Ethological profiling defines pain behaviors and parses analgesia from drug side effects in a mouse model of complex regional pain syndrome

Supplemental Digital Content is Available in the Text.

Holistic assessment of naturalistic and nocifensive behaviors parses alpha2-mediated analgesia from sedation in mice after tibial fracture-cast injury.

Abstract

Complex regional pain syndrome (CRPS) is a form of chronic postinjury pain affecting the extremities with contributions from the somatic and sympathetic nervous systems. The mouse tibial fracture-cast model was developed to enable preclinical study of CRPS mechanisms and guide condition-specific drug development. Given the inherent limitations of reflex pain measures in mice, we sought to holistically characterize pain-like behaviors in this model in neutral and aversive environments using quantitative behavior analysis with LabGym, a user-friendly automation tool that requires no special equipment or extensive computational resources. This study shows that tibial fracture-cast injury causes distinct changes in naturalistic and nocifensive behaviors in male and female mice in neutral and aversive environments, demonstrating reliable behavioral categorization using robust learning-based holistic assessment. As proof-of-concept for therapeutic testing, we leveraged this behavioral evaluation to characterize the peripheral vs central effects of targeting alpha-2 receptors (α2-AR) with dexmedetomidine (DEX), a selective α2-AR agonist with analgesic, sedative, and anxiolytic properties. We found that DEX reduced mechanical allodynia primarily through central α2-ARs. Dexmedetomidine also decreased motion metrics, grooming, and rearing in an open field and distinctly affected the quality and quantity of grooming in an aversive environment, and systemic α2-AR antagonists did not suppress this effect. Importantly, we also determined that the sedative effects of DEX were attenuated in injured compared with uninjured mice, consistent with known sympathetic nervous system activation in CRPS. Overall, this study highlights the use of automated behavioral testing to parse analgesia from sedation in a unique preclinical pain model.

1. Introduction

Complex regional pain syndrome (CRPS) is a form of chronic postinjury pain that develops in the extremities after minor trauma,⁶² affecting 50,000 new patients in the United States every year.⁶⁵ Casting of injured limbs or prolonged immobilization are established risk factors for CRPS development.^18,60 Hence, the widely used mouse tibial fracture-cast model was developed to enable preclinical study of CRPS mechanisms³ as it produces CRPS-like signs in mice that replicate key clinical features of the condition,⁸⁰ including touch, heat and cold hypersensitivity, limb swelling and inflammation, osteopenia, and loss of limb function.^3,21,34 The model also serves as a platform to test potential therapies for this debilitating condition.^3,29

As with all preclinical pain models, reflex measures (eg touch sensitivity) are reliably performed, but provide a limited evaluation of the “pain experience,” perhaps more specifically providing a readout of “nociception” after injury with limited translational validity.^3,55,73 Nonstimulus-evoked methods, including grimace scales, locomotion, weight bearing, and gait analysis, can also be evaluated^{3,14,16,21,29,34,35,73} and specific to the CRPS model, peripheral signs such as temperature changes in the injured limb can be measured.^3,21,35 Although these methods are useful in the preclinical setting, it is paramount that scientists develop and utilize reliable and sensitive objective measures of spontaneous pain behavior to improve translatability of preclinical findings.⁵⁵ Although it has not hitherto been implemented in a CRPS model, pain may also be evaluated by scoring changes in naturalistic behaviors such as grooming and rearing, in which pain-induced changes occur in other models.^16,19,76 The advent of quantitative behavior analysis tools, like DeepLabCut,⁵³ DeepPoseKit,²⁸ MoSeq,⁸¹ SLEAP,⁵⁹ LabGym,³³ SimBA,²⁷ B-SoiD,³² and VAME⁴⁸ allows detailed and relatively unbiased evaluation of naturalistic behaviors after injury.⁴ LabGym,³³ in particular, is user-friendly and allows for holistic assessment and quantification of diverse aspects of behavior.

Altered sweating, color, and skin temperature in the affected limb represent known sympathetic nervous system (SNS) involvement in CRPS.⁷ Adrenergic receptors are the main effectors of SNS activation, and, particularly, the expression and efficacy of alpha-2 receptors (α2-AR) are altered in the peripheral and central nervous system after nerve injury.⁶¹ The net effect of α2-AR activation is antinociception⁵⁰; however, sedation also results from action at α2-ARs centrally. In physiologic conditions, the primary site of α2-AR agonist analgesic action is at the spinal cord,⁴² but this may change after injury. Limited clinical case series demonstrate that topical α2-AR agonists can decrease hyperalgesia, suggesting a peripheral effect,^23,74 and central administration of α2-AR agonists reduces pain in patients with CRPS, suggesting a central effect.⁵⁶ However, the relative contributions of α2-AR signaling at each potential site of action to analgesia and sedation in a sympathetically maintained pain model have not been explored.

The goals of this study were thus 2-fold: (1) to holistically evaluate the tibial fracture-cast model using LabGym to assess changes in spontaneous and naturalistic behaviors in neutral and aversive environments, and (2) to evaluate the peripheral and central effects of α2-AR agonism on these behavioral metrics as proof-of-concept for parsing analgesia from sedation for therapeutic testing.

2. Materials and methods

2.1. Animals

Ten- to twelve-week-old adult male and female C57BL/6J (Jax stock #00664) mice, weighing approximately 25 g, were used for all experiments. Each cage housed 5 mice in a room maintained on a 12-hour light/dark cycle in a temperature-controlled environment with ad libitum access to food and water.

2.2. Drugs and routes of administration

Dexmedetomidine hydrochloride (Sigma-Aldrich, St Louis, MO, #SML0956-10 MG or #SML0956-50 MG), Yohimbine hydrochloride (Abcam, Fremont, CA, # ab120239), Vatinoxan/MK-467 hydrochloride (MedChemExpress, Monmouth Junction, NJ, #130466-38-5), and BRL-44408 maleate (Abcam, # ab120806) were dissolved in sterile water as directed by the manufacturer at a concentration of 10 mg/mL, 5 mg/mL, 5 mg/mL, and 5 mg/mL, respectively. Injured mice received a single dose of dexmedetomidine hydrochloride intraperitoneally (IP, 1.25 µg in 100 µL; 50 µg/kg). For antagonist experiments, injured mice were injected intraperitoneally with either yohimbine hydrochloride (0.05 mg in 100 µL; 2 mg/kg), Vatinoxan/MK-467 hydrochloride (0.05 mg in 100 µL; 2 mg/kg), or BRL-44408 maleate (0.05 mg in 100 µL; 2 mg/kg) 30 minutes before injecting dexmedetomidine hydrochloride (1.25 µg in 100 µL; 50 µg/kg) intraperitoneally. Drug doses were selected based on previous studies using dexmedetomidine,^58,67 yohimbine,^17,42,51,72 BRL-44408 maleate,⁸ and MK467⁴² in mice.

2.3. Tibial fracture-cast model of complex regional pain syndrome

Mice were anesthetized with isoflurane and underwent a closed right distal tibia fracture followed by casting. The right hind limb was wrapped in gauze, and a hemostat was used to make a closed fracture of the distal tibia. The hind limb was then wrapped in casting tape (3M ScotchCast Plus, Amazon, Seattle, WA) from the metatarsals of the hind paw up to a spica formed around the abdomen to ensure that the cast did not slip off. The cast over the paw was applied only to the plantar surface with a window left open over the dorsum of the paw and ankle to prevent constriction when postfracture edema developed. Mice were inspected throughout the postoperative period of cast immobilization to ensure that the cast was properly positioned. At 3 weeks postfracture, mice were briefly anesthetized, and casts were removed with cast shears. For behavioral assessment, mice were tested beginning 1 day after cast removal at 3 weeks until 6 days after cast removal. Complex regional pain syndrome model generation and behavioral testing were conducted following well-established methods for evaluating mouse behavior in the tibial fracture-cast model of CRPS.^{3,14,21,29,34,35}

2.4. Behavioral testing

All testing was conducted within the same time window on each testing day in an isolated, temperature- and light-controlled room. Mice were acclimated for 60 minutes in the testing environment within custom clear plastic cylinders (4″ D) on a raised metal mesh platform (24″ H). To ensure blinding of the experimenter, mice were either randomly placed in a cylinder and identified after testing, or mice were assigned to a group and injected with a coded treatment whose identity was revealed to the experimenter after the experiment was complete and the data were processed.

2.5. Mechanical nociception assays

To evaluate mechanical reflexive hypersensitivity, we used a logarithmically increasing set of 8 von Frey filaments (Stoelting, Wood Dale, IL), ranging in gram force from 0.007 to 6.0 g. These were applied perpendicular to the plantar hind paw with sufficient force to cause a slight bending of the filament. A positive response was characterized as a rapid withdrawal of the paw away from the stimulus filament within 4 seconds. Using the up-down statistical method, the 50% withdrawal mechanical threshold scores were calculated for each mouse and then averaged across the experimental groups.

2.6. Paw edema, unweighting, and temperature measurements

Hind paw edema was determined by measuring the hind paw dorsal-ventral thickness over the midpoint of the third metatarsal with digital calipers while the mouse was restrained in a tube. Hind paw thickness data were analyzed as the difference between the fracture side and the contralateral intact side and averaged across experimental groups. Paw edema was measured at 3 weeks postfracture 1 hour before drug injection and 2 hours after drug injection.

An incapacitance device (IITC Inc Life Science, Woodland Hills, CA) was used to measure hind paw unweighting. Mice were manually held in a vertical position over the apparatus with the hind paws resting on separate metal scale plates, and the entire weight of the mouse was supported on the hind paws. The duration of each measurement was 6 seconds, and 6 consecutive measurements were taken at 60-second intervals. Six readings were averaged to calculate the bilateral hind paw weight-bearing values. Unweighting was measured at baseline and then again at 3 weeks postfracture, and 1 and 2 hours post dexmedetomidine (DEX) injection. Data were analyzed as a ratio between the right hind paw weight divided by the sum of right and left hind paw values [2R/(R + L)] × 100%.

The temperature of the hind paw was measured using a fine-gauge thermocouple wire (Omega, Stamford, CT). Temperature testing was performed over the hind paw dorsal skin between the first and second metatarsals (medial), the second and third metatarsals (central), and the fourth and fifth metatarsals (lateral). The measurements for each hind paw were averaged for the mean paw temperature. Data were expressed either as temperature in degrees Celsius (°C) or as the average difference between the ipsilateral and contralateral hind paws within an experimental group. Paw temperature was measured at 3 weeks postfracture before and after (1 and 2 hours) drug injection.

2.7. Thermal nociception assays

To evaluate thermally induced reflexive responses, we used the Hargreaves and hotplate tests.

For the Hargreaves test, mice were placed in their glass-bottomed enclosures, and a radiant heat source aimed at the plantar surface of the mouse's right hind paw was positioned beneath the animal. The time taken to withdraw from the heat stimulus was manually recorded as the paw withdrawal latency in seconds. A cut-off time of 30 seconds was set to prevent thermal injury to the paw. Paw withdrawal latencies were recorded at baseline and 3 weeks postfracture.

For the hotplate test, the plate was set to 50°C, and mice were placed on the plate and recorded for 45 seconds using a high-definition DFK 22AUC03 video camera (60 fps). Only 1 exposure to the hotplate was applied in a given testing session to prevent behavioral sensitization that can result from multiple noxious exposures. In cases where multiple measurements were performed on the same cohort of mice, at least 2 days of rest were left between testing sessions. Measurements were performed at baseline and 3 weeks postfracture.

2.8. Open field video recordings and analysis

All open field recordings were taken overhead using a high-definition DFK 22AUC03 video camera (30 fps) in 2 side-by-side 45 cm × 45 cm × 45 cm boxes. Mice were recorded at baseline, postfracture, and 1-hour postdrug injection for 30 minutes at each timepoint. Before any analysis was performed, videos were cropped so that each video contained 1 box with 1 animal.

2.8.1. Mouse tracking and region of interest analysis

Mice were tracked, and region of interest (ROI) analysis was performed using AnyMaze (https://www.any-maze.com/). In brief, videos were imported into AnyMaze, and 2 ROIs were defined and manually drawn on each video. The first ROI was a square measuring 25 cm × 25 cm representing the center zone, and the second ROI was a larger square measuring 30 cm × 30 cm demarcating the peripheral thigmotaxis zone. We then proceeded with ROI analysis and generated CSV files containing descriptive statistics for distance traveled (m), average speed, time spent immobile, number of entries into each ROI, and the time spent in each ROI.

2.8.2. Behavior stratification using LabGym

Open field videos were also analyzed using LabGym.³³ To analyze different behaviors of interest, models were trained as outlined on the LabGym GitHub page (https://github.com/umyelab/LabGym). A detailed behavior analysis pipeline is depicted in Figure 1. To analyze different behaviors of interest, example frames were generated using LabGym and uploaded to Roboflow (app.roboflow.com). Ninety frames were annotated in Roboflow and used to train a detector for open field videos in LabGym. The trained open field mouse detector was used to generate behavior bout examples (15 frames per bout), which were manually sorted into the following behavior classes: body grooming, face grooming, walking, rearing, and idling. Sorted examples were then used to train a categorizer whose performance was tested on other sets of independently sorted data, as shown in Supplemental Figure 1, http://links.lww.com/PAIN/C479. The trained categorizer was used to generate behavioral categorizations with 40% uncertainty that were then fed into the quantifier to compute specific quantitative measurements for different aspects of each behavior as defined in Hu et al.³³

Figure 1.

The behavior video analysis pipeline. (A) Behavior videos are recorded. (B) A representative sample of frames is extracted from recorded videos. (C) Extracted frames are annotated using Roboflow. (D) Annotated frames are extracted and imported into LabGym. (E) Annotations are used to train a mouse detector in LabGym. (F) The mouse detector is used to generate behavior examples from recorded videos (10 or 15 frames per behavior example). (G) Behavior examples are manually sorted into behaviors of interest. (H) Sorted behavior examples are used to train a behavior categorizer. (I) Test videos are analyzed using the trained detector and categorizer, and behaviors of interest are quantified.

Quantitative behavior measures generated in this study as defined by LabGym:

• Count: The frequency of a behavior for the entire duration of analysis.³³

• Duration: The total time spent performing a specific behavior.³³

• Latency: The time it takes for a behavior to occur for the first time during analysis.³³

• Intensity: The summary of how intense a behavior is, calculated as the proportional changes of the animal’s body area between frames during a behavior divided by the time window used for categorizing the behaviors.³³

• Magnitude: The summary of the magnitude of motion, calculated as the maximum proportional change in animal body area when performing a specific behavior.³³

• Vigor: The summary of how vigorous a behavior is, calculated as the magnitude divided by the time during which a behavior bout takes place.³³

2.9. Hotplate video analysis

As with open field video recordings, hotplate videos were analyzed using LabGym.³³ In brief, example frames were generated using LabGym and uploaded to Roboflow (app.roboflow.com). A total of 914 frames were annotated to locate the mouse using instance segmentation in Roboflow. Annotations were exported from Roboflow, imported into LabGym, and used to train a highly precise and accurate detector in LabGym. The trained detector was used to generate behavior bout examples (10 frames per bout), which were manually sorted into the following behavior classes: grooming, jumping, paw withdrawal, paw licking, rearing, and other (representing miscellaneous behaviors observed). One thousand five hundred forty sorted bout examples were then used to train a categorizer, whose performance was also evaluated on independently sorted data. The trained categorizer was used to generate behavioral categorizations with 40% uncertainty, which were then quantified, exported, and plotted using GraphPad Prism version 10.2.2.

2.10. Quantification and statistical analysis

Measurements of cohort sizes were determined based on historical data from our laboratory using a power analysis to provide >80% power to discover 25% differences with P < 0.05 between groups, requiring a minimum of 4 animals per group for all behavioral outcomes. All experiments were randomized by cage and performed by a blinded researcher. Researchers remained blinded throughout behavioral assessments. Groups were unblinded at the end of each experiment after statistical analysis. All data are expressed as the mean ± SEM. Statistical analysis was performed using GraphPad Prism version 10.2.2 (GraphPad Software, Boston, MA). Where appropriate, data were analyzed using ordinary one-way ANOVA with a Tukey correction, ordinary two-way ANOVA with a Tukey or Sidak correction, and 2-tailed Student's t tests. No data were excluded from analyses.

2.11. Study approval

All procedures were approved by the Stanford University Administrative Panel on Laboratory Animal Care in accordance with American Veterinary Medical Association guidelines and the International Association for the Study of Pain.

3. Results

3.1. The tibial fracture-cast model results in a complex regional pain syndrome-like syndrome in mice that reproduces clinical signs

Before deep ethological studies, we sought to confirm the development of common signs of CRPS in our experimental mice by evaluating mechanical sensitivity, thermal sensitivity, weight bearing, paw thickness, and temperature, as shown in Figure 2A. As previously established in the lab,^21,35 we found that tibial fracture-cast injury resulted in a significant increase in mechanical sensitivity indicated by a decreased paw withdrawal threshold to mechanical stimuli (Fig. 2B). Tibial fracture-cast injury also reduced the paw withdrawal latency to a focused thermal stimulus when tested 3 weeks postfracture, indicating an increase in thermal sensitivity in injured mice (Fig. 2C). Injured mice also placed less weight on the injured than the uninjured paw (Fig. 2D). They exhibited an increase in injured paw width (Fig. 2E) and temperature (Fig. 2F) compared with the uninjured paw, indicating postinjury edema and inflammation.

Figure 2.

Tibial fracture-cast injury reproducibly results in CRPS-like signs and motion changes in male mice. (A) Schematic diagram showing experimental timeline for behavior testing (n = 10 male mice). Tibial fracture-cast injury increases mechanical (B) and thermal sensitivity (C) while decreasing weight bearing on the injured paw (D) and increasing paw edema (E) and temperature (F) at 3 weeks postfracture. (G) Schematic diagram showing experimental timeline for open field behavior testing (n = 10 male mice). (H) Representative path plots showing postinjury changes in motion in injured mice. Tibial fracture-cast injury reduces distance traveled (I) and average speed (J) in an open field 3 weeks postfracture. Tibial fracture-cast injury also reduces time spent in the center zone (K) while increasing time spent in the thigmotaxis zone (L) of the open field, and time spent immobile (M) 3 weeks postfracture. All behavior data shown in this figure were analyzed using Student's t-test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Data are expressed as the mean ± SEM. CRPS, complex regional pain syndrome.

We next characterized the effect of tibial fracture-cast injury on motion in an open field to evaluate mobility/function as outlined in Figure 2G. We observed that CRPS mice traveled shorter distances (Figs. 2H and I) at lower average speeds (Fig. 2J) after injury compared with their preinjury baseline. Injured mice also spent less time in the center zone (Fig. 2K) and more time in the thigmotaxis zone (Fig. 2L) of the arena. In general, mice also spent more total time immobile after injury (Fig. 2M).

We therefore concluded that tibial fracture-cast injury results in a CRPS-like presentation that recapitulates common signs of the condition, including paw edema, increased paw temperature, mechanical allodynia, thermal hyperalgesia, anxiety-like behaviors, and impaired mobility.

3.2. Semi-supervised behavior analysis reveals distinct changes in naturalistic behaviors in a neutral environment after tibial fracture-cast injury

First, we evaluated our LabGym open field model on 3 manually sorted datasets, including a naïve/uninjured animal dataset, an injured animal dataset, and a mixed (naïve/uninjured and injured) dataset. The LabGym open field model performed well on all datasets, achieving high precision, recall, and F1 scores for all behaviors of interest as shown in Supplemental Figure 1, http://links.lww.com/PAIN/C479.

Next, we conducted an in-depth analysis of specific naturalistic behaviors in an open field and the impact of tibial fracture-cast injury on those behaviors (Fig. 3A and B). We evaluated these naturalistic behaviors separately in injured male (Fig. 3, Supplemental Fig. 2, http://links.lww.com/PAIN/C479) and female mice (Supplemental Fig. 3, http://links.lww.com/PAIN/C479). We observed an increase in face and body grooming counts and duration in both sexes, with decreased intensity and vigor of the behaviors in male mice only (male mice: Figs. 3C and D; female mice: Supplemental Fig. 3b, c, http://links.lww.com/PAIN/C479). Both sexes also exhibited a decrease in walking duration and intensity and rearing intensity over the course of the recording (male mice: Figs. 3E and F; female mice: Supplemental Fig. 3d, e, http://links.lww.com/PAIN/C479). Interestingly, although tibial fracture-cast injury did not affect any behavior latencies in male mice (Supplemental Fig. 2a, http://links.lww.com/PAIN/C479), it increased face grooming latency and decreased walking latency in female mice (Supplemental Fig. 3b, d, http://links.lww.com/PAIN/C479).

Figure 3.

Tibial fracture-cast injury in male mice causes distinct changes in naturalistic behaviors in an open field. (A) Schematic diagram showing experimental timeline for open field behavior testing and analysis (n = 10 male mice). (B) Representative raster plots showing proportions of mice performing each behavior over the course of the analysis. (C) Tibial fracture-cast injury increases face grooming frequency and duration and reduces face grooming intensity and vigor. (D) Tibial fracture-cast injury increases the frequency and duration of body grooming and decreases body grooming intensity and vigor. (E) Tibial fracture-cast injury decreases walking bout count, duration, intensity, and vigor. (F) Tibial fracture-cast injury decreases rearing count, duration, intensity, and vigor. All behavior data shown in this figure were analyzed using Student's t-test (*P < 0.05, **P < 0.01, ***P < 0.001, ***P < 0.0001). Data are expressed as the mean ± SEM.

These results show that tibial fracture-cast injury distinctly affects the quality and quantity of naturalistic behaviors in male and female mice in a neutral environment. Overall, mice focus more on grooming and less on exploration-type behaviors, and these changes in behavior can be reliably identified using robust learning-based holistic assessment.

3.3. Semi-supervised behavior analysis reveals distinct changes in naturalistic and nocifensive behaviors in a noxious heat environment after tibial fracture-cast injury

As was performed with naturalistic behaviors in an open field, we leveraged the power of machine learning to evaluate the effect of a noxious heat stimulus on naturalistic (rearing, grooming) and pain-related behaviors (paw withdrawal, paw licking, jumping) (Fig. 4A). Our LabGym hotplate analysis model achieved high levels of accuracy for each of our behaviors of interest, as shown in Supplemental Figure 4, http://links.lww.com/PAIN/C479. The animals in this cohort did not jump during the hotplate exposure for the duration of the recording; therefore, this behavior was excluded from further analysis.

Figure 4.

Tibial fracture-cast injury in male mice causes distinct changes in pain-related behaviors in a noxious heat environment. (A) Schematic diagram showing experimental timeline for behavior testing (n = 10 male mice). (B) Representative raster plots showing proportions of mice performing each behavior over the course of the analysis. (C) Tibial fracture-cast injury reduces rearing count, duration, intensity, and vigor in a noxious heat environment. (D) Tibial fracture-cast injury increases grooming count, duration, intensity, and vigor in a noxious heat environment. (E) Tibial fracture-cast injury increases paw licking count and duration but does not affect paw licking intensity and vigor in a noxious heat environment. (F) Tibial fracture-cast injury reduces paw withdrawal count and duration but does not affect paw withdrawal intensity and vigor in a noxious heat environment. All behavior data shown in this figure were analyzed using Student's t-test (*P < 0.05, **P < 0.01, ***P < 0.001, ***P < 0.0001). Data are expressed as the mean ± SEM.

We evaluated naturalistic and nocifensive behaviors in a noxious heat environment separately in injured male (Fig. 4, Supplemental Fig. 5, http://links.lww.com/PAIN/C479) and female mice (Supplemental Fig. 6, http://links.lww.com/PAIN/C479). We observed an increase in count and duration of grooming and licking in male mice (Figs. 4D and E) with only an increase in duration of grooming in female mice (Supplemental Fig. 6c, http://links.lww.com/PAIN/C479). Furthermore, male mice robustly decreased count, duration, intensity, and vigor of rearing and count and duration of paw withdrawal (Figs. 4C and F), while female mice displayed no significant change in any of these behaviors (Supplemental Fig. 6b, e, http://links.lww.com/PAIN/C479). Latency to first bout for all the behaviors remained unchanged in both male (Supplemental Fig. 5a, http://links.lww.com/PAIN/C479) and female mice (Supplemental Fig. 6b-e, http://links.lww.com/PAIN/C479).

Collectively, these results show that tibial fracture-cast injury produces distinct changes to individual nocifensive behaviors in male mice, for example, opposite effects on paw licking (increased) and paw withdrawal (decreased).

3.4. Alpha-2-adrenergic receptor agonism alleviates mechanical allodynia after tibial fracture-cast injury through central action

We next sought to evaluate whether this comprehensive behavioral analysis could parse analgesia from other effects of potential pain therapeutics. We selected DEX, an imidazole compound that displays specific and selective α2-AR agonism, because it is a commonly used sedative agent with analgesic and anxiolytic properties.^26,71,85

First, we investigated the effect of intraperitoneal DEX injection on a standard measure of behavioral sensitivity (mechanical allodynia) and classic signs of CRPS (edema, weight bearing, paw temperature; male mice: Fig. 5A; female mice: Supplemental Fig. 7a, http://links.lww.com/PAIN/C479). We found an increase in mechanical withdrawal thresholds starting at 30 minutes and lasting up to 120 minutes after intraperitoneal DEX injection in male mice (Fig. 5B), indicating reversal of mechanical allodynia. There was also a small but inconsistent increase in mechanical withdrawal threshold in injured female mice after intraperitoneal injection, but it was not statistically significant (Supplemental Fig. 7b, http://links.lww.com/PAIN/C479). These DEX effects on mechanical withdrawal threshold were also observed in uninjured male and female mice up to 2 hours post-DEX injection (Supplemental Fig. 7c, http://links.lww.com/PAIN/C479). We also noted that intraperitoneal DEX increased paw withdrawal threshold in injured male mice to a greater degree (about 50% more) than in injured female mice (Supplemental Fig. 7d, http://links.lww.com/PAIN/C479).

Figure 5.

Systemic α2-AR agonism reduces reflex behaviors and paw warmth through central action in male mice. (A) Schematic diagram showing experimental timeline for open field behavior testing and analysis. (B) Intraperitoneal dexmedetomidine (DEX) reverses mechanical allodynia 3 weeks postfracture (2 treatment groups: saline or DEX; n = 10 male mice per group). Intraperitoneal DEX does not affect percentage weight bearing on the injured paw (C) and paw width (D) but reduces paw temperature (E) at 3 weeks postfracture (single treatment group: DEX; n = 10 male mice). (F) Schematic diagram showing experimental timeline for the α2AR antagonist behavioral experiments (n = 10 male mice per group). (G) Neither yohimbine (YOH), BRL44408 maleate, nor MK467 alone affects mechanical allodynia in injured mice. (H) Schematic diagram showing experimental timeline for the α2AR antagonist + DEX behavioral experiments (n = 10 male mice per group). (I) Administering DEX in the presence of YOH and BRL44408 maleate, but not MK467, reverses DEX-mediated effects on mechanical allodynia in injured mice. All behavior data in this figure were analyzed using an ordinary two-way ANOVA with Tukey correction (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Data are expressed as the mean ± SEM. α2-AR, alpha-2 receptors.

We next investigated the effect of α2-AR agonism on paw edema, weight bearing, and paw temperature 3 weeks postfracture in male mice. We were intrigued to find that systemic DEX had no effect on postfracture weight bearing (Fig. 5C) and paw edema (Fig. 5D). Dexmedetomidine did, however, significantly reduce the temperature of the injured paw 1 hour postinjection (Fig. 5E).

Next, we further interrogated the site of action and the relative contributions of central and peripheral α2-AR to the observed anti-allodynic effects using a systemic α2-AR antagonist (yohimbine hydrochloride), a systemic α2A-AR antagonist (BRL-44408 maleate), or a peripherally restricted α2A-AR antagonist (Vatinoxan/MK-467) (male mice: Fig. 5F). None of the adrenergic antagonists affected mechanical allodynia 3 weeks postfracture when administered alone (Fig. 5G). We then administered either yohimbine hydrochloride, BRL-44408 maleate, or Vatinoxan/MK-467 30 minutes before intraperitoneal DEX injection (male mice: Fig. 5H; female mice: Supplemental Fig. 7e, http://links.lww.com/PAIN/C479). When administered with DEX, yohimbine hydrochloride and BRL-44408 maleate reversed DEX's anti-allodynic effects in male and female mice (Fig. 5I and Supplemental Fig. 7f, http://links.lww.com/PAIN/C479, respectively). Interestingly, Vatinoxan/MK-467 did not affect DEX-mediated improvement in mechanical allodynia in male mice (Fig. 5I), suggesting that the analgesic site of action of systemic DEX might be in the central nervous system. However, Vatinoxan/MK-467 enhanced DEX-mediated improvement in allodynia in injured female mice 1 hour post-DEX injection (Supplemental Fig. 7f, http://links.lww.com/PAIN/C479).

Having evaluated DEX effects on mechanical sensitivity and CRPS signs, we next investigated the effects of α2-AR agonism on motion dynamics in an open field (male mice: Supplemental Fig. 8; female mice: Supplemental Fig. 9, http://links.lww.com/PAIN/C479). Compared with predrug, DEX administration after injury significantly increased immobility time only in injured male mice (Supplemental Fig. 8h, http://links.lww.com/PAIN/C479) with no effect in female mice predrug to postdrug in any motion parameters (Supplemental Fig. 9b-f, http://links.lww.com/PAIN/C479). By contrast, we found that intraperitoneal DEX had more pronounced sedative effects in uninjured mice as evidenced by reduction in the distance traveled and average speed, and increase in center time and time spent immobile in an open field in uninjured male and female mice (male mice: Supplemental Fig. 10b-f; female mice: Supplemental Fig. 10g-k, http://links.lww.com/PAIN/C479).

Taken together, these results reveal centrally mediated α2-AR effects on mechanical allodynia after injury in both male and female mice. They also demonstrate attenuated sedative effects on motion dynamics of systemic DEX administration in injured compared with uninjured mice, a possible indicator of sympathetic activation from CRPS-like injury.

3.5. Alpha-2 receptor agonism alters naturalistic behaviors in an open field after tibial fracture-cast injury with attenuated sedation

The next step was to leverage our semi-supervised behavior analysis to evaluate the effect of targeting α2-AR on naturalistic behaviors in an open field in males (vs. predrug: Fig. 6A and Supplemental Fig. 12, http://links.lww.com/PAIN/C479; vs uninjured: Supplemental Fig. 11, http://links.lww.com/PAIN/C479). In general, DEX treatment did not alter behavior latencies after injury except for a delay in walking (Supplemental Fig. 12a, http://links.lww.com/PAIN/C479). We also found that administering DEX to injured mice reduced the frequency of face grooming, body grooming, rearing, and walking over the course of analysis (Figs. 6B–F). Systemic DEX generally decreased duration, intensity, and vigor of these behaviors and also increased idle time (Supplemental Fig. 12b, http://links.lww.com/PAIN/C479). Overall, these findings suggest increased sedation with DEX; however, this effect was attenuated in injured compared with uninjured mice, consistent with observations on motion dynamics (Supplemental Fig. 11, http://links.lww.com/PAIN/C479).

Figure 6.

Systemic α2-AR agonism affects naturalistic behaviors in an open field after injury in male mice. (A) Schematic diagram showing experimental timeline for behavior testing and analysis (n = 10 male mice). (B) Representative raster plots showing proportions of mice performing each behavior at each time point over the course of the analysis. (C) Intraperitoneal DEX reduces face grooming counts, intensity, and vigor, but does not affect face grooming duration in injured mice. (D) Intraperitoneal DEX reduces body grooming count and duration but does not affect body grooming intensity and vigor in injured mice. (E) Intraperitoneal DEX further reduces walking frequency and duration but does not affect post injury walking intensity and vigor. (F) Intraperitoneal DEX further reduces rearing count, duration, intensity, and vigor. All behavior data in this figure were analyzed using an ordinary one-way ANOVA with Tukey's correction (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Data are expressed as the mean ± SEM. α2-AR, alpha-2 receptors; DEX, dexmedetomidine.

The effect of DEX on naturalistic behaviors in an open field in female mice was also evaluated (vs. predrug: Supplemental Fig. 13; vs uninjured: Supplemental Fig. 14, http://links.lww.com/PAIN/C479). We found effects similar to those observed in male mice, with DEX reducing injury-related behaviors compared with predrug (Supplemental Fig. 13, http://links.lww.com/PAIN/C479), but these effects were attenuated compared with the DEX effects in uninjured mice (Supplemental Fig. 14, http://links.lww.com/PAIN/C479).

These findings further suggest that DEX treatment affects the quality and quantity of naturalistic behaviors in an open field but that the sedative effects are attenuated in the context of CRPS-like injury.

3.6. Alpha-2 receptor agonism differentially affects paw licking and grooming in a noxious heat environment after tibial fracture-cast injury

Next, we conducted an in-depth analysis of α2-AR-mediated effects on injury-induced changes in nocifensive and naturalistic behaviors in a noxious heat environment (male mice: vs saline: Fig. 7 and Supplemental Fig. 15; vs uninjured: Supplemental Fig. 16, http://links.lww.com/PAIN/C479). Systemic DEX injection reduced the proportion of injured mice grooming over the course of analysis (Fig. 7B) and also decreased grooming count, duration, intensity, and vigor (Figs. 7C–G). There were no further changes observed in other miscellaneous behaviors, rearing, and paw withdrawal after DEX administration (Supplemental Fig. 15, http://links.lww.com/PAIN/C479). Although the increase in paw licking latency was reversed by DEX (Fig. 7H), no changes were observed in paw licking count, duration, intensity, and vigor after injury (Supplemental Fig. 15d, http://links.lww.com/PAIN/C479). Overall, these findings suggest less nocifensive and naturalistic behaviors with DEX in injured mice compared with a control saline group; however, when compared with uninjured mice (Supplemental Fig. 16, http://links.lww.com/PAIN/C479), the effect of DEX was attenuated, suggesting again that its effects are state-dependent.

Figure 7.

Systemic α2-AR agonism attenuates heat-evoked grooming and paw licking after tibial fracture-cast injury in male mice. (A) Schematic diagram showing experimental timeline for measuring heat-evoked pain responses following DEX treatment in injured mice (n = 10 male mice). (B) Representative raster plots showing proportions of mice performing each behavior at each time point over the course of the analysis. Intraperitoneal DEX reduces grooming count (C), duration (D), intensity (F), and vigor (G) but does not affect grooming latency (E) in injured mice. (H) Intraperitoneal DEX restores baseline paw licking latency in injured mice. All behavior data in this figure were analyzed using an ordinary 2-way ANOVA with Tukey's correction (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Data are expressed as the mean ± SEM. α2-AR, alpha-2 receptors; DEX, dexmedetomidine.

The effect of DEX on nocifensive and naturalistic behaviors in a noxious heat environment in females was also evaluated (vs. saline: Supplemental Fig. 17; vs uninjured: Supplemental Fig. 18, http://links.lww.com/PAIN/C479). We found effects similar to those observed in male mice, with DEX reducing grooming-related behaviors compared with saline (Supplemental Fig. 17, http://links.lww.com/PAIN/C479); however, these effects were similar to the DEX effects in uninjured mice (Supplemental Fig. 18, http://links.lww.com/PAIN/C479).

To further delineate the site of action and the relative contributions of central and peripheral α2-AR to the observed effects in our comprehensive behavioral analysis in a noxious heat environment, we administered α2-AR and α2A-AR antagonists with DEX (Fig. 8A). We demonstrate that co-administration of DEX and yohimbine did not result in a reversal of DEX-induced decrease in grooming counts, duration, intensity, and vigor in male mice (Fig. 8B and Supplemental Fig. 19, http://links.lww.com/PAIN/C479). Similar results were observed for grooming in a noxious heat environment when DEX was injected in the presence of BRL-44408 maleate (Fig. 8C and Supplemental Fig. 20, http://links.lww.com/PAIN/C479) and MK-467 (Figure 8D and Supplemental Fig. 21, http://links.lww.com/PAIN/C479). Interestingly, antagonist treatment seemed to enhance the effect of DEX, as suggested by a further decrease in grooming intensity in the presence of MK467 (Fig. 8D) and paw licking intensity and vigor (Yohimbine: Fig. 8E; BRL-44408 maleate: Fig. 8F; MK-467: Fig. 8G). α2-AR and α2A-AR antagonists administered with DEX in female mice after injury had minimal effects on any of the measured parameters (Yohimbine: Supplemental Fig. 22; BRL-44408 maleate: Supplemental Fig. 23; MK-467: Supplemental Fig. 24, http://links.lww.com/PAIN/C479).

Figure 8.

Systemic α2-AR antagonism enhances DEX-induced attenuation of nocifensive and naturalistic behaviors after injury in male mice. (A) Schematic diagram showing experimental timeline for measuring heat-evoked pain responses following DEX and antagonist treatment in injured mice. (B) Administering yohimbine (YOH) with DEX does not alter DEX's effect on grooming frequency, duration, intensity, and vigor in injured male mice. (C) Administering BRL44408 maleate (BRL) with DEX does not alter DEX's effect on grooming frequency, duration, intensity, and vigor in injured male mice. (D) Administering MK467 with DEX does not alter DEX's effect on grooming frequency, duration, and vigor but enhances DEX's effect on grooming intensity in injured male mice. (E) Administering YOH with DEX alters DEX's effects on paw licking count, intensity, and vigor in injured male mice. (F) Administering BRL44408 maleate (BRL) with DEX alters DEX's effects on paw licking intensity and vigor in injured male mice. (G) Administering MK467 with DEX enhances DEX's effect on paw licking intensity and vigor in injured male mice. All behavior data in this figure were analyzed using an ordinary one-way ANOVA with Tukey's correction (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). Data are expressed as the mean ± SEM. α2-AR, alpha-2 receptors; DEX, dexmedetomidine.

These findings show that, in contrast to the effects of systemic α2A-AR antagonism on reflex behaviors, it generally does not affect DEX-mediated changes in naturalistic (grooming) behaviors, but in fact may amplify DEX attenuation of paw licking in injured male mice in a noxious environment.

4. Discussion

4.1. A unique ethological signature in the tibial fracture-cast model of complex regional pain syndrome

Chronic postinjury pain has peripheral sensory and central affective components.^16,41,83 Postinjury pain, including CRPS, alters activity in brain regions involved in pain processing, such as the anterior cingulate cortex and the amygdala.⁷⁹ This, in turn, can alter stereotypical behaviors such as grooming, locomotor and exploratory behavior, and nesting.⁶⁶ The ability of preclinical models to predict the success of novel analgesics is limited.¹³ One of the proposed reasons is that historically, preclinical studies have evaluated nociceptive reflex pathways rather than full circuit responses reflective of “pain.”^13,52 Automated behavior analysis with tools like MoSeq,⁸¹ ARBEL,² and LUPE,³⁷ provides better precision for detecting subtle behavioral changes stemming from full circuit responses in pain models. However, there are significant barriers to integrating automated behavior analysis tools into routine behavior analysis pipelines in pain research labs. Most automation tools require considerable computational expertise, upfront investment in hardware such as high-speed cameras, training personnel, and construction of specialized testing apparatuses. We therefore sought a more user-friendly option for holistic ethological assessment of injured mice. LabGym requires no special equipment, is broadly applicable across behavioral paradigms, and requires little to no coding skills,³³ making it easier to deploy in labs without extensive computational resources. Importantly, an expert can “teach” LabGym how to categorize pain behaviors of interest, and in turn, LabGym can accurately mimic expert behavior categorization and allow assessment of the quality and the quantity of identified behaviors.³³

Injury-induced changes in stereotypical behaviors reflect coordinated alterations in how behaviors are selected, organized, and executed by the individual. In this study, we used semi-supervised behavioral analysis to resolve these changes across both neutral and noxious heat environments in the tibial fracture-cast model of CRPS. Our findings reveal a distinct ethological signature of injury characterized by shifts in behavioral engagement—how often and how long behaviors occur—and behavioral quality—how those behaviors are executed. Specifically, tibial fracture-cast injury increased engagement in grooming behaviors while suppressing exploratory behaviors such as walking and rearing in a neutral open-field environment. In comparison, reduced grooming is noted in a model of oxaliplatin-induced neuropathic pain,²² and increased grooming is noted in mouse models of inflammatory pain⁴⁵ and trigeminal pain.⁶³ Under noxious heat conditions, injury induced divergent nocifensive strategies, with increased paw licking and grooming accompanied by reduced paw withdrawal and rearing. These context-dependent changes suggest that injury reorganizes behavioral priorities rather than uniformly suppressing activity.

Furthermore, in our study, mice exhibit reduced rearing and locomotion in an open field after injury. Similar findings have been shown in other mouse fracture models in which the fracture impairs limb function and weight bearing, leading to reduced rearing and walking.^49,57 This can be likened to patients with CRPS who often avoid use of their injured limb.^31,69 In this context, increased paw licking and grooming may represent alternative nocifensive or stress-related coping strategies rather than simple indicators of pain intensity. We also observed more frequent, intense, and vigorous grooming in injured mice when they were placed on a hotplate, reflecting heightened stress.

This study provides a comprehensive evaluation of evoked and spontaneous pain in a mouse tibial fracture-cast injury model of CRPS. It combines previously documented changes in reflexive pain and autonomic signs with a detailed analysis of the changes in stereotypical mouse behaviors in neutral and noxious heat environments after injury. By integrating reflexive, spontaneous, and ethological measures, this approach provides a more nuanced framework for evaluating pain-related behavior and therapeutic efficacy in the preclinical models of chronic pain.

4.2. Peripheral vs central action of dexmedetomidine after tibial fracture-cast injury

We used DEX to characterize the effect of targeting α2-ARs as a proof-of-concept for using ethological profiling to evaluate drug action. By reducing the release of norepinephrine, DEX dampens neuronal excitability in pain pathways, resulting in analgesia, anxiolysis, and, at higher doses, sedation.⁸⁵ As in previous studies in other models of pain,^{20,39,44,46,85,86} DEX reduced mechanical allodynia in the tibial fracture-cast model of CRPS. Dexmedetomidine also reduced paw temperature after injury. Both temperature increase and decrease have been associated with DEX in humans, and the underlying mechanisms in either case are still unclear.¹⁵ In rodents, DEX reduces body temperature primarily through modulating thermoregulation in the hypothalamus.^9,47,54 However, peripheral vascular α2-ARs have also been implicated in DEX-mediated hypothermia^1,25 and could contribute to our finding. Dexmedetomidine also has anti-inflammatory properties that could indirectly result in paw temperature reduction.⁸⁴ The analgesic and anxiolytic effects of DEX may also normalize naturalistic behaviors that are altered by chronic postinjury pain.¹¹ Using comprehensive behavioral analysis, we sought to parse analgesia from the sedative effects of DEX as a tool compound with direct effects on both behaviors. Although grooming-related behaviors were reduced under both neutral and noxious heat conditions, exploratory behaviors such as locomotion and rearing were not restored following DEX treatment. Importantly, these effects occurred in the context of attenuated sedative responses of DEX after injury. Sedation is a well-documented effect of DEX,^26,36,75,78 and it has been shown in mice given doses starting from 30 µg/kg to 100 µg/kg.^{10,12,12,64,77} Our study revealed that DEX-mediated sedation is attenuated in injured mice compared with uninjured mice, implying that DEX's sedative effect is state-dependent. We hypothesize that this attenuation in mice with tibial fracture-cast injury is due to the increase in sympathetic tone from the injury, evidenced by paw edema and increased paw temperature in this study and several others.^14,35,68 Other possible adaptive changes in the peripheral and central autonomic nervous system that may account for this attenuated sedation include an increase in peripheral and spinal α2-ARs and changes in their efficacy.⁶

We next interrogated the relative contribution of peripheral and central α2-ARs to DEX action on mechanical allodynia and naturalistic behaviors in a noxious heat environment. We hypothesized that the modulating effects of DEX in our study resulted from combined activation of central and peripheral α2-ARs and used selective antagonists to test this hypothesis. We discovered that mechanical allodynia after tibial fracture-cast injury is primarily mediated by central α2A-ARs. Conversely, systemic and peripheral antagonism of α2A-ARs did not mitigate DEX-mediated effects on naturalistic behaviors but instead amplified DEX attenuation of paw licking in a noxious heat environment. This could indicate that reflexive pain after tibial fracture-cast injury is mediated by central α2-AR, while naturalistic behaviors may be mediated in a more complex manner. The persistence of DEX-mediated ethological effects despite α2-AR antagonism suggests that behavioral modulation by DEX extends beyond classical adrenergic analgesic pathways. Emerging evidence suggests that, in addition to α2A-AR activation, DEX may engage additional mediators and modulatory pathways to achieve analgesia. For example, DEX has been reported to interact with imidazoline receptors that participate in central sympathetic modulation and might help dampen the stress response after injury.^30,38,70 Dexmedetomidine also acts as an anti-inflammatory by reducing the release of pro-inflammatory cytokines and modulating neuroinflammatory responses,^43,82,84 a response that may be an indirect effect of DEX action on sympathetic tone. Similarly, DEX has been shown to exert a neuroprotective role by modulating glial activity,⁵⁸ which could also contribute to DEX analgesia. Finally, DEX might influence other neurotransmitter systems such as opioid systems,^5,24,40 which further complicates the delineation of a precise mechanism for DEX analgesia. This is in no way an exhaustive list, and further investigation is warranted to provide deeper insight into DEX's mechanism of action, which in turn would allow optimization of DEX's therapeutic benefits.

This unique ethological signature (summarized in Table 1) incorporates aspects of behavior that are reflective of the human condition and provides a platform for development for mechanism-based treatment, as demonstrated by our proof-of-concept using DEX.

Table 1.

Summary of injury and dexmedetomidine-induced changes in ethological and pain-related behaviors across environments and sexes.

A. Neutral environment (open field)
Behavior	Injury (male)	Injury (female)	Injury + DEX (male)	Injury + DEX (female)
Face grooming	↑ Engagement, ↓ Quality	↑ Engagement	↓ Quality; ↓ Engagement (count, intensity, vigor)	Mixed (variable by metric)
Body grooming	↑ Engagement, ↓ Quality	↑ Engagement	↓ Engagement (count, duration)	↓ Engagement
Walking	↓ Engagement, ↓ Quality	↓ Engagement, ↓ Quality	Not significantly altered	Not significantly altered
Rearing	↓ Engagement, ↓ Quality	↓ Quality	↓ Engagement	Not significantly altered

B. Aversive environment (noxious heat)
Behavior	Injury (male)	Injury (female)	Injury + DEX (male)	Injury + DEX (female)
Rearing	↓ Engagement, ↓ Quality	Not significantly altered	Not significantly altered	Not significantly altered
Grooming	↑ Engagement, ↑ Quality	↑ Engagement (duration only)	↓ Engagement, ↓ Quality	Not significantly altered
Paw licking	↑ Engagement (count, duration); Quality NS	Not significantly altered	↓ Quality (latency restored to baseline)	Not significantly altered
Paw withdrawal	↓ Engagement (count, duration); Quality NS	Not significantly altered	↓ Quality	Not significantly altered

C. α2-adrenergic receptor mechanism
Manipulation	Behavioral Effect	Interpretation
DEX alone	Reduces behavioral engagement and quality across contexts	α2-AR-mediated modulation of behavior; sedative effects are injury state-dependent
DEX + Yohimbine (nonselective α2-AR antagonist, systemic)	Reverses DEX anti-allodynic effects; grooming reduction persists	DEX analgesia is α2-AR-dependent
DEX + BRL-44408 (selective α2A-AR antagonist, systemic)	Reverses DEX anti-allodynic effects; grooming reduction persists	Central α2A-AR mediates mechanical analgesia
DEX + MK-467 (peripherally restricted α2A-AR antagonist)	Does not reverse DEX anti-allodynic effects; enhances DEX effects on nocifensive behaviors	Analgesia is centrally (not peripherally) mediated via α2A-AR

↑ Engagement = increased frequency and/or duration of behavior. ↓ Engagement = decreased frequency and/or duration. ↑/↓ Quality = increased/decreased intensity and vigor. NS = not statistically significant.

α2-AR, alpha-2 adrenergic receptor; CRPS, complex regional pain syndrome; DEX, dexmedetomidine.

5. Study limitations

First, mice were tested at a single timepoint after tibial fracture-cast surgery, which was selected given the known acute-to-chronic or peripheral-to-central transition at that timepoint. Performing such analyses at multiple timepoints after injury would provide a more holistic view of the evolution of behaviors after injury. Second, drug experiments involved a single drug treatment followed by behavior assessments. Continuous infusions or repeated dosing paradigms may have different effects on pain behaviors in this mouse model.

Conflict of interest statement

The authors have no conflict of interest to declare.

Supplementary Material

jop-167-1453-s001.pdf ^{(26.3MB, pdf)}

Appendix A. Supplemental digital content

Supplemental digital content associated with this article can be found online at http://links.lww.com/PAIN/C479.

Supplemental video content

A video abstract associated with this article can be found on the PAIN Web site.