Unitary recordings were made from a total of 75 muscle spindle afferents in 40 subjects. Discharge activity in 58 muscle spindle afferents showing spontaneous activity was recorded from relaxed leg muscles while voluntary muscle contraction was required to activate the remaining 17 afferents. Based on behavioural criteria 50 afferents were classified as supplying primary endings (group Ia afferents) and 25 as supplying secondary endings (group II afferents).
Muscle spindle firing driven by arterial pulsations
In total, we encountered seven muscle spindles discharging at a relatively low discharge rate (≤3 imp/s), apparently driven by arterial pulsations, e.g., showing no other spontaneous activity than spikes generated with a clear cardiac rhythmicity. Four muscle spindles fired one spike per cardiac cycle, corresponding approximately to the ascending section of the pulse wave (anacrotic limb or upbeat), which occurs about 250 ms after the peak of the R-wave of the ECG (). R-wave triggered post-stimulus time histograms revealed that every spike generated in these afferents was locked to the R-wave and thus exclusively driven by the arterial pulse, as shown for the three units in . We also encountered 3 muscle spindles that fired two or three spikes per cardiac circle. As shown in , the second spike usually closely followed the first. Two of those muscle spindles () showed a second cluster centred at the latency corresponding to the dicrotic notch of the pulse wave (cf. ).
Example of muscle spindle discharge locked to the arterial pulsations.
Pulse-wave driven muscle spindles.
Phase locking of ongoing spindle discharge
For the majority of spontaneously active afferents the ongoing discharge was primarily evoked by muscle stretch properties other than the pulse wave. However, the effects of cardiac rhythmicity could be discerned also in those afferents. The most pronounced effect was phase locking of individual spikes. In the literature such kind of behaviour is referred to as “resetting
,” and is described as an afferent that “discharged or failed to discharge at a fixed interval after the pulse” 
. Thus, such behaviour was predicted and searched for 
, but has not been previously found. No formal criteria have been developed to distinguish such afferents, therefore we relied on visual inspection of spike alignment in rasterplots and peak-and-trough patterns in density histograms. Such behavior is rarily seen in muscle spindle afferents, however three afferents clearly stood out. For example, the primary spindle afferent shown in discharged spontaneously at a low rate, around 10 imp/s. At 300 ms after the R-wave, corresponding to the ascending portion of the pulse wave, its instantaneous discharge rate exceeded 30 imp/s. After the peak there was a silent period of about 200 ms when the discharge virtually ceased. This is a very typical characteristic of dynamically sensitive group Ia muscle spindle afferents – they usually exhibit a silent period following an excitatory stimulus 
. Spiking activity resumed after 650 ms; this latency was constant across trials. In this subject, the R-R interval was 0.95 s and, due to the heart rate variability, the raster plot showed substantial jitter at the onset of the next systolic wave. Thus, in this case, phase locking of spiking activity resulted from post-stimulus depression.
Phase-locking and resetting of ongoing muscle spindle discharge.
In another Ia afferent phase locking was achieved through a different mechanism. As shown in the example illustrated in , phase locking occurred because of the excitatory effect of the upstroke of the pulse wave (). Three spikes generated at the time of the upstroke of the pulse wave showed slightly increased discharge rate and tight phase locking regardless of the timing of the preceding spike. The first inter-spike interval following the triplet was slightly prolonged, after which regular spiking activity resumed. R-triggered raster plots showed that these spikes were well aligned, although there was more jitter than during the pulse wave period.
By contrast, the third phase locking mechanism in spontaneously active afferents was based on inhibition by the upstroke of the pulse wave (). After being silenced for about 150 ms, approximately 250 ms after the R-wave the afferent resumed its firing with two phase-locked spikes. The following spikes showed considerable jitter and phase locking was not maintained.
Muscle spindle discharge modulated by arterial pulsations
In previous sections we demonstrated recordings from muscle spindles whose spiking activity was essentially driven by arterial pulsations and spontaneously active afferents in which some spikes apparently became phase-locked to the arterial pulsations. However, for the majority of tonically active muscle spindles – activated by the existing static stretch of the relaxed muscle – a more subtle modulation of discharge activity by the pulse wave was present. In these afferents, a modulation pattern could be discerned in the R-wave triggered average firing rates. This reflected a tendency to increase or decrease their instantaneous discharge rate while individual spikes were not locked to the arterial pulsations. The following analyses were undertaken to identify a proportion of afferents with significant pulse wave modulation, the proportion of total variance explained by the pulse wave, and the common modulatory pattern in the firing of muscle spindle afferents (see Methods).
Cardiac rhythmicity detected in the background discharge of muscle spindle afferents
Analyses in this section refer to data obtained from 51 muscle spindle afferents discharging in the absence of active muscle contraction, i.e. in relaxed muscles. The mean discharge rate of the muscle spindles ranged from 3.5 to 23.8 imps/s (8.3 imps/s, median) and the coefficient of variation ranged from 2.0 to 84.7% (8.2%, median). For each muscle spindle afferent we calculated the R-wave triggered average to identify the modulatory effect of the pulse wave on discharge frequency. shows examples of individual traces and averages from 10 muscle spindle afferents (eight innervating primary and two secondary muscle spindle endings). Five of those (A–E) showed increases in discharge rate (positive modulation) at the time of the upbeat of the pulse wave, while five showed an initial decrease (F–J). Note that some afferents had a biphasic modulation profile. shows the equivalent analysis in the frequency domain for the same afferents. The power spectra revealed significant peaks at the cardiac frequency and its harmonics, reflecting periodic modulations of discharge rate at the heart rate frequency. The higher harmonics represent the non-sinusoidal modulation as evidenced in . To assess the amount of pulse wave modulation we expressed the variance of the R-wave triggered average as a ratio of the total variance of the discharge rate. The amount of variance explained by arterial pulsations ranged from 0.3 to 61.7% (4.2% median). In 16 out of 51 muscle spindle afferents the variance explained by arterial pulsations was higher that 10%. To statistically assess whether the discharge rate modulation effect was indeed caused by the arterial pulsation - and not by other periodic components in the signal - the level of explained variance was compared to a set of permuted surrogate data (see Methods).
Examples of R-wave triggered discharge rate averages in pulse-wave modulated muscle spindles.
Examples of spectral analyses for the same ten afferents illustrated in , respectively.
Afferents with statistically significant pulse wave modulation
In 53% (27/51) of the afferents, pulse wave modulation was significant when evaluated against possible random effects. This effect was statistically significant in 60% (18/30) of the tested Ia afferents and in 43% (9/21) of the group II muscle spindle afferents. For the Ia afferents, the pulse modulation on average accounted for 19.1% of the variance in discharge rate; for group II afferents this was 9.4% (medians 15.3 vs. 7.6%). The largest modulatory effect within the cardiac circle peaked at about 225–300 ms (median 268 ms) after the R-wave. The mean amplitude of the peak modulation was 10.5% of the mean discharge rate. For the group Ia afferents the peak modulation was 14.3%, while for group II afferents it was 2.7% in (corresponding medians for group Ia and II afferents were 9.1% and 3.0% respectively). Regardless of the afferent type, the same number of afferents showed a positive or negative initial peak (14 vs. 13 afferents respectively). In 70% (19/27) of the afferents the modulation was biphasic and the initial peak was followed by a second peak of opposite polarity. The median amplitude of the second peak was 68% of the initial peak.
The relative size of the peak modulation correlated inversely with the mean discharge rate (rs
−0.54, p<0.05), while the absolute size of modulation expressed as imp/s was not influenced by discharge rate (rs
−0.13, p>0.05); this indicates that pulse wave modulation was largely additive. The relative size of the peak modulation showed a positive correlation with the coefficient of variation of discharge rate (rs
0.80, p<0.05; Spearman's rank correlation test). There was a weak relationship between cardiovascular parameters and the size of the modulatory effect across different subjects and afferents: the relative size of the peak modulation showed a weak correlation with mean blood pressure (rs
0.46, p<0.05), but not with pulse pressure (rs
0.15, p>0.05). The amount of variance explained by arterial pulsations was not correlated either with mean discharge rate (rs
−0.10, p>0.05) or coefficient of variance (rs
Effects at the level of muscle spindle afferent population assessed by PCA
To identify common features of the modulatory effect we used principal component analysis (PCA) to compare discharge rate modulation patterns across afferents. Most of the modulatory pattern features were explained by the first principal component (p<0.05): 48% of modulatory effect was accounted by the same pattern sharing common features between different afferents (). These analyses revealed a tri-phasic modulation pattern with three stable peaks reflecting increases and decreases in discharge rate. The distribution of the corresponding eigenvector coefficients confirmed that the arterial pulse wave had either an excitatory or an inhibitory effect on different afferents, as reflected by a positive and negative coefficient, respectively (). Presumably, the orientation of the muscle spindle ending with respect to a nearby blood vessel may result in loading (stretch), thereby increasing its firing rate, or unloading, and hence decreasing its discharge rate. The second principal component explained 21% of the variance (not shown) and was a modulation of the first principal component that captured the difference in latency of the modulation effect across afferents.
Principal component analysis (PCA) of 55 spontaneously active afferents in control (resting) conditions.
PCA of the power spectra further demonstrated the effect of arterial pulsations on muscle spindle discharge at the population level. The first principal component (p<0.05) explained 43% of the variance and revealed stable periodic modulation at the heart rate frequency and its harmonics (2f and 3f). The second principal component (explained variance 12%) was again a modulation of the first principal component, reflecting differences between afferents (not shown). For both time- and frequency-domain analyses, the afferents showing significant pulse wave modulation (determined from data resampling using circular permutation, as described in Methods) had larger eigenvector coefficients, as illustrated by the black histograms in .
Effect of physiological and cardiovascular challenges on the nature of pulse wave modulation of muscle spindles
An important question we intended to address is how much the strength of the pulse wave modulatory effect changes during physiological challenges that cause specific changes in cardiovascular parameters, sympathetic outflow or gamma motor neuron activity. In the following section our focus is on possible population effects and converging inputs from muscle spindles. Accordingly, we primarily report mean values while acknowledging that distributions might be skewed (see Methods).
Inspiratory capacity apnoea
This physiological manoeuvre - a maximal inspiratory breath-hold - is known to cause a sustained increase in muscle sympathetic nerve activity. The manoeuvre typically caused an initial fall in systolic blood pressure typically for up to 25 mmHg and a sustained fall in pulse pressure by 9 mmHg in average.
Muscle spindle discharge parameters were quantified during 30 s intervals measured in 22 spindle afferents during rest and apnoea. Ongoing discharge rate and discharge variability of muscle spindles were not influenced by the manoeuvre. This was true for nine primary and thirteen secondary afferents analysed separately (p>0.05; Wilcoxon test). Moreover, the overall size of discharge rate variability explained by arterial pulsations did not change (7.0 vs. 7.4% during control condition and apnoea respectively; p>0.05 Wilcoxon test), nor did modulation strength of single afferents (p>0.05; Wilcoxon test). Indeed, the average discharge modulation was similar during the apnoea and control period for both positively () and negatively modulated afferents (). There were no differences when group Ia and II afferents were analysed separately. Also, the strength of peak modulation was not influenced by the manoeuvre (6.0% vs. 6.1%; p>0.05; Wilcoxon test). The same conclusion was supported by analyses conducted on only those afferents that were significantly modulated (9/22) were selected: neither the amount of explained variance (14.1 vs. 15.1%) nor the size of peak modulation (8.6 vs. 8.3%, for control and apnoea respectively) were influenced by the apnoea (p>0.05; Wilcoxon test).
Comparison of R-wave triggered discharge rate averages between different physiological conditions.
We investigated whether pulse wave modulation was affected by acute muscle or skin pain. In 12 afferents investigated with muscle pain eight afferents showed significant modulation by arterial pulsations. Thirteen afferents were tested with skin pain, from which seven afferents were significantly modulated. On average, there was a small decrease in discharge rate during muscle pain (9.9 vs. 9.4 imp/s) and there was no effect of skin pain on the mean discharge rate of the muscle spindles (9.6 vs. 9.5 imp/s). The amount of variance explained by arterial pulsations was not affected by muscle pain (16.0% vs. 15.6% for control and pain respectively; p>0.05, n
12; Wilcoxon). With skin pain the amount of variance explained by arterial pulsations on average decreased from 8.3 to 5.3%. A decrease was observed for eight out of 13 afferents but the overall effect was not significant at the population level (p>0.05; Wilcoxon; ). However, from seven significantly modulated afferents the amount of explained variance decreased in six afferents and this effect was significant (p<0.05, n
7; Wilcoxon). No changes were detected in the amount of overall discharge variability between background and muscle- or skin-pain conditions (p>0.05; Wilcoxon). Changes in the pulse wave modulatory effect observed during the pain were not correlated with changes in heart rate (p>0.05) or changes in pulse pressure (p>0.05, n
25; Pearson correlation). In sum, these analyses indicate that no or only very weak effects from muscle or skin pain could be detected.
For seventeen muscle spindle afferents (14 primary and 3 secondary) cardiac rhythmicity was investigated during moderate strength voluntary muscle contraction. Twelve afferents (71%) showed significant modulation of discharge rate by arterial pulsations. All afferents were silent in relaxed muscles, but became engaged during active contraction most likely due to co-activation of gamma (fusimotor) neurones. Mean discharge rate during the active muscle contraction was comparable to the mean discharge rate observed in spontaneously active muscle spindles (10.2 vs. 9.6 imp/s, respectively); however, the coefficient of variance was higher during active contraction (21.7 vs. 10.8%). Due to this higher overall discharge variability the size of the explained variance was relatively small (). That is, only 2.4% of total variance was explained by arterial pulsations. In comparison, for spontaneously active afferents in relaxed muscles the arterial pulsations accounted for 9.4% (n
51) of the total variance. During active contraction the size of the net modulatory effect at the population level was considerable - at the peak it was 8.0% of the mean discharge rate. For comparison, the strength of discharge rate modulation in afferents responding to static stretch was 7.2%.