Integrated effects of altered action potentials and calcium release on skeletal muscle force generation in transgenic Huntington’s disease mice

Abstract

Huntington’s disease (HD) is a movement disorder commonly recognized as being neurodegenerative. An increasing number of studies also show primary HD dysfunction in multiple tissues, consistent with the widespread expression of the mutated huntingtin gene. Studies of HD skeletal muscle have revealed membrane hyperexcitability and prolonged action potentials due to Cl^– and K⁺ channel dysfunction as well as decreased Ca²⁺ release from the sarcoplasmic reticulum (SR) due to ryanodine receptor dysfunction. However, neither mechanism alone explains HD skeletal muscle function. To address this, we simultaneously recorded action potentials and SR Ca²⁺ release in model HD muscle and quantitatively linked the concerted mechanism to force generation. We discovered that the reduced SR Ca²⁺ release does not cause weakness in model HD muscle as expected because of the prolonged SR Ca²⁺ release (due to wider action potentials) and altered expression of Ca²⁺-binding proteins. The resulting integrated mechanism helps explains the surprisingly normal specific twitch force in model HD muscle and reveals a precarious balance that we show begins to disintegrate under very mild repetitive stimulation. The interplay of pathways also explains the resistance to myotonia in model HD muscle despite the substantial reduction in Cl⁻ current. By examining a concerted multi-molecular mechanism, we are able to understand tissue-level function in model HD muscle. This detailed study of twitch responses lays the foundation to examine the more complex integration of pathways during repetitive activity in HD muscle as well as in other normal and disease states that would benefit from the multi-molecular approach.

Introduction

Huntington’s disease (HD) is a monogenic disorder caused by an aberrant CAG repeat expansion within the coding region of the huntingtin gene that affects about 1 in 10,000 individuals worldwide, mostly of European descent [14, 51, 57]. The huntingtin mutation causes a host of cellular defects in multiple tissues that manifest in neurodegeneration, muscle degeneration, and systemic metabolic dysfunction that leads to cognitive and movement dysfunction, among other symptoms [18, 41, 62, 72]. The precise mechanisms by which the huntingtin mutation leads to these symptoms remains unclear, contributing to the lack of effective treatment options for HD patients. While most HD research has focused almost exclusively on central neurodegeneration, it has recently become more apparent that the impairment of movement in HD is in part related to skeletal muscle dysfunction [4, 15, 20, 23, 41, 62, 72].

Previously reported dysfunction in HD skeletal muscle include metabolic and mitochondrial defects [36, 42, 71], atrophy [25, 59, 66], weakness [15, 28, 43], and altered expression of genes needed for muscle differentiation [37, 69]. Additionally, our group observed neuromuscular and skeletal muscle defects in the R6/2 transgenic mouse model for HD that may contribute to impaired movement [33, 44, 45, 60, 74]. Namely, we found that R6/2 muscle fiber hyperexcitability was caused by reduced chloride channel (ClC-1) and inwardly rectifying K⁺ channel (Kir) currents and further showed that the reduced ClC-1 currents resulted from a progressive lack of Clcn1 splicing maturation [45, 74]. By examining R6/2 muscle membrane properties, we also found changes to the action potential (AP) waveform that include a low peak amplitude and prolonged falling phase that worsened during repetitive firing [44]. The prolonged AP repolarization in R6/2 muscle was explained by a decrease in K_V1.5 and K_V3.4 potassium channels [44]. Other groups have shown defects in excitation–contraction (EC) coupling in HD mouse muscle that would be expected to cause weakness [11, 17, 23]. Perhaps a bit surprising, R6/2 skeletal muscle does not function as predicted based on these repeatedly demonstrated mechanisms. For example, the dramatic reduction in ClC-1 currents does not produce clear myotonia (delayed relaxation) and the reduced Ca²⁺ release from the sarcoplasmic reticulum (SR) does not produce a decrease in specific force (force normalized to muscle mass), both of which can be seen in the R6/2 force records in this study.

To address this lack of translation of molecular mechanisms to physiology, this study took an integrative approach. We simultaneously examined the effects of altered R6/2 AP waveforms on Ca²⁺ release from the SR, and then assess the resulting effects on isometric twitch force. We observed a reduction in peak SR Ca²⁺ release flux in R6/2 muscle, consistent with previous reports [11, 17, 23] and further discovered an increase in the duration of SR Ca²⁺ release. By controlling the AP waveform with voltage-clamp, we demonstrate that the longer duration of the SR Ca²⁺ release was caused by the prolonged falling phase of R6/2 APs. The increased SR Ca²⁺ release duration helped compensate for the reduction in R6/2 peak SR Ca²⁺ release flux, as shown by similar levels of Ca²⁺ available for contraction (free Ca²⁺) between R6/2 and control fibers.

To obtain a more complete picture of the Ca²⁺ available for contraction, we examined key genes that modulate myoplasmic Ca²⁺ concentrations using Western blot and transcriptomics and used an equilibrium-based model to estimate the amount of Ca²⁺ that would bind troponin C and thus activate crossbridge cycling. The process was completed by measuring isometric muscle force. Despite a reduction in peak SR Ca²⁺ release, the specific twitch force of severely diseased R6/2 muscle was not lower than controls because the changes in the AP waveform and Ca²⁺-binding proteins. The severely altered AP and EC coupling mechanisms in R6/2 muscle actually function in concert to produce normal specific force in response to single APs. Obtaining this understanding required an integrated, multi-molecular approach. We further show that the delicate balance maintaining single twitch force [Scheme 1] does not maintain EC coupling or force during a simple train of 10 APs at 0.3 Hz.

Scheme 1.

The multi-molecular influence of Huntington’s disease on skeletal muscle function, generated in BioRender

Materials and methods

Animals

All animal procedures were performed in accordance with the policies of the Animal Care and Use Committee of Wright State University. To study HD skeletal muscle, we used the R6/2 mouse model, which is transgenic for exon 1 of the human huntingtin gene [38]. An R6/2 mouse breeding colony was established at the Wright State University Laboratory Animal Resources facility. Each R6/2 breeding pair consisted of a wild-type B6CBA female with an ovarian transplant (hemizygous for Tg(HDexon1)62Gpb) and a wild-type male (B6CBAF1/J) purchased from The Jackson Laboratory, stock #002810 (RRID: IMSR_JAX:002810). Tail samples were obtained between 7 and 14 days of age and sent to Laragen Inc. (Culver City, CA) for genotyping. Diseased mice were housed with wild-type littermates (control) in sex-matched cages after weaning at ∼14 days of age. Additional male and female control mice were obtained by crossing C57BL/6 J female (Stock No: 000664) with CBA/J male (Stock No: 000656) mice from the Jackson Laboratory. Additionally, to examine the effect of widening the action potential with tetraethylammonium ions (TEA) on SR Ca²⁺ release under voltage-clamp conditions, we used C57BL/10 J mice from Jackson Laboratory, stock #:000665 (RRID:IMSR_JAX:000665). All mice were age-matched and between 83 and 111 days of age.

Environmental conditions for housing the mice were maintained with a 12-h light/dark cycle and constant temperature (21–23 °C) and humidity (55 ± 10%). The cages contained corncob bedding (Harlan Teklad 7902) and environmental enrichment (mouse house and cotton nestlet). Mice were supplied with dry chow (irradiated rodent diet; Harlan Teklad 2981) and water ad libitum. Beginning at 10 weeks of age, all cages housing R6/2 mice were supplied with a Petri dish containing moist chow (dry chow soaked in water) to ensure adequate nutritional intake in the symptomatic mice. The health and phenotype severity assessment of R6/2 mice were performed weekly for mice 8–10 weeks of age, and daily for mice ≥ 11 weeks of age, as described previously [45, 74]. Assessment categories included physical condition, activity level, and weight loss. Of the Behavioral measures, mouse weight has been a reliable and easy to obtain measurement to assess disease progression. We previously showed that the body weight of R6/2 mice does not change from 6 to 10 weeks of age and decreases after 10 weeks of age (Miranda et al., 2017). Mice were euthanized by inhalation of a saturating dose of isoflurane (∼2 g/L) or CO₂, followed by cervical dislocation before harvesting muscles for experimentation.

Action potential and Ca²⁺ measurements

Hind limb flexor digitorum brevis (FDB) and interosseous (IO) muscles were dissected and enzymatically dissociated at 35 °C under mild agitation for ∼1 h using 1,000 U/mL of collagenase type II (Worthington Biochemical). Collagenase was dissolved in an extracellular solution (below). For the TEA experiments, the collagenase was dissolved in Leibovitz’s L-15 containing 10% foetal bovine serum (FBS) to improve the fiber preparation procedure. Dissociation was completed using mild trituration in the absence of collagenase. The fibers were allowed to recover at 21–23 °C for 1 h before electrical measurements were recorded.

Fibers were visualized on an Olympus IX71 microscope equipped with 10x (UPlanFL N/0.30∞/-/FN26.5) and 20x (UPlanFL N/0.50∞/0.17/FN26.5) non-immersion objectives, and a 40x (UPlanFLN/1.3 Oil ∞/0.17/FN26.5) oil-immersion objective (Center Valley, PA). Electrical properties were measured under standard current-clamp conditions at 21–23 °C using two borosilicate intracellular microelectrodes (Sutter Instruments, Novato, CA), an Axoclamp 900 A amplifier, a Digidata 1550 digitizer, and pClamp 11 data acquisition and analysis software (Molecular Devices, San Jose, CA). A reference electrode, grounded to the HSx1 headstage was placed into a cup containing 3 M KCl and connected to the extracellular fluid via agar bridges. Electrodes were impaled ~ 10 µm apart from each other. For current-clamp experiments, the voltage-sensing electrode was connected to an Axoclamp HSx1 headstage and the current-passing electrode was connected to an Axoclamp HSx10. For voltage-clamp experiments, the current-passing electrode was connected to an Axoclamp HSx2 (modified from a × 1 headstage). Data were acquired at 50 kHz, current and voltage records were low-pass filtered with the internal Axoclamp 900 A filters at 6 kHz.

Both the current-passing and voltage-sensing electrodes were filled with the same internal solution containing (in mM): 90 K-methanesulfonate, 2 Ca(OH)₂, 3 or 5 MgCl₂, 5 ATP disodium, 5 phosphocreatine disodium, 5 glutathione, 20 MOPS, 20 EGTA, pH 7.2 (KOH), and 0.02 fluo-4. The intracellular buffer had an osmolarity of 300 ± 5 mOsm. Three extracellular buffers were used for the current-clamp experiments: 0, 1 mM, and 10 mM TEA. Each extracellular buffer contained the following (in mM): 144 NaCl, 4 KCl, 1.2 CaCl₂, 0.6 MgCl₂, 5 glucose, 1 NaH₂PO₄, 10 MOPS, and pH 7.4 (NaOH). The 0 TEA buffer contained 50 µM BTS and 0.2% FBS to aid in electrode impalement. The 1 and 10 mM TEA buffers were made with TEA-OH. D-mannitol was added to the 0 and 1 mM TEA buffers so that osmolarity of all the extracellular buffers ranged from 300–305 mOsm. The extracellular buffer for the voltage-clamp experiments was modified to reduce or eliminate ionic currents in order to better control the AP waveform while measuring the Ca²⁺ transients. This buffer contained (in mM): 130 TEA-OH, 5 CsOH, 0.1 EGTA, 0.5 CaCl₂, 0.5 MgCl₂, 5 glucose, 10 MOPS, and pH 7.4 (HCl) + 0.2% FBS. This buffer substituted extracellular Na⁺ and K⁺ with TEA and Cs⁺. Additionally, Na⁺ currents were eliminated with 500 nM tetrodotoxin, K⁺ current was reduced with extracellular 4-aminopyridine (1 mM), Cl^– current was inhibited with extracellular 9-anthracene-carboxylic acid (400 µM); and ionic Ca²⁺ current across the membrane was blocked with extracellular Cd²⁺ (1.5 mM).

Upon impalement of each fiber, the intracellular solution was allowed to equilibrate with the sarcoplasm for 20 min while the baseline membrane potential was maintained between –85 and –80 mV by injecting a constant negative current. Fibers that exceeded –25 nA of holding current were excluded from the study. Current clamp action potentials (APs) were elicited by a 0.2 ms current pulse with an amplitude equal to 1.1 × the threshold for firing an AP. APs and intracellular Ca²⁺ were measured simultaneously. For each current-clamped fiber, APs and Ca²⁺ were first recorded in 0 TEA, followed by 1 mM TEA, followed by 10 mM TEA. Fluorescence due to Ca²⁺ binding fluo-4 was detected through the 40 × oil-immersion objective. A 470 nm wavelength light-emitting diode (Thorlabs, M470L3) with a 474/27 filter (Semrock FF01-474/27–25, Rochester, NY) was used for excitation. Emission was captured by a photomultiplier tube (PMT) (PMM02-1, Thorlabs, Newton, NJ) after passing through a 525/45 filter (Semrock FF01-525/45–25, Rochester, NY). The fluorescence signal from the PMT was low-pass filtered at 6 kHz (Brownlee Precision Model 440, Palo Alto, CA) and digitized with pClamp11. For each fiber, the background fluorescence of the dish was recorded under identical PMT and microscope settings and subtracted from the fluorescent signal for analysis.

Action potential analysis

APs were analyzed for the peak amplitude, 40% decay, and 80% decay as described previously [44]. The peak amplitude was the peak voltage above 0 mV. The 40% and 80% decay were the duration between the peak amplitude and the voltage at which the falling phase of the AP decays to 40% or 80% of the total voltage; the total voltage being the difference between the peak amplitude of each AP and the baseline membrane potential (~ –85 mV).

Ca²⁺ analysis

The time course of free [Ca²⁺] was estimated from the fluo-4 fluorescence signal as described previously for conditions in which the resting [Ca²⁺] is set by the buffering conditions of the pipette solution [29, 53, 54, 61]. One advantage of fluo-4 for this and previous studies is its high signal-to-noise ratio and low needed excitation intensities that reduce photobleaching and photodamage. A downside of fluo-4 is its slow fluorescence decay that may be due to complexation with other intracellular components or protons [61]. Despite this, the rising phase is unaffected, allowing a good approximation of the trajectories of the Ca²⁺ release flux in conditions with a high intracellular Ca²⁺ chelator (EGTA). For our calculations, first, the background-subtracted fluorescence signal (F) was divided by the resting fluorescence (F₀) to obtain F/F₀. F/F₀ was converted to the time course of free [Ca²⁺] with the following equation:

1

where k_off,fluo4 is the off-rate constant of fluo-4 (90 s⁻¹), and [Ca²⁺]_rest is the resting myoplasmic Ca²⁺ concentration. [Ca²⁺]_rest was determined using the MAXCHEALTOR program and the concentrations of Ca²⁺, Mg²⁺, ATP, and EGTA used in our intracellular solution (see above). It was assumed that 70% of the intracellular solution equilibrated with the sarcoplasm of the muscle fiber and the concentration of Ca²⁺ already in the sarcoplasm was 70 nM, as reported previously in R6/2 skeletal muscle [11]. Thus, 70% of the [Ca²⁺] calculated using MAXCHEALATOR was added to 30% of the [Ca²⁺] reported previously by Braubach et al., to obtain [Ca²⁺]_rest = 36.4 nM in the presence of 5 mM Mg²⁺. The time course of free [Ca²⁺] estimates the Ca²⁺ released into the myoplasm and accounts for the Ca²⁺ bound to fluo-4. However, the time course of free [Ca²⁺] does not account for the Ca²⁺ that binds to EGTA.

Under conditions of 20 mM EGTA, we assume that EGTA is the primary binding site for Ca²⁺ in the myoplasm. With EGTA in the intracellular solution, the following equation provides a good approximation of Ca²⁺ release flux:

2

where [Ca:EGTA] is the concentration of Ca²⁺ bound to EGTA, k_on,EGTA is the on-rate constant of EGTA (15 µM⁻¹ s⁻¹), k_off,EGTA is the off-rate constant of EGTA (1.2 s⁻¹), and [EGTA]_total is the concentration of EGTA in the intracellular solution (20 mM). The initial concentration for Ca²⁺ bound to EGTA ([Ca:EGTA] = 1399.93 µM) was determined by using MAXCHEALTOR. Based on Eq. 1 and Eq. 2, it is assumed that Ca²⁺ binds primarily to the dye (Eq. 1) and EGTA (Eq. 2), bypassing endogenous Ca²⁺ buffers, including parvalbumin, ATP, troponin C (EGTA also blocks contraction), and the sarco/endoplasmic reticulum Ca²⁺ ATPase (SERCA) [31, 55, 56].

With the gadget integration tool in OriginPro 2021 (OriginLab Corporation), the integral of the Ca²⁺ release flux was used to assess the peak Ca²⁺ release flux, full-width half-maximum (FWHM), and the Ca²⁺ available for contraction. The baseline for the integral/area under the curve (AUC) was set to “y = 0.” The peak Ca²⁺ release flux was the maximum height of the AUC, the FWHM was the duration of the AUC 50% from the maximum height, and the Ca²⁺ available for contraction was the AUC value.

Protein measurements

For Western blot analysis, FDB/IO and EDL muscles were lysed in RIPA buffer containing 10 mM Tris–HCl (pH 7.4), 30 mM NaCl, 1 mM EDTA, and 1% Nonidet P-40, supplemented with proteinase and phosphatase inhibitors. Protein concentrations were determined for each sample, with no fraction of the sample discarded. Equal amounts of protein were boiled at 95 °C for 5 min in 1% SDS sample buffer. Proteins were then separated by 10–15% SDS-PAGE. Muscle lysates were serially diluted and loaded at 0.5 µg, 1 µg, 2 µg, and 4 µg of total protein to calibrate protein loading and ensure that protein amounts were in a linear range for detection and to avoid saturation, as described previously [46, 48]. Based on those results, analysis for each protein was completed with 1 µg of total protein per lane. Proteins were transferred to PVDF membranes (Millipore, IPVH00010), which were then blocked with 5% non-fat milk (Cell Signaling Technologies, 9999) for 1 h and incubated overnight at 4 °C with primary antibodies in 5% non-fat milk in tris-buffered saline with Tween 20. Following incubation with secondary antibodies for 1 h at room temperature, protein bands were detected using an Amersham Imager 600 (GE Healthcare Life Sciences). Ponceau S staining was initially performed to evaluate total protein for normalization. However, since only faint Ponceau S bands were visible with the low protein amount that we used (1 µg), we needed to use a housekeeping gene for normalization. We have previously found that the expression of common normalizing genes such as β-tubulin and glyceraldehyde-3-phosphate (GAPDH) varied between R6/2 and control in a tissue specific manner [60]. For example, GAPDH expression appeared similar in R6/2 and control gastrocnemius but not the quadriceps femoris, whereas β-tubulin appeared to differ in both muscles [60]. For this study, we found that GAPDH expression appeared consistent in both the FDB/IO and EDL muscles, but β-tubulin expression appeared to differ between R6/2 and control in these muscles. Thus, the densitometry values were normalized to the GAPDH loading control (Cell Signaling Technologies, 2118, 1:1000), densitometry values were obtained from the same membrane. Primary antibodies that have been successfully used previously were selected and include SERCA1 (Thermo Fisher Scientific, MA3-912, 1:1000) [12, 26, 39], SERCA2 (Thermo Fisher Scientific, MA3-919, 1:1000) [13, 76], parvalbumin (Thermo Fisher Scientific, PA1-933, 1:10000) [47], and troponin C (Santa Cruz, sc-48347, 1:500) [40]. The secondary antibodies used include HRP-conjugated anti-mouse IgG (Cell Signaling Technology, 7076, 1:5000) and HRP-conjugated anti-rabbit IgG (Cell Signaling Technology, 7074, 1:5000).

Estimate of Ca²⁺ bound to troponin

To quantify the amount of myoplasmic calcium bound to troponin, we used an equilibrium-based model that accounted for key protein interactions of Ca²⁺ released from the SR in response to stimulation. We considered the reversible binding of Ca²⁺ to parvalbumin, cooperative binding to two regulatory sites on each troponin molecule, and two sites on SERCA. Reaction schemes, protein concentrations, and rate constants were based on previous reports[5, 6]. Data from our voltage-clamp analysis was used for the amount of free Ca²⁺ ().

For the reversible binding of Ca²⁺ to parvalbumin (Parv), we obtain the following equation at equilibrium

3

where are the concentrations of free (unbound) Ca²⁺, parvalbumin not bound to calcium, and parvalbumin bound to calcium, respectively, and is the dissociation constant for the binding/unbinding reactions. Similarly, Ca²⁺ binding to troponin (Trop) results in

4

5

where are the concentrations of troponin that are unbound to calcium, have one site occupied by calcium, and have both sites occupied by calcium, respectively, with and being the dissociation constants for the corresponding reactions. Finally, Ca²⁺ binding to SERCA results in the following equations at equilibrium

6

7

with being the concentrations of SERCA not bound to Ca²⁺, has one site bound to Ca²⁺, and has both sites bound to Ca²⁺. In addition to these five equations, we have four additional equations

8

9

10

11

where , and are the total concentrations of parvalbumin, troponin, SERCA, and Ca²⁺, respectively. Equations 3 – 4 were solved in Mathematica to obtain the calcium bound to troponin ( and ).

RNA-seq analysis

Raw count data from the RNA-sequencing study (GSE81367, [9]) were downloaded from the Gene Expression Omnibus (GEO). To control sample-level quality, Principal Component Analysis (PCA) was performed, identifying and eliminating one outlier sample. Log2 fold change values of differential gene expression were generated using DESeq2 (10.1186/s13059-014-0550-8). Multiple comparison testing by Benjamini–Hochberg procedure was used to assess significance; genes with an adjusted p-value of 0.05 were considered significant.

Isometric force measurements setup

The extensor digitorum longus (EDL) muscle was dissected from the mice and bathed in a solution containing (in mM) 118 NaCl, 3.5 KCl, 1.5 CaCl₂, 0.7 MgSO₄, 1.7 NaH₂PO₄, 26.2 NaHCO₃, 5.5 glucose and pH 7.4 that was continuously gassed with 95% O₂ and 5% CO₂. Additionally, the EDL was incubated in 300 nM α-bungarotoxin for 1 h to eliminate changes in neuromuscular transmission contributing to the force production. Force experiments were completed in a custom 3D-printed recording chamber at 21–23 °C. The proximal tendon was attached to the chamber with a 6–0 calibre silk suture. The suture was also used to tie the distal tendon to a hook, which was then attached to a 300D-300C-LR force transducer that was connected to a 305 C two-channel controller (Aurora Scientific). An S-900 pulse generator and S-910 Stimulus Isolation Unit (Dagan) were used to elicit 2.0 ms voltage pulses with an amplitude of 100 V via two platinum electrodes. All equipment was controlled and data was acquired using a Digidata 1550B digitizer and pClamp 10 (Molecular Devices). The optimal length of the muscle was determined by adjusting the tension of the muscle until the maximal twitch force was achieved. Force data was analyzed using pClamp 11 and OriginPro 2021 or 2022 software.

Chemicals

Chemicals were purchased from Fisher Scientific; exceptions include CaCl₂ and MgCl₂ stock solutions (TekNova), ATP di-Na (Sigma-Aldrich).

Statistics

We compared the action potential and Ca²⁺ release flux properties between control and R6/2 muscle fibers using linear mixed models. One model was used per dependent variable (AP peak, AP maximum rate-of-rise, AP D40, AP D80, peak Ca²⁺ release flux, Ca²⁺ release full-width half-maximum, and free Ca²⁺), each of which included mouse IDs as a random effect to account for multiple fibers being tested per mouse. We calculated p-values deriving degrees of freedom using Satterthwaite’s method to account for unequal variances [64].

Linear mixed models were used to compare the AP and Ca²⁺ release flux parameters in the presence of 0 mM, 1 mM, and 10 mM TEA. Each of the models contained one of the six dependent variables [AP peak, AP D40, AP D80, peak Ca²⁺ release flux, Ca²⁺ release full-width half-maximum, and free Ca²⁺) with the TEA level as fixed effect and mouse as random effect. We further compared the effect of the three levels of TEA in pairwise comparisons, with p-values adjusted by the Tukey method [70].

The voltage clamp analysis of action potential and Ca²⁺ release flux properties was assessed using linear mixed models with genotype (control and R6/2 muscle), AP waveform, and their interaction as fixed effects, as well as mouse as random effect. The method of Nakagawa and Schielzeth’s [50] was used for obtaining R2 measures. This yields the marginal R2 related to fixed effects only, the conditional R2 for fixed and random effects combined, and by taking the difference also an R2 related to the random effect of mouse only. We conducted posthoc pairwise comparisons only between Control:Control (Control fiber with a Control AP waveform) vs Control:R6/2; R6/2:Control vs R6/2:R6/2; and Control:Control vs R6/2:R6/2. We used Kenward and Roger’s method [32] for estimating degrees of freedom and a p-value adjustment for three simultaneous tests [67].

Western blot results for SERCA1, SERCA2, Parvalbumin, Troponin C between control and R6/2 were compared using t-tests. We investigated differences in individual twitch force parameters and EDL weights between control and R6/2 using linear models.

To examine differences in trains of twitches, we used linear mixed effects models to compare the effect of control vs. R6/2 fibers on six different variables over 10 stimuli. For each of the six models we used treatment (control vs. R6/2), stimulus, and their interaction as fixed effects and mouse as random effect. Note that we treated stimulus as a categorical variable as there was no expectation of a particular functional form in the relationship between the variables and the number of the stimulus. Also, we wanted to contrast the effect of treatment at each individual stimulus. We implemented the latter with a posthoc test relying on Kenward and Roger’s method [32] for estimation of degrees of freedom and the mvt method as implemented in the R package emmeans [35] for adjusting the p-value for multiple tests. The twitch train force measures were obtained from whole muscles and thus didn’t require the inclusion of mouse as a random effect. Consequently, we used linear models with treatment, stimulus and their interaction to analyze the three different variables, followed by a posthoc analysis for contrasting the effect of treatment at each individual stimulus again using the mvt adjustment for multiple tests.

We implemented all statistical analyses in the statistical programming environment R [58] with use of the additional packages lme4 [3], MuMIn [2], and emmeans [35].

Results

AP-evoked Ca²⁺ release

Our previous work demonstrated that APs are substantially prolonged in R6/2 skeletal muscle [44, 74]. To determine whether prolonged R6/2 muscle APs alter EC coupling, single APs and intracellular Ca²⁺ were measured simultaneously from late-stage R6/2 and age-matched noncarrier sibling FDB/IO (control) muscle fibers under current clamp conditions. Myoplasmic Ca²⁺ was detected with the fluo-4 Ca²⁺ indicator dye, which was loaded into the fibers via the voltage-sensing and current-passing microelectrodes (Fig. 1A). The intracellular solution in the electrodes contained 20 mM EGTA to help block contractions. Representative control and R6/2 traces showing an AP, intracellular fluorescence signal, and current stimulus are shown in Fig. 1B. The Ca²⁺ release flux was calculated from the AP-evoked intracellular fluorescence signal (Fig. 1C, left) as described previously [56]. Briefly, the fluorescence signal was normalized to the baseline fluorescence (Fig. 1C, middle) then corrected kinetically to obtain the Ca²⁺ release flux signal (Fig. 1C, right, Eq. 1 and 2 in methods).

Fig. 1.

Microscope setup, representative raw traces, and conversion of fluorescence to Ca²⁺ release flux. A Microscope setup for intracellular fluorescence detection. Fluo-4 Ca²⁺ indicator dye contained within intracellular microelectrode, excited with 470 nm light emitting diode (LED), and 525 nm emission captured by photomultiplier tube (PMT). B Representative control and R6/2 action potential, fluo-4 signal, and current stimulus. C Representative control fluorescence signal (F) showing baseline fluorescence (F0) converted to F/F0 then Ca²⁺ release flux

Averaged control and R6/2 APs and Ca²⁺ release flux signals are shown in Fig. 2A. We have previously reported that R6/2 APs were prolonged and had lower rise times and peak amplitudes compared to control [44]. Our previous study was completed without fluo-4 or other comparable Ca²⁺ indicators. In this study, R6/2 APs were significantly prolonged, with an increased 40% decay time (D40; control: 0.66 ± 0.13 ms, R6/2: 1.62 ± 0.14 ms, p = 1.212 × 10^–5) and 80% decay time (D80; control: 1.94 ± 0.90 ms, R6/2: 8.05 ± 0.96 ms, p < 0.0161). The R6/2 AP peak amplitude (control: 41.8 ± 3.0 mV, R6/2: 29.2 ± 3.1 mV, p = 0.0220) and the maximum rate-of-rise (control: 634.5 ± 39.2 mV/ms, R6/2: 415.4 ± 39.4 mV/ms, p = 3.792 × 10^–5) were also significantly decreased in R6/2 muscle compared to control. These results confirm that R6/2 muscle APs are prolonged with reduced amplitudes compared to control, as shown previously, and that fluo-4 at the concentrations used did not affect the AP waveform.

Fig. 2.

R6/2 action potential-evoked Ca²⁺ release flux. A Average control (black trace, n = 11 fibers, 4 mice) and R6/2 (magenta trace, n = 12 fibers, 4 mice) action potential and corresponding Ca²⁺ release flux. B Average R6/2 Ca²⁺ release flux signal with peak Ca²⁺ release flux, full-width half-maximum (FWHM), and Free Ca²⁺, aka., Ca²⁺ available for contraction (area under the curve shaded in gray) labeled. Box and whisker plots for C peak Ca²⁺ release flux, D FWHM, and E estimation of Ca²⁺ available for contraction. Box and whisker plots show the 25th and 75th percentile (box), mean (white box), median (line), and 1.5 interquartile range (error bars). A magenta * indicates a significant difference compared to control (p < 0.05)

Our analysis of SR Ca²⁺ release (Fig. 2B) showed that the peak R6/2 Ca²⁺ release flux (68.8 ± 22.3 µM/ms) was significantly lower compared to control (171.8 ± 22.2 µM/ms; p = 0.0161) (Fig. 2C), which previous work suggested was due to a dysfunctional ryanodine receptor (RyR1) in HD muscle [23]. However, we also found that the duration of the AP-triggered Ca²⁺ release flux (FWHM) was significantly increased in R6/2 muscle (1.72 ± 0.17 ms) compared to control muscle (1.08 ± 0.17 ms; p = 0.0315) (Fig. 2D). This prolonged Ca²⁺ release duration helps compensate for the reduced peak Ca²⁺ release flux because there was no significant difference in the free Ca²⁺ (Ca²⁺ available for contraction) between R6/2 and control muscles. Still, considering the R6/2 free Ca²⁺ (103.29 ± 26.81 µM) was trending lower compared to control (185.31 ± 26.63 µM/ms; p = 0.0654), the compensation appears incomplete (Fig. 2E). Overall, this data suggest that prolonged R6/2 APs increase the duration of RyR1-mediated Ca²⁺ release from the SR. Although, with this data alone it cannot be ruled out that disease-related changes in RyR1 lead to the increased Ca²⁺ release duration, independent of AP duration.

Blocking K⁺ channels increases Ca²⁺ release duration

To further investigate the association between prolonged APs and Ca²⁺ release duration, we used tetraethylammonium ions (TEA) to partially block voltage-gated K⁺ channels [30, 68] and prolong APs in healthy wild-type (WT) muscle. The goal was to acutely prolong muscle APs independently of a disease state and changes in the AP rising phase. APs were recorded under current-clamp conditions in the presence of 0 mM, 1 mM, and 10 mM TEA in the same WT muscle fiber. Statistical analysis accounted for the paired effect of measuring the 3 concentrations of TEA in each fiber (see Methods). Figure 3A shows averaged AP traces and Ca²⁺ release flux signals for all three concentrations of TEA, with the peak amplitude, D40, and D80 of the AP labelled. A full listing of the mean AP and Ca²⁺ release values as well as p-values for the TEA data is shown in Table 1.

Fig. 3.

Effect of prolonged APs on control muscle Ca²⁺ release flux. A Average action potential and corresponding Ca²⁺ release flux in 0 mM (black), 1 mM (green), and 10 mM (blue) TEA (n = 12 fibers, 5 mice). Box and whisker plots for action potential B peak, C D40, and D D80; and E peak Ca²⁺ release flux, F FWHM, and G Ca²⁺ available for contraction (Free Ca²⁺) in 0, 1, and 10 mM TEA. Box-and-whisker plots show the 25th and 75th percentile (box), mean (dashed white line), median (solid black line), and 1.5 interquartile range (error bars). Significant differences (p < 0.05) are shown as horizontal bars, the full list of p-values are shown in Table 1

Table 1.

Summary of the effects of 0, 1 mM, and 10 mM TEA on an action potential (AP) and SR Ca²⁺ release event. Shown in the top of the table are the mean values ± SEM of the AP peak amplitude (AP Peak), AP 40% decay time (AP D40), AP 80% decay time (AP D80), peak Ca²⁺ release flux (Peak Ca²⁺ Rel. Flux), full-width half-maximum duration of the Ca²⁺ release event (FWHM), and the estimated Ca²⁺ available for contraction (Free Ca²⁺). At the bottom of the table are the p-values for the comparisons of 0 vs 1 mM TEA, 0 vs 10 mM TEA, and 1 mM vs 10 mM TEA for each of the AP and Ca²⁺ release event properties, significant differences (p < 0.05) are shown in bold. Notably, the SEM values were the same for each characteristic because in our mixed models the variance was pooled across groups, which, especially when using the Kenward-Roger approximation for degrees of freedom, can lead to similar standard errors across different contrasts

[TEA]	AP Peak (mV)	AP D40 (ms)	AP D80 (ms)	Peak Ca2+ Rel. Flux (µM/ms)	FWHM (ms)	Free Ca2+ (µM)
mean ± SEM
0 mM	43.2 ± 2.1	0.57 ± 0.09	2.17 ± 0.30	157 ± 50.9	1.05 ± 0.02	163 ± 57.5
1 mM	41.1 ± 2.1	0.77 ± 0.09	2.37 ± 0.30	218 ± 50.9	1.10 ± 0.02	241 ± 57.5
10 mM	38.8 ± 2.1	1.28 ± 0.09	3.99 ± 0.30	313 ± 50.9	1.15 ± 0.02	375 ± 57.5
p-values
0 vs 1 mM	0.1903	0.1005	0.7628	0.3280	0.0462	0.2432
0 vs 10 mM	0.0025	< 0.0001	< 0.0001	0.0031	0.0002	0.0005
1 vs 10 mM	0.1353	< 0.0001	< 0.0001	0.0828	0.0770	0.0253

The average peak amplitude of the APs in 10 mM TEA was significantly lower than in the absence of TEA; otherwise, there were no other significant effects of TEA on peak AP amplitude (Fig. 3B). As expected, there was a greater effect of TEA on the repolarization phase of the APs. The D40 of the APs in 10 mM TEA was significantly prolonged compared to both 1 mM TEA and 0 mM TEA (Fig. 3C). Similarly, the duration of the late falling phase of the AP (D80) in 10 mM TEA was significantly prolonged compared to 1 mM TEA and 0 mM TEA (Fig. 3D). These results confirm that TEA prolongs muscle APs by slowing the repolarization phase of the AP.

Prolonging the APs with TEA had a significant effect on the resulting Ca²⁺ transient (Fig. 3A, bottom). The average peak Ca²⁺ release flux in response to 10 mM TEA was significantly larger than in the absence of TEA (Fig. 3E and Table 1). However, there were no significant differences between the peak Ca²⁺ release flux in response to 0 and 1 mM TEA or between 1 and 10 mM TEA (Fig. 3E). The FWHM significantly increased from 0 to 1 mM TEA and from 0 to 10 mM TEA, but was not significantly different from 1 to 10 mM TEA (Fig. 3F). The Ca²⁺ available for contraction increased between 0 to 10 mM TEA and between 1 to 10 mM TEA (Fig. 3G), but not between 0 to 1 mM TEA.

A limitation of the graphs in Fig. 3 is that they do not illustrate the changes in AP waveform and Ca²⁺ release flux with increasing [TEA] for each fiber. Thus, plots with lines connecting data points from each fiber with increasing [TEA] are shown in Fig. 4. Notably, changes in the D40 and D80 (Fig. 4B and 4 C, respectively) do not vary as much as the Ca²⁺ release flux parameters (Fig. 4D-F). Overall, the TEA experiments support the hypothesis that prolonged APs increase the Ca²⁺ release duration and the Ca²⁺ available for contraction. This is consistent with our previous report showing that AP-induced increases in myoplasmic Ca²⁺ depend more tightly on the area-under-the-curve than the peak of an AP [8, 73].

Fig. 4.

TEA action potential-evoked Ca²⁺ release flux for each fiber. A Action potential peak, B D40, and C D80; and Ca2+ release flux D peak, E FWHM, and F Ca²⁺ available for contraction in 0, 1, and 10 mM TEA (n = 5 mice, 12 fibers). Each line connects the same fiber across TEA concentrations, and each color represents the same fiber across all plots

Mechanisms of altered Ca²⁺ release via voltage-clamp

To examine the effect of prolonged R6/2 APs on Ca²⁺ release more mechanistically, we measured the SR Ca²⁺ release in response to both control and R6/2 AP waveforms (Fig. 5A) in individual fibers from control and R6/2 (Fig. 5B) mice. This was done under voltage clamp using the average control and R6/2 AP waveforms as the voltage command signals (Fig. 5A is same as Fig. 2A). By comparing the responses to both control and R6/2 waveforms in the same fiber, we could determine whether the genotype and/or the AP waveform influenced the Ca²⁺ release. Results are described with genotype followed by the AP waveform. For example, Control:Control indicates a control fiber with a control AP and Control:R6/2 indicates a control fiber with a R6/2 AP.

Fig. 5.

Voltage Clamp AP and Ca²⁺ release. A Average control (black trace, n = 15 fibers, 3 mice) and R6/2 (magenta trace, n = 19 fibers, 2 mice) action potential recorded under current clamp conditions used as the voltage command waveform (same as Fig. 2A). B The average Ca²⁺ release fluxes in response to the control (black trace) and R6/2 (magenta trace) AP waveforms in control (left panel) and R6/2 (right panel) fibers. C-E Statistics for the peak Ca²⁺ release flux, (C) FWHM, and (E) Free Ca²⁺ (Ca²⁺ available for contraction). Box-and-whisker plots show the 25th and 75th percentile (box), mean (dashed white line), median (solid black line), and 1.5 interquartile range (error bars). Significant differences (p < 0.05) are shown as horizontal bars, p-values are shown in the Results text

For the peak SR Ca²⁺ release flux (Fig. 5C), our statistical examination showed that 47.14% of the variance in the data was explained by the genotype, AP waveform and their interaction, as well as that differences between individual mice accounted for 17.04% of the variance. The remaining variance is ascribed to normal random variation. Overall, there was a significant effect of genotype on the peak SR Ca²⁺ release flux (p = 0.0492). In contrast, there was not a significant effect of AP waveform on the peak SR Ca²⁺ release flux (p = 0.1472) or in the interaction between genotype and AP (p = 0.5376). The later indicates that the effect of genotype does not depend on AP waveform and vice versa. Thus, the significantly reduced peak SR Ca²⁺ release flux in R6/2 fibers compared to control (Fig. 5C), in response to both control and R6/2 AP waveforms, was dependent on genotype but not AP waveform. A posthoc comparison of means found no significant differences for Control:Control (93.8 ± 13.8 µM/ms) vs Control:R6/2 (108.5 ± 13.8 µM/ms, p = 0.4232); R6/2:Control (32.6 ± 16.0 µM/ms) vs R6/2:R6/2 (38.6 ± 16.0 µM/ms, p = 0.8932); or Control:Control vs R6/2:R6/2 (p = 0.1817).

Compared to the peak release flux, there were greater differences in the duration of SR Ca²⁺ release (full-width half-maximum, FWHM) (Fig. 5D). The genotype, AP waveform and their interaction explained 63.54% of the variance in the FWHM data and differences between individual mice accounted for only 1.25% of the variance. Overall, there was a significant effect of genotype (p = 0.0360) and a dramatic effect (1.137 × 10^–12) of the AP waveform on the FWHM. A significant interaction between genotype and AP (p = 0.0395) was found for the FWHM, indicating that the effect of genotype depends on AP waveform and vice versa. Thus, differences in the FWHM of SR Ca²⁺ release depend heavily on the AP waveform and to a lesser extent on genotype. This suggests that the disease-mediated prolonging of the AP repolarization primarily drove the increased SR Ca²⁺ release flux time in R6/2 fibers. A posthoc comparison of means found significant differences in Control:Control (1.60 ± 0.10 ms) vs Control:R6/2 (2.19 ± 0.10 ms, p = 0.0001); R6/2:Control (1.90 ± 0.11 ms) vs R6/2:R6/2 (2.86 ± 0.11 ms, p < 0.0001); and Control:Control vs R6/2:R6/2 (p = 0.0010).

The estimates of free Ca²⁺ (Ca²⁺ available for contraction) (Fig. 5E) provide an indication of the extent to which the prolonged AP and resulting longer duration SR Ca²⁺ release event compensate for the reduced peak SR Ca²⁺ release flux in R6/2 muscle. We found that 49.98% of the variance in free Ca²⁺ was explained by the genotype, AP waveform and their interaction. Differences between individual mice accounted for 17.34% of the variance in free Ca²⁺. Overall, there was a non-significant but trending effect of genotype on free Ca²⁺ (p = 0.0637). Indeed, the significant decrease in peak R62 Ca²⁺ release that was attributed to genotype will affect free Ca²⁺. More evident, the AP waveform had a clear significant effect on free Ca²⁺ (p = 2.27 × 10^–6). There was not a significant interaction between genotype and AP (p = 0.1295) on the on free Ca²⁺. Therefore, overall differences in free Ca²⁺ between R6/2 and control were driven primarily by changes to the AP waveform. A posthoc comparison of the means found significant differences in Control:Control (157.4 ± 27.0 µM) vs Control:R6/2 (247.5 ± 27.0 µM, p = 0.0001) as well as R6/2:Control (62.8 ± 31.5 µM) vs R6/2:R6/2 (112.0 ± 31.5 µM, p = 0.0219), indicating that the prolonged R6/2 AP waveform increased the free Ca²⁺ in both control and R6/2 fibers. In contrast, there was no significant difference between Control:Control and R6/2:R6/2 (p = 0.7143), indicating that the prolonged R6/2 AP and resulting longer duration SR Ca²⁺ release event indeed compensates for the reduced peak SR Ca²⁺ release flux in R6/2 muscle.

There was also an apparent delay in the peak of the Ca²⁺ transient in response to the R6/2 AP waveform compared to the control AP waveform. Because this delay occurred in both control and R6/2 fibers, the effect is likely independent of genotype. This delay may be attributed to the lower and delayed peak of the R6/2 AP compared to the control AP.

Ca²⁺-binding protein expression

As noted above, the free Ca²⁺ indicates the amount of released SR Ca²⁺ that is considered available for contraction. Other factors, such as calcium-binding proteins and transporters that buffer myoplasmic Ca²⁺ and re-sequester Ca²⁺ in the SR, respectively, help define the relationship between released Ca²⁺ and muscle force. To determine whether HD affects proteins involved in Ca²⁺ handling, we measured protein expression levels of the SR Ca²⁺ ATPase types 1 and 2 (SERCA1 and SERCA2), parvalbumin, and the fast isoform of troponin C (TNNC2). SERCA1 is a P-type ATPases that acts as the main removal mechanism of Ca²⁺ from the myoplasm of fast-twitch skeletal muscle [10, 75]. SERCA2 is primarily expressed in cardiac and slow-twitch skeletal muscle and can be an indicator of fiber type switching in fast-twitch skeletal muscle [34, 77]. Parvalbumin is considered a slow Ca²⁺ buffer in the myoplasm of fast-twitch muscle fibers [7] that is correlated with the rate of muscle relaxation [27]. A decrease in either SERCA1 or parvalbumin activity or expression would increase the amount of Ca²⁺ in the myoplasm available to bind troponin C and drive muscle contraction. Levels of these proteins were measured in control and R6/2 FDB/IO muscles, the primarily fast-twitch muscles that we used for AP and Ca²⁺ analyses, and were normalized to GAPDH levels (Fig. 6A&B). In R6/2 muscle, there were significant reduction in SERCA1 (–53.0%, p < 0.0001) and parvalbumin (–69.6%, p < 0.0001) compared to control, whereas fast troponin C expression remained unchanged (–1.1%, p = 0.9141) (Fig. 6A&B). SERCA2 was not detected in the FDB/IO muscle.

Fig. 6.

Western blot analysis of Ca²⁺ -handling proteins. Immunoblots (A and C) and quantitation (B and D) of Ca²⁺ handling proteins in FDB/IO (top panels) and EDL (bottom panels) muscle of control (n = 5 mice) and R6/2 (n = 5 mice) mice. Box-and-whisker plots show the 25th and 75th percentile (box), mean (white box), median (line), and 1.5 interquartile range (error bars). An * indicates a significant difference in R6/2 compared to control (p < 0.05), the p-values are listed in the Results text

Similarly, we examined the expression level of the same Ca²⁺-binding proteins in extensor digitorum longus (EDL) muscle in control and R6/2 mice (Fig. 6C). The EDL is a fast-twitch muscle commonly used for ex vivo muscle contracture studies, as we did later in this study. In R6/2 EDL muscle, the expression of SERCA1 (–63.4%, p < 0.0001) and parvalbumin (–63.9%, p = 3.931 × 10^–4) were significantly decreased compared to controls, the expression of fast troponin C expression (34.7%, p = 0.0917) was not significantly different from controls but was trending lower (Fig. 6C&D), and SERCA2 was not detected.

Next, we sought to complement our measurements of protein expression with analysis of excitation–contraction (EC) coupling genes in a publicly available transcriptomic dataset (GSE81367) [9]. We examined 31 genes from control and R6/2 quadriceps muscle (n = 10, per group) in this dataset (Fig. 7). Genes were selected based on a previous proteomic analysis of EC coupling genes [22]. A full list of the protein names, Ensembl codes, and full transcriptomics results for each EC coupling gene are included in the Harvard Dataverse, https://doi.org/10.7910/DVN/4LCITP. We found an overall reduced expression of EC coupling genes in R6/2 muscle compared to control (Fig. 7A). For example, directly related to this study, Pvalb (parvalbumin), Casq1 (calsequestrin-1), Atp2a1 (SERCA1), Tnnc2 (troponin C fast), and Cacna1s (Ca_V1.1) all had –log₂ fold changes that were > 1.0. The only EC coupling gene with a significant increase in expression in R6/2 quadriceps was Casq2 (calsequestrin-2), which had log₂ fold change > 1.0. The transcriptomics data were also shown as a volcano plot to illustrate the relationship between the statistical significance (–log₁₀ padj) and the magnitude (log₂ fold change) of the change in gene expression (Fig. 7B). Genes with a –log₁₀ padj of > 2 are shown as a black circle in Fig. 7B.

Fig. 7.

Transcriptomic analysis of quadriceps muscles derived from a publicly available data set. A Bar graph of differential gene expression (log2 fold change) for a subset of excitation–contraction coupling genes in R6/2 quadriceps muscles compared to control mice from Bondulich et al. The padj values are shown next to each bar, significant decreases are shown in blue, non-significant changes are shown in purple and significant increases are shown in red. B A graph of the -log10 padj plotted as function of the log2 fold change to help illustrate padj and relative gene expression changes. Genes that changed expression with a -log10 padj greater than 2 are shown as black circles labeled with the gene name and -log10 padj value. Genes that changed expression with a -log10 padj less than 2 are shown as open circles and include Ryr2, S100a1, Stim1, Ryr3, Mcu, Fkbp1a, Tnnc1, Atp2a2, Atp2b2, Slc8a1, and Calm4

Model estimate of Ca²⁺ bound to troponin

To begin to quantitatively assess the integrated effects of the reduced Ca²⁺ release flux, prolonged Ca²⁺ release, and altered expression of Ca²⁺ binding proteins on muscle contraction, we used an equilibrium-based model that captured the interactions of myoplasmic Ca²⁺ after stimulation with key proteins to estimate the amount of Ca²⁺ that would bind to troponin and thus drive muscle contraction. Included in the model were the reversible binding of Ca²⁺ to parvalbumin, the cooperative binding of two Ca²⁺ molecules to troponin, and the binding of Ca²⁺ to two sites on SERCA1 [5, 6].

For control muscle, we used protein concentrations and rate constants that have been reported previously [5, 6]. The concentration of free Ca²⁺ was the empirical value obtained from our voltage clamp experiments of the Control-Control conditions. The parameters for control muscle were, and and the dissociation constants were . With these conditions, the total control Ca²⁺ bound to troponin was 0.0408 µM ( and ).

Modelling the R6/2 free Ca²⁺ (117 µM) from the voltage clamp conditions (R6/2-R6/2) using the same protein concentrations as were used for control muscle, we found that the total Ca²⁺ bound to troponin in R6/2 muscle was 0.0288 µM ( and ).

To examine the effect of the reduced expression of R6/2 Ca²⁺-binding proteins, we ran the model using relative changes in R6/2 protein concentrations that were significantly different compared to control in our Western blot analyses. R6/2 free Ca²⁺ was kept at 117 µM and rate constants were assumed to be unchanged in R6/2 muscle.

In R6/2 FDB muscle, we found significant decreases in SERCA1 (0.470 of control) and parvalbumin (0.304 of control). Thus, we ran the model for R6/2 FDB using (instead of ) and (instead of the ) and found that the total Ca²⁺ bound to troponin in R6/2 FDB muscle increased to 0.1406 µM ( and ). This value of Ca²⁺ bound to troponin in R6/2 FDB is substantially greater than the value estimated for control (0.0408 µM).

In R6/2 EDL muscle we also found significant decreases in SERCA1 (0.366 of control) and parvalbumin (0.361 of control) and thus ran the model for R6/2 EDL using (instead of ) and (instead of the ). This simulation showed that the total Ca²⁺ bound to troponin in R6/2 EDL muscle was 0.1137 µM ( and ), which was also substantially greater than the value estimated for control.

R6/2 muscle twitch force

To better determine the physiological significance of R6/2 AP and SR Ca²⁺ release alterations, we measured the twitch force generated in response to isolated APs from whole dissected EDL muscle from control and R6/2 animals. Isometric twitches were elicited via direct muscle stimulation and force was measured with a transducer under ex vivo conditions at 21–23 °C to more closely match the conditions of the AP and Ca²⁺ release experiments. The EDL was stimulated 10 times at a frequency of 0.3 Hz. This is a relatively low-intensity stimulation protocol that allows for recovery between each stimulation. For example, it allowed us to observe changes without the depolarization of the resting membrane potential that has been shown at higher AP frequencies during longer stimulation trains [44]. To ensure that we examined a twitch response that was not affected by any other stimuli of the 0.3 Hz train, we assessed the first twitch of the train (Fig. 8). The averaged absolute (i.e., not normalized to muscle size) force of the first twitch for control and R6/2 muscles are shown in Fig. 8A (top panel). The peak of the absolute twitch force of R6/2 muscle was significantly lower than control (p = 0.0135) (Fig. 8B). A decrease in R6/2 absolute twitch force was expected, not only because reduced absolute peak twitch force has been reported in R6/2 EDL previously [28, 43], but also because of the muscle atrophy that is a hallmark characteristic of R6/2 muscle [38, 63]. Indeed, we found that R6/2 EDL muscle weights were significantly reduced and about half the size of control EDL muscles (p = 4.08 × 10^–5) (Fig. 8C). To account for this atrophy, we normalized the absolute force to the EDL weight to obtain the specific force (Fig. 8A, bottom). The results showed that the R6/2 peak specific force was nearly indistinguishable from that of control (p = 0.4197) (Fig. 8D). Additionally, there were no significant differences in the time-to-peak measured as the time from 10 to 90% of the peak (p = 0.1214) (Fig. 8E), half-width (p = 0.6354) (Fig. 8F), or half-maximum relaxation time (p = 0.5757) (Fig. 8G). These results suggest that the increased duration of Ca²⁺ transients caused by the prolonged APs as well as the reduced SERCA1 nearly fully compensate for the reduced rate of SR Ca²⁺ through RyR1 for twitch contractions.

Fig. 8.

Twitch force. A Average absolute and specific (E) EDL twitch force for control (n = 7 muscles, 4 mice) and R6/2 (n = 9 muscles, 5 mice). Box and whisker plots for B absolute peak C EDL weight, D specific peak, F time-to-peak, G half-width, and H half-max relaxation time. Box and whisker plots show the 25th and 75th percentile (box), mean (white box), median (line), and 1.5 interquartile range (error bars). A magenta * indicates a significant difference compared to control (p < 0.05)

R6/2 Twitch Force & Ca²⁺ Release Upon Repeated Stimulation

The ability of R6/2 skeletal muscle to generate near-normal specific twitch force despite drastic changes to EC coupling is striking. However, physiologically, nearly all muscle contractions are orchestrated by trains of APs. To begin to determine the extent to which the altered EC coupling in R6/2 muscle maintains approximately normal function, we examined the remaining 9 EDL twitch responses from the 0.3 Hz train, as well as measured APs and Ca²⁺ release in response to a stimulation frequency of 0.3 Hz for ten pulses (Fig. 9). As above, APs and Ca²⁺ were recorded from FDB/IO fibers. Our main objective was to determine if there were differences between control and R6/2 muscle during this mild stimulation. The p-values for differences between control and R6/2 for all of the parameters shown in Fig. 9 are included in Table 2.

Fig. 9.

Low frequency twitch train. The action potential (AP), Ca²⁺ and contractile responses to 10 stimuli at 0.3 Hz in control (shades of black, n = 21 fibers, 7 mice) and R6/2 (shades of magenta, n = 14 fibers, 6 mice) muscle. Responses are overlayed and color coded going from darkest (1st stimuli) to lightest (10th stimuli) trace. Traces are shown in panels A, C, & E on the right. Responses normalized to the first stimuli are shown in panels B, D, & F on the right. A The average APs in control (top panel) and R6/2 (bottom panel) FDB muscle. B The average AP peak (top panel), 40% decay time (D40, middle panel), and 80% decay time (D80, bottom panel). C The average Ca²⁺ reflux recorded simultaneously with the APs in control (top panel) and R6/2 (bottom panel) FDB muscle. D The average peak Ca²⁺ release flux (top panel), full-width at half maximum (FWHM, middle panel), and free Ca²⁺ (bottom panel). E The average specific muscle force generation (twitches) in control (top panel) and R6/2 (bottom panel) EDL muscle. F The average peak force (top panel), rise time (middle panel), and half-width (bottom panel). A black horizonal bar indicates a significant difference between control and R6/2 (p < 0.05). A full list of p-values is shown in Table 2

Table 2.

Summary of the p-values for comparisons between control and R6/2 during the twitch train of 10 stimuli at 3 Hz shown in Fig. 9. Parameters of the action potential, SR Ca²⁺ release flux, and isometric force were compared. As shown in Fig. 9, the data were normalized to the 1st stimuli. The p-values for comparisons of the response to the 1st stimuli for the action potentials and SR Ca²⁺ release flux may have slightly shifted from 1.000 because accounting for the random effect of the mouse variable allowed for some adjustments due to the estimating of individual means for mice. Significant differences (p < 0.05) are shown in bold

	p-values for differences between control and R6/2 in twitch train
	Action Potential			Ca2+ Release Flux			Isometric Twitch
Stimuli	peak	D40	D80	Peak	FWHM	Free Ca2+	Peak	Rise-Time	Half-Width
1	0.9997	1.0000	1.0000	0.9991	1.0000	0.9959	1.0000	1.0000	1.0000
2	1.0000	0.7992	1.0000	1.0000	0.6754	0.9987	0.4504	1.0000	0.6308
3	0.9699	0.1097	1.0000	0.9999	0.4767	1.0000	0.0294	1.0000	0.0302
4	1.0000	0.0099	0.9960	1.0000	0.0153	1.0000	0.0117	0.9946	0.0001
5	0.9325	0.0123	0.9366	0.9991	0.0080	1.0000	0.0208	0.9997	< 0.0001
6	1.0000	0.0004	0.5158	0.9619	0.0001	1.0000	0.0004	1.0000	< 0.0001
7	0.9808	0.0035	0.4188	0.9775	0.0072	1.0000	0.0272	0.9994	< 0.0001
8	1.0000	0.0003	0.2023	0.7913	< 0.0001	0.9919	0.0230	1.0000	< 0.0001
9	1.0000	0.0006	0.1068	0.4961	0.0006	0.9852	0.0049	1.0000	< 0.0001
10	1.0000	0.0001	0.0198	0.4561	< 0.0001	0.9211	0.0759	0.9959	< 0.0001

APs during the twitch train did not undergo dramatic changes in control or R6/2 muscle (Fig. 9A). The measured AP parameters (peak, D40, and D80) were normalized to the 1 st stimulation to show relative changes (Fig. 9B). The AP peak was essentially constant during the 10 stimuli in both control and R6/2 muscle and there were no significant differences between control and R6/2 (Fig. 9B, top panel and Table 2). However, the D40 increased during stimulation, particularly in R6/2 muscle and was significantly increased compared to control in stimuli 4–10 (Fig. 9B, middle panel and Table 2). There was also an increase in the D80, but there was less of a difference between the control and R6/2 D80 during the train, with a significant difference occurring only during stimulation 10 (Fig. 9B, bottom panel and Table 2).

The Ca²⁺ release flux triggered by the APs (Fig. 9C) underwent more dramatic changes during the twitch train than the APs. For example, the peak Ca²⁺ release flux decreased to nearly 60% at the 10th stimuli vs the 1 st, with no significant difference between control and R6/2 muscle (Fig. 9D, top panel and Table 2). At the same time, the Ca²⁺ release FWHM increased for both control and R6/2 muscle. However, the FWHM increased significantly more in R6/2 than control fibers during stimuli 4–10 (Fig. 9D, middle panel and Table 2), which is likely explained by the prolonged AP D40 in R6/2 fibers. As a result of the large decreases in Ca²⁺ release flux, the free Ca²⁺ decreased in both control and R6/2 fibers during the ten stimuli to a similar extent (Fig. 9D, bottom panel and Table 2).

The changes in EC coupling correlate with altered force generation in EDL muscle (Fig. 9E). There was a slight ± 10% decrease in peak force during the train, significantly more so in control than R6/2 muscle during stimuli 3–9 (Fig. 9F, top panel and Table 2). The rise time of the twitches decreased slightly during the train in a similar fashion in control and R6/2 muscle (Fig. 9F, middle panel and Table 2). We found that the twitch half-width decreased during the brief train in both R6/2 and control muscle. However, the R6/2 half-width declined at a significantly slower rate than controls during stimuli 3–10 (Fig. 9F, bottom panel and Table 2). The large half-width of the R6/2 twitches appears to correlate with the increased FWHM of the R6/2 Ca²⁺ release flux. As a consequence of the changes, the twitches were actually more stable in R6/2 than control muscle. Broadly, this examination of APs, SR Ca²⁺ release, and muscle force during a train of twitches reveals that the balance of defects producing near normal force in response to a single, isolated stimuli is not sustainable even under very mild repetitive stimulation that is more physiologically relevant.

Discussion

Overall, this study integrated defects in R6/2 action potentials (APs), sarcoplasmic reticulum (SR) Ca²⁺ release and the expression of excitation–contraction (EC) coupling genes to explain twitch force [Scheme 1]. Viewed in isolation, each of these defects would be expected to alter muscle contractility differently. Yet, the actual consequences for muscle force generation depend on the combined effects of these alterations. The disconnect between individual molecular mechanisms and cell/tissue function became apparent when measuring R6/2 muscle force in response to a single stimulus (twitch). Despite dramatic changes in AP properties, SR Ca²⁺ release, and EC coupling gene expression, twitch force that was normalized to muscle weight was unexpectedly indistinguishable from control force. However, this apparent stabilization of force was unsustainable. If R6/2 muscle was pushed even slightly with very mild repeated stimulation (train of twitches), differences in EC coupling and perturbations in force generation between control and R6/2 emerge.

Simultaneous action potentials and Ca²⁺ release under current clamp

To begin to examine the relationship between APs and SR Ca²⁺ release, we first examined simultaneous recordings of APs and the resulting Ca²⁺ transients using intracellular electrodes under current clamp conditions. The R6/2 APs were considerably prolonged, which we have previously shown was associated with reduced expression of K_V1.5 and K_V3.4 channels [44]. SR Ca²⁺ release was significantly reduced in R6/2 fibers compared to control, consistent with previous reports [11, 17, 23]. Viewed alone, the reduced Ca²⁺ release would suggest weakness in R6/2 fibers. However, the duration of AP-induced Ca²⁺ release was longer in R6/2 muscle compared to control, increasing the total Ca²⁺ released and helping to compensate for the reduced Ca²⁺ release. The longer duration of SR Ca²⁺ release can help explain the normal specific EDL force observed in our study and a study of RyR dysfunction in R6/2 and Q175 mice [23]. This data alone did not reveal the mechanism underlying the prolonged SR Ca²⁺ release. The increased duration of R6/2 Ca²⁺ release could have been the result of altered RyR1 inhibition or the prolonged APs. While Braubach and colleagues previously showed the time course of relaxation of AP-induced Ca²⁺ signals were slower in R6/2 muscle, this was not linked to AP duration. We have recently demonstrated that AP properties influence both the onset and duration of SR Ca²⁺ release in skeletal muscle [8, 73].

We first tested the possibility that prolonged APs increase the duration of R6/2 Ca²⁺ release by blocking K_V channels in control fibers to mimic the results in R6/2 fibers. The application of 1 and 10 mM TEA, an established blocker of K_V channels [30, 68], caused a dose-dependent increase in AP repolarization time and Ca²⁺ transient duration in control fibers, supporting the idea that reduced K_V channels current can increase the duration of SR Ca²⁺ release via prolonged APs. This supports previous work suggesting that the ability of TEA to potentiate twitch force is a result of the prolonged APs [21]. An important implication of this experiment is that pharmacological inhibition of K_V channels may provide a means to therapeutically modulate muscle Ca²⁺ and force generation.

Simultaneous action potentials and Ca²⁺ release under voltage clamp

A more mechanistic examination of the role that the AP repolarization plays in SR Ca²⁺ release was achieved by controlling the AP waveform using voltage clamp. By using the average measured AP waveform (control or R6/2) as the voltage clamp command, we were able to record the Ca²⁺ release in response to a control or R6/2 AP in the same fiber. For example, we found that the peak Ca²⁺ release flux was dramatically reduced in R6/2 fibers in response to both control and R6/2 AP waveforms, indicating that Ca²⁺ release is reduced in R6/2 fibers independent of the AP waveform. In contrast, the duration of Ca²⁺ release depended on the AP waveform but not the fiber genotype, indicating that the increased duration of the R6/2 Ca²⁺ release was caused by the prolonged R6/2 AP. Overall, this and previous studies [23, 44] indicate that R6/2 peak Ca²⁺ release is decreased due to perturbations in RyR and that the duration of R6/2 Ca²⁺ release is increased because of prolonged AP repolarization (due to reduced K_V1.5 and K_V3.4 expression). By increasing the duration of Ca²⁺ release, the reduced K_V1.5 and K_V3.4 currents help maintain muscle force generation. Indeed, the free Ca²⁺ available for contraction after a single AP in the voltage clamp experiments were not significantly different in control fibers with a control AP compared to R6/2 fibers with a R6/2 AP. The benefit of reduced K_V1.5 and K_V3.4 currents is likely not limited to Ca²⁺ handling. The reduction in R6/2 muscle ClC-1 currents [45, 74] would be expected to result in myotonia. However, only very subtle signs of delayed relaxation are present in R6/2 muscle. The reduced K_V1.5 and K_V3.4 currents combined with the fiber atrophy likely prevent the build-up of K⁺ in the t-tubules that is thought to contribute to the onset of myotonia [1, 49]. The mechanism we examined here is likely not unique to skeletal muscle. Recently, the increase in Ca²⁺ release duration compensating for reduced peak Ca²⁺ release flux was also shown R6/2 pyramidal neurons [52].

It is worth noting that the method we employed was designed to provide a mechanistic assessment of Ca²⁺ release from the SR in response to action potentials. However, estimates the reuptake of Ca²⁺ by the SERCA pump from this data may be hindered by the high concentration of EGTA in our fibers. To better assess SERCA pump activity, future studies will need to examine Ca²⁺ dynamics without adding EGTA to the inside of the fibers, likely by relying only on optical measures of Ca²⁺ and voltage.

Ca²⁺-binding protein expression levels

Another factor that determines the amount of Ca²⁺ available for contraction is the expression level of Ca²⁺-binding proteins. We used Western blot to measure the protein levels of SERCA1&2, parvalbumin, and fast troponin C in control and R6/2 muscle. The significantly decreased SERCA1 and parvalbumin in FDB/IO and EDL muscle provides an additional mechanism (along with the prolonged SR Ca²⁺ release) to increase free myoplasmic [Ca²⁺] during activity. This is because SERCA1 is the primary mechanism for re-sequestering Ca²⁺ in the SR of fast-twitch skeletal muscle after activity [10, 75]. Parvalbumin acts as a sarcoplasmic Ca²⁺ buffer allowing for faster twitch relaxation times and blunting the rise in free sarcoplasmic Ca²⁺ during the twitch [7, 27]. The decrease of parvalbumin expression we detected would allow myoplasmic Ca²⁺ levels to rise to a higher level in R6/2 FDB/IO muscle and perhaps prolong the Ca²⁺ transient. Together, the decreases in SERCA1 and parvalbumin in R6/2 muscle would be expected to allow increased binding of Ca²⁺ to troponin C and subsequent cross-bridge formation to be elevated, resulting in enhanced force development. SERCA2 is typically expressed in cardiac and slow-twitch skeletal muscle [34, 77]. The lack of detection of SERCA2 expression would argue against a fast- to slow-fiber type shift in R6/2 muscle and the lack of a significant change in troponin C expression in R6/2 FDB/IO and EDL would suggest that Ca²⁺-activation of the myofilaments may be unaffected in these in R6/2 muscles.

Model of Ca²⁺ binding

To better predict the effects of the altered expression level of Ca²⁺-binding proteins and the reduced SR Ca²⁺ release flux, we ran an equilibrium-based model of Ca²⁺ binding in skeletal muscle. The model included our empirical measures of myoplasmic Ca²⁺ in control and R6/2 muscle as well as the significantly changed Ca²⁺-binding proteins in R6/2 FDB/IO and EDL muscle. We first estimated the effect of only the reduced SR Ca²⁺ release flux in R6/2 muscle, which resulted in 30% less binding of Ca²⁺ to troponin C compared to control. Including the significant changes in SERCA1 and parvalbumin in R6/2 FDB/IO and EDL muscle resulted a dramatic increase in the expected binding of Ca²⁺ to troponin, such that the more troponin would be bound by Ca²⁺ in R6/2 muscle than in control muscle. This alone would suggest that specific twitch force in R6/2 could be greater than in control. The near normal R6/2 EDL specific twitch force suggests that additional factors such as SERCA1 and parvalbumin activity as well as possible changes in crossbridge cycling may influence contractility in R6/2 muscle. Moreover, the substantial decrease in SERCA1 and parvalbumin without a corresponding change in R6/2 twitch relaxation rate suggests that myofilament Ca²⁺ sensitivity may be increased in R6/2 muscle. Increased Ca²⁺ sensitivity would allow a reduced peak [Ca²⁺] to generate near-normal peak force while remaining low enough to be cleared at a normal rate despite reduced SERCA1 and parvalbumin expression. Extending our empirical and modeling studies to these factors will allow us to better predict muscle force generation in future work. It will also be important to extend our model to trains of stimulation. For example, decreases in SERCA1 and parvalbumin expression may have greater effects on the more sustained increases in myoplasmic Ca²⁺ that occur during trains of stimulation. Future experiments with trains of stimulation in the absence of EGTA will provide better data to model myoplasmic [Ca²⁺] and the reuptake of Ca²⁺ into the SR by SERCA1. The data here on the responses during a single stimulus provides the foundation to build more advanced models of muscle [Ca²⁺] and contractility.

Transcriptomics

To obtain a broader examination of the changes in R6/2 EC coupling genes, we used transcriptomics. A previous transcriptomics report revealed that one-half of the genes were dysregulated in R6/2 quadriceps and tibial anterior muscles at 12 weeks of age compared to their wild-type counterparts, and gene ontology analysis revealed that affected transcripts included those related to muscle function and energy metabolism [9]. We revisited this data set and our differential gene expression analysis revealed an overall decrease in the expression of genes related to EC coupling in R6/2 muscle. The most dramatic decreases in expression were found for SERCA1 and parvalbumin, supporting our Western blot results. Also directly related to this study, the transcriptomics revealed decreased expression of R6/2 calsequestrin-1, troponin C fast, and Ca_V1.1 (–log₂ fold change > 1.0) and RyR1 (–log₂ fold change < 1.0) in R6/2 quadriceps. There were also significant decreases in Ca²⁺-binding proteins that are newer targets of skeletal muscle investigations, such as Ca²⁺-binding protein 7 and sarcalumenin [19, 24]. The only gene with a significant increase in expression in R6/2 muscle was calsequestrin-2, which is the cardiac form of calsequestrin [65]. We also found gene changes consistent with our previous work showing disrupted transverse tubules and the potential uncoupling of Ca_V1.1 and RyR1 in R6/2 skeletal muscle [60]. The decreased expression of junctophilin 1 & 2 as well as triadin in the transcriptomics analysis are consistent with the disrupted transverse tubules and Ca_V1.1/RyR1 complex in R6/2 skeletal muscle. The wealth of data available from the transcriptomic analysis will help support and guide future physiological studies of Ca²⁺ regulation in HD skeletal muscle.

Consideration of muscle fiber type

Our data regarding R6/2 muscle fiber type composition initially appear contradictory. Increased calsequestrin-2 expression suggests a slow-to-fast fiber type transition, yet the absence of SERCA2 upregulation argues against this interpretation. Previous studies have suggested a shift toward a more oxidative fiber type in R6/2 muscle [43, 59, 69]. Supporting this, we observed reduced myosin heavy chain 2b (MyHC-2b) expression accompanied by elevated MyHC-1 and MyHC-2 × mRNA levels in the tibialis anterior muscle [45]. However, examination of the soleus muscle revealed increased MyHC-2b and decreased MyHC-1 expression, consistent with a slow-to-fast transition [16]. Furthermore, we previously found that R6/2 muscle maturation is likely disrupted, evidenced by increased embryonic and neonatal MyHC expression in late-stage R6/2 muscle [45]. The later find may clarify the matter somewhat since less mature R6/2 muscle may acquire an intermediate fiber phenotype, with fast muscles like the TA becoming slower and slow muscles like the soleus becoming faster. Importantly, these fiber type transitions were partial rather than complete, with transitioning fibers retaining substantial expression of the expected control MyHC isoforms. Taken together, these findings suggest that R6/2 muscle cannot be accurately classified according to canonical fiber type categories. The elevated calsequestrin-2 expression coupled with unchanged SERCA2 levels reflects a unique protein isoform profile distinct from typical muscle fiber types, likely resulting in contractile properties that do not match any standard fiber type classification.

Multi-molecular mechanisms

Physiologically, the widespread decrease in EC coupling genes shown by the transcriptomics data may offset the increase in R6/2 muscle membrane excitability that we have previously identified [44, 45, 74], pointing to pathological mechanisms that could on balance produce a state of apparent normal function. Indeed, despite the numerous changes to gene expression, AP properties, and SR Ca²⁺ release, our measures of specific twitch force, which account for muscle mass and the characteristic atrophy of HD muscle, were normal for R6/2 EDL muscle. Although it is arguably remarkable that R6/2 muscle can maintain normal twitch force in the face of dramatic changes to EC coupling, normal muscle function in the animal requires sustained force during trains of activity. To begin to assess the ability of R6/2 to maintain normal specific muscle force, we measured APs, Ca²⁺ and twitch force during a very mild stimulation protocol of 10 isolated APs at 0.3 Hz. The altered AP and Ca²⁺ kinetics in R6/2 muscle that were evident with a single stimulation increased during the repeated twitch stimulation and resulted significantly altered twitch force kinetics. The delicate balance of defects required to produce normal specific single twitch force in R6/2 muscle is thus not sustainable during even very mild repeated stimulation. With this work as a foundation, we can begin to mechanistically understand the disruptions to R6/2 muscle function during tetanic stimulation. It is becoming increasingly clear that peripheral defects are present in HD patients and need to be examined. Moreover, R6/2 skeletal muscle and neurons share many analogous mechanisms of K⁺ and Ca²⁺ channel dysfunction. Examinations of the more accessible skeletal muscle is expected to help elucidate muscle as well as neuronal mechanisms of disease in HD.

More broadly, this study demonstrates a multi-molecular approach that we needed to employ in order to understand R6/2 muscle contractility. Studied alone, the changes in EC coupling would strongly suggest that R6/2 specific force should be weak. Yet, the ability of the prolonged APs to increase the duration of SR Ca²⁺ release combined with the decrease in Ca²⁺-binding proteins work to counter the decreased rate of SR Ca²⁺ release to produce the unexpectedly normal specific twitch force in R6/2 skeletal muscle. Taking the same integrative approach provides an explanation for the lack of myotonia in model HD muscle despite the hyperexcitability caused by the substantial decrease in chloride currents. R6/2 muscle is hyperexcitable but likely not subject to myotonia because there is less build-up of K⁺ in R6/2 transverse tubules due to the reduced K⁺ currents. A treatment that normalizes the K⁺ currents may inadvertently result in myotonia (due to reduced chloride currents) and/or weakness (due to reduced SR Ca²⁺ release). Targeting any of the individual molecular mechanisms without examination of concurrent mechanisms my result in unexpected and unpredictable phenotypes. The work suggests that detailed examinations of function in native cells with synchronized pathways may improve the translation of molecular mechanism to whole tissue and animal physiology.

Author contributions

D.R.M.: conceptualization, methodology, investigation, writing—original draft, writing—review & editing, and visualization. S.R.A.B.: investigation, writing—original draft and visualization. P.H.M.K.: investigation, writing—original draft and visualization. J.L.R.: software, validation, formal analysis, investigation, writing—original draft, and visualization. J.K.: investigation. H.B.: writing—review & editing. R.J.T.: conceptualization and methodology. G.M.W.: conceptualization, methodology, software, data curation, writing—review & editing, supervision, project administration, and funding acquisition. A.S.: conceptualization, software, writing—review & editing, and funding acquisition. V.B.: methodology, software, formal analysis, and writing—review & editing. H.R.: methodology, formal analysis, writing—review & editing, and supervision. E.O.H.: conceptualization, methodology, writing—review & editing, visualization, supervision, project administration, and funding acquisition. A.A.V.: conceptualization, methodology, software, validation, formal analysis, data curation, writing—original draft, writing—review & editing, supervision, project administration, and funding acquisition. All authors approve the final version of the manuscript and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Last, all persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This project was supported by NIH/NINDS R15 NS099850 (A.A.V.), NIH/NIAMS R01 AR075726 (E.H.O), NIH/NIAMS R01 AR077574 and DoD Idea Development Award W81XWH2110679 (H.R.), NIH/NCI R21 CA247647 (G.M.W.), and NIH/NIDCD R01DC019268 (A.S.).

Data availability

The data that support the findings of this study are available in the Harvard Dataverse at https://doi.org/10.7910/DVN/4LCITP.

Declarations

Competing interests

The authors declare no competing interests.