Acetylcholine receptors in the equatorial region of intrafusal muscle fibres modulate mouse muscle spindle sensitivity

Abstract

Key points

• Acetylcholine receptors are aggregated in the central regions of intrafusal muscle fibres.

• Single unit muscle spindle afferent responses from isolated mouse extensor digitorum longus muscle were recorded in the absence of fusimotor input to ramp and hold stretches as well as to sinusoidal vibrations in the presence and absence of the acetylcholine receptor blockers d‐tubocurarine and α‐bungarotoxin.

• Proprioceptive afferent responses to both types of stretch were enhanced in the presence of either blocker.

• Blocking acetylcholine uptake and vesicular acetylcholine release by hemicholinium‐3 also enhanced stretch‐evoked responses.

• These results represent the first evidence that acetylcholine receptors negatively modulate muscle spindle responses to stretch.

• The data support the hypothesis that the sensory nerve terminal is able to release vesicles to fine‐tune proprioceptive afferent sensitivity.

Abstract

Muscle spindles are complex stretch‐sensitive mechanoreceptors. They consist of specialized skeletal muscle fibres, called intrafusal fibres, which are innervated in the central (equatorial) region by afferent sensory axons and in both polar regions by efferent γ‐motoneurons. Previously it was shown that acetylcholine receptors (AChR) are concentrated in the equatorial region at the contact site between the sensory neuron and the intrafusal muscle fibre. To address the function of these AChRs, single unit sensory afferents were recorded from an isolated mouse extensor digitorum longus muscle in the absence of γ‐motoneuron activity. Specifically, we investigated the responses of individual sensory neurons to ramp‐and‐hold stretches and sinusoidal vibrations before and after the addition of the competitive and non‐competitive AChR blockers d‐tubocurarine and α‐bungarotoxin, respectively. The presence of either drug did not affect the resting action potential discharge frequency. However, the action potential frequencies in response to stretch were increased. In particular, frequencies of the dynamic peak and dynamic index to ramp‐and‐hold stretches were significantly higher in the presence of either drug. Treatment of muscle spindle afferents with the high‐affinity choline transporter antagonist hemicholinium‐3 similarly increased muscle spindle afferent firing frequencies during stretch. Moreover, the firing rate during sinusoidal vibration stimuli at low amplitudes was higher in the presence of α‐bungarotoxin compared to control spindles also indicating an increased sensitivity to stretch. Collectively these data suggest a modulation of the muscle spindle afferent response to stretch by AChRs in the central region of intrafusal fibres possibly fine‐tuning muscle spindle sensitivity.

Key points

• Acetylcholine receptors are aggregated in the central regions of intrafusal muscle fibres.

• Single unit muscle spindle afferent responses from isolated mouse extensor digitorum longus muscle were recorded in the absence of fusimotor input to ramp and hold stretches as well as to sinusoidal vibrations in the presence and absence of the acetylcholine receptor blockers d‐tubocurarine and α‐bungarotoxin.

• Proprioceptive afferent responses to both types of stretch were enhanced in the presence of either blocker.

• Blocking acetylcholine uptake and vesicular acetylcholine release by hemicholinium‐3 also enhanced stretch‐evoked responses.

• These results represent the first evidence that acetylcholine receptors negatively modulate muscle spindle responses to stretch.

• The data support the hypothesis that the sensory nerve terminal is able to release vesicles to fine‐tune proprioceptive afferent sensitivity.

Introduction

Coordinated movements, including locomotion and their control, require proprioceptive information, i.e. information about muscle tone as well as position and movement of the extremities in space (Dietz, 2002). Muscle spindle afferents are the primary proprioceptive sensory receptors. They detect how much and how fast a muscle is lengthened (Proske & Gandevia, 2012; Kröger, 2018). Muscle spindle afferent feedback is required for a stable posture and directed movement but also for the realignment of fractured bones (Blecher et al. 2017a), for the maintenance of spinal alignment (Blecher et al. 2017b), and for basic locomotor recovery and circuit reorganization after spinal cord injury (Takeoka et al. 2014). In adult mice, muscle spindles are 200–400 μm long and consist of three to eight specialized intrafusal muscle fibres. Intrafusal fibres lie in parallel with extrafusal muscle fibres and are surrounded by a connective tissue capsule (Banks, 1994; Bewick & Banks, 2015; Kröger, 2018). Three types of intrafusal fibres can be distinguished: nuclear bag1, nuclear bag2 and nuclear chain fibres. Sensory neurons innervating bag1 fibres respond maximally to the velocity of changes in muscle fibre length (dynamic sensitivity) and those innervating bag2 fibres as well as nuclear chain fibres respond maximally to the amount of stretch (static sensitivity; Banks, 1994).

The central (equatorial) part of intrafusal muscle fibres is innervated by two types of afferent proprioceptive sensory neurons (termed ‘type Ia afferents’ and ‘type II afferents’ according to their axonal conduction velocity; Banks, 2015). Type Ia afferents form so called annulospiral sensory nerve endings whereas type II afferents flank the Ia afferents (Schroder et al. 1989; Sonner et al. 2017). Afferent neurons generate action potentials with frequencies that are proportional to the size of the stretch and to the rate of stretching (De‐Doncker et al. 2003). The cell bodies of these sensory neurons constitute a minor fraction of all neurons in the dorsal root ganglion (DRG) and their peripheral endings can be labelled by antibodies against the vesicular glutamate transporter 1 (VGluT1; Wu et al. 2004).

In addition to the sensory neurons, intrafusal muscle fibres are innervated by efferent γ‐motoneurons, the so‐called fusimotor innervation (Banks, 1994). These neurons constitute about 30% of all motoneurons in the ventral horn of the spinal cord. Axons of γ‐motoneurons enter the spindle together with the sensory fibres in the central region of the spindle but innervate intrafusal muscle fibres exclusively at both ends (polar regions) where they form a cholinergic synapse that appears functionally similar to the neuromuscular junction formed by α‐motoneurons on extrafusal muscle fibres. γ‐Motoneurons adjust the sensitivity of muscle spindle afferents by inducing contraction in the polar region of the muscle spindle to exert tension on the central equatorial region of the muscle fibre (Banks, 1994; Proske, 1997). This allows for continuous control of the mechanical sensitivity of spindles over the wide range of lengths and velocities that occur during normal motor behaviours. While the general functional properties of muscle spindles are rather well documented, the molecular basis of this function and potential modulatory activities remain mostly unknown (Kröger, 2018).

In a previous study, we analysed the development of synapse‐like specializations in muscle spindles (Zhang et al. 2014). We reported a concentration of fetal (γ‐subunit‐containing) as well as adult (ε‐subunit‐containing) type acetylcholine receptors (AChRs) in the equatorial region of intrafusal fibres at the contact site with the sensory nerve endings. Similarly, high concentrations of other proteins characteristic of cholinergic synapses, including the vesicular acetylcholine transporter, choline acetyltransferase and the AChR‐associated protein rapsyn, were detected at this nerve‐to‐muscle contact site (Zhang et al. 2014). The function of these cholinergic specializations, however, remained unknown. In the present study, we analysed the response of individual proprioceptive sensory afferent neurons to ramp‐and‐hold stretches as well as to sinusoidal vibrations in the presence and absence of a competitive (d‐tubocurarine) or a non‐competitive (α‐bungarotoxin) AChR blocker. Moreover, we inhibited the high‐affinity vesicular ACh transporter using hemicholinium‐3. We show that all drugs increased muscle spindle sensitivity to stretch, demonstrating that AChRs in the central region of intrafusal muscle fibres modulate the sensitivity of muscle spindle sensory afferents.

Methods

Ethical approval

Use and care of animals and all experimentation were approved by German authorities and according to national law (§7 TierSchG; license Az.: 55.2‐1‐54‐2532.8‐160‐13) and conducted in accordance with the guidelines of the Ludwig‐Maximilians‐University. Experimental protocols were designed to minimize suffering and the number of animals used in the study. The work complies with the ethical principles under which The Journal of Physiology operates. At most five adult animals were housed in a sterile cage on a 12 h light–dark cycle. A total of 19 male C57Bl/6JRj mice aged 10–15 weeks with a weight of 22–28 g were used for the electrophysiological analyses. In the case of α‐bungarotoxin and d‐tubocurarine, five muscle spindles each from a different mouse were analysed for each of the drug conditions. In the case of hemicholinium‐3, four muscle spindle afferents, each from a different mouse, were recorded. In addition, we used five muscle spindle recordings, each from a different mouse, as controls. The values of the control group were compared to all three drugs. The immunohistochemical analyses was performed on soleus, quadriceps and extensor digitorum longus (EDL) muscles from three additional male C57Bl/6JRj mice.

Immunofluorescence

Immunofluorescence staining was performed as described previously (Zhang et al. 2014). To obtain muscle tissue for immunohistochemistry, mice were deeply anaesthetized using ketamine (Pfizer, Berlin, Germany) and xylazine (Bayer AG, Leverkusen, Germany). After transcardial perfusion with phosphate‐buffered saline (PBS) followed by 4% paraformaldehyde, the EDL muscle was dissected. Muscles were post‐fixed in 4% paraformaldehyde for 2 h and then incubated in 30% sucrose in PBS overnight at 4°C. Fixed muscles were cryo‐preserved in Tissue‐Tek O.C.T. Compound (Sakura Finetek Europe, AJ Alphen an den Rijn, Netherland) and cryo‐sectioned along the longitudinal axis at 20–30 μm thickness.

Frozen sections were blocked in PBS containing 0.2% Triton X‐100 (Sigma‐Aldrich Chemie GmbH, Taufkirchen, Germany) and 1% bovine serum albumin (Carl Roth GmbH, Karlsruhe, Germany; blocking solution) for 30 min at room temperature and incubated with the primary antibody in blocking solution at 4°C overnight. Sensory nerve terminals were stained using antibodies from guinea pig against the vesicular glutamate transporter 1 (VGluT1) diluted 1:1000 (AB5905, Millipore, Darmstadt, Germany). Primary antibodies were detected by incubating the sections for 1 h with Alexa647‐conjugated donkey anti‐guinea pig secondary antibody diluted 1:500 (AP1493SD, Millipore). AChRs were visualized using Alexa488‐conjugated α‐bungarotoxin diluted 1:1000 in blocking solution (B13422, Life Technologies, Darmstadt, Germany). The nuclei were routinely stained using 4′,6‐diamidino‐2‐phenylindole (Carl Roth) at a concentration of 2 μg/ml in blocking solution.

After staining the sections were embedded in Mowiol mounting medium (Carl Roth) and analysed using a Zeiss LSM 710 laser scanning confocal microscope (Carl Zeiss AG, Oberkochen, Germany). Sequentially scanned confocal Z‐stacks of whole muscle spindles were obtained using 1 μm optical sections and compiled using the ZEN2009 software (Zeiss). Laser power levels, photomultiplier gain levels, scanning speed and the confocal pinhole size were kept constant between experimental and control specimens. Digital processing of entire images, including adjustment of brightness and contrast, was performed using the Java image processing program software package Fiji (Schindelin et al. 2012).

Extracellular muscle spindle afferent recordings

Proprioceptive sensory neuron responses to stretch were assayed using an isolated muscle–nerve preparation as previously described (Wilkinson et al. 2012; Franco et al. 2014). Mice were sacrificed by cervical dislocation and the EDL muscle together with the deep peroneal branch of the sciatic nerve were dissected and placed in an oxygenated tissue bath containing artificial cerebrospinal fluid (ACSF; Wilkinson et al. 2012). The tendons were sutured to a fixed post and at the other end to a lever arm, connected to a dual force and length controller (300C‐LR, Aurora Scientific, Dublin, Ireland) allowing the simultaneous recording of muscle tension, muscle length and muscle spindle afferent discharges. Baseline muscle length (L _o) was defined as the length at which maximal twitch contractile force was generated. Sensory activity was sampled using a suction electrode (tip diameter 50–70 μm) which was connected to an extracellular amplifier (Model 1800, A&M Systems, Elkhart, IN, USA). A signal was classified as being from a putative muscle spindle sensory afferent if it displayed a characteristic instantaneous frequency response to stretch as well as a pause during twitch contraction (Wilkinson et al. 2012). For every muscle spindle afferent recording, triplicates of ramp‐and‐hold stretches (L _o plus 7.5% of L _o; ramp speed 40% L _o s⁻¹; stretch duration: 4 s with 45 s intervals between each stretch) were recorded and averaged. In addition, 16 sinusoidal vibrations (amplitudes: 5, 10, 50 and 100 μm, each with a frequency of 10, 25, 50 and 100 Hz) were recorded for 9 s and the values were expressed as impulses per 9 s (imp/9 s). Then, 30 μM d‐tubocurarine (T2379, Sigma‐Aldrich Chemie GmbH), 1.25 μM α‐bungarotoxin (B1601, Life Technologies, Carlsbad, CA, USA) or 30 μM hemicholinium‐3 (HC‐3; Sigma‐Aldrich) was added to the oxygenated ACSF. At this concentration, α‐bungarotoxin and tubocurarine completely inhibit neuromuscular transmission (Ganguly et al. 1978; Chang et al. 1989; Wenningmann & Dilger, 2001). For control measurements 100 μL ACSF was added instead of a drug. Initial control experiments showed no significant differences in the drug activity after equilibration times between 1 and 4 h. Therefore, an equilibration time of 1 h for ACSF, d‐tubocurarine and HC‐3, respectively, and 3 h for α‐bungarotoxin were chosen. After equilibration, the response of the same muscle spindle afferent neuron to the same series of three repetitions of ramp‐and‐hold stretches and to the vibrations was determined. This allowed a precise comparison of the stretch‐evoked responses of individual single unit muscle spindle afferents in the absence and presence of the drug. The resting discharge (average baseline firing rate) as well as the dynamic peak (highest firing rate during ramp – baseline firing rate), the dynamic index (dynamic peak – firing rate 0.45–0.55 s into stretch – baseline firing rate) and the static response (firing rate 3.25–3.75 s into stretch – baseline firing rate) and the time of silence (time after the end of the ‘ramp‐and‐hold’ stretch and the first action potential) were determined (Wilkinson et al. 2012; see Fig. 2).

Figure 2. Inhibition of AChRs increases stretch‐evoked sensory afferent instantaneous frequencies.

A and B, action potentials of the same muscle spindle proprioceptive afferent during a ramp‐and‐hold stretch in the absence (A) and presence (B) of α‐bungarotoxin. C, a ramp‐and‐hold stimulus of 4 s with a ramp speed of 40% L _o s⁻¹ and an overall 7.5% L _o length change of the same spindle as in A and B. D, the response of a different muscle spindle afferent to stretch to illustrate the five different parameters that were analysed at different time points before, during and after the stretch: resting discharge (RD), dynamic peak (DP), dynamic index (DI), static response (SR) and time silenced (TS). E, a representative image of the instantaneous frequency of a single proprioceptive sensory unit response to stretch before (green dots) and after (grey dots) the addition of α‐bungarotoxin (α‐Btx). The presence of α‐bungarotoxin did not influence the resting discharge (the small variability shown in E is not significant), but increased the instantaneous frequencies at all other time points during the stretch. [Color figure can be viewed at wileyonlinelibrary.com]

For data analysis, individual sensory neurons were identified by spike shape and the interspike interval using the Spike Histogram feature of Lab Chart (ADInstruments, Sydney, Australia) and spindle afferent baseline firing rate, dynamic peak, dynamic index as well as the static stretch response were determined as described above. Only recordings in which an individual muscle spindle afferent unit could be unambiguously identified were included in the analysis. Action potentials from additional potential muscle spindles that appeared during the stretch were not scored. No attempt was made to discriminate type Ia from type II afferents (see Wilkinson et al. (2012) for a detailed discussion).

At the end of each recording, muscle health was ensured by determining the maximal contractile force during a direct tetanic stimulation with paddle electrodes embedded in the incubation chamber (500 ms train at 120 Hz frequency and 0.5 ms pulse length, supramaximal voltage; Grass S44 stimulator; Wilkinson et al. 2012). The diameter of the EDL muscle was determined at L _o and with this information, the peak force was calculated and compared to the previously reported peak force of the EDL of 23,466 N/cm² (Larsson & Edstrom, 1986; Brooks & Faulkner, 1988). Neither the maximal directly evoked muscle force after tetanic stimulation nor the resting spindle discharge varied significantly during the experimental period, suggesting that the spindle afferents were not undergoing time‐dependent changes in firing properties and that the muscle was returning to resting length following each stretch.

Statistics

Only single unit spindle afferent responses that could be recorded without interruption throughout the entire experiment, i.e. before and after the addition of ACSF, d‐tubocurarine, α‐bungarotoxin or hemicholinium‐3, were included in the analysis. For ramp‐and‐hold stretches, baseline values for all parameters (baseline firing rate, dynamic peak, dynamic index and the static stretch response) were determined as an average of three stretches before drug addition. Likewise, post‐drug addition firing frequencies were determined after equilibration from an average of three stretches. Frequencies of individual muscle spindle afferents before drug addition were subtracted from the frequencies after drug addition. The same time points within a ramp‐and‐hold stretch were used to calculate the change in firing in our no‐drug control group. The mean of the overall changes in firing rate of all three drug groups were compared statistically using one‐way ANOVA (factor: drug) with Dunnett's post hoc test to the no‐drug (ACSF) control group. Values are reported as mean of the frequency change from pre‐ to post‐drug addition (Δmean; imp/s) in a dot plot with each dot representing an individual muscle spindle afferent.

For the sinusoidal vibrations, the total number of action potentials per 9 s (imp/9 s) for each of the 4 frequencies and 4 amplitudes before drug addition were subtracted from the corresponding values (imp/9 s) after drug addition. Data are shown as Δ imp/9 s in a dot plot with each dot representing an independent experiment. Statistical significance was determined using the three‐way ANOVA (factors: drug, amplitude and frequency). Additionally, all the response to the 16 different vibrations in the absence and presence of all four drugs were compared individually with one‐way ANOVA (factor drug; Dunnett's multiple comparisons test). All analyses were performed using Prism (v8; GraphPad Software, Inc., La Jolla, CA, USA). The level of significance (P‐value) for all statistical tests was set at ^* P < 0.05, ^** P < 0.01, ^*** P < 0.001.

Results

AChRs are concentrated at the equatorial region of the muscle spindle and colocalize with the sensory nerve terminal

Previously, the presence of mRNA coding for the α‐ and ε‐subunit of the AChR in the central part of intrafusal fibres and the concentration of α‐, γ‐ and ε‐subunit‐containing AChRs at the contact site of the sensory nerve terminal and intrafusal muscle fibres was reported (Sanes et al. 1991; Hippenmeyer et al. 2002; Zhang et al. 2014). To investigate the precise distribution of the AChRs with respect to the sensory nerve terminal, we analysed sections stained with fluorescently labelled α‐bungarotoxin and anti‐VGluT1 antibodies. High‐resolution confocal microscopic analyses of the central region of intrafusal muscle fibres demonstrated a precise codistribution of the AChRs with the VGluT1 immunoreactivity (Fig. 1). A similar overall distribution of the AChRs was observed in muscle spindles from EDL, quadriceps and soleus muscles (data not shown). As discussed previously (Zhang et al. 2014), this labelling represents a concentration of the AChRs in the intrafusal fibre plasma membrane at the contact region with the sensory nerve terminal. Little α‐bungarotoxin labelling was detected in the area of the intrafusal muscle fibre membrane between the annulospiral endings. A similar spatial distribution of AChRs was observed in nuclear bag and nuclear chain fibres from several different muscles. Based on the staining intensity, the concentration of the AChRs at the contact site between sensory neuron and intrafusal fibre appeared lower, compared to the concentration of AChRs at neuromuscular junctions on extrafusal fibres or at γ‐motoneuron endplates (data not shown).

Figure 1. AChRs and sensory nerve terminals colocalize in the central region of muscle spindles.

A and B, high‐resolution confocal z‐scan of the distribution of the AChR (green channel, A) and the sensory nerve terminal indicated by anti‐VGluT1 immunoreactivity (red channel, B) in the central region of a nuclear chain fibre of a soleus muscle. Panel C shows a merged image. Note the precise overlap of both staining patterns and the concentration of AChR labelling at the contact site between sensory neuron and intrafusal muscle fibre. [Color figure can be viewed at wileyonlinelibrary.com]

d‐Tubocurarine, α‐bungarotoxin and hemicholinium‐3 increase stretch‐evoked action potential frequencies of muscle spindle afferents

To investigate the function of the AChRs at the contact region between the sensory nerve terminal and the intrafusal muscle fibre, we analysed individual muscle spindle afferent unit responses to a ramp‐and‐hold stretch (L _o plus 7.5% of L _o; ramp speed 40% L _o s⁻¹; Fig. 2 C), before (Fig. 2 A) and after (Fig. 2 B) the addition of two AChR blockers (d‐tubocurarine and α‐bungarotoxin, respectively) and a choline uptake inhibitor (hemicholinium‐3). All drugs increased the firing frequency of stretch‐evoked action potentials during ramp‐and‐hold stretches when compared to the frequencies of the same muscle spindle afferents before drug addition (Fig. 2 E for α‐bungarotoxin). Five different time points examining firing rates before, during and after the stretch were analysed in detail (Fig. 2 D). Analysis of the overall effect of the drugs using one‐way ANOVA (factor: drug) revealed no differences for the resting discharge (P = 0.0610), but statistically significant differences for the dynamic peak (P = 0.0100), dynamic index (P = 0.0204) and static response (P = 0.0459).

For better comparison of the specific effect of the individual drug, the results were expressed as Δmean, which represents the mean of the difference of the frequency (imp/s) before compared to after drug addition. Neither AChR inhibitor had an effect on the resting discharge (Δmean control: 2.092 imp/s; Δmean d‐tubocurarine: 1.346 imp/s; Δmean α‐bungarotoxin: 4.758 imp/s; Dunnett's post hoc test relative to ACSF control: P = 0.9664, control vs. d‐tubocurarine; P = 0.4313, control vs. α‐bungarotoxin; Figs 2 E and 3 A) or the time silenced (data not shown). However, dynamic peak (Δmean control: −12.04 imp/s; Δmean d‐tubocurarine: 9.992 imp/s; Δmean α‐bungarotoxin: 26.53 imp/s; Fig. 3 B) and dynamic index (Δmean control: −5.413 imp/s; Δmean d‐tubocurarine: 2.78 imp/s; Δmean α‐bungarotoxin: 5.799 imp/s; Fig. 3 C) were increased compared to the same muscle spindle afferent before addition of the drug. The difference was statistically significant for the dynamic peak (P = 0.0442, control vs. d‐tubocurarine; P = 0.0027, control vs. α‐bungarotoxin; Fig. 3 B) and for the dynamic index (P = 0.0412, control vs. d‐tubocurarine; P = 0.0073, control vs. α‐bungarotoxin; Fig. 3 C). α‐Bungarotoxin also significantly increased the static response (Δmean control: −5.581 imp/s; Δmean α‐bungarotoxin: 18.69 imp/s; P = 0.0169, control vs. α‐bungarotoxin; Fig. 3 D). d‐Tubocurarine did not affect the static response (Δmean d‐tubocurarine: 8.276 imp/s; P = 0.2122, control vs. d‐tubocurarine). Overall, our results demonstrate a higher sensitivity of muscle spindle responses to stretch in the presence of either AChR inhibitor.

Figure 3. d‐Tubocurarine, α‐bungarotoxin and hemicholinium‐3 have no effect on resting discharge frequencies but increase the firing rate during ramp‐and‐hold stretches.

The effect of stretches of 7.5% of L _o on single unit proprioceptive afferents was determined before and after addition of d‐tubocurarine, α‐bungarotoxin or hemicholinium‐3. Each circle represents the change of the instantaneous frequency (impulses per second) of a single muscle spindle afferent unit where the impulses before drug addition were subtracted from the values after drug addition. The mean of the change is indicated as horizontal line. Five spindles, each from a different mouse, were evaluated in the presence and absence of α‐bungarotoxin or d‐tubocurarine (n = 5). Four mice were evaluated in the presence and absence of hemicholinium‐3 (n = 4). Control values represent muscle spindle activities where ACSF was added instead of a drug. Addition of either drug increased the firing frequencies during the dynamic peak (B) and the dynamic index (C) whereas the resting discharge frequency (A) remained unchanged. α‐Bungarotoxin also affected the static response (D). Statistical significance was evaluated using one‐way ANOVA (factor: drug) with Dunnett's post hoc corrections. The small increase (A) or decrease (B–D) of the control frequency after ACSF addition compared to the frequency before ACSF addition was not statistically significant.

Hemicholinium‐3 (HC‐3) is a blocker of the high‐affinity uptake of choline into nerve terminals including the α‐motoneuron terminal at the neuromuscular junction (Yu & Van der Kloot, 1991). Accordingly, application of HC‐3 to neuromuscular junctions results in an inhibition of ACh synthesis and release (Carpenter & Woodruff, 1987). We used HC‐3 to investigate if the blockade of vesicular acetylcholine uptake and release influenced the proprioceptive afferent responses to stretch. We observed no effect of HC‐3 on the change of the action potential frequency before and after drug addition at rest (Δmean control: 2.092 imp/s; Δmean HC‐3: −1.612 imp/s; P = 0.235; Fig. 3 A) and on the time silenced (not shown). This demonstrates that HC‐3, similar to both AChR inhibitors, did not activate γ‐motoneuron endplates and had no effect on muscle spindle sensory afferents at rest. However, in response to ramp‐and‐hold stretches, the change of the frequency of muscle spindle afferents during the dynamic peak (Δmean control: −12.04 imp/s; Δmean HC‐3: 26.7 imp/s; Fig. 3 B) and dynamic index (Δmean control: −5.413 imp/s; Δmean HC‐3: 16.35 imp/s; Fig. 3 C) were increased. The increase of the frequency was statistically significant (Dunnett's post hoc test relative to ACSF control; dynamic peak P = 0.0111; dynamic index P = 0.0071). We observed no effect of HC‐3 on the static response (Δmean control: −5.581 imp/s; Δmean HC‐3: 8.867 imp/s; static response P = 0.228; Fig. 3 D). These results demonstrate that ACh uptake, synthesis and release modulate muscle spindle afferent firing rates in response to stretch.

The absolute values of the mean muscle spindle afferent frequencies before and after drug addition together with the corresponding standard deviations are given in Table 1. Comparing these values in the presence and absence of either drug revealed an increase in the response to stretch of approximately 20% in the presence of either AChR inhibitor and of approximately 30% in the presence of HC‐3. Collectively, our results demonstrate a role of AChRs during muscle spindle afferent responses to stretch and suggest a modulatory function for AChRs during proprioception.

Table 1.

Original values (frequencies in imp/s) of the mean responses to ramp‐and‐hold stretches used to determine the Δmean in Fig. 3

	Mean (imp/s)
	Control		α‐Bungarotoxin		d‐Tubocurarine		HC‐3
Stages during ramp‐and‐hold stretch	Before	After	Before	After	Before	After	Before	After
RD	13.0 ± 6.0	15.1 ± 5.8	18.4 ± 4.4	23.2 ± 5.8	16.4 ± 3.6	17.8 ± 2.4	20.8 ± 1.5	19.2 ± 2.8
DP	94.2 ± 20.6	88.0 ± 31.5	56.8 ± 21.6	83.3 ± 34.5	60.5 ± 23.5	70.5 ± 7.8	50.5 ± 14.7	77.2 ± 12.8
DI	37.3 ± 9.3	34.2 ± 11.6	27.9 ± 10.3	33.7 ± 10.4	27.3 ± 8.6	30.1 ± 9.2	26.1 ± 5.3	42.5 ± 5.4
SR	42.2 ± 12.9	40.3 ± 18.3	17.7 ± 7.4	36.4 ± 22.9	21.5 ± 10.5	29.8 ± 16.2	15.6 ± 1.8	24.4 ± 5.5

The values are means ± SD with n = 5 for control, α‐bungarotoxin and d‐tubocurarine and n = 4 for HC‐3. DI, dynamic index; DP, dynamic peak; RD, resting discharge; SR, static response.

α‐Bungarotoxin increases muscle spindle responses during sinusoidal vibration stimuli

To analyse the function of the AChRs specifically during the dynamic response to stretch, we determined the effect of both AChR blockers and of HC‐3 on muscle spindle afferent responses to sinusoidal vibrations varying in both displacement and frequency. Stimuli included four different frequencies (10, 25, 50 and 100 Hz) and four different amplitudes (5, 10, 50 and 100 μm). Compared to the ramp‐and‐hold stimuli, small amplitude sinusoidal vibrations test dynamic responsiveness at much smaller length changes and, thus, provide more specific information about the dynamic sensitivity of muscle spindle afferents (Brown et al. 1967). We observed an increase in the firing frequency in response to sinusoidal vibrations in the presence of all three drugs. Statistical analysis with three‐way ANOVA (factors: drug, amplitude and frequency) showed that the change of the frequency before and after drug addition was statistically significant for the factor drug (P = 0.0003) but not for the factors amplitude (P = 0.6465) or frequency (P = 0.0602). The effect was also significant for the drug by frequency analysis (P = 0.0176) but not for the drug by amplitude (P = 0.5672), amplitude by frequency (P = 0.7768) or drug by amplitude and frequency analysis (P = 0.522).

Additionally, each of the 16 vibrational stimuli was analysed individually before and after drug addition with one‐way ANOVA and Dunnett's multiple comparison test. This analysis revealed that the frequencies were significantly increased in the presence of α‐bungarotoxin particularly at high frequencies (50 and 100 Hz) and small amplitudes (5 and 10 μm; Fig. 4 B and C) but not at higher amplitudes (50 and 100 μm; Fig. 4 D and E) compared to the ACSF control. We observed no significant change in the presence of d‐tubocurarine or HC‐3 (Fig. 4 B–E). The absolute values (imp/9 s ± SD) for the responses to sinusoidal vibrations are shown in Table 2. In summary, the analysis of the effect of the drugs on muscle spindle afferent responses to vibrations further support a role for AChRs in modulating muscle spindle function during the dynamic phase of a stretch.

Figure 4. Response of muscle spindle sensory afferents to sinusoidal vibration stimuli in the presence and absence of cholinergic inhibitors.

A, individual action potentials of a representative muscle spindle afferent in response to sinusoidal length changes displayed below (stretch is represented as upwards deflection). While in the absence of any drug (left side), this muscle spindle afferent entrained with every 4th vibration, addition of d‐tubocurarine (right side of A) increased the firing frequency so that entrainment was observed every second to third vibration. B–E, quantification of 16 different sinusoidal vibration stimuli (5, 25, 50 and 100 μm amplitude with each of the frequencies 10, 25, 50 and 100 Hz). Each circle represents the difference of the impulses per 9 s (Δimp/9 s) where the values before drug (or ACSF in the case of controls) addition were subtracted from those after drug addition. The mean of the 5 (ACSF, α‐bungarotoxin or d‐tubocurarine) or 4 (hemicholinium‐3) individual experiments is indicated as red line. Asterisks represent statistically significant differences between the muscle spindles before and after addition of the drug, as determined using one‐way ANOVA (factor: drug) with Dunnett's post hoc test relative to the no drug control group. The non‐normalized values for each condition ± SD are shown in Table 2. Note the increased firing frequencies in the presence of α‐bungarotoxin particularly at small amplitudes and high frequencies. [Color figure can be viewed at wileyonlinelibrary.com]

Table 2.

Original values (impulses per 9 s) of the mean responses of muscle spindle afferents to sinusoidal vibrations of different amplitudes and frequencies

		Amplitude
Frequency		5 μm	10 μm	50 μm	100 μm
Control
10 Hz	Before	119.5 ± 61.6	148.3 ± 80.2	155.9 ± 83.8	176.3 ± 56.7
	After	135.2 ± 49.6	133.8 ± 46.8	130.4 ± 42.7	191.6 ± 60.1
25 Hz	Before	122.9 ± 65.4	207.5 ± 68.1	221.5 ± 39.1	234.5 ± 13.4
	After	143.2 ± 58.2	211.8 ± 26.4	229.4 ± 8.3	267.6 ± 76.9
50 Hz	Before	125.2 ± 67.5	241.4 ± 89.6	433.2 ± 257.1	448.4 ± 71.4
	After	143.0 ± 56.3	234.4 ± 38.6	358.6 ± 79.8	454.6 ± 8.7
100 Hz	Before	124.2 ± 66.9	232.8 ± 95.2	455.8 ± 239.1	887.9 ± 231.2
	After	138.8 ± 56.4	236.4 ± 43.3	484.6 ± 220.3	716.2 ± 210.8
α‐Bungarotoxin
10 Hz	Before	182.5 ± 29.8	144.8 ± 43.9	162.0 ± 35.5	162.4 ± 36.2
	After	208.6 ± 57.4	191.8 ± 51.2	195.2 ± 52.7	207.4 ± 66.8
25 Hz	Before	158.8 ± 60.4	202.2 ± 44.6	206.8 ± 36.9	224.8 ± 0.4
	After	207.8 ± 49.3	203.2 ± 45.1	202.8 ± 44.4	246.2 ± 55.1
50 Hz	Before	156.2 ± 68.3	221.4 ± 69.6	344.8 ± 131.0	386.6 ± 88.7
	After	217.4 ± 54.0	355.8 ± 128.7	390.0 ± 119.5	405.4 ± 90.2
100 Hz	Before	155.6 ± 57.1	233.2 ± 79.4	385.0 ± 159.8	729.0 ± 290.4
	After	228.6 ± 62.2	378.6 ± 149.8	651.4 ± 303.1	767.0 ± 267.5
d‐Tubocurarine
10 Hz	Before	147.2 ± 32.4	137.2 ± 40.0	141.6 ± 40.9	168.6 ± 22.8
	After	152.6 ± 31.2	148.6 ± 39.4	161.4 ± 24.6	180.2 ± 1.0
25 Hz	Before	155.2 ± 36.0	198.2 ± 44.4	207.8 ± 36.9	227.0 ± 1.7
	After	170.8 ± 26.7	211.2 ± 28.1	227.4 ± 1.6	227.0 ± 2.1
50 Hz	Before	161.4 ± 38.6	223.6 ± 56.8	337.0 ± 99.4	450.4 ± 0.5
	After	181.8 ± 34.9	279.4 ± 97.2	403.0 ± 62.0	450.2 ± 0.4
100 Hz	Before	157.6 ± 34.7	227.4 ± 65.2	378.4 ± 91.9	637.8 ± 215.0
	After	178.0 ± 32.5	294.4 ± 92.2	509.2 ± 207.7	672.4 ± 211.5
HC‐3
10 Hz	Before	190.7 ± 12.8	209.7 ± 7.3	212.3 ± 17.4	207.3 ± 17.9
	After	242.3 ± 65.2	255.0 ± 52.2	251.7 ± 43.9	267.0 ± 46.6
25 Hz	Before	183.7 ± 20.0	220.7 ± 8.0	330.3 ± 1.2	317.7 ± 2.0
	After	229.7 ± 60.8	267.0 ± 46.8	358.0 ± 36.1	365.7 ± 34.9
50 Hz	Before	174.0 ± 18.7	228.7 ± 82.4	379.3 ± 93.2	600.3 ± 87.0
	After	229.3 ± 60.9	258.7 ± 88.1	400.7 ± 75.8	656.0 ± 25.6
100 Hz	Before	181.7 ± 16.5	228.3 ± 73.6	381.7 ± 240.9	704.0 ± 247.7
	After	227.0 ± 71.6	266.7 ± 102.1	436.7 ± 210.7	826.7 ± 134.2

The values were used to determine the frequency change as a percentage in Fig. 4B–E. The values are means ± SD with n = 5 for control (ACSF), α‐bungarotoxin and d‐tubocurarine and n = 4 for HC‐3.

Discussion

In this study, we blocked AChR function using a non‐competitive (α‐bungarotoxin) as well as a competitive (d‐tubocurarine) antagonist and tested the reaction of muscle spindles to stretch by using two different kinds of protocols, i.e. ramp‐and‐hold stretches and sinusoidal vibrations. During both kinds of stretch protocols, the action potential frequency was increased in the presence of either drug, whereas no change of the post‐ramp time silenced, or in the discharge frequency at resting length was detectable. We observed a similar effect, i.e. increase in muscle spindle sensitivity during stretch, after inhibiting the high‐affinity choline uptake system using HC‐3. These results provide the first evidence for a function of AChRs in the equatorial region of the muscle spindle.

Several studies have previously analysed the effect of cholinergic agonists and antagonists on muscle spindle firing (for review see Carr & Proske, 1996). For example, application of cholinergic agents (including ACh, succinylcholine or nicotine) to muscle spindles induced strong excitatory activity in cat soleus muscle in vivo, which could be blocked by d‐tubocurarine (Granit et al., 1953; Albuquerque & Smith, 1964; Rack & Westbury, 1966; Smith & Albuquerque, 1967). This activity was concluded to be entirely due to a contraction of the polar regions of intrafusal fibres rather than activity acting directly on sensory nerve terminals. Likewise, intravenous injection of d‐tubocurarine into anaesthetized cats followed by fusimotor stimulation resulted in a small transient increase of the discharge frequencies in 55%, a decrease in the frequencies in 29% and an unchanged discharge frequency in 15% of the cases during the first 20 min after d‐tubocurarine administration (Smith & Albuquerque, 1967). In these and several other studies (reviewed by Carr & Proske, 1996), the drugs were either applied systemically, for example by intravenous infusion, or their direct effect on stretch‐evoked action potentials was not recorded. Therefore, these studies could not distinguish effects of the drugs at neuromuscular junctions on extrafusal fibres, γ‐motoneuron endplates or AChRs in the central region of intrafusal fibres.

Our study differs from previous studies in several other aspects. We specifically analysed the effect of antagonists instead of agonists. This excludes direct activation of AChRs at α‐ and γ‐motoneuron endplates. We analysed stretch‐induced afferent changes instead of activity induced by γ‐motoneuron stimulation. This allowed the selective analysis of AChR function in the equatorial region of muscle fibres. Finally, we used single unit muscle spindle afferent recordings with a much higher sensitivity allowing us to detect small modulatory activities, which apparently have escaped detection in previous studies (Smith & Albuquerque, 1967; Ganguly et al. 1978; Akoev, 1980).

Addition of d‐tubocurarine, α‐bungarotoxin or hemicholinium‐3 did not change the resting discharge frequencies, strongly suggesting that neither drug directly activated γ‐motoneuron endplates. In addition, the absence of an effect of either drug on the resting discharge levels makes a potential agonistic activity of d‐tubocurarine on intrafusal muscle fibres unlikely (Ziskind & Dennis, 1978; Takeda & Trautmann, 1984). The effect of d‐tubocurarine, α‐bungarotoxin and hemicholinium‐3 specifically on stretch‐evoked responses also makes it unlikely that another sites of action of the drugs, including the sympathetic innervation (Santini & Ibata, 1971) or nociceptive fibres (Lund et al. 2010b), affected our analysis. It is therefore very likely that we examined the effect of the drugs exclusively on the AChRs concentrated at the contact site between intrafusal muscle fibres and proprioceptive sensory neuron.

Since all drugs caused an increase in the dynamic as well as in the static response to stretch, all three types of intrafusal fibres are likely to be affected. In agreement with this hypothesis, AChRs were detected at the contact site between sensory neuron and nuclear bag as well as nuclear chain fibres. However, more refined experiments are needed to specifically distinguish if the drugs affect all three different intrafusal fibre types to the same extent.

The effect of α‐bungarotoxin on muscle spindle responses to ramp‐and‐hold stretches as well as to vibrations was stronger compared to the effect of d‐tubocurarine. For example, α‐bungarotoxin affected the static response and small amplitude vibrations, whereas d‐tubocurarine did not. It is unlikely that we used non‐saturating conditions of either AChR inhibitor, since increasing the concentration of α‐bungarotoxin or d‐tubocurarine did not affect muscle spindle responses. One potential explanation is that both drugs affect the fetal and adult AChR, which are both concentrated in the central region of intrafusal fibres (Zhang et al. 2014), to a different extent. It has previously been shown that d‐tubocurarine is sixfold less effective in blocking ACh‐elicited current from fetal AChR compared to α‐bungarotoxin (Kopta & Steinbach, 1994). This reduced efficacy in blocking the fetal AChR might be sufficient to reduce the effect of d‐tubocurarine below statistical significance.

Previous studies have shown that in addition to the AChR, the acetylcholine receptor‐associated protein rapsyn is concentrated in the equatorial region precisely at the contact site of the intrafusal fibre and the sensory neuron (Zhang et al. 2014), suggesting a similar anchoring of the AChR in the subsarcolemmal cytoskeleton as at the neuromuscular junction. The sensory nerve terminals also contain the vesicular acetylcholine transporter and choline acetyl transferase, the key enzyme of acetylcholine synthesis (Zhang et al. 2014). In addition, synaptic‐like vesicles (Bewick et al. 2005) and molecules required for their exocytosis can be detected in sensory nerve terminals. These include the presynaptic cytomatrix protein bassoon, the vesicle clustering protein synapsin I, as well as synaptophysin, synaptotagmin, VAMP/synaptobrevin I and II, and syntaxin (De Camilli et al. 1988; Li et al. 1996; Aguado et al. 1999; Bewick et al. 2005; Simon et al. 2010; Zhang et al. 2014; Bewick, 2015). In addition, previous studies using the styryl dye FM1‐43 demonstrated vesicle exo‐ and endocytosis within the sensory nerve terminal (Banks et al. 2002; Bewick et al. 2005). This vesicle turnover was increased fourfold in response to stretch and depended on the presence of extracellular calcium ions (Bewick et al. 2005). Collectively, these previously published results together with the present study suggest that the terminals of the stretch‐sensitive muscle spindle afferents have molecular specializations reminiscent of cholinergic synapses and contain the molecular machinery required to release synapse‐like vesicles.

The high‐affinity choline transporter is essential for proper signalling at cholinergic synapses, including the neuromuscular junction, due to the transporter's rate‐limiting reuptake of choline that is required to sustain ACh synthesis and release (Carpenter & Woodruff, 1987; Ferguson et al. 2004; Lund et al. 2010a). Accordingly, treatment of neuromuscular junctions with the high‐affinity choline transporter antagonist hemicholinium‐3 reduced the level of ACh in presynaptic vesicles suggesting that this drug leads to an impaired ability to maintain an adequate amount of releasable pools of ACh (Yu & Van der Kloot, 1991). We demonstrate that incubation of muscle spindles with hemicholinium‐3 resulted in an increase in the firing frequencies during the dynamic peak and dynamic index by about 30%. Thus, the effect of blocking AChRs in the central region of intrafusal fibres or inhibiting ACh uptake into the sensory nerve terminal were similar, further supporting the hypothesis that the ACh‐mediated signalling is part of a stretch‐dependent modulatory feedback system regulating muscle spindle sensitivity. Since at the neuromuscular junction more than 90% of the HC‐3‐sensitive high‐affinity choline transporter is present on synaptic vesicles (Nakata et al. 2004) and since the high‐affinity choline uptake is rate limiting for ACh synthesis and its release (Ferguson et al. 2004), our results are also consistent with a vesicular release of acetylcholine from sensory neurons.

As discussed in detail previously (Zhang et al. 2014), several lines of evidence strongly suggest that AChRs in the central region of intrafusal fibres are concentrated in the muscle fibre plasma membrane, rather than in the sensory nerve terminal. Thus, it appears likely that α‐bungarotoxin and d‐tubocurarine affect AChRs located in the plasma membrane of intrafusal fibres rather than in the sensory nerve terminal. Based on the results of the present study, we propose that some vesicles in the sensory nerve terminal contain acetylcholine. The stretch‐dependent release of this ACh could activate AChRs in the intrafusal muscle fibre membrane directly opposite to the sensory nerve terminal. Whether this release is calcium‐dependent (Bewick et al. 2005) and involves the stretch‐sensitive calcium channel, which has been described in the sensory nerve terminal (Hunt et al. 1978; Kruse & Poppele, 1991), or the Piezo2 channel (Woo et al. 2015), is currently unknown. Since we did not observe an effect of d‐tubocurarine or of α‐bungarotoxin on the firing rate of resting muscle spindles, we assume that acetylcholine is released primarily during mechanical activity. This was unexpected since at least some glutamate‐containing vesicles are released constitutively (Bewick et al. 2005). Our results are therefore consistent with the possibility that both neurotransmitters are stored within separate populations of vesicles. However, future studies are needed to further characterize the glutamate‐ and ACh‐containing vesicle populations and their modulation of muscle spindle sensitivity.

Binding of sensory neuron‐derived acetylcholine to the AChR would most likely result in a depolarization of the intrafusal fibre due to cation influx via the AChR. If this leads to a graded potential or to an action potential is unknown. In any case, since nuclear bag and nuclear chain fibres contain a cylinder of myofilaments directly underneath the plasma membrane in their equatorial region (see, for example, Ovalle, 1972), the AChR‐mediated depolarization could lead to a contraction of the sarcomeres in the central part of the intrafusal muscle fibres. The function of this contraction, however, remains speculative. In principle, a contraction of sarcomeres in the equatorial region would lead to a lengthening of the polar regions and to a reduced sensitivity to stretch of the equatorial region. Alternatively, the AChR‐mediated depolarization in the equatorial region could sum with the graded potentials or action potentials induced by γ‐motoneuron activity in the polar regions leading to an even stronger contraction. Thus, the mechanism of how AChRs in the equatorial region modulate the stretch‐induced muscle spindle afferent firing rate during stretch remains unclear and more experiments are needed to substantiate this hypothesis. In any case, our results further support the idea that in addition to sending afferent information about muscle length changes to the central nervous system, muscle spindle afferent terminals also release neurotransmitters that may control intrafusal fibre tone.

A termination of the AChR activation could be mediated by acetylcholinesterase, although there have been conflicting results regarding the presence of this enzyme in the central region of intrafusal fibres (Giacobini, 1959; Schober & Thomas, 1978; Zhang et al. 2014). Alternatively, a long‐lasting presence of ACh (due to the absence of acetylcholinesterase in the central region of intrafusal fibres) might generate a background intrafusal muscle fibre tone, counteracting the mechanical tension generated by the fusimotor activity in the polar regions.

Proprioceptive sensory afferents contain glutamate‐filled synapse‐like vesicles suggesting that they might release glutamate. This glutamate might, via an unusual phospholipase D‐linked metabotropic glutamate receptor within the sensory nerve terminal, enhance stretch‐induced sensory afferent excitability (Bewick et al. 2005; Simon et al. 2010; Bewick, 2015; Bewick & Banks, 2015). The results from our study suggest that ACh has a different effect. In contrast to glutamate, ACh is likely to modulate only stretch‐induced responses via the nicotinic AChRs in the intrafusal muscle fibre plasma membrane. Moreover, ACh acts as a negative modulator, reducing stretch‐induced sensitivity. Our results suggest that mechanical activity not only releases glutamate‐ but also ACh‐containing vesicles, and that both modulate the proprioceptive sensory neuron's response to stretch.

Additional information

Competing interests

All authors declare no conflict of interest.

Author contributions

S.K. and K.A.W. conceived the project and designed the experiments. L.G. and C.H. performed, analysed and interpreted the experiments. S.K. wrote the first draft of the manuscript. All authors revised the manuscript and approved the final version submitted for publication. All authors agree to be accountable for all aspects of the work in ensuring that questions relating to the accuracy or integrity of any part are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed as authors.

Funding

The work was supported by grants from the Deutsche Forschungsgemeinschaft (DFG; grant KR1039/16‐1), the Friedrich‐Baur‐Association, and the Deutsche Gesellschaft für Muskelkranke (DGM) . We are particularly grateful to the Graduate School of Systemic Neuroscience Munich (GSN) and the Munich Centre for NeuroSciences – Brain and Mind (MCN) for their generous financial support.

Biography

Laura Gerwin obtained her MSc degree in biology with a specialization in medical biology at the Technical University of Munich, Germany. She then joined the Institute of Stem Cell Research at the Helmholtz Centre Munich, and worked in the group of Prof. Stephan Kröger at the Biomedical Centre of the Ludwig‐Maximilians‐University, Munich, Germany, to obtain her PhD. She is interested in the communication between brain and muscles, specifically in the function of muscle spindles under healthy and pathological conditions.