CatWalk XT Enables Sensitive Detection of Locomotor Deficits after C2 Hemisection and Integrates Readily into Ongoing Experimental Workflows in Adult Female Rats

Abstract

Over half of all spinal cord injuries (SCIs) in the United States occur at the cervical level and often cause both life-threatening breathing dysfunction and profound locomotor deficits. Although the C2 hemisection (C2Hx) model is vital for study of breathing dysfunction after experimental SCI, post-C2Hx locomotor deficits have only been assessed by metrics such as the Martínez Scale and ladder tests, which require time-intensive scoring by human observers. However, the CatWalk XT Gait Analysis system (Noldus Information Technology) delivers automated, quantitative assessment of interlimb coordination, gait cycle, and paw function, providing robust locomotor data in a time-efficient and practical manner. CatWalk’s efficiency may render it more feasible to incorporate locomotor analysis into studies that would not otherwise include such data, but it has not yet been applied to C2Hx. Thus, we conducted a proof-of-principle analysis using C2Hx-injured, adult female rats and uninjured naïve controls from two separate ongoing experiments. We hypothesized that C2Hx would produce primarily left-sided impairments measurable by CatWalk as predicted both by prior literature and the neuroanatomical specifics of the injury model. Indeed, CatWalk reliably identified pronounced deficits in left forepaw use and weight bearing, as well as impairments in stepping efficiency, interlimb coordination, speed, and overall gait stability across the 4-week period following injury. In summary, our findings demonstrate that CatWalk sensitively detects locomotor dysfunction after C2Hx and can be readily integrated into ongoing studies, thereby enhancing experimental efficiency and maximizing the scientific yield from existing animal cohorts.

Introduction

Spinal cord injury (SCI) poses a significant public health concern, affecting ∼300,000 individuals within the United States. ¹ SCI has an annual incidence of approximately 18,000 cases, more than half of which involve the cervical spinal cord and often result in life-threatening respiratory motor paresis and locomotor impairment.^1,2

Experimental murine models of high cervical SCI lead to locomotor deficits ranging from paresis of below-level limbs to dyscoordinated stepping and can improve with time and therapeutic intervention.^3–14 The C2 hemisection (C2Hx) model has been a preclinical tool fundamental to our understanding of SCI-induced breathing deficits since 1895, but the locomotor sequelae of the model are less thoroughly described. ¹⁵ Previous work indicates that C2Hx decreases forced running capacity and diminishes performance on both the Basso Beattie Bresnahan-derived Martinez’s Scale and ladder climbing tests, but such assessments require time-intensive scoring by human observers.^{6–8,16–20}

The CatWalk system (Noldus) measures interlimb coordination, gait cycle, paw function, and more based upon automated analysis of paw positioning during ambulation but has not yet been applied to C2Hx.^9,19,21,22 CatWalk allows for measurement of more locomotor-related parameters than any prior single approach and only relies upon human observation for oversight of paw print classification.^9,19,21,22 Thus, the system’s efficiency may render it feasible for incorporation into studies that would not otherwise measure locomotor data.^9,19,21,22

Thus, we conducted a proof-of-principle analysis to compare CatWalk-measured locomotor behavior between C2Hx-injured and uninjured adult female rats through 4 weeks post-injury. We drew the C2Hx and control groups from separate ongoing experiments to reduce animal use in accordance with the “3Rs” of animal research and to evaluate the practicality of integrating CatWalk into existing study designs. ²³ We hypothesized that CatWalk would reveal general and ipsilesional locomotor deficits, as indicated by prior literature and the neuroanatomical specifics of the injury model.^6,12,24,25 Overall, this work evaluates a novel locomotor assessment tool for the C2Hx model and highlights its practical utility for improving research efficiency and maximizing data obtained from existing animal cohorts.

Methods

Animal subjects

Animal use was compliant with the Institutional Animal Care and Use Committee guidelines at the University of Kentucky. Thirty-one female, retired breeder, Sprague Dawley® outbred rats (Inotiv, West Lafayette, IN; pre-injury weight range 255–380 g, 10–14 months of age) were housed in a dedicated vivarium on a 12/12 light/dark cycle and fed a normal chow diet ad libitum.

CatWalk data from subjects assigned to the saline-treated control group of a separate study were reused here as the C2Hx-injured group (n = 11). ²⁶ That group of animals originally contained 22 subjects, but 3/22 (∼14%) were lost due to attrition acutely following C2Hx, leaving 19. Initially incorrect acquisition parameters prevented data collection for eight of those subjects. Of the remaining 11 subjects, 1 failed to complete a compliant CatWalk trial at the 1-week post-injury (1 WPI) timepoint due to severity of locomotor deficit.

A separate, naïve cohort of animals (n = 9) was utilized as the uninjured control for this study and after its conclusion was reused in an unpublished experiment examining peripheral organ dysfunction. As a result, no true randomization of subjects to injured vs. uninjured groups occurred, though all subjects were matched according to strain, sex, and age. ²⁷

Surgical procedure and aftercare for C2Hx

C2Hx surgeries were conducted under isoflurane anesthesia according to previously published procedure.^27–30 Aftercare included subcutaneous saline administration and analgesia with bupivacaine instillation along the surgical site, in addition to subcutaneous administration of buprenorphine (two doses) and carprofen (three doses). Because of their inclusion as a control group in a separate study, subjects receiving C2Hx surgery also were infused with 2 mL of intravenous saline via the tail vein on days 1, 3, and 6 post-injury. ²⁷

CatWalk XT

Locomotor assessments were made using the CatWalk XT Gait Analysis system (Noldus Information Technology b.v., Wageningen, The Netherlands), which utilizes a green-lighted walkway against a red overhead light to identify and record rat foot placements during ambulation from one end of the walkway to the other.^21,31 Gait analysis was conducted by a single researcher in a dark room, with settings of 11.0 dB camera gain, 0.10 intensity threshold, 17.5 V red ceiling light, and 15.0 V green walkway light. A minimum green intensity value of 85 was utilized to identify pixels representing real paw prints, and the video camera was adjusted to 29 inches beneath the walkway.

Subjects were acclimated to the walkway prior to the first day of data acquisition. Afterwards, locomotor assessments were recorded 1–4 days prior to C2Hx injury and every 6 to 7 days thereafter until 4 weeks post-injury. There was approximately 1 week between timepoints.

Runs of duration outside of the 0.5–12 sec range and/or those with maximal variation exceeding 90% were excluded. Three compliant runs were required for successful trial completion on each day, and subjects’ data were excluded from that timepoint if they failed to complete at least 1 compliant run within 15 min of first walkway placement. After automatic paw-print classification by the CatWalk software, any remaining unidentified signals were manually reviewed by a researcher unblinded to treatment group, due to throughput constraints and the low likelihood of bias in this context.

Data analysis

CatWalk variables relying on interactive footprint measurements or missing values were excluded. We selected 17 variables based on their relevance to the C2Hx injury neuroanatomy and survey of the literature.^{9,19,21,22,31–43} These were systematically chosen to evaluate general and limb-specific locomotor function across all four limbs, favoring aggregate measures over individual components.^{9,19,21,22,31–43}

Secondarily, we used the Classification Learning Automated Software System (CLASSify, University of Kentucky Center for Applied Artificial Intelligence) to assess whether other, less intuitive CatWalk variables might better capture C2Hx-driven locomotor differences. ⁴⁴ Results did not indicate any necessary changes to the originally determined set of variables.

Statistical analysis

A longitudinal mixed model was used to test for between-group differences in CatWalk parameters over time. All models included a within-subjects factor of time (baseline, 1-, 2-, 3-, and 4-week post-injury), a between-subjects factor of C2Hx injury status (C2Hx injury, uninjured control), and a time/group interaction. The variables of regularity index and the proxy measure of each forepaw’s maximal weight-bearing ability (Max Contact Max Intensity) were non-normal and transformed prior to analysis using Box-Cox transformation. Significant interactions were followed up with post hoc pair-wise comparisons among C2Hx injury groups at each timepoint. A p value of less than 0.05 was utilized as the significance cutoff with no corrections for multiple comparisons. All data graphs were generated with GraphPad Prism (GraphPad Software, Boston, MA), and data analysis was performed using SAS v 9.4 (SAS Institute INC, Cary, NC).

Results

Analysis of locomotor recovery revealed a significant interaction of C2Hx by time for all selected CatWalk variables except for the forepaw base of support (Fig. 1Bi). As a result, pairwise comparisons were generated for all timepoints of all other variables (Table 1). Asterisks within Figure 1 and bolded values within Table 1 indicate significant pairwise comparisons. Briefly, C2Hx induced an increased hindpaw (Fig. 1Ci), but not forepaw (Fig. 1Bi) base of support, increased the number of steps/trial (Fig. 1Aii), and diminished regularity index (Fig. 1Bii), as well as both overall speed (Fig. 1Cii) and speed attributable to each of the four limbs (body speed, Fig. 1Ciii–vi). Left forepaw print area and maximal weight-bearing (max contact max intensity) were significantly diminished by injury across the post-injury timeline (Fig. 1Aiii,1Biii). Subjects’ left hindpaw and both right-sided limbs’ print area and maximal weight-bearing (Fig. 1Aiv–vi, Biv–vi) were much less affected by injury, except possibly at 3 and 4 WPI for the left hindlimb and right forelimb.

FIG. 1.

C2Hx injury induces locomotor deficits which are measurable by CatWalk XT. The figure legend (Ai) and base of support metrics for forelimbs (Bi) and hindlimbs (Ci) are within the first row (i) while overall locomotor metrics are within the second row (ii) and limb-specific parameters are depicted in order from left forelimb, left hindlimb, right forelimb, and right hindlimb in rows iii through vi, respectively. The “base of support” represents the distance between either the forepaws (Bi) or hindpaws (Ci) and is implicated in previous studies of spinal cord injury. The “regularity index” (Panel Bii) indicates interlimb coordination. In rows iii–vi, column A depicts print area for each paw, column B contains the variable of “maximum contact maximum intensity” for each paw, and column C contains measurements of “body speed” for each limb. Print area is a metric describing the surface area of each paw contacting the platform during locomotion, as an indication of functional paw use, seen in rows iii–vi, column A. Maximum contact maximum intensity is a proxy measure for maximal weight-bearing ability for each paw depicted in rows iii–vi, column B. Finally, “body speed” is a measure of the component of overall walking speed attributable to each limb, rows iii–vi, column C. In all panels, the −1 timepoint on the x-axis depicts pre-injury. In summary, C2Hx significantly increases the mean number of steps/trial (Aii) but diminishes interlimb coordination (Bii), mean speed (Cii), and particularly left forelimb print area (Aiii) and maximum weight-bearing (Biii) in rats. p values <0.05 are bolded where present and asterisks represent significant pair-wise comparisons by post-hoc testing at indicated timepoints.

Table 1.

Means and p Values Reported for Pairwise Comparisons of CatWalk Variables Between C2Hx-Injured and Uninjured Animals at Pre-Injury (−1 Weeks Post-Injury [WPI]), 1 WPI, 2 WPI, 3 WPI, and 4 WPI. p Values Listed as NA Where Overall Interaction Did Not Attain Statistical Significance

Outcome	Group	−1 WPI mean, p		1 WPI mean, p		2 WPI mean, p		3 WPI mean, p		4 WPI mean, p
Number of steps (# steps)	C2Hx	48.2727	0.8221	67.7922	0.0001	60.5455	0.3386	59.3636	0.008	55.7273	0.0295
	Control	49.0000		46.4444		53.6667		48.5556		44.6667
Regularity Index (% transformed to A.U.)	C2Hx	3.8673	0.6323	2.1946	<.0001	2.0454	0.0393	2.5431	0.0171	2.8423	0.5367
	Control	3.635		4.3799		3.082		4.0916		3.1976
Mean speed (cm/s)	C2Hx	32.5502	0.681	19.6576	0.0004	24.9966	0.9578	24.5025	0.0474	27.3642	0.7556
	Control	30.6303		36.2476		25.2209		31.7206		26.0720
Forepaw base of support (cm)	C2Hx	2.2851	NA	2.5266	NA	2.7927	NA	2.6021	NA	2.6033	NA
	Control	2.3941		2.1347		2.2881		2.3625		2.3123
Hindpaw base of support (cm)	C2Hx	3.0069	0.3224	3.5908	0.1784	4.1468	0.0034	3.9246	0.0017	3.8794	0.0728
	Control	3.2393		3.0669		3.2437		3.2068		3.3814
Left forepaw (LF) print area (cm2)	C2Hx	1.8917	0.8111	0.7198	<.0001	1.0498	0.0019	0.9073	<.0001	1.3356	<.0001
	Control	1.8543		1.4643		1.7210		2.0710		2.2517
LF maximal weight-bearing (Intensity transformed to A.U.)	C2Hx	7.9564	0.7082	3.4147	0.0476	4.3637	0.0102	4.0589	<.0001	5.3364	0.0003
	Control	8.3469		5.543		7.5934		9.0458		9.7000
LF attributable speed (cm/s)	C2Hx	31.9596	0.8705	21.4766	0.0034	25.3504	0.8529	25.6799	0.0771	27.9697	0.6077
	Control	31.2367		35.8412		26.1186		32.0165		26.0091
Left hindpaw (LH) print area (cm2)	C2Hx	2.3786	0.7367	0.9437	0.8395	1.6091	0.3720	1.2665	0.0001	1.7474	0.0002
	Control	2.2647		0.9853		1.9185		2.3880		2.9674
LH maximal weight-bearing (intensity transformed to A.U.)	C2Hx	4.0021	0.2409	1.8538	0.7124	2.8582	0.2951	2.7346	0.0066	3.4464	0.0063
	Control	4.7339		1.6246		3.6784		4.6090		5.0817
LH attributable speed (cm/s)	C2Hx	31.6531	0.9094	18.2778	<.0001	25.3852	0.5731	24.2679	0.029	28.6807	0.796
	Control	31.0920		36.6579		27.8887		32.2944		27.6756
Right forepaw (RF) print area (cm2)	C2Hx	1.8717	0.9029	1.4767	0.7802	1.8864	0.3976	1.5868	<.0001	1.9982	0.1996
	Control	1.8872		1.4369		1.7512		2.0807		2.2105
RF maximal weight-bearing (intensity transformed to A.U.)	C2Hx	5.4589	0.5535	3.9323	0.3857	5.624	0.7620	4.1138	0.0018	6.5323	0.2008
	Control	6.0404		3.2076		5.2683		7.1622		8.0634
RF attributable speed (cm/s)	C2Hx	32.1384	0.807	19.2975	0.0004	25.3198	0.9939	25.3940	0.0809	27.7493	0.8955
	Control	30.9848		35.8652		25.3513		31.5902		27.2294
Right hindpaw (RH) print area (cm2)	C2Hx	2.2640	0.8074	1.7120	0.0026	2.1512	0.3303	1.8804	0.2374	2.2141	0.0646
	Control	2.1850		1.0163		1.8546		2.2355		2.8147
RH maximal weight-bearing (intensity transformed to A.U.)	C2Hx	2.7769	0.44	2.1572	0.0072	2.6753	0.1687	2.2594	0.1826	2.8729	0.2395
	Control	3.1473		0.968		1.8859		3.0007		3.5470
RH attributable speed (cm/s)	C2Hx	31.9484	0.8369	19.8942	0.0002	25.6262	0.7546	24.5956	0.0524	28.6975	0.77
	Control	30.9541		36.6019		26.9635		31.7590		27.5305

Bolded values indicate significance by p < 0.05 cutoff.

Discussion

In support of hypothesis, CatWalk following C2Hx revealed ipsilesional forepaw deficits with increased hindlimb base-of-support and reduced stepping efficiency, coordination, and speed (Fig. 1ii to iii, Bii to iii, and Ci–vi), consistent with past data derived from observer-based manual scoring.^6,12,24,25 Practically, CatWalk was easily incorporated into the ongoing experimental workflow that supplied the current study’s data. ²⁷ This, combined with the ability to use animals across multiple projects, highlights CatWalk as a feasible way to expand locomotor assessment in SCI research while optimizing stewardship of experimental resources.

Although injured subjects showed greater contralesional hindpaw use and weight-bearing than controls at 1 WPI (Fig. 1Avi, Bvi), this largely reflected a sharp baseline-to-1 WPI decline in the control group, despite absence of injury. This control group variability over time may have obscured injury-related compensation and also suggests that the surgical procedure itself may reduce locomotor consistency, potentially due to postoperative pain.^45,46 Consequently, including a sham surgical C2Hx group as a control would strengthen future investigations.

Because we lack a definitive explanation for the week-to-week variability seen in our control group’s metrics of hindlimb and contralesional function, we interpret those findings cautiously. Specifically, we suggest that CatWalk variables representing functions mediated by spinal tracts only indirectly affected by SCI are less informative than measures of motor systems directly denervated by injury.

No consensus standard set of CatWalk variables has yet been established for SCI research. While our approach for variable selection combined a literature review with AI-based oversight using CLASSify, a previous study used linear discriminant analysis to integrate CatWalk parameters after SCI into a single validated metric. ⁴⁷ However, because their dataset did not include high cervical SCI, future composite CatWalk variable development should address this for greater generalizability.

Finally, it is vital to acknowledge limitations of the current analysis with respect to its use of the C2Hx model and experimental design. First, while C2Hx is vital for the study of a controlled, unilateral functional deficit, cervical contusion models more closely approximate the heterogeneous pathology of human SCI.^48–51 We selected C2Hx in the current study to expand its utility and to demonstrate the integration of CatWalk into existing workflows, taking advantage of suitable subjects already available from concurrent experiments. Even so, future investigation of locomotor changes via CatWalk after cervical contusion would enhance the clinical relevance of our current findings and further strengthen the applicability of this method across experimental SCI models.

Secondly, using animals from parallel ongoing experiments introduces potential sources of bias. Because subjects were not prospectively randomized into C2Hx and control groups, subsequent experimental interventions occurred under conditions that were less controlled than would be achieved in a dedicated prospective study. This may explain some of the variability observed within the control cohort, particularly given that the C2Hx group received supplemental fluids and postsurgical analgesics not administered to control subjects. Although subjects were matched for strain, sex, age, housing environment, and testing timeline, future studies specifically designed to evaluate CatWalk after C2Hx will be important for more robust assessment of these metrics.

To support this, the present dataset will be made publicly available on ODC-SCI.org, enabling independent analysis by other research groups. Such open data access may facilitate pooled or comparative analyses across laboratories, improving statistical power and promoting a more comprehensive understanding of how CatWalk metrics apply to rodent SCI models, including those with limb or paw deficits less directly tied to the anatomical injury.

In conclusion, this study identifies key CatWalk metrics of locomotor impairment after C2Hx in rats and recommends using surgical sham controls and evaluating cervical contusion in future studies. Similar to other models of SCI, C2Hx reduced regularity index and increased hindpaw base of support. C2Hx also attenuated gross locomotor speed and efficiency and specifically diminished ipsilateral forepaw use and maximal weight-bearing. However, C2Hx did not profoundly affect ipsilateral hindpaw or contralateral compensatory function. Our work also demonstrates how using subjects across multiple projects can reduce experimental animal use, promoting more efficient use of resources. These data will help refine understanding of locomotor deficits following incomplete laceration SCI and encourage future studies and shared analyses to advance therapeutic development for people with SCI.

Transparency, rigor, and reproducibility

This study was not formally preregistered due to the opportunistic nature of its experimental design. The analysis plan was not formally preregistered, but Dr. Aaron Silverstein (first author) and Dr. Christopher McLouth (coauthor and biostatistician) certify that the analysis plan was hypothesis-driven and prespecified. Data drawn from a sample size of 31 total subjects were utilized based on availability of post-SCI subjects (22) co-utilized in a separate experiment and naïve animal subject availability (9) within our laboratory which served as our control group. All animals and data points are fully accounted for, including missing data and attrition, and these are transparently reported in the article’s methods section. Semiautomated CatWalk data collection was performed by experimenters aware of subjects’ injury status as fully described within this article’s methods section. Data for all subjects were collected using the same CatWalk system in the same laboratory, location, though with variance in time of collection between 0700 and 2200 h during the day across an 8-month period of time in multiple experimental cohorts. All statistical analysis was performed in collaboration with a biostatistician and did not include correction of p values for multiple comparisons, but included rigorous evaluation of the assumptions underlying our longitudinal mixed model and transformations of dependent variables as appropriate and described in our article’s methods section. No replication or external validation studies are currently planned or ongoing to our knowledge. All raw data have been uploaded to the Alilain Lab space on the FAIR-compliant repository ODC-SCI.org and will be made publicly available as soon as possible. Analysis scripts and metadata are available upon reasonable request to the corresponding authors, and the same authors are also available for technical assistance or advice related to rodent models of SCI, the CatWalk system, and its applicability to experimental SCI. The authors agree or have agreed to publish the article using the Mary Ann Liebert Inc. “Open Access” option under appropriate license.

Authors’ Contributions

A.S.: Conceptualization, methodology, software, formal analysis, investigation, data curation, writing—original draft, writing—review and editing, supervision, and visualization; C.C.: Methodology, investigation, and writing—review and editing; C.M.: Software, formal analysis, resources, writing—original draft, and writing—review and editing; J.G.: Conceptualization, methodology, resources, writing—review and editing, visualization, supervision, project administration, and funding acquisition; W.A.: Conceptualization, methodology, resources, writing—review and editing, visualization, supervision, project administration, and funding acquisition.