Early Changes in Respiration and Motor Activity After Spinal Cord Injury Predict Pain-Related Outcomes in Mice

Subjects and surgical procedures

Adult male and female C57BL/6J mice (The Jackson Laboratory, #000664) were used. Mice were anesthetized with isoflurane (5%, gas; lowered to 2–3% upon reaching stable anesthesia). Under sterile conditions, the spinal cord was exposed following skin incision and dorsal laminectomy (sham and SCI mice), and in SCI mice, the injury was administered as previously described. ³¹ Briefly, mice received a ∼70 kDyne (moderate to severe) impact contusion SCI (IH-0400 Impactor, Precision Systems and Instrumentation, Fairfax Station, VA, USA) delivered to the lower thoracic (T) spinal cord at level T10. In all mice, the overlying skin was sutured shut and the wound area treated with a topical ointment. Mice were weighed daily following surgery, and their bladders were expressed by researchers twice daily for the duration of experiments. Animals were included in the study only if they exhibited a Basso Mouse Scale for Locomotion score of 0–1 at 1 day postoperation (dpo), consistent with a successful acute contusion injury, and if weight support was gained by 28 dpo for behavioral tests. Three cohorts of SCI mice were used in this study. The first cohort (n = 7) was sacrificed 9 days postsurgery for use in other experiments. As a result, tests occurring at the 4-week postsurgical time point only involved the second (n = 6) and third (n = 4) cohorts. All mice were administered meloxicam (5 mg/kg, subcutaneous [SC]) and buprenorphine (0.05 mg/kg, SC) prior to surgery for acute pain management, then left to recover on a heated pad. The same meloxicam dosage was delivered twice daily for 2 days following surgery. Sterile saline (0.9%, intraperitoneal [IP]) was administered daily for the first 48 h after surgery, with subsequent injections given as needed. Findings presented at 1–2 dpo (meloxicam administered) and 3 dpo (no meloxicam) were consistent, suggesting that analgesic effects did not play a major role.

Breathing and motor activity

Mice were placed in square enclosures with two electric field sensors (EPIC, Plessey Semiconductors, Plymouth, UK; now commercially available at Level 42 AI, Inc, Mountain View, CA) affixed to the side of each enclosure to noninvasively record RR and its variability in mice acutely after injury (Fig. 1A).^32,33 Sensor recordings were captured as voltage traces through a customized interface in LabVIEW (National Instruments, Austin, TX). A lowpass Chebyshev filter at 12 Hz was used to remove electrical noise. Resting RR, RRVAR, and movement were quantified via threshold-based event detection in Clampfit (bandpass filter 1–7 Hz; Molecular Devices, San Jose, CA) in combination with derived spectrograms (Fig. 1B). Before surgery, each mouse underwent a 1.5-h acclimation session in the same enclosures as above, with the final 30-min period recorded used to calculate the baseline RR, RRVAR (the standard deviation of RR), ¹⁸ and time spent moving. This design was chosen following previous testing to ensure the reliability of environmental acclimation. The day after their baseline recordings were taken, mice received the surgery, with subsequent postacclimation 30-min recording sessions beginning at 1 dpo. Postsurgical time points included the first 3 dpo (acute time points), day 8 postoperation (subacute time point), and 4 weeks after operation (chronic time point). We were unable to obtain recordings from n = 4 SCI mice at 2 dpo due to data capture error, resulting in n = 13 SCI mice at this time point (n = 17 for others).

OPEN IN VIEWER

FIG. 1.

Electric field sensor recording apparatus with example captured data. (A) Mice were placed in acrylic chambers allowing for a full range of movement. Each mouse chamber had two affixed sensors that translate activity-based changes in the electric field to changes in voltage. (B) Example raw voltage recording (bottom) and transformed frequency spectrograms generated in Spike2 (Cambridge Electronic Design, Milton, England, top). Note the discrete power bands associated with respiration rate (frequency) when mouse is at rest (inactive) and that active motor epochs have larger voltage responses with broad increases in power across spectral frequencies displayed. Time spent moving, a spontaneous motor behavior, was calculated from spectrograms as total trial time minus total length of resting respiration epochs. Results were verified using simultaneous video capture in a subset of mice.

Mechanical and thermal sensitivity tests

The von Frey hindpaw mechanical sensitivity test was undertaken at baseline (before surgery) and at the 4-week postoperation time point, at which many studies have shown full development of SCI mechanical hypersensitivity associated with neuropathic pain.^19,31,34 For 3 days, mice were habituated to the behavioral testing room, including the red acrylic chambers where testing was performed. On testing days, following a 30-min acclimation period, individual animals were assayed for mechanical sensitivity according to the established up-down method, ³⁵ using calibrated von Frey hairs (NC12775-99, North Coast Medical, Inc., Morgan Hill, CA, USA). Right and left hindpaw withdrawal thresholds were averaged to determine overall mechanical sensitivity.

One day after the von Frey test and using the same acrylic chambers, the Hargreaves test was conducted to quantify thermal nociception. ³⁶ The test was run at 4 weeks postoperation but not at baseline to avoid habituation effects observed in previous cohorts. For this test, radiant heat was used to induce withdrawal responses, with stimulation directed at the hindpaw (Plantar Test Apparatus, IITC Life Science). The right and left paws were each tested 3–5 times, with at least 5 min between replicate measurements, and the closest two values from each set (i.e., per paw) were averaged to obtain withdrawal latencies. These were then averaged over both paws to compute overall thermal sensitivity.

Thermal Preference test

The Thermal Preference test was performed 1 day following the Hargreaves test. Mice were placed in an individual rectangular enclosure whose flooring was a thermal plate with temperature gradient controlled by Peltier heating and cooling elements across the length of the enclosure (constructed by William N. Goolsby). Tests applied a gradient range of 20 − 40°C to the thermal platform, centered around the mouse thermoneutral preference point of ∼30°C. ³⁷ Because the platform temperature followed a linear gradient, preferred temperature could be calculated using the animal’s body position (midline [rostro-caudal and left–right] location). Mice were allowed to explore for 30 min to 1 h, and their chosen resting spot after this time period was taken to indicate their individual thermal preference temperature.

Statistical analysis

All quantitative measurements are reported as mean ± standard error of the mean (SEM). To account for missing data points, mixed-effects analyses (with follow-up multiple comparisons tests) were performed for respiratory measures and activity, a paired t-test to assess pre–post von Frey score changes, and an unpaired t-test to assess thermal gradient preferences and Hargreaves latencies relative to a sham and naïve mouse cohort. To quantify the relationship between sensory tests and physiological variables (respiration and movement), correlation analyses were undertaken at time points corresponding to observed changes in SCI mice, with linear regression performed to assess statistical significance. To avoid statistical overtesting, correlations were performed based on the a priori hypothesis that the earliest changes after SCI would provide the greatest opportunity for revealing predictive relationships relevant to clinical translation and preventative interventions. Comparisons were made between changes in respiration and activity at acute (1 dpo) and subacute (8 dpo) time points and development of pain-related outcomes at 4 weeks post-SCI. In the case of significant results, further analysis was performed at 2 and 3 dpo to provide added clarification on the timeframe of effects. Statistics were performed using GraphPad Prism software, Version 10.2.1 for Windows (GraphPad Software, Inc.; San Diego, CA), with significance set at p < 0.05 and two-tailed tests.

RR, RRVAR, and activity levels

As shown in Figure 2A, RR did not significantly change from the baseline within the first 8 days after operation (mixed-effects model; F(2.015, 30.22) = 0.6317, p > 0.05). In contrast, RRVAR (Fig. 2B) changed after injury (mixed-effects model; F(2.525, 37.87) = 11.96, p < 0.0001), with significantly decreased values on the first 3 days after injury compared with baseline (p < 0.05, Dunnett’s multiple comparisons tests). RRVAR recovered by the eighth day postinjury. Time spent moving (Fig. 2C) also changed in the days after injury (mixed-effects model; F(3.018, 44.52) = 20.18, p < 0.0001), with post hoc tests revealing a significant decrease on each of the first 3 days after injury (p < 0.0001). Like RRVAR, time spent moving recovered by the eighth day postinjury. While there was a borderline significant correlation (p = 0.05) between decreases in RRVAR and percent time active from baseline to 1 dpo, these variables were not correlated within individual mice when averaged across 1–3 dpo (Fig. 2D). Compared with the sham control population, SCI mice were significantly less active across time points (mixed-effects model group effect; F(1, 22) = 26.16, p < 0.0001), with significantly less movement at 2 and 3 dpo (p < 0.05 in both cases).

OPEN IN VIEWER

FIG. 2.

Comparing respiratory rate (RR), RR variability (RRVAR), and movement changes in the early stages after SCI. These measures were recorded from 1 day preinjury (baseline) to 8 days postinjury in spinal cord injured (SCI) mice (n = 17, except 2 dpo [n = 13]) and a cohort of sham controls (n = 7). (A) RR did not change significantly throughout the experiment. (B) In SCI mice, RRVAR significantly decreased on the first through third days postoperation compared with baseline and recovered by 8 dpo. (C) In SCI mice, time spent moving significantly decreased on each of the first 3 dpo and recovered by 8 dpo. SCI mice also spent significantly less time moving than sham controls on 2 and 3 dpo (see the “Results” section). For (A)–(C), *p < 0.05, ***p < 0.005, and ****p < 0.0001; post hoc tests within SCI mice after mixed-effects analysis. (D) Average percent time active and RRVAR from 1 to 3 dpo were uncorrelated across SCI mice. dpo, days postoperation.

Mechanical and thermal pain assessments

The von Frey test was conducted at baseline and 4 weeks post-SCI, and the Hargreaves at 4 weeks post-SCI, to assess changes in threshold for paw withdrawal reflexes in response to mechanical and heat stimuli, respectively. As seen in previous studies,^19,31,34 the threshold filament force needed to induce a withdrawal response in the von Frey test decreased by 4 weeks after SCI, consistent with mechanical allodynia (Fig. 3A; paired t-test; t(9) = 7.356, p < 0.0001). This effect was also significant compared with sham controls at 4 weeks postoperation (unpaired t-test; t(15) = 2.883, p < 0.05).

OPEN IN VIEWER

FIG. 3.

Von Frey test scores and thermal preferences. N = 10 SCI. (A) von Frey scores for all mice tested at baseline (1 day preinjury) and 4 weeks postinjury. The lines between data points indicate changes within individual mice. There was a statistically significant change in von Frey scores 4 weeks postinjury (****p < 0.0001). (B) The Hargreaves Test for heat allodynia was run at 4 weeks postinjury in SCI mice. SCI mice withdrew their hindpaws significantly earlier than a control population of non-SCI (naïve and sham) mice (N = 15; unpaired t-test, ****p < 0.0001). (C) The thermal preference test across a temperature gradient of 20–40°C, undertaken at 4 weeks postinjury. The average thermal preference after SCI was significantly higher than the thermoneutral temperature in non-SCI mice (N = 24; unpaired t-test, **p < 0.01). Sham and naïve mice were combined for these analyses since the two groups were statistically indistinguishable. PWT, paw withdrawal threshold.

In the Hargreaves test (Fig. 3B), mice had a mean latency of 3.55 ± 0.59 sec to remove their hindpaw from the thermal stimulus, which was significantly lower than the withdrawal latency of 10.87 ± 1.17 sec in a control cohort of n = 15 mice without SCI (naïve and sham control) (t(23) = 4.832, p < 0.0001), consistent with heat allodynia.

In the Thermal Preference test, conducted at 4 weeks post-SCI (Fig. 3C), SCI mice preferred a temperature of 33.7 ± 1.2°C. This value was higher than the reported thermoneutrality point of 30°C³⁷ and significantly higher than the preferred temperature of 30.9 ± 0.4°C observed in a control cohort of n = 24 mice without SCI (naïve and sham) (unpaired t-test; t(32) = 3.033, p < 0.01), suggesting thermal dysregulation.

Correlations between early SCI changes and later pain-related outcomes

To understand the relationship between early measures of RRVAR and movement with the subsequent development of dysfunction, correlation analyses were performed. We assessed whether acute and subacute changes after SCI were associated with changes in measured mechanical and thermal sensitivity. A comprehensive table of correlations is included in Figure 5. Larger decreases in RRVAR from baseline to 1 day after SCI corresponded to less sensitivity to stimulation in the von Frey test 4 weeks after injury (greater withdrawal thresholds, p = 0.01). Follow-up analysis revealed that this effect was also significant from baseline to 3 dpo (p = 0.024) with the same trend at 2 dpo (p = 0.06). We also investigated the relationship between early changes in RRVAR after SCI and future behavioral outcomes (Fig. 4). The decrease in RRVAR at 1 dpo (compared with its recovery at 8 dpo) was positively correlated with increases in temperature preferences at 4 weeks postinjury (Fig. 4A), that is, early reductions in RRVAR that recovered by 8 dpo predicted a preference for hotter temperatures. Follow-up analyses revealed that this effect was also significant at 2 dpo (p = 0.009) and 3 dpo (p = 0.01) versus the 8 dpo time point.

OPEN IN VIEWER

FIG. 4.

Correlating early recovery of movement and RRVAR with pain outcomes in SCI mice. (A) Decreases in RRVAR at 1 compared with 8 dpo were significantly associated with higher preferred floor temperatures at 4 weeks postinjury. (B) Decreased % time active at 1 compared with 8 dpo was negatively correlated with von Frey scores (i.e., emergent mechano-sensitivity) at 4 weeks postinjury. N = 10; dpo, days postinjury; BL, baseline.

Exploring the relationship between movement and future hypersensitivity, recovery of activity (change in % time active from 1 to 8 dpo) was negatively correlated with von Frey scores at 4 weeks postinjury (Fig. 4B), that is, animals with greater recovery showed more mechanical hypersensitivity. Follow-up analyses revealed that there was a significant correlation between reduction in % time active at 1 dpo and 4-week von Frey score (p = 0.036), but this correlation was lost at 2 and 3 dpo.

In sham control mice, which did not show the development of mechanical or thermal dysfunction at 4 weeks postoperation (see above), none of these correlations reached statistical significance.