Calcium homeostasis in parvalbumin DRG neurons is altered after sciatic nerve crush and sciatic nerve transection injuries

Abstract

Reflex abnormalities mediated by proprioceptive sensory neurons after peripheral nerve injury (PNI) can limit functional improvement, leaving patients with disability that affects their quality of life. We examined postinjury calcium transients in a subpopulation of dorsal root ganglion (DRG) neurons consisting primarily of proprioceptors to determine whether alterations in calcium homeostasis are present in proprioceptors, as has been documented in other DRG neurons after PNI. Using transgenic mice, we restricted expression of the calcium indicator GCaMP6s to DRG neurons containing parvalbumin (PV). Mice of both sexes were randomly assigned to sham, sciatic nerve crush, or sciatic nerve transection and resuture conditions. Calcium transients were recorded from ex vivo preparations of animals at one of three postsurgery time points: 1–3 days, 7–11 days, and after 60 days of recovery. Results demonstrated that the post-PNI calcium transients of PV DRG neurons are significantly different than sham. Abnormalities were not present during the acute response to injury (1–3 days), but transients were significantly different than sham at the recovery stage where axon regeneration is thought to be underway (7–11 days). During late-stage recovery (60 days postinjury), disturbances in the decay time course of calcium transients in transection animals persisted, whereas parameters of transients from crush animals returned to normal. These findings identify a deficit in calcium homeostasis in proprioceptive neurons, which may contribute to the failure to fully recover proprioceptive reflexes after PNI. Significant differences in the calcium transients of crush versus transection animals after reinnervation illustrate calcium homeostasis alterations are distinctive to injury type.

NEW & NOTEWORTHY This study examines calcium homeostasis after peripheral nerve injury in dorsal root ganglion (DRG) neurons expressing parvalbumin, a group of large-diameter afferents primarily consisting of proprioceptors, using two-photon calcium imaging in the intact DRG. Our findings identify aberrant calcium homeostasis as an additional source of sensory neuron dysfunction following peripheral nerve injury, uncover differences between two injury models, and track how these changes develop and resolve over the course of recovery.

INTRODUCTION

Recovery of motor coordination after peripheral nerve injury (PNI) is limited by spinal reflexes, which rarely return to baseline (1). Without complete functional recovery, patients who experience PNI suffer disability, social and economic hardships, and decreased quality of life (2–4). The goal of this study is to better elucidate mechanisms underlying reflex abnormalities by investigating proprioceptive neuron function at multiple time points after different types of injury.

Peripheral proprioceptive neurons send feedback to the central nervous system that is critical for smooth movements and to maintain posture. Three major proprioceptive postinjury alterations have been identified as contributors to reflex abnormalities after PNI: 1) the morphologies of muscle spindles are altered after PNI (5, 6); 2) some injured neurons will regenerate to reinnervate the incorrect peripheral targets (7); and 3) there are permanent structural changes to spinal cord circuitry, including a reduction and reorganization of proprioceptive synapses onto motor neurons (8), as well as a decrease in presynaptic inhibitory synapses onto proprioceptive synaptic terminals in the spinal cord (9). The present study examines a fourth component of the peripheral proprioceptive system after PNI: calcium homeostasis in proprioceptive neurons.

In injury models of neuropathic pain, researchers have uncovered several injury-induced changes to calcium homeostasis in dorsal root ganglion (DRG) neurons, including decreased resting calcium (10) and diminished sarco/endoplasmic reticulum calcium ATPase (SERCA) activity (11). However, changes to calcium regulatory mechanisms vary depending on the type of DRG neuron and type of injury model (10, 12, 13). An important subpopulation of DRG neurons that have not yet been investigated express the calcium-binding protein parvalbumin (PV) (14). PV-expressing neurons make up less than 20% of the cells in the DRG and consist mainly of proprioceptors (afferents that innervate muscle spindles and Golgi tendon organs). A relatively small percentage (from 3% to 20% depending on the spinal level) of PV-expressing neurons are rapidly adapting low-threshold cutaneous mechanoreceptors, including Pacinian and Meissner afferents, which encode stimuli that contribute to the sense of touch (14, 15). This study tested the hypothesis that calcium homeostasis in PV-expressing DRG neurons is altered after PNI by analyzing electrically evoked calcium transients from transgenic mice with restricted expression of a genetically encoded calcium indicator, GCaMP6s.

Calcium imaging was performed following either sciatic nerve crush or transection, two injury models known to produce proprioceptive deficits. A nerve crush produces a milder injury and improved functional recovery (16, 17) compared with nerve transection (1, 18). In a crush injury, axons are severed, but most surrounding basal lamina tubes remain intact, providing a pathway for regenerating axons to follow to their original receptors. In nerve transection, the axons, basal lamina tubes, and nerve are completely severed. Without intact basal lamina tubes, regenerating neurons are more likely to regrow to the wrong receptor or incorrect muscle or to erroneously innervate the muscle freely (19). The impact of nonspecific reinnervation on calcium handling is explored here by comparing calcium homeostasis in these two injury models, known to vary in degree of nonspecific reinnervation.

The combined calcium imaging and transgenic mouse approach permits functional assessment of PV-expressing neurons at any postinjury time point, even before afferents have reconnected with target muscles or receptors. Specifically, we investigated an early time point (1–3 days postinjury) to characterize the acute reaction to injury, then evaluated calcium homeostasis during a time window where axon regeneration has been reported to be underway (7–11 days postinjury) (20). Lastly, calcium handling was analyzed late in the recovery process (>60 days postinjury), when extensive reinnervation of muscle targets has been reported (5, 6). Assessment of calcium homeostasis before nonspecific reinnervation occurs will determine whether mechanisms in addition to nonspecific reinnervation also contribute to the differing functional outcomes observed after crush and transection injuries. In addition, earlier time points allow us to pinpoint when calcium handling dysfunction begins, shedding light on potential therapeutic windows during which interventions may be most effective.

MATERIALS AND METHODS

Animals

All animal experimental procedures were conducted under the approval of the Wright State University Institutional Animal Care and Use Committee. Knock-in mice expressing Cre recombinase from the parvalbumin (PV) locus (JAX Stock 008069) (21) were crossed with mice that conditionally expressed the genetically encoded calcium indicator GCaMP6s (PV-cre/+;GCaMP/+; JAX Stock 024106) from the ROSA26 locus (22). PV-cre mice have previously been shown to drive cre-dependent reporter expression in >90% of PV-expressing neurons within the DRG (21). Although subsets of low-threshold mechanoreceptors also express PV, more than 85% of PV-expressing neurons in the lumbar DRGs used in this study (L5 and L6) have been shown to be proprioceptors by combinatorial expression with other proprioceptive markers (14, 15). Transgenic mice expressing the calcium indicator GCaMP6s in PV-expressing (PV+) neurons were used to enable optical recording of proprioceptive sensory neuron activity within the DRG. Although the vast majority of PV+ neurons in the DRG are proprioceptors, low-intensity electrical currents (<10 µA) (23) were used to minimize the likelihood of activating and imaging the subset of cutaneous low-threshold mechanoreceptors that also express PV but make up less than 15% of all PV+ neurons in the L5 DRG. In this study, 35 female and 30 male adult (>2 mo) PV-cre/+; GCaMP6s/+ mice were used.

Survival Surgeries

All surgeries were performed by the same veterinary technician to minimize procedural variability. Mice were anesthetized via inhaled 2%–5% isoflurane. Access to the sciatic nerve was obtained via a longitudinal incision along the dorsal thigh of the right hind limb. For sham procedures, the nerve was exposed via separation of the hamstrings by blunt dissection. To induce a crush injury, the exposed sciatic nerve was firmly compressed for 5 s with forceps. To induce a transection injury, the sciatic nerve was completely transected with microdissection scissors and then the two cut ends were rejoined with epineural sutures. In all injury and sham surgeries, the skin was closed with a single staple, and mice received postoperative analgesics to reduce pain or discomfort.

Ex Vivo Sciatic Nerve and DRG Preparation

At one of three postsurgical time points, animals were anesthetized via intraperitoneal injection of Euthasol (≥270 mg/kg ip; Virbac). Animals were perfused with ice-cold carbogenated (95% O₂; 5% CO₂) artificial cerebrospinal fluid (ACSF) containing (in mM): 127 NaCl, 1.9 KCl, 1.2 KH₂PO₄, 1 MgSO₄7H₂O, 26 NaHCO₃, 16.9 D(+)-glucose monohydrate, and 2 CaCl₂. Animals were then decapitated and the spinal column and right hind limb were dissected in a chamber filled with chilled, recirculating ACSF. In our hands, using chilled (16°C–18°C), but not ice-cold, ACSF during the subsequent dissection optimized preparation viability (24). A dorsal laminectomy was performed, and the dura was removed to expose DRGs to circulating ACSF. The right sciatic nerve was transected immediately proximal to the site of injury. The final preparation consisted of the proximal right sciatic nerve connected to the L5 DRG and L5 dorsal root. To optimize imaging quality, dura was removed from the L5 DRG using a collagen-breaking solution [12.5% collagenase type I (10 mg/mL; Sigma C0130), 12.5% thermolysin (1,000 units/mL; Sigma P1512), and 75% H₂O)] applied via a small-tipped glass pipette. Glass pipette suction electrodes (A&M Systems, WA) were fire polished and sized to fit the sciatic nerve and L5 dorsal root.

Calcium Imaging and Electrophysiology

A dissecting chamber containing the sciatic nerve, L5 DRG, and L5 dorsal root in recirculating ACSF were placed under an objective for two-photon imaging (Olympus FV1000-MPE; ×25 1.05 NA objective, 920 nm excitation wavelength; see Fig. 1A). All recordings were made at room temperature (∼22°C). The sciatic nerve was stimulated with low-current pulse trains for 1 s followed by 6-s rest intervals to identify fluorescing cells via epifluorescence [0.1 ms pulses at 50 Hz, stimulation current <10 µA; A360 stimulus isolation unit; World Precision Instruments (WPI)]. Stimulation at higher intensities always revealed additional responding cells, but these were not selected for imaging. Thus, only the PV-expressing neurons with the lowest threshold were selected for analysis. The vast majority of these cells (based on both abundance in the DRG and generally lower thresholds) are likely to be proprioceptors (15). However, some low-threshold cutaneous mechanoreceptors with unusually low thresholds may have been selected for imaging. It is equally likely that some proprioceptors with slightly higher thresholds than the cutoff were excluded from this study. Responding cells were systematically sampled using the following procedures. Beginning at the most superficial focal plane in the DRG, all responsive cells in a given focal plane were imaged, one at a time, before moving to a deeper focal plane to image the next group of responsive cells (see Fig. 1B). Line scans were used to maximize temporal sampling frequency. Line scan dimensions were held constant (9.3 µm, 76 pixels in our system) and were drawn across the brightest part of the cytosol, excluding the nucleus. The line was scanned 20,000 times (total scan time = 28.16 s; sampling frequency >700 Hz) to capture the entire time course of a calcium transient evoked by sciatic nerve stimulation. Detector sensitivity [photomultiplier tube (PMT) voltage] was held constant between and within experiments. Seventy-two cells were imaged using a lower laser intensity (2.5%–6.3%) than the standard laser intensity (6.5%) used for the other 724 cells to avoid oversaturation of fluorescent signal. For optical recordings, prestimulus baseline fluorescence was measured for 5 s to calculate F₀ before a 0.5 s pulse train (0.1 ms pulses at 50 Hz) was delivered to the sciatic nerve at supramaximal intensity (0.6 mA) to produce 25 action potentials (Fig. 1C).

Figure 1.

Summary of the approach. A: cartoon of ex vivo preparation of a sciatic nerve, dorsal root ganglia, and dorsal root under an objective for 2-photon imaging. The sciatic nerve was stimulated electrically (0.1 ms pulses at 50 Hz for 0.5 s at supramaximal intensity), whereas optical recordings were taken from fluorescing cell somas (green). Cells were imaged individually, one cell at a time. B: an image of a fluorescing soma 2 days after sham surgery. High temporal resolution of fluorescence changes was obtained using line scans (∼700 Hz scanning rate) along region of interest shown with the yellow line. C: heat map shows change in fluorescence (in arbitrary units, a.u.) along the scanned line (vertical axis) across time (horizontal axis). Baseline fluorescence measurements are collected for 5 s before stimulus. D: example calcium transient from cell shown in B illustrates parameters measured in this study: DT1, decay time 1; DT2, decay time 2; Peak, peak amplitude; RF, resting fluorescence; RT, rise time. Time window of electrical stimulation of the sciatic nerve (0.5 s) is indicated by the gray box. E: a representative train of compound action potentials recorded from the L5 dorsal root show that action potentials are observed with every electrical pulse (0.1 ms pulses at 50 Hz for 0.5 s). F: first, second, and last compound action potentials from train in E illustrate the stability of evoked signal throughout stimulus train. Stimulus artifacts have been removed from traces in E and F for clarity.

A recording electrode on the L5 dorsal root monitored stimulus efficacy and preparation viability throughout the experiment through extracellular recordings of compound action potentials evoked by each stimulus train (EX4-400 Quad differential amplifier, low-cut filter: 2 Hz, high-cut filter: 10 kHz, gain: 1,000×; Dagan Corp.; digitized using Clampex, version 10.7). A custom Clampex script synchronized the initiation of the optical recording and controlled the pattern of nerve stimulation. Optical recordings for each cell were repeated three times and averaged and analyzed offline.

Experimental Design and Statistical Analysis

Three time points after injury were chosen based on known pathophysiological events (see results for details). Animals were randomly assigned to sham, crush, or transection surgery conditions. Following surgeries, animals underwent experiments at one of the three time points. For data collection, sham animals were also subdivided into experiments at each of the three time points to control for potential time-dependent effects. However, for data analysis, sham animals from all three time points were grouped together as “sham.” Grouping sham animals together from all three time points for analysis purposes conserved the number of animals used and also limited the number of comparison groups in statistical analysis. In addition, a minimum of four males and four females in each group controlled for sex. Experimenters were not blind to the animal’s surgery assignment (sham, crush, or transection) during data collection and analysis. Table 1 shows the breakdown of animals used in each group and time point.

Table 1.

Animals used for calcium imaging

Postinjury Time Point	Surgery Group
	Sham		Crush		Cut
	F	M	F	M	F	M
A	2	2	4	4	4	4
Rg	2	2	6	4	5	4
Rn	2	2	5	4	5	4
Sex totals	6	6	15	12	14	12
Group totals	12		27		26

Distribution of animals used in calcium imaging experiments. Animals were randomly assigned to sham, crush, or transection surgery groups. Following surgery, calcium imaging experiments were completed on an animal during acute injury (A), regeneration (Rg), or reinnervation (Rn) stages. A combination of males (M) and females (F) controlled for sex.

Calcium imaging data were acquired using Olympus Fluoview software (version 4.0 b). Raw imaging data were processed and analyzed with custom scripts in MATLAB (version R2015a) to extract five parameters for each calcium transient: resting fluorescence, peak amplitude, rise time, decay time 1 (DT1) and decay time 2 (DT2; see Fig. 1D). The rising phase of the transient was fit using a polynomial function, whereas a second-order exponential decay function best fit the decay. DT1 corresponds to the time constant of the first element in the second-order decay function and the time required for the transient to decay from DT1 to 10% of resting fluorescence is given as DT2.

In SAS (version 9.4), a two-way mixed-effects ANOVA was used to investigate the independent variables of Sex and Group. The Group variable had seven factor levels to consider both injury condition and the time point after surgery: sham, crush acute, transection acute, crush regeneration, transection regeneration, crush healed, and transection healed. Multiple neurons were imaged from each animal. Therefore, the model included a random effect of animal to account for any correlation between the cells sampled and the animal, i.e., between-animal variation. This allows all of the data points to be included in the analysis, as opposed to using the averages for each animal. Five models were run to investigate the independent variables for each calcium transient parameter: resting fluorescence, peak amplitude, rise time, DT1, and DT2. α was set at 0.01 after a Bonferroni correction was applied to account for a potentially inflated type I error rate that can result from running multiple models (ɑ = 0.05, 0.05/5 models = 0.01). When the P values from an ANOVA suggested significant differences between Group factor levels, post hoc comparisons were made and P values for each comparison were adjusted via a step-down Bonferroni multiple comparison procedure (25).

RESULTS

Stability of Compound Action Potentials Recorded at the L5 Dorsal Root

To monitor the consistency of electrical stimulation of the sciatic nerve and potential changes in the responsiveness of low-threshold afferents following nerve injury, extracellular recordings were made on the L5 dorsal root (Fig. 1E). The amplitude of the compound action potential across the multiple current pulses in the stimulus train provides a measure of the uniformity of responses of the population of axons in the recording pipette. When normalizing to the amplitude of the first spike in a given 500-ms train of stimuli at 50 Hz, the reduction in spike amplitude of the second or last (25th) spike in the train was less than 10% for any injury status (sham, crush, or transection) or recovery time point measured, indicating that low-threshold axons respond robustly as a group at this stimulus frequency (Fig. 1F; second spike: sham 99.0 ± 1.1%, crush 2-day 96.2 ± 2.7%, crush 7- to 11-day 99.0 ± 1.8%, crush 60-day 95.2 ± 3.4%, transection 2-day 100.2 ± 2.0%, transection 7- to 11-day 99.1 ± 1.8%, transection 60-day 98.1 ± 1.2%; last spike: sham 94.2 ± 1.2%, crush 2-day 94.9 ± 6.1%, crush 7- to 11-day 93.3 ± 3.5%, crush 60-day 94.4 ± 3.5%, transection 2-day 98.4 ± 5.3%, transection 7- to 11-day 96.7 ± 2.1%, transection 60-day 94.6 ± 1.3%; mean normalized peak values in percent ± SD; n = 3 animals for each crush and transection group and n = 2 for sham). Spike shape also remained consistent between injury status groups and recovery time points, as indicated by the width at half-maximal amplitude of the averaged spikes in a train [sham (n = 12 animals) 0.88 ± 0.24 ms, crush 2-day (n = 7) 0.82 ± 0.21 ms, crush 7- to 11-day (n = 9) 1.04 ± 0.36 ms, crush 60-day (n = 7) 1.10 ± 0.29 ms, transection 2-day (n = 8) 0.96 ± 0.27 ms, transection 7- to 11-day (n = 9) 1.03 ± 0.40 ms, transection 60-day (n = 9) 1.12 ± 0.30 ms; means ± SD; F(1,6) = 1.15, P = 0.34, 1-way ANOVA].

Overview of Calcium Transients of Sham and Injury Animals

Calcium transients from 796 cells from 65 animals were recorded for this study. On average, 12 ± 4 (mean ± SD) cells were imaged per animal (6–26 range). Prestimulus resting fluorescence, as well as peak amplitude and rise time of the evoked transients, were measured from all cells. Transient decay parameters DT1 and DT2 were also obtained for all but 4 and 13 cells, respectively, as evoked transients in these few cells could not be fit by second-order exponential decay function or had not returned to baseline by the end of the scan (23 s poststimulus). For all five calcium transient parameters investigated, there were no significant two-way interactions between Sex and Group. Looking at the main effects for the variable of Sex, there were no differences between males and females in any of the five calcium transient parameters examined: resting fluorescence, P = 0.47; peak amplitude, P = 0.13; rise time, P = 0.36; DT1, P = 0.03; and DT2, P = 0.15. Looking at differences between animal groups, there were no differences in resting fluorescence [F(1,6) = 1.72, P = 0.13, 2-way mixed-effects ANOVA, see Fig. 2]. Thus, resting fluorescence did not differ between sham, crush, or transection animals at any of the time points examined. For the remaining four calcium transient parameters, significant main effects were found for Group as discussed in detail in the following sections.

Figure 2.

The resting fluorescence of parvalbumin (PV)-expressing cells from injured animals did not significantly differ from sham at any investigated time point (A, acute; Rg, regeneration; Rn, reinnervation). Each circle depicts the average resting fluorescence of all cells imaged from one animal (a.u., arbitrary units). Filled circles, female mice; open circles, male mice. The range and variance of resting fluorescence values appear similar across sexes, surgery group, and postsurgical time points. Two-way mixed effects ANOVA [F(1,6) = 1.72, P = 0.13]. Each group consists of 8–12 animals.

Acute Injury Responses (1–3 Days)

Immediately following injury, retrograde signals cue the neuronal soma that an injury has occurred. Within 24 h, the soma hypertrophies, Nissl bodies dissolve, and the nucleus moves to an eccentric position as mRNA metabolism and protein synthesis increase, all of which allow the cell to combat stress and survive (26, 27). The 2-day postinjury time point in this study provides time for these molecular changes to take place. Importantly, severed axons can still transmit signals from the periphery for up to 3 days in mice (28). Thus, at this stage, somas likely still receive activity-induced electrical signals from the periphery.

Calcium imaging experiments were performed on animals 1–3 days after surgery (2.2 ± 0.5, mean ± SD). At this time point, the calcium transients of imaged neurons from crush and transection animals were not significantly different from sham animals in any of the parameters measured: resting fluorescence, peak amplitude, rise time, DT1, or DT2. These parameters also did not significantly differ between crush and transection animals. Although not statistically significant, a trend for larger, longer duration calcium transients compared with sham is evident in both injury models (see Table 2). For example, 2 days after injury, the mean peak amplitude is 104, 131, and 152%ΔF/F, and the average DT1 is 2.7, 3.0, and 3.1 s for sham, crush, and transection animals, respectively.

Table 2.

Calcium transient parameters 1–3 days postinjury

	Sham (12)	Crush (8)	Transection (8)
Resting F, a.u.	415 ± 68	414 ± 97	426 ± 92
Peak, %ΔF/F	104 ± 64	131 ± 65	152 ± 63
Rise time, s	1.22 ± 0.20	1.41 ± 0.20	1.43 ± 0.19
DT1, s	2.66 ± 0.62	2.98 ± 0.63	3.05 ± 0.61
DT2, s	9.09 ± 0.75	9.71 ± 0.76	9.72 ± 0.73

Average calcium transient parameters for each group during the acute injury response phase (1–3 days postinjury). Data are presented as means ± SD with the number of animals given in parentheses. At this time point, no significant differences were detected between groups for any measured parameters using two-way mixed-effects ANOVA with step-down Bonferroni multiple comparison procedures. a.u., Arbitrary units; DT1, decay time 1; DT2, decay time 2.

Responses at 7–11 Days Postinjury

As the acute inflammatory response subsides, cellular efforts refocus from survival to regrowth and healing (26). There is a latent period before axonal sprouting begins, plus an additional lag time in regeneration as the distal axon degenerates and cellular debris is cleared by Schwann cells and macrophages (29). In total, these processes take ∼5–7 days before regrowing axons reach a constant regenerative rate (20, 30). Therefore, by 7 days postinjury, surviving neurons should be regenerating, but these nascent axons will not yet have reinnervated muscle targets.

Calcium imaging experiments were performed on animals 7–11 days postsurgery (9.2 ± 1.0, mean ± SD). At this time point, the calcium transients of neurons from crush and transection animals were significantly different versus sham (see Fig. 3). On average, rise time was 0.37 s longer in crush animals than in sham [P = 0.0011, 95% CI (0.05, 0.62); see Table 3] and 0.49 s longer in transection animals than in sham [P < 0.0001, 95% CI (0.20, 0.79)]. Although regeneration rise time was longer after transection than after crush, there was no significant difference between these two groups [P = 0.24, 95% CI (−0.15, 0.46)]. Also, DT1 was 1.20 s longer in crush animals than in sham [P = 0.0002, 95% CI (0.29, 2.11)] and 1.64 s longer in transection animals than in sham [P < 0.0001, 95% CI (0.71, 2.58)]. However, there was no significant difference in the DT1 of crush and transection animals at this stage [P = 0.61, 95% CI (−0.53, 1.41)]. In addition, DT2 was 1.27 s longer in crush animals [P = 0.0015, 95% CI (0.18, 2.36)] and 1.56 s longer in transection animals [P = 0.0001, 95% CI (0.44, 2.67)] compared with sham. Again, there was no significant difference between injury types with DT2 [P = 0.79, 95% CI (−0.87, 1.44)].

Figure 3.

Calcium transients from peripheral nerve injury (PNI) animals significantly differ from sham 7–11 days postinjury. A: representative traces illustrate the mean amplitudes of calcium transients observed in different groups at this stage postinjury. B: although peak amplitudes from injury animals trended higher than sham, there were no significant differences between groups. S, Sham (n = 12); C, Crush (n = 10); T, Transection (n = 9). C: representative calcium transients from each condition are plotted with amplitudes normalized to 100%ΔF/F to illustrate observed differences in rise time and decay parameters. Time scale is truncated to highlight transient fluorescence increase. D–F: rise time, DT1, and DT2 were significantly longer than sham for both injury groups. There were no significant differences between crush and transection injury groups. For each box plot: lower bar, 1st quartile; box bottom, 2nd quartile; horizontal line, median; x, mean; box top, 3rd quartile; top bar, 4th quartile; circles outside plot indicate data values outside 1.5 times the interquartile range (box). Two-way mixed-effects ANOVA with step-down Bonferroni multiple comparisons. Significant changes are indicated (**P < 0.01: ***P < 0.001; ****P < 0.0001). DT1, decay time 1; DT2, decay time 2.

Table 3.

Calcium transient parameters 7–11 days postinjury

	Sham (12)	Crush (10)	Transection (9)
Resting F, a.u.	415 ± 68	409 ± 79	494 ± 78
Peak, %ΔF/F	104 ± 64	179 ± 64	174 ± 63
Rise time, s	1.22 ± 0.20	1.56** ± 0.19	1.71**** ± 0.19
DT1, s	2.66 ± 0.62	3.86*** ± 0.62	4.30**** ± 0.61
DT2, s	9.09 ± 0.75	10.36** ± 0.74	10.65*** ± 0.73

Average calcium transient parameters for each group during the regeneration phase (7–11 days postinjury). Data are presented as means ± SD with the number of animals given in parentheses. Two-way mixed-effects ANOVA with step-down Bonferroni multiple comparisons. a.u., Arbitrary units; DT1, decay time 1; DT2, decay time 2. Significant differences from sham are indicated as: **P < 0.01, ***P < 0.001, ****P < 0.0001. Sham data from Table 2 is replicated here for ease of comparison.

Although the rise and decay times of transients significantly changed during this phase of recovery, the peak amplitudes of transients did not. The mean peak amplitude of transients from crush and transection animals trended higher than sham: 75%ΔF/F and 69%ΔF/F higher, respectively. However, these values did not reach significance for either injury group [crush: P = 0.0576, 95% CI (−19, 169); transection: P = 0.0958, 95% CI (−27, 165)]. Also, the average peak amplitudes for transients from crush animals did not significantly differ from transection animals [P = 1.0, 95% CI (−94, 104)].

Late-Stage Recovery (60 Days Postinjury)

It takes ∼21 days for axons from a crushed sciatic nerve to regenerate to the plantar muscles in mice (6). Forty days after crush, the stretch receptors in the plantar muscles are more densely innervated by afferents than they were at 21 days (6), demonstrating that reconnections with receptors may take several weeks. Remyelination begins at the injury site ∼21 days after injury and continues distally for some time (20). By 7 wk postinjury in rodents (either crush or transection), behavioral studies demonstrate that further functional improvements do not occur with additional time (18, 31). Thus, 60 days after injury represents late-stage recovery where reinnervation and functional recovery have largely plateaued to maximum levels.

Calcium imaging experiments for this time point were completed a minimum of 60 days and up to 71 days postsurgery (63 ± 3.3, mean ± SD). At this time point, calcium transients in crush-injured animals returned to normal values and there were no significant differences between sham and crush animals. However, calcium transient parameters from transection-injured animals remained elevated compared with sham, even at this late-recovery stage (see Fig. 4). Peak amplitude, rise time, DT1, and DT2 in transients from transection animals were all significantly higher than sham [peak amplitude: P = 0.0029, 95% CI (11, 203); rise time: P = 0.0003, 95% CI (0.09, 0.68); DT1: P = 0.0004, 95% CI (0.24, 2.11); DT2: P = 0.0027, 95% CI (0.11, 2.34); see Table 4]. At 60 days postsurgery, transients from transection animals were also significantly different versus crush. DT1 was on average 1.10 s longer in transection animals compared with crush at 60 days [P = 0.0023, 95% CI (0.10, 2.11)]. Other transient parameters did not significantly differ between crush and transection animals, even though peak amplitude and rise time differences did approach significance [peak amplitude: P = 0.0307, 95% CI (−12, 194); rise time: P = 0.0429, 95% CI (−0.06, 0.57)].

Figure 4.

Calcium transients from transection injured animals are significantly different than sham and crush injured animals during late-stage recovery. A: representative traces depicting the average calcium transient for each group approximately 60 days postsurgery. B: peak amplitude of transection injury animals remained significantly elevated compared with sham animals after 60 days of recovery. C: representative calcium transients plotted with peak amplitudes normalized to 100%ΔF/F for ease in comparing rise time and decay parameters. Time scale is truncated to focus on transient fluorescence increase. D–F: transient parameters from transection animals were significantly different than sham. DT1 was significantly different in transection vs. crush. S, Sham (n = 12); C, Crush (n = 9); T, Transection (n = 9). For each box plot: lower bar, 1st quartile; box bottom, 2nd quartile; horizontal line, median; x, mean; box top, 3rd quartile; top bar, 4th quartile; circles outside plot indicate data points outside 1.5 times the interquartile range (box). Two-way mixed-effects ANOVA with step-down Bonferroni multiple comparisons. Significant changes are indicated (**P < 0.01; ***P < 0.001). DT1, decay time 1; DT2, decay time 2.

Table 4.

Calcium transient parameters 60–70 days postinjury

	Sham (12)	Crush (8)	Transection (8)
Resting F, a.u.	415 ± 68	399 ± 75	461 ± 74
Peak, %ΔF/F	104 ± 64	121 ± 72	211** ± 63
Rise time, s	1.22 ± 0.20	1.35 ± 0.22	1.60*** ± 0.19
DT1, s	2.66 ± 0.62	2.72 ± 0.70	3.83***†† ± 0.61
DT2, s	9.09 ± 0.75	9.55 ± 0.83	10.32** ± 0.73

Average calcium transient parameters for each group during late-stage recovery (>60 days postinjury). Data are presented as means ± SD with the number of animals given in parentheses. Two-way mixed-effects ANOVA with step-down Bonferroni multiple comparisons. a.u., Arbitrary units; DT1, decay time 1; DT2, decay time 2. Significant differences from sham are indicated as **P < 0.01, ***P < 0.001, ****P < 0.0001, and from crush as ††P < 0.01. Sham data from Table 2 are replicated here for ease of comparison.

DISCUSSION

This study compared the calcium transients evoked in the PV-expressing lineage of DRG neurons from animals that had undergone sham, crush, or transection injuries at multiple postsurgical time points (see Fig. 5 for summary of findings). There were no significant differences in the resting fluorescence of these cells from animals in any condition or time point. Factors contributing to the fluorescence of a cell at rest include autofluorescence from the GCaMP6s indicator in its unbound form and fluorescent GCaMP6s bound to cytosolic calcium at rest. Accordingly, the level of fluorescence measured at rest provides an indirect measure of GCaMP6s expression levels and resting calcium concentrations. The fact that resting fluorescence did not significantly change at any postinjury time point, in either injury model, suggests that GCaMP6s expression and resting calcium concentrations did not significantly change after injury. Alternatively, cell-to-cell variability in GCaMP6 expression may have limited the sensitivity of the indicator to relatively small changes in intracellular calcium concentrations (32). Interestingly, a study using spinal nerve ligation, a different injury model than used here, found that large DRG neurons from animals exhibiting an increased sensitivity to pain (hyperalgesia), demonstrated a significant 50% reduction in resting calcium concentrations (10). However, in other animals in the same cohort that did not display hyperalgesia, no significant changes in resting calcium were detected. As proprioceptive neurons are not known to contribute to hyperalgesia, it is possible that reported changes in resting calcium were caused by nonproprioceptive large DRG neurons and that PV-expressing neurons respond differently to injury. Another possibility is that the lack of changes in resting fluorescence in the present study highlights another distinction between spinal nerve ligation and the injury models used here.

Figure 5.

Summary of postinjury changes to calcium transients over time. Representative calcium transients for each group at three key postsurgical time points are shown at top. Transients were evoked by electrical stimulation (gray box on gray bar). Transients from crush animals trended slightly larger than sham, whereas transients from transection animals trended larger than crush and sham groups. Summary of statistical differences between groups using two-way mixed-effects ANOVA with step-down Bonferroni multiple comparisons are shown at bottom. Dashes reflect no significant differences from sham. Arrows depict statistically significant differences from sham at time points corresponding with above traces. Additional statistical differences between transection and crush conditions are depicted with **P < 0.01. C, Crush; DT1, decay time 1; DT2, decay time 2; RT, rise time; T, Transection.

Calcium transients in individual neurons were evoked using a standardized stimulation at 50 Hz. Previous reports showed a linear relationship between the number of action potentials and the amplitude of the evoked calcium transient in PV-expressing cells (24), but the kinetics of the GCaMP6s reporter are too slow to resolve calcium changes elicited by each spike in the train. The compound action potential recorded at the dorsal root measures the population response to the stimulus train, but calcium imaging precludes independent verification of the action potential in individual cells. Importantly, the amplitude of the compound action potential remained stable throughout the stimulus train suggesting consistent responses from the population of responding neurons. If a significant portion of the cells are unable to fire one-to-one with the 50 Hz stimulus across the 0.5 s train, the amplitude of the compound action potential will decrease, as it represents the sum of all individual action potentials from axons in the dorsal root. The difference between the amplitude of the first, second, and last (25th) compound action potential was less than 10% for sham or injured neurons (crushed or transected) regardless of recovery time, indicating that responses were robust throughout the train.

The hyperexcitability of injured neurons (33) could result in additional action potentials being elicited during the stimulus train, which could influence the calcium transient amplitude. If a significant fraction of PV-expressing neurons fired additional action potentials in response to the stimulus, this would broaden the compound action potential recorded at the dorsal root. However, we found that the width of the action potential at half-maximum amplitude did not change in recordings from injured animals. Two other factors make it unlikely that hyperexcitability is responsible for the changes observed. First, short current pulses (0.1 ms) were used to minimize the time axons would be above threshold, even if that threshold is reduced in hyperexcitable neurons, to fire action potentials. Second, recordings at room temperature strongly bias PV-expressing neurons to fire single action potentials in response to depolarizing current pulses, even when these current pulses are long in duration (800 ms) (34). Moreover, calcium transient decay parameters (DT1 and DT2) are not influenced by action potential frequency, even at frequencies much higher than used here (24). The fact that decay parameters change after injury suggests that factors beyond firing frequency are responsible for changes in calcium handling.

Mice were first examined ∼2 days after injury to determine whether calcium homeostasis is disrupted in proprioceptive neurons during their acute reaction to injury and inflammation. During this acute injury phase, there were no significant differences in the calcium transients from animals after sciatic crush or transection injuries compared with sham. By 3 days after injury, gene regulation and protein synthesis are extensively modified to promote survival of the injured neuron and initiate degeneration of the distal axon (35). In addition to these intracellular changes, the breakdown of the blood-nerve barrier enhances the infiltration of extracellular signals (26). Considering the molecular transformation injured neurons are experiencing at this time, it is remarkable that injured PV-expressing neurons can regulate calcium similarly to neurons from sham animals.

Axon regeneration toward the periphery has been reported beginning 7 days after injury in mice (20). At this stage, rise and decay times of calcium transients from both injury models were significantly longer than sham. Although the average rise and decay trended toward longer times in transection animals compared with crush, there were no statistically significant differences between crush and transection animals during the regenerative phase. Therefore, it is likely that the changes to calcium homeostasis during this time period reflect similar physiological processes in both injury models, i.e., regrowth. During the regrowth phase, neurons are electrically stimulated before they have established connections with peripheral targets, and optical recordings take place in the DRG. Thus, proprioceptive neurons are being functionally evaluated in complete isolation from peripheral and central connections and still demonstrate alterations. Therefore, this finding that calcium homeostasis is abnormal 10 days after injury provides evidence that intrinsic changes to proprioceptive neurons occur after nerve injury.

Notably, calcium homeostasis deviated from sham animals at some point between acute injury and regenerative phases. Hence, in addition to finding that proprioceptive neurons are intrinsically altered after nerve injury, this study identified a window when abnormalities first appear: between 3 and 7 days after sciatic nerve crush or transection. This could have applicability to therapeutic approaches, such as when calcium-modifying drugs, released at the site of injury via a subdermal osmotic pump, were shown to improve functional recovery in rats following PNI (36). Fine-tuning the treatment window to consider the timeline of calcium homeostatic alterations may yield even better functional outcomes.

We found that 60 days after injury, calcium transients from crush animals returned to normal, whereas calcium transients from transection animals were significantly different from sham. Furthermore, one of the four measured parameters, DT1, was significantly different in transection animals compared with crush animals 60 days after injury. It is noteworthy that this parameter was not significantly different at the two earlier time points, providing further evidence that crush and transection injuries follow distinct regeneration trajectories.

Reinnervation of target muscles occurs before this late-stage recovery time point. In fact, muscle spindles begin to be reinnervated 21 days after a sciatic nerve crush injury in mice (6) and by 26 days in cats that underwent a crush injury at the same distance from the target muscles as in mice (5). By 33 days after the injury, 78% of muscle spindles have primary (Ia afferent) endings (5), and although these endings are morphologically less robust than in naïve animals, 90% of recorded afferents display sensitivity to stretch and continue to fire for at least part of the hold phase of a stretch stimulus (16). These studies suggest that full morphological recovery is not required for stretch sensitivity, and that functional sensitivity is observed in parallel with the reinnervation of the muscle spindle.

Less is known regarding the time course of recovery from transection injuries, and as transected nerves have no intact basal lamina tubes to guide regenerating axons, nonspecific reinnervation has been suggested to contribute to the difference in functional outcomes between crush and transection injuries (19). Moreover, some proprioceptive afferents remain unresponsive to muscle stimuli and presumably have not reinnervated muscle sensory organs, even months after the injury (7). In the present study, direct measurement of somatic calcium handling eliminates confounds and differences in peripheral and central connections, nonspecific reinnervation or other mechanisms that differ between injury types. Yet, abnormalities within transected proprioceptive neurons persist even 2 mo after injury. The potential differences in the status of reinnervation of muscle sensory organs between crush- and transection-injured neurons raises the possibility that neurons that fail to reconnect with muscle sensory targets persist in an “injured” state in terms of calcium handling, whereas neurons that successfully reconnect revert to basal calcium responses. Without knowing the reinnervation status of each neuron imaged in this study, this remains a possibility. However, for all parameters with significant differences at 60-day postinjury, values were normally distributed with no indication of bimodal distribution that would be suggestive of subsets of neurons with “normal” versus elevated “injured” responses. This supports the conclusion that transection differs from crush injuries in terms of calcium handling in these cells.

Depolarization-induced calcium transients represent the summation of several simultaneous calcium-regulatory processes: intrusion, extrusion, sequestration, and release. In PV-expressing neurons, the extrusion of calcium via plasma membrane calcium ATPases affects all parameters of the calcium transient, whereas SERCA activity influences only rise time and decay constants (24). Mitochondrial buffering and extrusion of calcium through the sodium-calcium exchanger also significantly affect the calcium transients of DRG neurons (37, 38). Changes in the activity of these calcium regulatory mechanisms would likely contribute to the changes in transient parameters observed here. Recent investigations of gene expression changes among lumbar DRG neuron subtypes at multiple time points after sciatic nerve crush and transection injuries offers insight into potential changes in calcium regulatory systems after injury (39). No significant changes were noted among presumptive proprioceptive neurons in the expression of Atp2a2 or Slc8a1, the genes that encode SERCA and the sodium-calcium exchanger, respectively. However, the gene that encodes PMCA activity in presumptive proprioceptors, Atp2b3, was downregulated 40%–50% at both 3 and 7 days postinjury (39).

Reductions in the endogenous buffering capacity of PV-expressing neurons via calcium-binding protein expression could also contribute to altered calcium handling after injury. Although proprioceptors and low-threshold mechanoreceptors all express PV, additional calcium-binding proteins are expressed in subsets of PV-expressing neurons. Calb1 (encoding calbindin) and Calb2 (encoding calretinin) are both enriched in proprioceptive neurons that supply muscle spindles (40). Although both PV and Calb2 expression remain stable after injury (39, 41), Calb1 expression is reduced by ∼50% at both 3 and 7 days postinjury (39). The fact that we observed trends, but not significant changes in calcium signals, during the acute injury response (1–3 days postinjury), suggests that disruptions in calcium handling develop over time after injury. Further research is needed to confirm the calcium regulatory mechanisms responsible for altering calcium transients after PNI and to determine the direct effect of altered calcium homeostasis on peripheral encoding and integration of proprioceptive signals in the spinal cord.

GRANTS

This work was supported by NIH Grant NS072454. M.C.W. is supported by the Biomedical Sciences PhD program at Wright State University.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

M.C.W. and D.R.L. conceived and designed research; M.C.W. performed experiments; M.C.W. analyzed data; M.C.W. and D.R.L. interpreted results of experiments; M.C.W. prepared figures; M.C.W. drafted manuscript; M.C.W. and D.R.L. edited and revised manuscript; M.C.W. and D.R.L. approved final version of manuscript.