Mice with uPyX engage in significant voluntary complex wheel running rehabilitation.

Mice had access to wheels for voluntary rehabilitation for 26 days post-lesion (A). Schematic of uPyX (B) and representative histological verifications of a sham and uPyX animals are shown (C, C’). Wheel running data was collected using the REVS platform (D). Representative image of a mouse running, with inset showing forelimb rung grasping (E). Side-profile of mouse running (F). Mean runtime from baseline to 28 dpi showed an acute deficit as revealed by an interaction effect between lesion and day (F(2,374) = 5.295, p < 0.0001, ANOVA). Bonferroni-corrected post hoc testing revealed a significant difference at 2 dpi (p=0.01, t-test) (G). Mean speed (H) and distance (I) also showed significant interaction effects (speed: (F(2,374) = 4.469, p < 0.0001; distance: (F(2,374) = 5.552, p < 0.0001; ANOVAs) and deficits at 2 dpi (speed: p = 0.03; distance: p = 0.00006; t-tests). Total cumulative distance ultimately showed no difference between lesioned and sham mice (F(1,17)=0.990, p=0.334, ANOVA) (J). Dotted red line indicates lesion timepoint 0 dpi. Error bars represent +/− SEM. * = p < 0.01, *** = p < 0.001. Scale bar = 200 µm.

Rehabilitation metrics with individual data points.

Bar plot showing mean runtime at – 2, 2, 7, 14, 21, and 28 dpi with individual data points. There was a significant deficit at 2 dpi (p=0.01, t-test) (A) between sham (gray) lesioned (pink) mice. Mean speed (B) and distance (C) also showed significant deficits at 2 dpi (speed: p = 0.03; distance: p = 0.00006; t-tests). Individual lines for each animal show total cumulative distance throughout the experiment for sham (gray) and lesioned (pink) mice. Thicker line shows group means and error bars represent +/− SEM. There was no significant effect of lesion (F(1,17)=0.990, p=0.334, ANOVA) (D). Dotted red line indicates lesion timepoint, 0 dpi. * = p < 0.01, *** = p < 0.001.

Rehabilitation promotes forelimb fine motor function recovery in uPyX mice.

Complex ladder rung testing was completed on –2, 2, 7, 14, 21, and 28 dpi (A). Representative image of a mouse crossing the complex ladder with examples of quantified errors (B). Total number of injured forelimb (FL) steps significantly increased in lesioned animals (F(1,33) = 14.445, p = 0.0006, ANOVA). This lesion effect persisted through 28 dpi (F(1,32) = 13.339, p = 0.0009, ANOVA), but there was no effect of rehabilitation (F(1,32) = 0.772, p = 0.402, ANOVA) (C). Injured FL footfalls significantly increased in lesioned animals (F(1,33) = 20.782, p = 0.00007, ANOVA) and this lesion effect persisted through 28 dpi (F(1,32) = 12.090, p = 0.001, ANOVA). There was additionally a significant effect of rehabilitation (F(1,32) = 7.938 p = 0.008, ANOVA). Bonferroni-corrected post hoc tests revealed a trend of uPyX+Rehab having improved performance relative to uPyX-Rehab at 7 dpi (p = 0.071, t-test) and 14 dpi (p = 0.052, t-test). There was a significant difference between the two groups at 21 dpi (p = 0.003, t-test) and 28 dpi (p = 0.035, t-test) (D). The intact FL also showed a significant lesion-induced footfall deficit (F(1,33) = 4.719, p = 0.037, ANOVA) at 2 dpi, but it did not persist through 28 dpi (F(1,32) =2.186, p = 0.149, ANOVA). Although there was no long-term lesion effect, there was a main effect of rehabilitation (F(1,32) = 4.37, p = 0.045, ANOVA) (E). The injured FL had a significant increase in reach errors at 2 dpi (F(1,33) =15.125, p = 0.00046, ANOVA). This effect persisted through 28 dpi (F(1,32) =11.483, p = 0.002, ANOVA), and there was additionally a significant effect of rehabilitation (F(1,32) = 9.171, p = 0.005, ANOVA). Bonferroni-corrected post hoc tests revealed that uPyX+Rehab significantly improved performance relative to uPyX-Rehab for injured FL reaches at 7 dpi (p = 0.00178, t-test), 14 dpi (p = 0.0114, t-test), 21 dpi (p = 0.00296, t-test), and 28 dpi (p = 0.00943, t-test) (F). Injured HL footfalls had a significant lesion-induced increase at 2 dpi (F(1,33) = 10.603, p = 0.003, ANOVA), which persisted through 28 dpi (F(1,32) = 9.558, p = 0.004, ANOVA). There was a trend of rehabilitation improving injured HL footfalls (F(1,32) = 3.347, p = 0.077, ANOVA) (G). Intact HL footfalls significantly increased with lesion at 2 dpi (F(1,33) = 4.317, p = 0.046, ANOVA) and this effect persisted through 28 dpi (F(1,33) = 4.506, p = 0.042, ANOVA). There was a trend of rehabilitation improving intact HL footfalls (F(1,32) = 3.510, p = 0.070, ANOVA) (H). Principal component analyses revealed that all 4 groups clustered together at –2 dpi (I). At 2 dpi, the two lesion groups separated from the sham groups, driven most strongly by differences in time to cross, step count, and injured FL errors (J). By 28 dpi, the uPyX+Rehab separated from uPyX-Rehab and re-clustered with the sham groups (K). Error bars represent +/−SEM. * = p < 0.05, = p<0.10.

Fine motor function metrics with individual data points.

Bar graphs at –2, 2, 7, 14, 21, and 28 dpi showing the total number of injured forelimb (FL) steps with individual data points. (A). Bar graphs showed that injured FL footfalls significantly increased in lesioned animals (F(1,33) = 20.782, p = 0.00007, ANOVA) and this lesion effect persisted through 28 dpi (F(1,32) = 12.090, p = 0.001, ANOVA). There was additionally a significant effect of rehabilitation (F(1,32) = 7.938 p = 0.008, ANOVA). Bonferroni-corrected post hoc tests revealed a trend of uPyX+Rehab having improved performance relative to uPyX-Rehab at 7 dpi (p = 0.071, t-test) and 14 dpi (p = 0.052, t-test). There was a significant difference between the two groups at 21 dpi (p = 0.003, t-test) and 28 dpi (p = 0.035, t-test) (B). Bar graphs showing the total number of intact forelimb (FL) errors with individual data points (C). Bar graphs showing that the injured FL had a significant increase in reach errors at 2 dpi (F(1,33) =15.125, p = 0.00046, ANOVA). This effect persisted through 28 dpi (F(1,32) =11.483, p = 0.002, ANOVA), and there was additionally a significant effect of rehabilitation (F(1,32) = 9.171, p = 0.005, ANOVA). Bonferroni-corrected post hoc tests revealed that uPyX+Rehab significantly improved performance relative to uPyX-Rehab for injured FL reaches at 7 dpi (p = 0.00178, t-test), 14 dpi (p = 0.0114, t-test), 21 dpi (p = 0.00296, t-test), and 28 dpi (p = 0.00943, t-test) (D). Bar graphs showing the total number of injured hindlimb (HL) errors with individual data points (E). Bar graphs showing the total number of intact HL errors with individual data points (F).

Individual and mean paw trajectories.

Pooled individual injured forelimb footfalls from Sham-Rehab (light grey), Sham+Rehab (dark grey), uPyX-Rehab (green) and uPyX+Rehab (pink) groups at -2 dpi (A), 2 dpi (B), and 28 dpi (C) show consistent maximum displacement beneath the rung across days and groups. Thicker lines represent the mean typical footfall and shading represents the 99% confidence interval around the mean. Pooled individual injured forelimb reaches show typical reach height at -2 dpi (D), aberrant overreaches in the lesioned groups at 2 dpi (E), and recovery to baseline for rehabilitated mice at 28 dpi (F). Thicker lines represent the mean typical reach and shading represents the 99% confidence interval around the mean.

Rehabilitation functionally recruits intact CSNs.

Schematic of intersectional tracing approach to target CSNs for labeling in M2 and M1 of mouse cortex (A). There was a significant main effect of rehabilitation on percentage of cFos+ CSNs in M2, with rehab groups showing an increased colocalization (p=0.00538, linear mixed-effect model). Filled and open circles represent values from CFA– and RFA-traced cells respectively (B). Representative images of M2 CSN and c-Fos labeling. Red arrows indicate a c-Fos-CSN and white arrows indicate a c-Fos+ CSN (C, D). There was a significant main effect of rehabilitation on percentage of c-Fos+ CSNs in M1, with rehab groups showing increased colocalization (p=0.00389, linear mixed-effect model). Filled and open circles represent values from CFA– and RFA-traced cells respectively (E). There was no effect of tracing origin on c-fos activation for both M2 (p=0.142, linear mixed-effect model) and M1 (p=0.079, linear mixed-effect model). Representative images of M1 CSN and c-Fos labeling. Green arrows indicate a c-Fos-CSN and white arrows indicate a c-Fos+ CSN (F, G). Scale bar = 100 microns. Error bars represent +/− SEM. *** = p < 0.001.

Large-scale reorganization of intact corticospinal tract projections in the lesioned and rehabilitated mouse brain.

Representative intact corticospinal tract projections at 0.0 AP, –2.0 AP, –4.0 AP, and –6.0 AP from bregma. High-level organization of mouse brain regions are color-coded: blue = cerebrum, interbrain = green, midbrain = purple, pons = orange, medulla = pink, cerebellum = yellow. Traced projections are shown in black (A). Quantification of normalized projection density in 15 functional subregions in control, uPyX-Rehab, and uPyX+Rehab mice. Closed and open circles represent values from CFA– and RFA-traced CSNs respectively. There was no significant effect of tracing origin on projection density (p=0.2447, linear mixed-effect model). The uPyX+Rehab mice had reduced projection density in the non-motor isocortex compared to control animals (p = 0.005587, linear mixed-effect model). uPyX-Rehab and uPyX+Rehab had increased projection density in the motor medulla compared to control animals (uPyX-Rehab vs. control, p = 0.0129; uPyX+Rehab vs. control, p = 0.000031; linear mixed-effect models). There were no significant differences in the other regions. (B). Correlation between rehabilitation index and plasticity index for spinally-projecting medulla-motor regions in uPyX+Rehab and uPyX-Rehab mice. There was a significant correlation in uPyX+Rehab mice (R2 = 0.161, p = 0.0341, Pearson’s correlation), but not uPyX-Rehab mice (R2 = 0.004, p = 0.762, Pearson’s correlation) (C, D). In spinally-projecting pons-motor regions, there were no significant correlations in uPyX+Rehab (R2 = 0.03, p = 0.399, Pearson’s correlation), nor uPyX-Rehab animals (R2 = 0.002, p = 0.811, Pearson’s correlation). Filled and open circles represent values from CFA– and RFA-traced cells respectively (E, F). Error bars represent +/− SEM. * = p < 0.05, ** = p < 0.01, *** = p <0.001. Thal-sm=Sensory-motor thalamus, Thal-other=Thalamus other, Hyp=Hypothalamus, Mb-mot=Motor midbrain, Mb-sen=Sensory midbrain, Mb-other=Midbrain other, P-mot=Motor pons, P-sen=Sensory pons, P-other=Pons other, My-mot=Motor medulla, My-sen=Sensory medulla, My-other=medulla other, Cb-nuc=Cerebellar nuclei.

Subregion specific reorganization of intact corticospinal tract projections in the medulla of lesioned and rehabilitated mice.

Normalized projection density (NPD) increased in the lateral paragigantoreticular nucleus (LPGi) of uPyX+Rehab mice compared to control mice (t=2.91, p=0.01, Tukey-adjusted post hoc). There was no main effect of tracing origin (t=0.850, p=0.408; linear mixed-effect model) (A). Normalized projection density increased in both uPyX+Rehab (t = 6.67, p < 0.0001, Tukey-adjusted post hoc) mice compared to controls, and uPyX+Rehab mice compared to uPyX-Rehab mice (t = 3.98, p = 0.002, Tukey-adjusted t-test) in the alpha part of the gigantoreticular nucleus (GiA). There was no main effect of tracing origin (t=0.851, p=0.407; linear mixed-effect model) (B). Normalized projection density increased in both uPyX+Rehab (t = 7.11, p < 0.0001, Tukey-adjusted post hoc) mice compared to controls, and uPyX+Rehab mice compared to uPyX-Rehab mice (t = 3.70, p = 0.001706, Tukey-adjusted t-test) in the ventral part of the medullary reticular nucleus (MdV). Normalized projection density also increased in uPyX-Rehab mice compared to controls (t = 3.34, p = 0.004967, Tukey-adjusted t-test). There was no main effect of tracing origin (t=1.493, p=0.157; linear mixed-effect model) (C). Schematic of the rostral medulla (D) and representative images of intact CST density in a control (E), uPyX-Rehab (F), and uPyX+Rehab (G) mouse. Schematic of the caudal medulla (H) and representative images of intact CST density in a control (I), uPyX-Rehab (J), and uPyX+Rehab (K) mouse. The red box indicates the zoomed-in area of interest. Correlations between percentage of injured forelimb footfall recovery and normalized projection density were not significant in the LPGi (R2 = 0.012, p = 0.654, Pearson’s correlation) nor GiA (R2 = 0.087, p = 0.221, Pearson’s correlation) (L, M). Correlation between percentage of injured forelimb footfall recovery and normalized projection density was significant in the MdV (R2 = 0.327, p = 0.0105, Pearson’s correlation). Closed and open circles represent values from CFA– and RFA-traced CSNs respectively (N). Error bars represent +/− SEM. * = p < 0.05, ** = p < 0.01, *** = p < 0.001. Scale bar = 200 microns.