Transient configurations of baroresponsive respiratory-related brainstem neuronal assemblies in the cat

Abstract

The regulation of gas exchange requires coordination of the respiratory and cardiovascular systems. Previous work suggested that medullary raphe neurones transform and transmit information from baroreceptors to neurones in the ventral respiratory group. This study tested the hypothesis that distributed brainstem neuronal assemblies are transiently reconfigured during the respiratory cycle and baroreceptor stimulation. Blood pressure was perturbed by intravenous injection of an alpha1-adrenergic receptor agonist, unilateral pressure changes in the carotid sinus, or occlusion of the descending aorta in 14 Dial-urethane anaesthetized, vagotomized, paralysed, artificially ventilated cats. Neurones were monitored simultaneously with microelectrode arrays in two or more of the following sites: n. raphe obscurus, n. raphe magnus, rostral and caudal ventrolateral medulla, and the nucleus tractus solitarii. Transient configurations of baroresponsive assemblies were detected with joint pericycle-triggered histograms, the gravitational representation, and related pattern detection methods. Data were also analysed with cycle-triggered histograms, peristimulus-time and cumulative sum histograms, cross-correlograms, spike-triggered averages of efferent phrenic activity, and joint impulse configuration scatter diagrams (snowflakes). Five to nine simultaneously recorded spike trains from control expiratory phases were compared with data from interleaved equal-duration time blocks from control inspiratory phases. In each of seven animals, significant impulse synchrony detected by gravity analysis was confined to one phase of the respiratory cycle. Repeated patterns of distributed synchrony confined to periods of altered baroreceptor activity were detected and involved neurones that individually did not change firing rates during stimulation. Snowflakes and logical cross-correlation analysis provided evidence for the cooperative actions of impulses in concurrently active parallel channels. In 12 of 17 pairs of neurones with at least one baroresponsive cell, joint pericycle-triggered histograms detected synchrony indicative of shared inputs or functional excitatory interactions that varied as a function of time in the respiratory cycle. Neurones in four of the pairs had no respiratory modulation of their individual firing rates. Data from eight other pairs were indicative of fluctuations in inhibition during the respiratory cycle. The results demonstrate repeated transient configurations of baroresponsive neuronal assemblies during the respiratory cycle, without concomitant firing rate changes in the constituent neurones, and suggest distributed network mechanisms for the modulation of baroreceptor-mediated reflexes.

Figure 1. Spike train data segmentation processes used in this study

Illustrative example showing firing times of five simultaneously recorded neurones presented together with arterial blood pressure, integrated efferent phrenic nerve activity (Phrenic; 200–3000 Hz bandpass, time constant 200 ms), and E-pulses that mark the onset of the expiratory phase. A, in some gravity analyses, data were drawn from selected time blocks, represented by the shaded regions of the figure, and concatenated for subsequent analysis. In some cases, data were selected from a particular phase of successive respiratory cycles; samples from control and stimulus periods were then compared. Alternatively, control data from inspiratory and interleaved expiratory phases were compared. B, other analyses compared longer continuous samples, indicated by the horizontal arrows, that included control or stimulus intervals, or both. Joint pericycle-triggered histograms were triggered by E-pulses or other markers of phases of the respiratory or cardiac cycles.

Figure 2. Changes in neuronal assembly synchrony during baroreceptor stimulation

A, positions of eight particles, each representing one simultaneously recorded neurone, projected from the gravity N-space to three dimensions at 1.25 s into the analysis of concatenated spike train data recorded during baroreceptor stimulation. B, final particle positions at the end of the gravity calculations; acceptor and effector charge decays forward; decay time constant 10.0 ms. Numbers of spikes in sample: 1, 639; 2, 83; 3, 162; 4, 725; 5, 145; 6, 798; 7, 112; 8, 219. C, initially each particle was equidistant (100 arbitrary units) from all other particles. The red line (arrow) is the distance between particles 2 and 5 as a function of time in the analysis. The top and bottom thin blue lines define the maximum and minimum distances, respectively, for the particle pair found in 20 data sets in which each spike train was shifted relative to the others by different multiples of the basic time block to obliterate short time scale correlations. These lines served as empirical confidence limits (P < 0.05) for rejecting the null hypothesis that the particles did not aggregate more than expected for spike trains without such correlations. The nearly horizontal black line indicates the mean particle distance at each time step for 20 control calculations. D, grey regions in each row indicate times of significant aggregation for the indicated pairs (P < 0.05); the red row (arrow) corresponds to the particle distance as a function of time (PDFT) plot in C. E, significant particle aggregation in data from concatenated interleaved control intervals; same gravity parameters as A–D. Arrow indicates no aggregation in control period for neurones 2 and 5. Numbers of spikes in sample: 1, 668; 2, 47; 3, 152; 4, 727; 5, 123; 6, 813; 7, 112; 8, 210.

Figure 3. Respiratory phase-dependent variations in synchrony and asynchrony of baroresponsive neurones measured with joint pericycle-triggered histograms

A, variations in the synchronous discharge of two caudal raphe neurones indicated by a non-uniform distribution of statistically significant white bins along the diagonal of the surprise matrix (left) calculated from the normalized JPCTH triggered by onset of expiratory phase; 1110 cycles averaged. Horizontal and vertical cycle-triggered histograms for neurones CM5 (22991 spikes) and CM6 (11527 spikes), respectively, frame the surprise matrix for the JPCTH. The right panel shows a diagonal Gaussian smoothed histogram that is a tally of the paradiagonal bins, smoothed using a Gaussian of 4 bins, as well as a normalized cross-correlogram (upper right) calculated by summing of paradiagonal bins and corrected for different numbers of bins along the diagonal in the matrix. Subsequent similar figures follow the same format and labelling conventions; these have been detailed elsewhere (Aertsen et al. 1989). B, phase-dependent variations in the decreased firing probability of an RV E-Decr neurone following spikes in a caudal mid-line E-Decr neurone; 1013 cycles averaged. Horizontal CTH for neurone RV4; 25047 spikes. Vertical CTH for neurone CM6; 2885 spikes. Averages were triggered by expiratory phase onset. Tallies of paradiagonal bins are plotted in the diagonal coincidence histogram on the right; the histogram partially obscures the primary trough in the ordinary cross-correlogram at the upper right.

Figure 10. Peristimulus firing rates during repeated transient increases in carotid sinus pressure and neuronal assembly correlations in an animal near the apnoeic threshold

A, firing rate histograms of a set of nine simultaneously recorded brainstem neurones, integrated efferent phrenic activity and arterial blood pressure during elevated carotid sinus pressure (continuous lines). Arrows near blood pressure trace indicate responses to occlusion of carotid sinus region before sinus pressure was elevated. Abrreviation: N, neurone recorded in NTS; other abbreviations as in Methods. B, stack of respiratory cycle-triggered histograms; 728 control cycles triggered by the onset of the inspiratory phase were averaged. Note that indicated discharge patterns were based on averages triggered by expiratory phase onset (not shown). Rate scale does not apply to average integrated phrenic multineurone efferent activity. Number of action potentials for indicated neurone: 1, 19290; 2, 8921; 3, 416; 4, 12672; 5, 2262; 6, 21040; 7, 11602; 8, 35367; 9, 8600. C, particle distance as a function of time plot from gravity analysis of the same neurones. Data included all spikes in 328 concatenated 1.5 s samples that followed the onset of control inspiratory phases. All pairs aggregated significantly (P < 0.05) except for 2–1, 3–1, 4–1 and 9–1. Distance plots for three pairs of particles with the greatest degree of condensation are labelled. Gravity parameters: acceptor charge decay forward and effector charge decay backward; charge decay time constant 5 ms. Numbers of spikes: 1, 6657; 2, 1248; 3, 65; 4, 1443; 5, 506; 6, 4120; 7, 2965; 8, 7897; 9, 335.

Figure 13. Cross-correlograms and inferred connectivity

A, subset of cross-correlograms with primary features indicative of paucisynaptic interactions among neurones represented in Figs 10 and 11. Top, central peak: DI, 19.4; S, 0.40; HW, 52.5 ms; offset trough with positive lag of 52.5 ms; DI, 5.3; S, 0.11; HW, 52.5 ms; 14254 reference and 28446 target spikes. Middle, offset peak with positive lag of 17.5 ms: DI, 17.2; S, 0.95; HW, 52.5 ms; 14181 reference and 3132 target spikes. Bottom, central peak (DI, 5.2; S, 0.70; HW, 52.5 ms) and multiple peaks with positive lags. B, summary of dynamic connectivity inferred from JPCTH in Fig. 3.C, schematic diagrams of functional connectivity for neurones in group 3 inferred from data. Circled numbers label connections inferred from corresponding cross-correlograms in Figs 10, 11, 12 and 13A. Changes in firing rates during baroreceptor stimulation indicated by arrows. See Discussion for details.

Figure 11. Multiple measures of coincident and nearly coincident spikes in a baroresponsive neuronal assembly

A, repeated pattern of distributed synchrony detected in gravity data represented in Fig. 10C. Aggregation velocities were averaged over successive 491 ms intervals. Spark patterns matched under criterion 2. B, connected pairs of labelled circles indicate vectors in the spark patterns. C, snowflake calculated for spike trains of three of the neurones represented in the spark pattern. See text for details. Time span from centre to edge of snowflake is 615.0 ms. Numbers of spikes: 6, 28087; 7, 14181; 9, 17101.

Figure 12. Logical cross-correlograms

A, cross-correlogram with offset peak; positive lag of 17.5 ms; DI, 4.2; S, 0.20; HW, 70.0 ms; 21887 reference and 2262 target spikes. B, tutorial figure to illustrate calculation of logical AND and logical NOT cross-correlograms. See text. C, logical AND cross-correlogram in upper plot shows increased firing probability of neurone 5 following spikes in neurone 6 that were nearly coincident (±35.5 ms) with spikes in neurones 7 and 9. The number of events above background in the peak (299) divided by the number of trigger events (3104) was 0.096. The number of events above background in the peak of the lower logical NOT correlogram (139) divided by the number of trigger events (24946) was 0.0056. D, logical AND cross-correlogram shows firing probability of neurone 5 following spikes in neurone 6 that were nearly coincident (±35.5 ms) with time shifted spikes in neurones 7 and 9; see text. No significant primary feature was present.

Figure 4. Simultaneously recorded responses of raphe and VRG neurones and changes in synchrony during baroreceptor stimulation and the respiratory and cardiac cycles

A, top, labels to the left of each firing rate histogram in this and subsequent figures include recording location, indicated by the prefix of each neurone's identifier, type of respiratory modulation (see abbreviations in text) and direction of change in firing rate during perturbation. Firing rates in spikes s⁻¹ are shown on the right and refer to the largest bin in the corresponding plot. Blood pressure, integrated efferent phrenic nerve activity, and stimulus duration (red rectangle) are also shown. B, results of gravity analysis of three continuous data samples drawn from the two control and one stimulus intervals marked by rectangles in A. Dark regions in each row indicate times of significant aggregation for the indicated pairs (P < 0.05). Gravity parameters: acceptor charge decay forward, effector charge decay backward; charge decay time constant 15.0 ms. Numbers of spikes: (control 1) 1, 64; 2, 48; 3, 41; 4, 32; 5, 188; (baroreceptor stimulation) 1, 63; 2, 81; 3, 104; 4, 45; 5, 264; (control 2) 1, 65; 2, 36; 3, 42; 4, 40; 5, 166. C, surprise matrix of JPCTH shows respiratory phase-dependent fluctuations in synchrony for two rostral raphe E-Aug neurones; 294 cycles averaged. Horizontal CTH for neurone 4; 2341 spikes. Vertical CTH for neurone 3; 4930 spikes. Averages were triggered by expiratory phase onset. D, surprise matrix for same cells in a cardiac cycle-triggered JPCTH data set; 6944 cycles averaged.

Figure 5. Changes in neuronal synchrony during baroreceptor stimulation are not necessarily dependent on changes in firing rates

A, firing rate histograms and gravitational representations of eight simultaneously recorded rostral VRG and medullary raphe neurones during transient increases in blood pressure produced by inflation of an embolectomy catheter in the descending aorta. Stimulus intervals are indicated by changes in arterial blood pressure. Integrated efferent phrenic nerve activity indicates phases of the respiratory cycle. Neurones 1–3 and 5–8 were monitored in the nucleus raphe magnus; cell 4 was recorded in the ventrolateral medulla. Rectangles mark four samples of the data evaluated with the gravity method. B, times of significant particle pair aggregation during gravity analysis of four continuous data samples drawn from two stimulus intervals and interleaved control periods marked by rectangles in A. Dark regions in each row indicate times of significant aggregation for the indicated pairs (P < 0.01). Gravity parameters: acceptor and effector charge decays forward, charge decay time constant 10.0 ms. Numbers of spikes: (baroreceptor stimulation 1) 1, 219; 2, 15; 3, 79; 4, 282; 5, 61; 6, 296; 7, 44; 8, 105; (control 1) 1, 278; 2, 11; 3, 67; 4, 300; 5, 40; 6, 320; 7, 41; 8, 98; (baroreceptor stimulation 2) 1, 263; 2, 37; 3, 50; 4, 282; 5, 43; 6, 296; 7, 43; 8, 79; (control 2) 1, 259; 2, 25; 3, 58; 4, 279; 5, 53; 6, 331; 7, 45; 8, 91.

Figure 6. Peristimulus firing rates of ten simultaneously recorded neurones, integrated efferent phrenic activity and blood pressure during elevated carotid sinus pressure

A, two rostral VRG I-AUG neurones labelled 8 and 9 had reduced average firing rates during 6 baroreceptor stimulation intervals as compared to corresponding prestimulus control periods. Asterisks mark cycles with peak phrenic amplitudes more that 2 standard deviations below the mean of cycles preceding the baroreceptor stimulus interval. Dashed line marks peak systolic pressures before stimulation. B, phase graph documents decline in peak phrenic amplitude during and immediately following elevated pressure in carotid sinus. Brackets to the right indicate the mean ± 2 s.d. (dotted lines) of the peak amplitude of integrated phrenic activity measured for the 27 cycles preceding the stimulation. Asterisks mark peak values for cycles similarly labelled in A.

Figure 7. Measures of particle pair distances and aggregation velocities from gravity analysis of spike train data

A, PDFT plot for all pairs of neurones represented in the firing rate histograms of Fig. 6. Gravity parameters: acceptor and effector charge decays forward, charge decay time constant 5 ms. Numbers of spikes: 1, 95; 2, 1991; 3, 508; 4, 775; 5, 562; 6, 1641; 7, 1018; 8, 800; 9, 326; 10, 96. B, particle condensation profile for data in A. Successive elements in each row of this two-dimensional array are shaded to represent the relative average aggregation velocities of each pair of particles during an interval represented by the particular column (140 ms). C, dark regions in each row indicate times of significant aggregation for the indicated pairs represented in A and B (P < 0.01).

Figure 8. Spark plots indicate repeated transient configurations of neuronal assemblies during periods of baroreceptor stimulation or interleaved control periods

A, left, three examples of repeated patterns of synchrony detected in the data set represented in Figs 6 and 7. Aggregation velocities of the following particle pairs, calculated over 70 ms intervals, were represented in the three spark patterns: a, b and c. The patterns matched according to criterion 1. The stimulus period is indicated on the side panel. Time is represented along the z-axis, oriented from left to right in the figure. Right, pairs of labelled circles connected by lines indicate vectors in the corresponding spark patterns. B, left, three examples of repeated patterns of synchrony detected in the data set represented in Fig. 5. Aggregation velocities were calculated for each successive 491 ms interval. The three spark patterns labelled a, b and c matched according to criterion 1. Right, connected labelled circles indicate vectors in the spark patterns.

Figure 9. Cross-correlation histograms and spike-triggered average for neurones in assemblies represented in previous figures together with inferred functional connectivity

A, subset of cross-correlograms with primary features indicative of paucisynaptic interactions among neurones represented in Figs 6, 7 and 8A. Top left, offset peak with a positive lag of 8.5 ms; DI, 3.2; S, 0.05; HW, 34.0 ms; 36230 reference and 42911 target spikes. Histogram was scaled by subtraction of 90 % of the minimum bin counts from each bin; y-axis label includes bin counts. Top right, central peak with DI, 3.1; S, 0.06; HW, 47.5 ms; offset trough with negative time lag of 38.0 ms; DI, 3.8; S, 0.08; HW, 28.5 ms; 20057 reference and 42913 target spikes. Histogram was scaled by subtraction of 50 % of the minimum bin counts from each bin; y-axis includes bin counts. Bottom left, central peak; DI, 9.5; S, 0.18; HW, 66.5 ms; 36230 reference and 27316 target spikes. Bottom right, central peak in spike-triggered average of full-wave rectified efferent phrenic nerve activity sampled at 10 kHz; 42879 spikes of neurone 8 used as trigger events. B, subset of cross-correlograms with primary features indicative of paucisynaptic interactions among neurones represented in Figs 5 and 8B. Top, offset trough with a positive lag of 7.5 ms; DI, 3.6; S, 0.55; HW, 60.0 ms; 1677 reference and 4989 target spikes. Bottom, central peak with DI, 5.4; S, 0.31; HW, 30.0 ms; offset trough with positive lag of 45.0 ms; DI, 3.2; S, 0.19; HW, 15.0 ms; 10253 reference and 4989 target spikes. C and D, schematic diagrams of functional connectivity for correlational assemblies inferred from data in this and preceding figures. Circled numbers label connections inferred from corresponding cross-correlograms in A and B, respectively. Changes in firing rates during baroreceptor stimulation indicated by arrows. See Discussion for details.