Blockade of 5‐HT2 receptors suppresses motor unit firing and estimates of persistent inward currents during voluntary muscle contraction in humans

Serotonergic neuromodulation contributes to enhanced voluntary muscle activation. However, it is not known how the likely motoneurone receptor candidate (5‐HT2) influences the firing rate and activation threshold of motor units (MUs) in humans. The purpose of this study was to determine whether 5‐HT2 receptor activity contributes to human MU behaviour during voluntary ramped contractions of differing intensity. High‐density surface EMG (HDsEMG) of the tibialis anterior was assessed during ramped isometric dorsiflexions at 10, 30, 50 and 70% of maximal voluntary contraction (MVC). MU characteristics were successfully extracted from HDsEMG of 11 young adults (four female) pre‐ and post‐ingestion of 8 mg cyproheptadine or a placebo. Antagonism of 5‐HT2 receptors caused a reduction in MU discharge rate during steady‐state muscle activation that was independent of the level of contraction intensity [P < 0.001; estimated mean difference (∆) = 1.06 pulses/s], in addition to an increase in MU derecruitment threshold (P < 0.013, ∆ = 1.23% MVC), without a change in force during MVC (P = 0.652). A reduction in estimates of persistent inward current amplitude was observed at 10% MVC (P < 0.001, ∆ = 0.99 Hz) and 30% MVC (P = 0.003, ∆ = 0.75 Hz) that aligned with 5‐HT changes in MU firing behaviour attributable to 5‐HT2 antagonism. Overall, these findings indicate that 5‐HT2 receptor activity has a role in regulating the discharge rate in populations of spinal motoneurones when performing voluntary contractions. This study provides evidence of a direct link between MU discharge properties, persistent inward current activity and 5‐HT2 receptor activity in humans.


Introduction
Serotonin (5-HT) is a monoamine that modulates the intrinsic properties of motoneurones. Cellular studies indicate that 5-HT can promote depolarization via facilitation of rectifying inward currents (Hsiao et al., 1997;), facilitation of low-voltage Ca 2+ currents ) and inhibition of the K + leak conductance (Elliott & Wallis, 1992;. Activation of 5-HT receptors on motoneurones can also cause an increase in firing rate by modulating slow afterhyperpolarisation (sAHP) and Ca 2+ -activated K + channels (Grunnet et al., 2004). Several experiments have been able to mimic the excitatory effects of 5-HT pharmacologically with agonism of somatodendritic 5-HT 2 receptors and abolished excitatory effects with 5-HT 2 receptor antagonism (Cotel et al., 2013;Elliott & Wallis, 1992;Hsiao et al., 1997;Jackson & White, 1990). Thus, the availability of 5-HT and 5-HT 2 receptor activity appears to be a key mechanism in modulating motoneurone discharge.
It has recently been demonstrated that enhancing the availability of 5-HT in the CNS increases the ability to activate biceps brachii voluntarily during high-intensity elbow flexions in humans (Henderson et al., 2022;Kavanagh et al., 2019;Thorstensen et al., 2020). Enhancement of voluntary activation involves 5-HT 2 receptor activity, because the administration of a 5-HT 2 antagonist causes a reduction in elbow flexion force by ∼3% during maximal contractions (Thorstensen et al., 2021). Although these studies provide insight to how serotonergic neuromodulation contributes to voluntary muscle activation, it is still largely unknown how 5-HT 2 activity influences the recruitment and firing rate of motor units (MUs) responsible for activating the muscle. Of the data that are available, MUs in tibialis anterior fire an average of 12% slower with 5-HT 2 receptor blockade when performing brief rapid dorsiflexions, whereby MUs firing at higher discharge rates are more affected by 5-HT effects than MUs firing at lower discharge rates (Goodlich et al., 2021). This not only aligns with the aforementioned voluntary activation findings but also supports the viewpoint that serotonergic drive in the CNS might be scaled to the level of motor activity (Jacobs & Fornal, 1997;Jacobs et al., 2002;Veasey et al., 1995). Despite our understanding of how serotonergic neuromodulation contributes to MU activity during brief, rapid contractions, the role that 5-HT plays in the regulation of MUs during steady-state muscle activation in humans is unknown.
Persistent inward currents (PICs) on somatodendritic surfaces of motoneurones are strongly influenced by monoamines and make key contributions to motoneurone behaviour. Activation of PICs provides an additional intrinsic source of depolarizing current to synaptic inputs to motoneurons that is thought to be proportional to the monoaminergic drive (Hultborn et al., 2003;Lee & Heckman, 1998, 2000. Persistent inward current activity also promotes self-sustained firing after the removal of synaptic inputs that depolarize the motoneurone (Hounsgaard et al., 1988;Lee & Heckman, 1998). This is reflected by hysteresis in MU discharge characteristics, whereby the derecruitment of motoneurones occurs at lower levels of synaptic input compared with the recruitment of motoneurones (Gorassini et al., 2002). Although there is evidence that facilitation can occur during sustained activation, whereby an increase in firing rate occurs in response to repetitive injected currents, there has been limited investigation of constant synaptic input to motoneurons in the form of a steady-state muscle contraction. A growing number of human investigations are reporting concomitant changes to estimates of PICs and MU discharge rate, where 5-HT mechanisms are frequently cited as a mediating factor (Hassan et al., 2021;Mesquita et al., 2022;Orssatto, Borg, et al., 2021). However, a direct link between MU discharge, 5-HT 2 receptor activity and PIC activity is yet to be demonstrated during voluntary muscle contractions. We propose that a pharmacological approach that modifies the effectiveness of 5-HT on MU activity will clarify this relationship.
The purpose of this study was to determine whether 5-HT 2 receptor activity contributes to human MU behaviour during voluntary contractions of differing intensity. A 5-HT 2 competitive antagonist was administered to participants, in whom MU activity was extracted from the tibialis anterior by high-density surface EMG (HDsEMG) pre-and post-drug ingestion. Ramped contractions were used to investigate MU firing characteristics at recruitment, derecruitment, and during steady-state muscle activation. It was hypothesized that MU discharge rate immediately after recruitment and during the steady-state contractions would be reduced with 5-HT 2 antagonism. It was also hypothesized that estimates of PIC amplitude calculated from the discharge characteristics of paired MUs [i.e. change in frequency ( F)] would be reduced with 5-HT 2 antagonism, thus reflecting a decline in self-sustained firing owing to removal of serotonergic effects.

Experimental design
The study used a human, placebo-controlled, double-blind, two-way, cross-over design. Participants attended two testing sessions separated by 1 week. The order of drug administration was counterbalanced. A number generator randomly assigned the participants an identification number between 1 and 14. Counterbalancing was achieved by administering cyproheptadine to the even numbered participants in their first session and odd numbered participants in their second session. Blinding was achieved by having a separate investigator responsible for drug administration and labelling data according to session. Only after decomposition, visual inspection and cleaning of MU filters were the data stratified into cyproheptadine or placebo sessions.

Participants and ethical approval
Fourteen healthy, recreationally active individuals (age 23.9 ± 2.5 years, seven female) were recruited to the J Physiol 601.6 study. Before enrolment, participants were screened using a medical history questionnaire containing exclusion criteria specific to the administration of cyproheptadine, in addition to acute or chronic neuromuscular injury. Participants were instructed to refrain from any stimulants or depressants, such as caffeine or alcohol, and from moderate-to high-intensity exercise in the 12 h before testing. Approval for testing procedures was obtained via the Human Research Ethics committee of Griffith University (GU ref. no. 2020/264), and all procedures were performed in accordance with the Declaration of Helsinki except for registration in a database. Written informed consent was obtained from all participants before testing.

Drug administration
Baseline neurophysiological measurements were obtained before a single oral dose of cyproheptadine (8 mg) or a placebo (containing Avicel filler) was administered to each participant. Two hours after pill ingestion, experimental testing took place, in which neurophysiological measurements were again made. The timing of testing aligned with high plasma concentrations of cyproheptadine (D' Amico et al., 2013;Wei et al., 2014) and with the testing window reported in previous cyproheptadine studies (Goodlich et al., 2021;Thorstensen et al., 2021). The dose of cyproheptadine is consistent with previous studies that have used cyproheptadine as an intervention to assess serotonergic effects on the motor system (D' Amico et al., 2013;Goodlich et al., 2021;Thorstensen et al., 2021;Wei et al., 2014). Cyproheptadine binds with high affinity to 5-HT 2A/B/C receptors (Boess & Martin, 1994;Honrubia et al., 1997), attenuating serotonergic effects via competitive antagonism. The cyproheptadine and placebo were compounded in opaque capsules to ensure blinding of the drug conditions. Drug administration and experimental testing were performed at similar times in the morning for all participants. There were no adverse effects of cyproheptadine ingestion; however, moderate levels of drowsiness were reported by all participants ∼4 h after pill ingestion.

Experimental set-up and torque measurement
Participants sat comfortably in a motorized therapy chair, which was adjusted for each participant to position their right hip, knee and ankle at 90°of flexion in the sagittal plane. The right foot was secured with a non-compliant, ratchet-type binding to a custom-designed footplate, which incorporated a commercially available torque sensor (capacity = 565 N m; model 2110-5K; Honeywell International, Charlotte, NC, USA). The footplate was mounted on a bespoke aluminium frame, which was secured to the therapy chair, with the torque sensor axis of rotation aligned to participant's malleoli (Fig. 1A). Ankle torque was sampled at 2000 Hz using a Power 1401 interface with Spike2 software (v.7, Cambridge Electronic Design, Cambridge, UK). Feedback for the unfiltered torque signal was displayed on a computer monitor positioned ∼1 m in front of the participant, with dorsiflexion torque presented as a positive inflection in torque on the screen.

Experimental protocol
Determination of maximal voluntary contraction. For all maximal dorsiflexion contractions in this study, care was taken to ensure that participants minimized the use of muscles other than the tibialis anterior to generate isometric dorsiflexion torque. An investigator provided instructions and a demonstration on performing the task before participants attempting the task. During testing, surface EMG (sEMG) activity from synergistic and antagonistic muscles was monitored constantly (see Electromyography section below), and any trial with excessive activity from synergist or antagonist muscles was deemed a mistrial and repeated. After familiarization, participants performed five maximal effort dorsiflexions lasting ∼3 s in duration, with rest periods of ≥3 min between contractions. The trial that generated the highest magnitude of torque was determined to be the participant's maximal voluntary contraction (MVC). This value was then used to set the submaximal dorsiflexion targets for each participant. Dorsiflexion MVC was determined at four time points: pre-placebo, post-placebo, pre-cyproheptadine and post-cyproheptadine.
Contraction protocol. After determining dorsiflexion MVC, participants performed submaximal, trapezoidal contractions to 10, 30, 50 and 70% of MVC in a randomized order (Fig. 1B). Each trapezoid had a duration of 24 s and a rate of torque development and decline of 10% MVC/s. The duration of contraction and rate of torque development were fixed in an effort to keep MU spike frequency adaptation and spike threshold accommodation consistent across intensities (Orssatto, Mackay, et al., 2021;Powers & Heckman, 2015;Vandenberk & Kalmar, 2014). After a period of familiarization, two trials were performed at each target intensity. Participants were given 2 min of rest after the lower intensities and 5 min of rest after higher intensities, to minimize fatigue.

Electromyography
Muscle activity for the tibialis anterior was measured using a semi-disposable 64-channel HDsEMG grid electrode (8 × 8) with a 10 mm interelectrode distance (OTBioelettronica, Torino, Italy). After skin preparation (shaving, abrasion and cleansing with 70% isopropyl alcohol), the position and orientation of the electrode grid were determined by an experienced investigator via palpation of the right tibialis anterior muscle belly (Fig. 1A). Electrodes were fixed to the middle of the muscle belly using a bi-adhesive, perforated foam layer and conductive paste (SpesMedica, Battipaglia, Italy). A dampened strap earth electrode (OTBioelettronica) was positioned over the right ankle malleoli. The HDsEMG signals were recorded in monopolar mode and converted to a digital signal by a 16-bit wireless amplifier (Sessantaquattro; OTBioelettronica). The HDsEMG signals were recorded and visualized using OTBioLab+ software (v.1.3.0.; OTBioelettronica).
To ensure that muscle activity during dorsiflexion MVCs was predominantly from tibialis anterior, bipolar sEMG electrodes were attached to several muscles of the test leg. These muscles were monitored for unintended activity that could contribute to MVC torque. Specifically, 24 mm Ag/AgCl electrodes (Kendall ARBO; Cardinal Health, Dublin, OH, USA) were placed over the medial gastrocnemius, soleus and vastus lateralis muscles. Electrodes were aligned parallel to the underlying muscles fibres, with an interelectrode distance of 24 mm. EMG signals were differentially amplified (×1000) by a NL844 pre-amplifier and bandpass filtered (10-500 Hz) by a NL135 low-pass filter and NL144 high-pass filter (Digitimer, UK). The sEMG was sampled at 2000 Hz via a Power 1401 interface with Spike2 software (v.7; Cambridge Electronic Design).

High-density surface EMG analysis
Before decomposition, monopolar HDsEMG signals were digitally bandpass filtered at 20−500 Hz with a second-order Butterworth filter. The HDsEMG signals were decomposed into individual MU action potentials using blind source separation, via the convolutive kernel compensation method (Holobar & Zazula, 2007). This method has been validated previously for a broad range of contraction intensities of the tibialis anterior muscle Holobar et al., 2014;Negro et al., 2016). Extracted MUs were tracked pre-to post-pill ingestion, within the same contraction intensity, for both the placebo and cyproheptadine testing sessions. This was achieved by concatenating the HDsEMG recordings from preand post-ingestion, then applying the individual MU filters (action potential waveform shapes) to the entire concatenated data set. The accuracy of decomposition was assessed using pulse-to-noise ratio during each individual contraction (Holobar et al., 2014), and decomposed spike trains showing pulse-to-noise ratios <28 dB were discarded from the analysis, as described by Del Vecchio, Holobar, et al. (2020). All MU pulse trains were inspected manually by an investigator experienced in MU analysis, and only pulse trains with a reliable discharge pattern were considered for analysis. The MU recruitment threshold and derecruitment threshold were calculated as the torque values corresponding to the first and last MU firing, respectively. The MU discharge rate was calculated at recruitment (the average of the first four action potentials), during the plateau phase of the trapezoidal contraction (the average discharge rate of the first Figure 1. Electrode orientation and contraction protocol A, the right foot of the participant was secured to a footplate that incorporated a torque sensor. The HDsEMG electrodes were fixed over the TA muscle belly and oriented in the estimated direction of muscle fibres. B, the order of the two laboratory visits (placebo in blue and cyproheptadine in red) was counterbalanced. Maximal dorsiflexions and submaximal, trapezoidal dorsiflexions were performed at two time points: pre-(grey) and post-(black) pill ingestion. Submaximal contractions were performed to 10, 30, 50 and 70% of MVC; the order of which was randomized. Submaximal contraction duration was 24 s for all intensities, and the ramp up/down rate was fixed at 10% of MVC per second. Abbreviations: CYP, cyproheptadine; HDsEMG, high-density surface EMG; MVC, maximal voluntary contraction; PLA, placebo; TA, tibialis anterior. [Colour figure can be viewed at wileyonlinelibrary.com] J Physiol 601.6 10 interspike intervals during steady contraction at the target intensity) and at derecruitment (the average of the last four action potentials).

Estimation of persistent inward currents
The effect of PICs on motoneurone firing was estimated by investigating MU onset-offset hysteresis. This was done by calculation of F, using the paired MU technique (Gorassini et al., , 2002. Briefly, F is the difference in the smoothed firing rate of a lower-threshold control unit between the onset and offset timings (i.e. recruitment and derecruitment) of a higher-threshold test unit (Gorassini et al., 2002). In the present study, F values are presented as 'unit-wise' averages for all suitable test-control unit pairs, thus reducing the number of F values to one per test unit (Hassan et al., 2021). When individual test units were paired with multiple control units that met inclusion criteria, the average F value was obtained (Hassan et al., 2021;Mesquita et al., 2022;Trajano et al., 2020). Inclusion criteria for MU pairs were as follows: (i) test MUs were recruited >1 s after control units to enhance the likelihood of full PIC activation; (ii) test MUs ceased firing >1.5 s before control units to reduce the possibility of overestimating F; and (iii) test units were activated continuously for ≥2 s (Hassan et al., 2020). Previously, trapezoidal and triangular contractions ≤30% of MVC have been used for the successful calculation of F with the paired MU method (Afsharipour et al., 2020;Orssatto, Mackay, et al., 2021;Wilson et al., 2015). Calculations of F for 50 and 70% MVC intensities have not been validated and are not reported in the present study. Calculation of F requires low-threshold MUs to approximate synaptic input, and these units are difficult to decompose successfully at higher contraction intensities owing to superimposition of larger, higher-threshold MUs in the HDsEMG signal.

Estimated coherence
Given that 5-HT receptors are expressed at different levels of the nervous system, we also sought to estimate the common synaptic input at different frequency bands that are associated with cortical and spinal oscillatory activity (Bräcklein et al., 2022;Del Vecchio, Sylos-Labini, et al., 2020;Negro et al., 2016). Pooled coherence between identified MU spike trains from the tibialis anterior muscle was estimated within each contraction intensity and drug session and compared between pre-and post-pill ingestion. In each trial, analysis was performed on pools of spike trains consisting of permutations of all possible combinations of three MUs or until completion of 100 permutations. Coherence was estimated with Welch's averaged periodogram with a non-overlapping Hanning window of 1 s in duration (Del Vecchio, Sylos-Labini, et al., 2020). The areas associated with the peak values of coherence in the δ (0-5 Hz), α (5-15 Hz) and β (15-35 Hz) bands were identified.

Statistical analysis
All statistical analysis was performed in R, using RStudio (v.4.1.1; R Foundation for Statistical Computing, Vienna, Austria). To assess test-retest reliability of MU characteristics from pre-to post-placebo, two-way mixed effects, consistency and single measurement intraclass correlation coefficients (ICC 3,1 ) were computed for recruitment threshold, average discharge rate during the plateau and derecruitment threshold. Two-way, repeated-measures ANOVA was used to assess MVC torque, synergist/antagonist muscle activity and MU yield pre-and post-pill ingestion for both the placebo and cyproheptadine testing sessions. Separate linear mixed-effects models were used to evaluate the effects of drug, contraction intensity and their interaction on each outcome measure of interest. Models were developed using the nlme package (Pinheiro et al., 2017), and comparisons were made pre-and post-pill ingestion in order to account for day-to-day variability in MU activity. The outcome measures of interest were MU recruitment threshold, derecruitment threshold, discharge rate at recruitment, discharge rate during the plateau, discharge rate at derecruitment, and F. Linear mixed-effects models were used for MU analysis because they allow for the inclusion of all units, whilst accounting for the hierarchical nature of the data (i.e. higher correlation for units within subjects compared with between subjects; Tenan et al., 2014;Yu et al., 2021). Models were developed by iteratively adding predictor variables or interaction effects, and the fit of models was compared using an ANOVA. Drug state, contraction intensity and the interaction between them were considered fixed effects, with a random intercept for each subject [e.g. discharge rate ∼ drug × intensity + (1 | subject ID)]. Significance was calculated using the lmerTest package in R (Kuznetsova et al., 2017), which uses Satterthwaite's method to approximate degrees of freedom and generate P-values for mixed-effects models by comparing the full model (with the effect of interest) against a null model (excluding the effect of interest). In the event of a significant drug effect, post hoc tests were conducted to examine estimated marginal means with 95% confidence intervals between pre-and post-pill ingestion using the emmeans package (Lenth & Lenth, 2018). Additionally, in the event of a significant drug effect, Cohen's d effect sizes were calculated using both the raw data (d unadjusted ) and the modelled data (d adjusted ) to account for the clustering of MUs within individual subjects. In the event of a significant drug-by-contraction intensity interaction effect, post hoc tests were conducted to examine pre-and post-pill ingestion differences within each contraction intensity. Kenward-Roger approximation was used for estimating degrees of freedom for the post hoc pairwise comparisons. Bonferroni corrections were applied to account for multiple comparisons. To assess the strength of bivariate correlations between preand post-pill ingestion, across all intensities, a Pearson product-moment correlation was used. For all statistical comparisons, an α-value of P < 0.05 was considered statistically significant.

Participant characteristics and MVC torque
Three individuals (three female) were excluded from statistical analysis owing to failure to identify and track MUs reliably. There was no significant difference in maximal dorsiflexion torque amplitude between preand post-pill ingestion (F 1,10 = 0.217, P = 0.652), and there were no between-session differences in maximal dorsiflexion torque amplitude (F 1,10 = 0.004, P = 0.951; Table 1). The level of muscle activity in antagonist and synergist muscles was consistent for all experiments, because there were no differences identified in EMG amplitude pre-and post-pill ingestion for gastrocnemius (  Figure 2A and B illustrates the dorsiflexion torque trajectories and the discharge timings for one subject at 10 and 70% of MVC during the placebo and cyproheptadine conditions, respectively. Figure 2B highlights the consistency of recruitment and derecruitment order in the cyproheptadine session of the same MU matched from pre-to post-pill ingestion. The instantaneous discharge rate for both the placebo and the cyproheptadine sessions exhibited an initial acceleration in firing during the ascending slope of the trapezoidal contraction, which was followed by a more linear firing pattern during the plateau phase. Decomposition of HDsEMG signals from the submaximal trapezoidal contractions yielded a total of 1175 MU spike trains: 588 units matched pre-to post-during the placebo session and 587 units matched pre-to post-during the cyproheptadine session. The average number of MUs per subject was 26.7 ± 12.2 during the placebo session and 26.7 ± 12.0 during the cyproheptadine session. Table 2 provides detail on the number of MU spike trains identified across each contraction intensity across all experiments. There were no significant differences identified between pre-and post-pill ingestion (F 1,10 = 2.09, P = 0.179), nor were there any significant differences between testing placebo and cyproheptadine sessions (F 1,10 = 0.001, P = 0.982).

Motor unit recruitment threshold and derecruitment threshold
Both the recruitment threshold (F 3,647.03 = 1263.6, P < 0.001; Fig. 3A) and derecruitment threshold (F 3,648.54 = 1505.7, P < 0.001; Fig. 3C) of decomposed MUs increased significantly with increasing contraction intensity. Antagonism of 5-HT 2 did not affect the MU recruitment threshold (F 1,645.28 = 3.70, P = 0.055; Fig. 3A). The derecruitment threshold was significantly higher after the ingestion of cyproheptadine (F  Fig. 3C). No contraction intensity-by-drug interaction was detected for either the MU recruitment threshold or the MU derecruitment threshold.
The relationship between recruitment threshold pre-cyproheptadine and the recruitment threshold post-cyproheptadine is presented in Fig. 3B (y = 1.00x + 1.11, r 2 = 0.96, Pearson, P < 0.01). Likewise, the pre-to post-cyproheptadine relationship for the derecruitment threshold is presented in Fig. 3D (y = 0.99x + 1.45, r 2 = 0.97, Pearson, P < 0.01). Data in these correlation analyses are not grouped according to a specific contraction intensity, but instead provide insight into MU properties across the full range of recruitment thresholds. This is important, because at high contraction levels, the HDsEMG method is not able to decompose MUs with a low recruitment threshold. As the contraction level increases, the MU that can be sampled change to higher threshold, as is evident in Fig. 4A and C. The regression lines in Fig. 3B and D maintain a gradient of approximately one, indicating the consistency of cyproheptadine effects (or lack thereof) across a broad sample of units with different thresholds. The R 2 values of 0.96 and 0.97 also indicate that the variance of post-drug ingestion values is well accounted for by the pre-ingestion recruitment and derecruitment thresholds, further highlighting the similarity of effects across the whole range of motoneurons.

Motor unit discharge rate during the plateau phase of submaximal isometric contractions
Across the range of submaximal intensities, the average discharge rate during the steady-state contraction reduced post-ingestion of cyproheptadine by an average of 6.1%. There was a significant decrease in average discharge rate during the plateau phase of the trapezoidal dorsiflexions identified from pre-to post-ingestion of cyproheptadine (F 1,642.54 Fig. 4B). The pre-to post-cyproheptadine relationship for discharge rate during the plateau phase is presented in Fig. 5 (y = 0.87x + 1.03, r 2 = 0.92, Pearson, P < 0.01). The regression line is located below the line of identity, which confirms that the MU discharge rate was reduced post-cyproheptadine compared with pre-cyproheptadine. With increasing MU discharge rate, the regression line deviates further away from the reference line, indicating that effects of cyproheptadine might be more pronounced with faster-firing MUs.

Estimate of persistent inward current amplitude
Motor unit pairs that met inclusion criteria were successfully identified in all 11 participants at 10% MVC and in 10 participants at 30% MVC. There were 214 unit-wise comparisons made at 10% MVC and 141 at 30% MVC. The mean F value pre-ingestion of cyproheptadine at 10 and 30% MVC was 3.09 ± 2.44 and 4.44 ± 1.73 Hz, respectively. At 10% MVC post-ingestion, subject-wise F averages decreased in seven participants. This was reflected by a significant decrease in unit-wise F from pre-to post-ingestion of cyproheptadine (F 1,175.33 Fig. 6A). At 30% MVC, there were also seven participants whose subject-wise F averages decreased. This was again reflected by a significant decrease in unit-wise F from pre-to post-ingestion of cyproheptadine (F 1,141.88 Fig. 6B).

Coherence
Average estimated coherence pre-and post-cyproheptadine at each contraction intensity is displayed in Table 3. Across the full range of contraction intensities, there were no significant differences between pre-and post-cyproheptadine coherence values in the δ, β or α bands (Table 3). Hence, cortical communication to the muscle was consistent between pre-and post-cyproheptadine.

Discussion
The purpose of this study was to determine whether 5-HT 2 receptor activity contributes to human MU behaviour during voluntary ramped contractions. The main findings of this study were that 5-HT 2 antagonism caused: (i) a reduction in MU discharge rate during steady-state muscle activation; (ii) an increase in MU derecruitment threshold that corresponded to decreases in discharge rate; and (iii) a reduction in estimates of PIC amplitude that aligned with 5-HT-related changes in MU firing behaviour. Overall, these findings are consistent with animal preparations, single-cell experiments and human MU experiments. However, we were able to extend this work by revealing how 5-HT influenced larger populations of motoneurones during voluntary contractions of differing intensities. By using a pharmacological intervention to target 5-HT 2 receptors we have provided further evidence to support a direct link between MU discharge properties, PIC activity and 5-HT 2 receptor activity in humans.

Antagonism of 5-HT 2 receptors caused a reduction in motor unit discharge rate during steady-state muscle activation
Motoneurones receive ionotropic input from descending cortical drive, peripheral afferents and spinal interneurons. In the present study, isometric voluntary contractions were consistently performed in the same conditions for each contraction intensity, and monoaminergic input to the motoneurone was modified by blocking 5-HT 2 receptors. The duration of contraction and rate of torque development were fixed to keep MU spike frequency adaptation and spike threshold accommodation consistent across contraction intensities (Orssatto, Mackay, et al., 2021;Powers & Heckman, 2015;Vandenberk & Kalmar, 2014). This approach provided an opportunity to investigate intensity-wise differences in steady-state activation of tibialis anterior MU in humans. Not surprisingly, the MU discharge rate increased with increasing contraction intensity. However, after 5-HT 2 Figure 5. Correlation plot comparing motor unit discharge rate during the plateau phase pre-and post-5-HT 2 antagonism Discharge rates are shown pre-5-HT 2 antagonism (on the ordinate) and post-5-HT 2 antagonism (on the abscissa). Correlation includes discharge rate data from MUs during all contraction intensities. Each subject is indicated by a different colour. A line of identity is provided to highlight where a one-to-one relationship would occur. Abbreviations: MU, motor unit; pps, pulses per second. [Colour figure can be viewed at wileyonlinelibrary.com] receptor blockade, there was a reduction in MU firing during the steady-state contractions that was independent of the contraction intensity being performed. This result corroborates our previous findings for very fast muscle contractions, whereby 5-HT 2 receptor blockade reduced the MU discharge rate and rate of torque development during rapid isometric dorsiflexions ranging from 30 to 70% MVC (Goodlich et al., 2021). Our new findings provide a direct link between 5-HT 2 receptor activity and the firing rate of MUs during steady isometric contractions.
In addition to a reduction in MU firing during steady-state contractions, a concomitant reduction in estimates of PIC amplitude was observed with 5-HT 2 antagonism. Cellular studies that use constant-current injection protocols have revealed that activation of somatodendritic 5-HT 2 receptors facilitates Na + PICs in rat motoneurones (Harvey et al., 2006) and Ca 2+ PICs in rat (Murray et al., 2011) and turtle  motoneurones. It is likely that a similar mechanism emerged in our human participants during steady-state contractions, where voluntary drive to motoneurones was kept relatively constant, and serotonergic effects were detected in F from pre-to post-ingestion of cyproheptadine. A reduction in the magnitude of PICs, as a result of attenuated 5-HT 2 receptor activity, might underpin changes in steady-state firing of human MUs. This has been demonstrated with dorsolateral funiculus stimulation in turtle spinal cord preparations, where enhanced 5-HT release was associated with facilitation of voltage-sensitive PICs (Perrier & Delgado-Lezama, 2005). This effect was inhibited by the administration of a selective 5-HT 2 receptor antagonist, which is consistent with the MU discharge and PIC findings in the present study.

Effects of 5-HT 2 antagonism on motor unit firing subsequent to recruitment and prior derecruitment
As the PIC activates, there is an initial acceleration in the rate of motoneurone firing during recruitment, typically lasting 1−2 s Hounsgaard et al., 1988;Lee & Heckman, 2000). In the present study, the MU discharge rate was investigated during the first four interspike intervals after recruitment, when this accelerated firing would be expected to occur. After 5-HT 2 receptor blockade, there was a small, but significant, reduction in MU discharge rate immediately after recruitment compared with pre-drug ingestion. We speculate that this finding reflects an attenuation of intrinsic sources of depolarizing current within the motoneurone, because lower levels of neuromodulatory drive (i.e. the availability of 5-HT) are associated with a reduction in the potency of the initial acceleration J Physiol 601.6 in firing rate (Lee & Heckman, 2000). A small, but significant, suppression in MU discharge was also present immediately before deactivation of the MU during the derecruitment phase of the contraction. Intracellular recordings of decerebrate cat motoneurones indicate that motoneurons derecruit at a much higher amplitude of injected current in the absence of 5-HT (Hounsgaard et al., 1988). The hallmark of PIC-induced self-sustained firing is discharge hysteresis, whereby derecruitment occurs at a lower level of synaptic input than was required to recruit the motoneurone initially Hounsgaard & Kiehn, 1989;Hounsgaard et al., 1988;Lee & Heckman, 1996). In the present study, we observed a reduction in F values after the drug intervention, indicative of a reduction in PIC activation. Less PIC activation at the same relative contraction intensity would be likely to result in derecruitment of motoneurones at a higher level of synaptic input, as was observed in the present study as a significant increase in the derecruitment threshold after 5-HT 2 receptor antagonism (Fig. 3C).

Considerations when attributing changes in discharge rate to estimates of PIC
Estimating PIC amplitude from paired MUs extracted from HDsEMG might provide a greater window into motoneurone populations compared with MUs extracted from intramuscular EMG. However, the greater number of MUs that are extracted from HDsEMG unavoidably inflates the amount of data processed (degrees of freedom) in statistical models. Hence, small changes in firing rate can reveal statistical significance. Recent interventional studies have demonstrated this, where firing rate can change significantly by ∼0.5 Hz through muscle stretching (Mazzo et al., 2021) and by 0.33-0.64 Hz with vibration-induced reciprocal inhibition (Mesquita et al., 2022;Orssatto et al., 2022) when using HDsEMG techniques. However, it is important to note that each reported finding is not exclusive to human models, but also consistent with experiments using animal preparations and cellular studies. Thus, findings from human experiments align with mechanisms that have been identified in in vitro and in situ experiments. Pharmacological interventions that manipulate 5-HT activity are also founded in animal and cellular studies, where drug effects on human MU firing rate appear to be greater than non-pharmacological interventions. The effect of blocking 5-HT 2 receptors (using chlorpromazine) on F in non-injured individuals has previously been described in spinal cord injury studies (fig. 5b of the paper by D' Amico et al. 2013). Although these data was from a small cohort (n = 4 participants, with an unknown number of MU pairs) there was still a demonstrable reduction in F to ∼60−65% of pre-drug values. The present study extends these findings by identifying a 35% reduction in F (0.99 Hz) after 5-HT 2 receptor blockade with cyproheptadine in a larger cohort with a greater representation of motoneurone populations. Our findings confirm previous work that suggests serotonergic effects on motoneurones are not dependent on the intensity of the voluntary muscle contraction being performed (Goodlich et al., 2021).

Considerations when attributing changes in torque to MU activity
Although it is tempting to relate MU discharge and recruitment results to muscle forces about the ankle,  Average estimated coherence pre-and post-cyproheptadine, shown as group means ± SD (n = 11). Abbreviation: CYP: cyproheptadine.
we approach this issue with caution. Our lack of drug-related differences in maximal torque generation might suggest that MU activity should not have been affected. However, it is important to note that drug effects on force were assessed only at MVC, during which all units might have been recruited, whereas MU activity was assessed only during submaximal contractions, where additional recruited units might not have been detected after decomposition and MU matching. Thus, direct relationships between submaximal and maximal contraction intensities are difficult to comment on, and neuromuscular modelling might be a useful tool to investigate these relationships. The modest effect sizes for many of the results in the present study might also suggest that 5-HT has a minimal a role in regulating the excitability of motoneurones in humans. However, it is crucial to note that human 5-HT studies must operate within a boundary of safety, thus limiting the dosage that can be administered to healthy individuals to exert receptor blockade. Administration of higher doses would probably be associated with more complete receptor blockade, yield greater drug effects, and possibly see the emergence of differences in maximal dorsiflexion torque. The lack of (8 mg) cyproheptadine-related differences in MVC results mirrors previous dorsiflexor investigations (Goodlich et al., 2021) but contrasts with the significant declines in maximal torque that have been observed for the elbow flexors (Thorstensen et al., 2021).

Cortical mechanisms of 5-HT effects on motoneurones
Although serotonergic effects would have been likely to act on 5-HT pathways in the cortex, we contend that the level of cortical involvement in our results was insignificant. Pooled coherence between identified MU spike trains from the tibialis anterior in the present study revealed that there were no pre-to post-drug differences in α, β (i.e. cortical; Bräcklein et al., 2022) or δ bands. These findings indicate that common synaptic input to motoneurones was consistent across testing sessions. Although a high density of 5-HT fibres has been identified in motor areas of the rat and primate cortex, the role of these rostral raphe projections in motor activity of humans has not been well defined. Moreover, the intensity of 5-HT release from the descending raphespinal pathway, but not the ascending raphe pathway to the cortex, is thought to correspond to the intensity of motor activity being performed (Jacobs & Fornal, 1997;Jacobs et al., 2002;Veasey et al., 1995).

Conclusion
The present study provides new evidence that 5-HT 2 receptor activity affects human MU discharge properties, whereby blockade of 5-HT 2 receptors with cyproheptadine increases MU derecruitment thresholds and reduces the steady-state discharge rate in the absence of changes in maximal dorsiflexion force. These findings were consistent irrespective of the intensity of the voluntary muscle contraction being performed, although there is evidence to suggest that the effects of 5-HT 2 receptor antagonism are greater at higher MU discharge rates. Concomitant reductions in F values indicate that altered PIC activity might underpin the observed changes in MU properties. Collectively, these results support the viewpoint that the 5-HT 2 receptor plays a role in regulating motor activity, whereby a PIC-based mechanism is involved in regulating the excitability of human motoneurones. Motor unit firing patterns during submaximal trapezoidal contractions. There was no significant change in discharge rate pre-to post-placebo ingestion immediately after recruitment (F 1,452 = 1.75, P = 0.187; Fig. A3A), during the plateau (F 1,510 = 3.63, P = 0.058; Fig. A3B) or immediately before derecruitment (F 1,486 = 0.279, P = 0.598; Fig. A3C). Discharge rate increased significantly with increasing contraction intensity immediately after recruitment (F 3,452 = 66.67, P < 0.001), during the plateau (F 3,510 = 507.45, P < 0.001) and immediately before derecruitment (F 3,486 = 5.73, P < 0.001). However, there was no interaction between contraction intensity and placebo ingestion for discharge rate immediately after recruitment (P = 0.052), during the plateau (P = 0.729) or immediately before derecruitment (P = 0.457). Estimates of persistent inward currents using F. There were no significant differences in estimates of persistent inward currents ( F) from pre-to post-ingestion of the placebo at an intensity of 10% of MVC (F 1,218 = 1.57, P = 0.1695; Fig. A4A) or at 30% of MVC (F 1,116 = 2.82, P = 0.0804; Fig. A4B).