are expiratory phases. As shown in this example, when the pressure pulse occurs during expiration, it noticeably delays the onset of the next inspiratory burst in PNA (i.e., prolongs expiration). It also causes a dip in thoracic sympathetic nerve activity (tSNA, red trace). Neural activity is recoded extracellularly by 16-channel multielectrode array. Examples of neuronal activity traces are shown in violet and pink. First three neurons exhibit post-inspiratory discharge pattern (pI) with stronger firing during the pressure pulse. In contrast, the fourth neuron (aug-E) that fires at the end of expiration reduces its activity during perfusion pressure excursion. Abbreviations: ADC, analog-digital converter; aug-E, augmenting expiratory; pI, post-inspiratory; PNA, phrenic nerve activity; PP, perfusion pressure; tSNA, thoracic sympathetic nerve activity; VNA, vagus nerve activity

FIGURE 5

The effects of pressure pulses delivered in different phases of the respiratory cycle on the respiratory cycle duration. We determined the phase of pressure pulse from its peak. In the in situ preparation, if the pressure pulse occurred in inspiration (I, n = 9), then it had no significant effect on cycle duration. But when delivered during first half (post-I, n = 11) or the second half of expiration (E2, n = 9), it prolonged the expiratory phase and thus increased cycle duration. Bars show the mean and error bars the SD

FIGURE 6

Effects of pressure pulses on firing of expiratory-modulated brainstem neurons. (a, b) Tracings from top: representative post-I neuron (a) and aug-E neuron (b), perfusion pressure (red) and integrated PNA (black). Grey thick curve in the top panel represents the cycle-triggered average of the firing rate of these neurons in unperturbed cycles. The pressure pulse was delivered at the time when the post-I neuron (a) would cease firing and when the aug-E neuron (b) would be augmenting. During baroreceptor stimulation induced by the transient pulse pressure, the firing rate of the post-I increased (a) whereas the aug-E neuron decreased its firing (b). (c) Group data summarizing the effect of pressure pulses on the activity of neurons of different firing phenotypes (I, n = 8; post-I, n = 5; aug-E, n = 14). When the pulse was delivered during inspiration, it had no significant effect on the average firing rate of the recorded inspiratory neurons. When the pulse was delivered during expiration, we registered significant increases in post-I neuron activity and decreases in aug-E activity

FIGURE 7

Model for cardio-respiratory interactions. The respiratory central pattern generator (CPG) is represented by interconnected populations of neurons in Bötzinger (BötC) and pre-Bötzinger complexes that contribute to the activity of the phrenic premotor population (ramp-I) in the rostral ventral respiratory group (rVRG). These neurons define the activity of the diaphragm and lung inflation. In the absence of ramp-I activity, the lungs passively deflate. The lung volume is decoded by pulmonary stretch receptors that send synaptic inputs to the pump cells located within the nucleus tractus solitarii through the vagus nerve. Pump cells excite BötC post-I neurons, which creates a negative feedback loop for off-switching inspiratory activity (Hering–Breuer reflex). Nucleus ambiguus contains a population of cardiac neurons that modulate heart rate by inhibitory inputs to the sinoatrial node. The cardiac neurons receive inputs from respiratory populations and/or pump cells so that their output becomes respiratory modulated and thus serves as a mechanism for respiratory sinus arrhythmia and blood pressure oscillations (Traube–Hering waves) in the model. Arterial baroreceptors encode the blood pressure value and send this information to the nucleus tractus solitarii second-order neurons in the baroreflex arc. The latter project to the post-I neurons in the BötC thus creating a beat-by-beat arterial pressure input to the respiratory CPG underlying cardio-ventilatory coupling. Through these mechanisms, the heartbeat can affect the timing of the next breath

FIGURE 8

Modelling the effect of transient baroreceptor stimulation. Traces from the top: activity of four neuronal populations representing the respiratory CPG: pre-I/I, early-I, aug-E and post-I (see Figure 7). Black traces represent unperturbed activity. After inspiration ends, the post-I population activates strongly and then adapts, gradually releasing the aug-E population from inhibition. This post-I adaptation eventually allows the inspiratory populations (early-I and pre-I/I) to activate. Dashed traces show the CPG activity in the presence of a baroreceptor stimulus (bottom trace). When it arrives centrally, the post-I population reactivates again, inhibits the aug-E population and prevents inspiration from starting until the baroreceptor activity wanes

FIGURE 9

Representative data from a subject (a, c) compared to model simulations (b, d). (a, c) Traces (30 s) of tidal volume, time stamp for ECG R-peaks and brachial intra-arterial pressure during baseline (a) and slow deep breathing (SDB, c). (b, d) Dynamics of the corresponding variables in the model mimicking baseline (b) and SDB (d) conditions (see text). We tuned the model in such a way that respiratory phase durations, HR, systolic and diastolic pressures as well as variabilities of all these metrics in both baseline and SDB conditions matched our previously published experimental group data (Barnett et al., 2020). We varied the CVC gain in the model to determine the range in which the model demonstrated heartbeat distributions similar to the experimentally observed ones (Figure 2)

FIGURE 10

The model qualitatively reproduces the heartbeat latency distributions as long as baroreceptor-to-respiratory network pathway is functional. Each panel shows the cumulative distribution function (CDF) of heartbeat latency before inspiration (blue lines) and corresponding probability density function (PDF) histograms (green bars) in the 0.5 s time interval preceding the inspiratory onset. Orange lines depict CDFs of uniform distributions. Red bars indicate maximal differences between the heartbeat latency CDFs from the model and uniform distributions. For the first row of the plots, the model included interactions between respiratory and cardiac systems in both directions, that is, RSA and CVC. Both conditions, baseline and SDB in human subjects, feature a 200–300 ms gap in heartbeat latency distributions. In the second row, we disrupted the CVC by setting the NTS-to-post-I synaptic weight to 0 (see Figure 7), which made the heartbeat latency distribution statistically indistinguishable from the uniform distribution (orange lines). The third row was constructed by removing the respiratory modulation of NA cardiac neurons (underlying RSA, see Figure 7) from the model while retaining CVC, which did not have a significant effect on the distributions (compare with the first and the third rows)