Cardiac modulation of alpha motoneuron discharges

Abstract

Recordings of alpha motoneuron discharges from branches of the intercostal and abdominal nerves in anesthetized cats were analyzed for modulation during the cardiac cycle. Cardiac modulation was assessed by the construction of cross-correlation histograms between the R-wave of the ECG and the largest amplitude efferent spikes. In all but two recordings (which were believed to have either no or few alpha spikes), the histograms showed relatively short duration peaks and/or troughs (widths at half amplitude 4–50 ms) at lags of 10–150 ms. These observations were deduced to result from activity in oligosynaptic pathways, probably from muscle spindle afferents, whose discharges are known to be synchronized to the cardiac pulse. The results suggest that onward transmission of the cardiac signal from thoracic muscle afferents (and possibly from other dynamically sensitive afferents) to other parts of the central nervous system is highly likely and that therefore these afferents could contribute to cardiac interoception.

NEW & NOTEWORTHY It has been recognized since 1933 that muscle spindles respond to the cardiac pulse, but it is unknown whether this cardiac signal is transmitted to other levels in the nervous system. Here we show that a cardiac signal, likely arising from muscle spindles, is present in the efferent activities of thoracic and abdominal muscle nerves, suggesting probable onward transmission of this signal to higher levels and therefore that muscle spindles could contribute to cardiac interoception.

INTRODUCTION

Muscle spindle afferents have long been recognized to show modulation of their activities by the cardiac pulse (Birznieks et al. 2012; Critchlow and von Euler 1963; Ellaway and Furness 1977; Kirkwood and Sears 1982a; Matthews 1933; McKeon and Burke 1981), as have some other dynamically sensitive peripheral receptors (Gammon and Bronk 1935; Macefield 2003). Surprisingly, however, apart from a single abstract (Ellaway et al. 1979), the possible transmission of such cardiac effects to motoneurons has not been investigated.

During the course of reanalysis of previously published data, we had the opportunity to further investigate this matter for the naturally occurring motoneuron discharges of a number of different thoracic and abdominal muscle nerves. The question is of particular interest in the light of the recent hypothesis by Birznieks et al. (2012) that the cardiac modulation of spindle discharges was unlikely to lead to onward transmission of a cardiac signal because of cancellation of positive and negative effects. However, for our recordings, cardiac modulation of the alpha motoneurons was a universal occurrence. These observations, which we suggest result from transmission in oligosynaptic pathways, therefore predict that somatosensory cardiac modulation is likely to have further influences on higher centers.

METHODS

The recordings examined came from experiments for which the main results have already been reported and were conducted according to UK legislation [Animals (Scientific Procedures) Act 1986] under Project and Personal Licenses issued by the UK Home Office. The data came from 17 cats of either sex, weighing 2.5 to 3.7 kg. Twelve of these came from the experiments reported by Saywell et al. (2007, 2011) and the other 5 from those reported by Road et al. (2013). All of the animals were anesthetized with sodium pentobarbitone (initial dose 37.5 mg/kg ip, then intravenously as required). Neuromuscular blockade was achieved by the use of gallamine triethiodide (subsequent to surgery, intravenously, repeated doses 24 mg as required) and the animals were artificially ventilated via a tracheal cannula with oxygen-enriched air, to bring the end-tidal CO₂ fraction initially to ~4%. CO₂ was then added to the gas mixture to raise the end-tidal level sufficient to give a brisk respiratory discharge in the midthoracic intercostal nerves (typically 6–7%). During neuromuscular blockade, anesthesia was assessed by continuous observations of the patterns of the respiratory discharges and blood pressure together with responses, if any, of both of these to a noxious pinch of the forepaw. Only minimal, transient responses were allowed before supplements (5 mg/kg) of pentobarbitone were administered. The animal was supported by vertebral clamps, a clamp on the iliac crest, and a plate screwed to the skull. Rectal temperature was maintained between 37°C and 38°C by a thermostatically controlled heating blanket. Mean blood pressures, measured via a femoral arterial catheter, were above 80 mmHg throughout.

Nerve Recordings

Nerve recordings were originally made either during spike-triggered averaging measurements or for cross-correlation measurements, while investigating the connections from expiratory bulbospinal neurons to motoneurons or to spinal interneurons. The recordings were made from the cut central ends of selected nerves via pairs of platinum wire electrodes and with conventional amplification (filter settings 300 Hz to 3 kHz). Thoracic nerves were maintained in a single paraffin oil pool constructed from skin flaps and the lumbar nerves were recorded under petroleum jelly, most often with a piece of thin plastic film separating the electrodes from underlying muscle. These recordings were stored on magnetic tape and subsequently acquired for computer analysis via a 1401 interface and Spike2 software (Cambridge Electronic Design, Cambridge, UK). Sampling rates were around 8–10 kHz, varying according to the number of channels sampled.

In the first group of 12 cats, the recordings examined consisted of one external intercostal nerve in either T5 or T6, from one experimental run in each cat, each run being a period used for intracellular recording from a motoneuron or an interneuron with no stimuli being delivered (1–23 min). To investigate cardiac modulation, an ECG signal was obtained from a cord dorsum recording, originally used for monitoring afferent nerve volleys during the initial testing of the motoneuron or interneuron (Ford et al. 2014; Saywell et al. 2011). The recording filters had been set at 300 Hz to 10 kHz, specifically to minimize the occurrence of an ECG signal (the magnetic tape recorder further reduced the bandwidth to 300 Hz to 3 kHz). Nevertheless, in the recordings selected for analysis here an ECG signal was still visible and could be made suitable for triggering by low-pass filtering within Spike 2 (Fig. 1B). A simple level-crossing detection gave the timing of the R-wave. Either polarity could be used, according to the signal-to-noise available. The signal itself (the QRS complex) also varied in time course between experiments, often being multiphasic, so the timing of the triggers varied by a few milliseconds between different experiments.

Fig. 1.

Cross-correlation analyses of external intercostal nerve discharges. A: extract from a recording from T5 external intercostal nerve. The horizontal line (*) indicates the threshold amplitude level for alpha motoneuron spikes. B: extract from a simultaneous recording (different time scale) of the cord dorsum recording at high gain: top, raw recording; middle, filtered version; bottom, triggered R-wave events. C, a–e: analysis of data shown in A and B. a, Cross-correlation histogram derived from the alpha discharges and the R-wave (reference) events. b, Same data, split into two independent time periods (1,496 and 1,813 R-wave events, respectively) and displayed over one cardiac cycle (250 ms). c, Autocorrelation histogram for R-wave events. d, Central part of the histogram in a, for comparison with the R-wave-triggered average of the external intercostal nerve recording (e). f and g, Cross-correlation histograms from T5 external intercostal nerve recordings from two further experiments. Numbers of R-wave events: a, 3,309; f, 1,139; g, 1,695. R-wave periods for f and g, 260 and 276 ms, respectively. Ordinate ranges a, b, and d, 0.75m–1.2m; f, 0.8m–1.2m; g, 0.2m–2.0m; (m is the mean count calculated over ± 0.6 s). Bin widths, 4 ms, except f, 10 ms. The calibrations for A and for Ce are in the same arbitrary units.

In the second group of 5 cats, longer runs of data (52–100 min) originally obtained for cross-correlation measurements were available (one run from each cat). Recordings from one external intercostal nerve (T5 or T6) and three to six internal intercostal nerves or nerve branches (T8–T11) from the left side of each cat were included. These branches included the following nerves on the left side of T8 and/or T9 (4 cats): 1) one of the filaments of the internal intercostal nerve, which are the naturally occurring branches that leave the nerve at intervals to innervate the internal intercostal muscle layer (Sears 1964a); 2) the lateral branch of the internal intercostal nerve, which innervates external abdominal oblique; 3) the distal remainder of the internal intercostal nerve, which innervates the more distal part of the internal and parasternal intercostal muscles, transversus abdominis, and rectus abdominis (see Meehan et al. 2004 for references). In the fifth cat, the whole internal intercostal nerves of T9 and of T11 were included, together with those of T9, T10, and T11 on the right side. In addition, in two of the animals, recordings from branches of the L1 ventral ramus were also available, including (in both animals) a branch innervating internal abdominal oblique and a distal remainder (for more details see Road et al. 2013).

In only one of this second group was there a cord dorsum recording that could be used to provide an ECG signal, but in the other four an ECG signal could be derived from one of the nerve recordings, in each case a nerve with a relatively modest efferent discharge. Again, low-pass filtering in Spike2, followed by level crossing was used to provide R-wave events (Fig. 3A). Even so, in some recordings, at the height of inspiration or expiration, extra events sometimes occurred as a result of efferent spikes remaining of comparable size to the R-wave. These were almost always readily detected by visual inspection of an instantaneous frequency display of the heart rate signal, and these spurious events were deleted. A few events that were ambiguous for timing because of superposition of the R-wave with efferent spikes were lost in this process.

Fig. 3.

Analyses of recordings from branches of internal intercostal nerves. A: simultaneous recordings, traces from above: T5 external intercostal nerve; low pass version of same signal (to reveal ECG); R-wave events, triggered from ECG; lateral branch of T8 internal intercostal nerve (T8 Lat); distal remainder of T8 internal intercostal nerve (T8 Dist); T9 internal intercostal nerve filament (T9 Fil). Horizontal lines (*) indicate the threshold amplitude levels for alpha motoneuron spikes. Note cardiac-related bursts of small spikes in the T8 Dist recording, particularly in late expiration and early inspiration. B: cross-correlation histograms for each of the three nerves illustrated above, as indicated. 14647 R-wave events. R-wave period, 275 ms. Ordinate ranges for histograms: T8 Lat, 0.5m–2.5m; T8 Dist, 0.7m–1.3m; T9 Fil, 0.8m–1.2m. Bin widths, 4 ms.

Analysis

This consisted of the construction of cross-correlation histograms between efferent spikes and the R-wave events (reference events) over a lag range of ± 0.6 s. Bin widths were usually 4 ms, but in a few instances where the run length was short or the number of efferent spikes was low, a bin-width of 10 ms was used instead.

RESULTS

Selection of Alpha Spikes

External intercostal nerves.

Seventeen recordings of the efferent discharges from the external intercostal nerves of T5 or T6 (each from one cat) were analyzed, a representative example being illustrated in Fig. 1A. The periodic discharge in inspiration is obvious. The largest spikes belong to alpha motoneurons, but in these recordings we do not have an exact criterion for the division between the spikes from alpha and gamma motoneurons. In the original recordings from intercostal nerve filaments (the naturally occurring intramuscular intercostal nerve branches), Sears (1964b) was able to say with confidence that the distinction between alpha and gamma discharges was obvious. Our present results confirm this (see section Internal intercostal nerve filaments), but the statement does not apply to whole nerve recordings, largely because so many more units were active. Moreover, because we used an elevated level of CO₂, the presumed inhibition of the gamma motoneurons during expiration in our recordings was stronger than in Sears (1964b) and the continuing discharge through expiration, which could be one of the identifying features, was considerably curtailed compared with those earlier recordings. We therefore adopted a somewhat arbitrary criterion for selecting alpha spikes. We reasoned that for a rostral segment such as T5 or T6 and with an elevated CO₂, at least some alpha motoneurons would be active at the start of inspiration. For confirmation of this, see, for instance, illustrations in Kirkwood et al. (1981) or Davies et al. (1985), both of which show external intercostal nerve filament discharges in the conditions of a relatively strong respiratory drive (neuromuscular blockade and artificial ventilation) as compared with Sears (1964b), but still a lower level of drive than the present experiments. The gamma discharges through expiration in the illustrations of these reports are clear, as are the alpha discharges at the start of inspiration. We therefore consistently set the level between alpha and gamma spikes as indicated in Fig. 1, a little below the amplitude of earliest inspiratory spikes. This is a conservative choice, i.e., there is little chance of gamma spikes above this level, but there would probably have been some alpha spikes in the range designated as gamma.

Internal intercostal nerves.

Recordings from five whole internal intercostal nerves were analyzed, all from one cat. Two of these are included in Fig. 2, A and B, together with a recording from the external intercostal nerve, whose main bursts of discharge defined inspiration. Four of the five showed strong expiratory discharges, together with additional inspiratory bursts which, in this recording, merged into an apparent tonic background. This background was assumed to include tonic gamma discharges, but these are hard to separate objectively from the phasic alpha discharges because there were hardly any clear gaps between the inspiratory and expiratory phases. Spike amplitude levels for the alpha spikes were assigned as indicated on the figure, largely by analogy with those used for the external intercostal nerves. The fifth nerve recording with no phasic bursts was assumed to include no alpha discharges.

Fig. 2.

Analyses of recordings from whole internal intercostal nerves. A: extract from simultaneous recordings from one external (T6 Ext) and two internal (T9L Int and T11L Int) intercostal nerves from the left side of the animal, together with R-wave events. B, same recordings as in A, from the period indicated by the bar under the left of T11L Int, shown with gains 2 × increased and time scale 8 × expanded. Horizontal lines (*) indicate the threshold amplitude levels for alpha motoneuron spikes. C, cross-correlation histograms as in Fig. 1. Top two panels apply to the nerve discharges illustrated in A and B. Bottom two panels apply to simultaneous recordings from two other internal intercostal nerves on the right side of the animal. 8698 R-wave events. R-wave period, 271 ms. Ordinate ranges 0.8m–1.2m for all histograms. Bin widths, 4 ms.

Lateral branch.

The lateral branch is one of the two main branches of the internal intercostal nerve. It innervates one muscle (external abdominal oblique) and also has a cutaneous component. Recordings from 6 such nerves (from 4 cats) were analyzed, one of those recordings being illustrated in Fig. 3A. A feature of 4/6 of the recordings from these nerves was that there was a very clear distinction between the alpha and gamma spikes, similar to the original illustrations for the filament discharges in Sears (1964b). In the particular recording of Fig. 3, the distinction is helped by the rather modest alpha discharge (2–3 units), whereas the gamma discharges were very busy.

Distal branch.

The distal remainder of the internal intercostal nerve (Saywell et al. 2007) is similar in size to the lateral branch. It innervates the internal intercostal and abdominal muscles as well as parasternal, interchondral muscle and has a cutaneous component (see methods). The discharges include an inspiratory burst (extent according to the segment), assumed to be destined for the parasternal muscle (De Troyer et al. 2005; Taylor 1960) and thus these discharges appear similar to those of the whole internal intercostal nerve, though with fewer spikes. The amplitude level for alpha spikes was set as for the whole nerve, as in Fig. 3A. Six such recordings were examined, from four cats.

Lumbar (L1) nerves.

Recordings from two branches of the L1 nerve, one innervating internal abdominal oblique muscle, one a more distal remainder (for description see Road et al. 2013), were each available in two cats. In all four instances phasic expiratory alpha and more tonic gamma spikes were distinguishable, in particular rather clearly for the internal abdominal oblique, as for the T8 lateral branch in Fig. 3, though with more alpha spikes.

Internal intercostal nerve filaments.

The separation between alpha and gamma spikes were particularly clear for these recordings (Fig. 3A), confirming the original observations of Sears (1964b), where the efferent spikes comprised “two distinct sizes.” Six recordings of filament discharges (including this one) were analyzed, of which five included alpha spikes.

Cardiac Modulation

To assess cardiac modulation we constructed cross-correlation histograms between the R-wave events and the events defined by the level crossings for the alpha ranges of spike amplitude (Figs. 1B, 2C, and 3B). A clear modulation was found to be present for 40 of the 42 individual nerve discharges examined. The two exceptions consisted of one internal intercostal nerve discharge (mentioned above), which was believed to include no alpha spikes, and one with a very low mean count (15) in the histogram bins, meaning that peaks similar to those observed in many of the other examples would have been undetectable. Thus cardiac modulation can be said to have been universally present. It consisted of a sequence of peaks and troughs, with widths at half amplitude in the range 4–50 ms.

No statistical tests were used to identify the presence or absence of individual peaks and we are not concerned here to define whether an individual histogram showed, for instance, three or four peaks. The number of the peaks showed considerable variation between different animals and/or nerves sampled. However, the data illustrated in Figs. 1–3 cover the range of peak amplitudes and bin counts involved and also include confirmation that most of the peaks detectable by eye were repeatable (Fig. 1Cb). Within that range of variation, however, the examples in Fig. 2 are of particular interest, where four recordings, all from nerves of the same category (whole internal intercostal) were made in the same animal. Each of the four nerves gave multiple peaks in the histograms. Note that the two from the left of the animal show at least four peaks at identical latencies. The two from the right side both show two peaks at similar latencies to those on the left, at lags of around 20 and 80 ms, but the weaker peaks that may also be present (again similar to each other) do not correspond to peaks in the histograms from the left.

A possible source of error in these experiments could be artifactual triggering via the presence of an ECG signal, which could summate with noise or smaller spikes to cross the threshold and contribute extra spike events. For the 17 animals of the first group, such a possibility was never detected. For instance, in the recording of Fig. 1A, R-wave-triggered averaging did detect an ECG signal (Fig. 1Ce), as a triphasic wave (lags −20 ms to 0), but this signal was very small, well within the baseline noise in the original recording (note the calibrations). The histogram of Fig. 1Ca is repeated on an expanded scale in Fig. 1Cd, showing no detectable peaks corresponding in time to the ECG signal. For the other five animals, a rather larger ECG signal was sometimes present in one or another nerve recording (hence our ability to use such a signal to generate the R-wave events), but this still only created artifacts in 5/25 histograms. These consisted of very narrow peaks around zero lag, corresponding to the averaged ECG signal in that nerve, and they were always distinct from the peaks and troughs measured here, which were all at lags clearly displaced from zero.

Another possible source of error might be thought to arise from the presence of sympathetic discharges, which are well known to show cardiac modulation. These discharges were indeed often present as the smallest group of spikelike waveforms (TW Ford and PA Kirkwood, unpublished observations), sometimes with cardiac modulation, visible as cardiac-synchronized bursts in the recordings (e.g., Fig. 2, all three nerves, and Fig. 3, T8 distal branch). When estimated as R-wave-triggered histograms, these discharges showed a completely different time course than those described here, approximating to a sinusoid at the cardiac frequency (cf. Fatouleh and Macefield 2013; Jänig et al. 1983; Sato and Schaible 1987) with a lag to its peak of around 200 ms. Although it is possible that some of the low amplitude slower components detectable in the baselines of some of the histograms (e.g., Fig. 1Cb or Fig. 3B, T8Dist) might have arisen in this way via summation of spikes, such components were not involved in the measurements here (see next section).

Quantitative Summaries

Measurements of the lags and amplitudes of the histogram peaks are illustrated in Fig. 4A. Amplitudes were measured peak-to-trough during the first cardiac cycle with a positive lag. The largest peak-to-trough value was measured, whether the trough was before or after the peak. This was most often the first peak to the right of zero (but could be the 2nd or 3rd, as in Fig. 4A). Amplitudes were expressed as a multiple of the mean count (m), measured over ± 0.6 s. The lag to the maximum value (or to the minimum if a trough was the dominant feature) was measured for each histogram. The cardiac cycle times, estimated from the R-wave autocorrelation histograms (e.g., Fig. 1Cc), ranged from 232 to 330 ms. Note the narrow peak in the autocorrelation histogram (2 bins), which was typical of the vagotomized group (5 examples with 1 bin, 6 examples with 2 bins and one with 3), the width varying according to the strength of respiratory sinus arrhythmia or slower drifts in heart rate. The nonvagotomized group of five animals showed somewhat wider peaks in their autocorrelation histograms (widths at half-amplitude of 2–4 bins, modal value 3).

Fig. 4.

Quantitative summary. A, example of measurements of lags (p) and amplitudes (q/m) of peaks in the histograms; m is the mean count, calculated over ± 0.6 s. When the histogram was dominated by a trough, the lag was measured to the minimum value of the trough. B and C: distributions of the lags and amplitudes of the peaks from all the recordings, according to the nerves recorded: Ext, external intercostal nerve; Int, whole internal intercostal nerve; Lat, lateral branch of that nerve; Dist, distal branch of that nerve; L1, branches of L1 ventral ramus; Fil, filament of internal intercostal nerve. The filled areas for lags refer to troughs. Note the log scale for the amplitude distributions, with a bin width equivalent to a factor of 2.

The measurements of lags are summarized in Fig. 4B. Despite the fact that in some instances peaks were detectable at a wide range of times in the cardiac cycle, the measured lags to the largest peak or trough were relatively restricted (10–150 ms, all but two less than 85 ms). No obvious differences were detected between the different categories of nerve, nor between the lags for dominant peaks vs. dominant troughs (open vs. filled bars in Fig. 4B). The measurements of amplitude are summarized in Fig. 4C. These covered a wide range, 0.15–1.98 (note the log scale in Fig. 4C), again with no obvious differences between the different nerve categories.

Gamma Motoneurons

The nerve filament recordings comprised the nerve category where the separation of gamma and alpha motoneuron discharges by spike amplitude was the clearest, as in Sears (1964b). R-wave-triggered histograms for the gamma spikes were almost featureless, but some small peaks at the same lags as those for the alpha spikes in the same filament could be seen. However, these peaks all had amplitudes less than 10% of that for the alpha population and were close to the limit of detection.

For the other nerves, the range of spike amplitude for the gamma discharges was even more approximately defined than it was for the alpha range because a lower border also had to be chosen to separate the gamma spikes from sympathetic discharges, which could be of considerable amplitude, as noted above (Figs. 2 and 3). It was therefore hard to define an amplitude range that was not at risk of overlap with spikes from either alpha or sympathetic ranges. Despite this uncertain definition it was considered worth measuring the cardiac modulation of this population for comparison with the alpha population and with the gamma population in the filaments.

Not surprisingly, the histograms for the gamma ranges often showed features common to the ranges on either side, i.e., peaks and troughs similar to those for the alpha range for the same nerve, or a near sinusoid waveform similar to that shown for sympathetic discharges, or a mixture of these. The features similar to the alpha range were never larger in amplitude than for the alpha range. Only 3/41 had amplitudes above 50%, and 19/41 had amplitudes less than 10%. Overall, therefore, the most parsimonious explanation is that synaptic effects equivalent to those revealed here for the alpha motoneurons did not occur in the gamma motoneurons and the effects seen in some histograms were a result simply of the overlap of amplitudes for the two populations of spikes.

DISCUSSION

The universal occurrence of synchronization here between the R-wave of the ECG and alpha motoneuron spikes for both of the intercostal layers and for all three abdominal layers (external oblique from T8, the other two at L1) comprises an important new observation. It may be similar to the effects reported as “occasionally seen” for forelimb and neck motor units in an abstract from Ellaway et al. (1979), but this paper is the first detailed description of the phenomenon.

At first sight one might be tempted to ascribe histograms with multiple, apparently periodic peaks, such as those in Fig. 2C, to the same mechanism that produces periodicities of around 10 Hz in sympathetic discharges, similarly linked to the cardiac cycle (Cohen and Gootman 1970). However, most of the peaks observed here did not appear to be periodic, except, of course for the cardiac period. Often, the histograms were dominated by individual peaks (e.g., Figures 1Cf or 3B, T8Lat) or troughs (e.g., Fig. 1C, a and b). We therefore suggest that the obvious source of the effect is the well-recognized cardiac modulation of afferents from muscle spindle primary endings (Birznieks et al. 2012; Critchlow and von Euler 1963; Ellaway and Furness 1977; Matthews 1933; McKeon and Burke 1981), in particular the modulation shown by spindle afferents from external intercostal muscle (Kirkwood and Sears 1982a). The afferents recorded in this last report came from one of the muscles whose nerve discharges were investigated here and it is because the time course of the afferent modulation was so similar to that seen here for the nerve discharges that these afferents are the prime candidates for the source of the effect.

The occurrence of multiple peaks could have several explanations. First, not only do individual afferents often show multiple peaks in their cross-correlation histograms with the R-wave, but also different afferents can fire at different times with respect to the R-wave (Birznieks et al. 2012; Kirkwood and Sears 1982a). Second, individual motoneurons might not only respond at different times in the cardiac cycle, but also be sufficiently well synchronized to the R-wave for their intrinsic firing frequencies to appear in the histograms. Birznieks et al. (2012) suggested that the diversity of synchronization patterns for individual afferents would mean that their postsynaptic effects would effectively cancel out, so as to be equivalent only to some extra noise. That is clearly not the case here, though it does appear, from the sharpness of the peaks, that all of the effects seen here are likely to be from afferents similar to those which Birznieks et al. (2012) described as “driven.” For Kirkwood and Sears (1982a) these would be those in the highest conduction velocity, most dynamically sensitive group. It is quite possible that the slower conducting, less dynamically sensitive thoracic spindle afferents were then like the majority of the afferents reported by Birznieks et al. (2012), as showing a more temporally dispersed, lower amplitude modulation. If so, then their effects on motoneuron firing may well cancel out, as predicted by Birznieks et al. (2012), because the histograms generated from the alpha spikes generally showed only relatively narrow peaks.

One might think that the most obvious route for transmission from spindle afferents to the motoneurons should be the monosynaptic link. However, the monosynaptic connections of thoracic spindle afferents to motoneurons are quite restricted, being limited to the same segment, together with minor contributions from the segments on either side (Kirkwood and Sears 1982b; Sears 1964c). In the present recordings, these connections could have contributed very little, because the segment concerned was deafferented by virtue of the nerve sectioning required for the efferent recordings. This often also applied to one or both of the adjacent segments. We therefore suggest that the effects we have described were instead mediated oligosynaptically. There is some existing evidence for such a connection (Kirkwood and Sears 1982a; Kirkwood et al. 1982) and the effects could easily arise via spindle afferents in the remaining undissected thoracic segments. Two other observations support this view. First, the presence of troughs in the histograms (sometimes without either preceding or following peaks) suggests likely inhibitory effects. Second, patterns of responses such as those in Fig. 2C, where the motoneurons of T9 and T11 on the left showed synchronization patterns that were remarkably similar to each other, as were those of T10 and T11 on the right, suggest an influence rather more distributed than that shown for monosynaptic connections.

Note that, because we are suggesting oligosynaptic connections, we do not know, for a given nerve, which muscle supplied the responsible afferents. The cardiac modulation of thoracic spindle afferents has only been described for the external intercostal muscle, but one must assume that such modulation is just as likely for the internal layer, and likely also for the abdominal muscles. It would be interesting to see how different the effects would be if recordings were made with most of the local afferents intact—most likely considerably stronger, on account of the monosynaptic links.

Two additional groups of receptors should be considered as possible contributors to the cardiac modulation, particularly because we believe the effects were evoked oligosynaptically. The first comprises Pacinian corpuscles, in particular those of the mesentery. Their discharges were shown to have prominent cardiac modulation by Gammon and Bronk (1935) and are well placed because their afferent fibers run in the splanchnic nerve and enter the spinal cord at thoracic levels (Bain et al. 1935). However, we regard these as less favorable candidates than the spindles because they did not seem to be involved in splanchnic-to-intercostal reflexes (Downman 1955), the thresholds of which were higher than necessary to recruit these afferents. The second group consists of cutaneous receptors. In contrast to the mesenteric Pacinian corpuscles, thoracic cutaneous receptors are certainly involved in local reflexes (Aminoff and Sears 1971; Downman 1955). Moreover, Macefield (2003) showed that 44% of the tactile receptors sampled from the human hand (mostly in the finger pads) showed cardiac modulation. However, as Macefield (2003) pointed out, the local environment of the finger pads is particularly well suited to the transmission of small movements such as the cardiac pulse and it is not known to what extent this will be the case for the skin of the thorax. In the hairy skin of the rabbit’s or cat’s hind legs Brown and Iggo (1967) reported “occasionally” finding afferents responding to the cardiac pulse, corresponding to the “few” similarly found by Hunt and McIntyre (1960). With only minimal positive evidence, we therefore also consider these receptors to be less favorable candidates than the spindles.

Functional Relevance

Ellaway et al. (1979) suggested that cardiac modulation of alpha motoneurons could promote tremor locked to the heart beat. That remains a possibility. Similarly, the effects could be regarded as a component of motoneuron synchronization occurring via synchronized motoneuron inputs. In addition to an effect on tremor, such synchronized inputs can provide additional motoneuron firing above that expected from the equivalent summation of randomly occurring EPSPs. This may, or may not, be an advantage for the animal, depending on how well that process is controlled by the central nervous system (CNS) (cf. Kirkwood et al. 1984).

This idea is rather similar to the suggestion of Birznieks et al. (2012) that the net effect of the cardiac modulation of spindle afferents could be considered simply as the addition of noise. However, we contend that the demonstration here of cardiac modulation of motoneuron discharges does represent meaningful onward transmission of a cardiac signal from thoracic afferents. On the assumption that the signal is transmitted oligosynaptically, it is also possible that what is seen by the motoneurons might be only a fraction of what is available for transmission to other parts of the nervous system. Nevertheless, it is worth considering what strength of connection might be represented in the histograms here. The peaks showed a range of amplitudes, broadly similar to the wide range of responses seen in cross-correlation histograms that measured individual (monosynaptic) connections between thoracic spindle afferents and motoneurons (Kirkwood and Sears 1982b), but the duration of the peaks here were around 3–30 times greater, suggesting an effect 3–30 times stronger. However, we are not suggesting that this level of excitation is necessarily important for the motoneurons, especially as it may be an underestimate, our recordings here being made under anesthesia and with considerable deafferentation. Rather, the observations simply show that the information can be transmitted over pathways of more than one synapse. The motoneurons in this respect provide a surrogate for more direct evidence of onward transmission elsewhere in the CNS. Note that in this respect it is an advantage that any monosynaptic effect was minimal, since the presence of such an effect would have likely occluded the effect we did measure and would have prevented us from making this deduction. However, the fact that the signal is available does not necessarily mean that it has a functional significance; more experiments will be required to judge this. One important possibility is that the signal could contribute to the various psychological/reflex effects related to the cardiac cycle that are most often ascribed to an input from arterial baroreceptors (e.g., Dworkin et al. 1994).

There are two aspects of this to be considered. First, in the interpretation of motor responses (e.g., reaction times), our results suggest that cardiac modulation of excitability of motoneurons or of spinal cord circuits cannot be ignored. Such modulation could itself have two alternative effects. One is the added excitability occurring at the times of the peaks in our histograms, but alternatively, the CNS might choose to initiate actions at other times in the cycle, when the noise level is lower. The second aspect is that the prevailing view in the literature is that the cardiac modulation of a very wide range of cognitive functions (e.g., Azevedo et al. 2017) is modulated exclusively by the baroreceptive afferents. Without denying at all the strong evidence supporting baroreceptive afferents in this role, we suggest that the participation of other afferents such as those from muscle spindles cannot be ruled out, especially now we have demonstrated synaptic transmission of their cardiac-related activities. Muscle spindle afferents do have a direct link to higher centers (Hore et al. 1976; Oscarsson and Rosén 1963). Support for this hypothesis comes from the reduction in cardiac awareness reported by Montoya and Schandry (1994) for patients with complete spinal cord injuries between C5 and T4.

It should be pointed out that, although we have concentrated on muscle spindles as the most likely origin of the cardiac modulation, the arguments here concerning a contribution to cardiac awareness and related phenomena could apply equally to any of the other thoracic somatosensory afferents mentioned earlier. However, the framing of these arguments in terms of muscle spindles makes the hypothesis particularly intriguing, as it then links the most classic of proprioceptors to the most classic interoceptive modality.

GRANTS

The experimental work was supported by the Jeanne Anderson Fund (Institute of Neurology), the Wellcome Trust (038027/Z/93/Z/1.5), the Canadian MRC, and the International Spinal Research Trust.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

T.W.F. and P.K. conceived and designed research; T.W.F. and P.K. performed experiments; T.W.F. and P.K. edited and revised manuscript; T.W.F. and P.K. approved final version of manuscript; P.K. analyzed data; P.K. interpreted results of experiments; P.K. prepared figures; P.K. drafted manuscript.