An Exercise Mimetic Approach to Reduce Poststroke Deconditioning and Enhance Stroke Recovery

Evidence supports early rehabilitation after stroke to limit disability. However, stroke survivors are typically sedentary and experience significant cardiovascular and muscular deconditioning. Despite growing consensus that preclinical and clinical stroke recovery research should be aligned, there have been few attempts to incorporate cardiovascular and skeletal muscle deconditioning into animal models of stroke. Here, we demonstrate in rats that a hindlimb sensorimotor cortex stroke results in both cardiovascular and skeletal muscle deconditioning and impairments in gait akin to those observed in humans. To reduce poststroke behavioral, cardiovascular, and skeletal muscle perturbations, we then used a combinatorial intervention consisting of aerobic and resistance exercise in conjunction with administration of resveratrol (RESV), a drug with exercise mimetic properties. A combination of aerobic and resistance exercise mitigated decreases in cardiovascular fitness and attenuated skeletal muscle abnormalities. RESV, beginning 24 hours poststroke, reduced acute hindlimb impairments, improved recovery in hindlimb function, increased vascular density in the perilesional cortex, and attenuated skeletal muscle fiber changes. Early RESV treatment and aerobic and resistance exercise independently provided poststroke benefits, at a time when individuals are rapidly becoming deconditioned as a result of inactivity. Although no additive effects were observed in these experiments, this approach represents a promising strategy to reduce poststroke behavioral impairments and minimize deconditioning. As such, this treatment regime has potential for enabling patients to engage in more intensive rehabilitation at an earlier time following stroke when mechanisms of neuroplasticity are most prevalent.

climb until rats reached exhaustion. The weight carried prior to exhaustion was recorded as the new maximal lifting capacity. Maximal lifting capacity was determined every 4 exercise sessions.
Maximal exercise test. Rats were habituated to the enclosed treadmill over 5 days by incrementally increasing the workload and duration of time spent in the chamber. Initially, a steel grid at the back of the treadmill delivered an electric stimulus (0.2mA, 1 stimulus of 200ms every second) to encourage running.
At the conclusion of each session, rats were rewarded with dark chocolate (100mg). Over time, the electrical stimulus was no longer required and was turned off to minimize stress. To calculate VO2 peak, rats acclimated to the treadmill chamber for 5 minutes, followed by a warmup period of 5 minutes at 15m/min and 0% incline. Subsequently, the incline was increased to 15% and treadmill speed was increased 2m/min every 2 minutes until exhaustion. The criteria for reaching VO2 peak was a leveling-off of oxygen consumption despite an increase in workload, a respiratory exchange ratio of 1.04 or higher, or when the rat could no longer maintain a sufficient pace (remaining at the back for longer than 3 seconds).
Ambient air was delivered to the chamber at a rate of 4.5ml/min and chamber air (200ml/min) was sampled every 30 seconds by oxygen and carbon dioxide sensors.
Vascular density quantification in peri-lesional cortex. Z-projections were created from approximately thirty optical sections imaged at 20x magnification and spaced 0.5 microns apart. A Gaussian smoothing filter with a standard deviation of 0.5 microns was applied to z-projection images. Blood vessels were then identified using a local adaptive thresholding operation. Pixels were classified as a blood vessel if their intensity was larger than the average intensity inside a circular region centered at the pixel. The radius used for the circular region was 20 microns. The resulting image contained only two colors: white, representing blood vessels, and black, representing the background. White connected components smaller than 100 microns² were removed from the image. The medial lines of the blood vessels were then identified using the Palágyi-Kuba thinning procedure. Pixels in a medial line having only one neighbor were associated with termination points and pixels with three or more neighbors were associated with bifurcation points. 1 The blood vessels were partitioned into a set of segments, where a segment was defined as a blood vessel between two termination or bifurcation points. Segments smaller than 3 microns were iteratively removed from the image.

Muscle fibres and CD31 labeled capillaries quantification.
To identify each muscle fibre type and CD31 labeled capillaries, the neural network software Ilastik (www.ilastik.org) was used to produce pixel segmentation maps that were a binary representation of each fluorescent channel ( Figure 1, supplemental materials). Prior to batch processing of images, Ilastik was manually trained to differentiate each muscle fibre type (I, IIa, IIb) and capillaries (CD31) in 10-15 images of plantaris muscle cross sections. To validate the accuracy of Ilastik to detect each fibre type, a separate 50 cross sectional images of plantaris muscle were manually counted and compared to Ilastik segmentations ( Figure 2C, supplemental materials). Validation was also completed to establish the accuracy of Ilastik to delineate the crosssectional area of each fibre type ( Figure 2D, supplemental materials). In 68 plantaris muscles a 700µm x 700µm grid was placed over fluorescence images and within every 32nd counting grid each fibre type was manually traced. Manual tracing of the cross-sectional area of each fibre type within the grid was compared to the accuracy of Ilastik to segment out the same muscle fibres. To quantify CD31 labeled capillaries, a similar process was conducted ( Figure 2E, supplemental materials). A 700µm x 700µm counting grid was placed on fluorescence images from 70 plantaris muscle cross-sections and within every 32nd counting grid each capillary was manually counted and compared to Ilastik segmentations.
Quantification of pixel segmentation maps to determine skeletal muscle fibre counts, cross-sectional area (CSA), and vascular density was conducted using a custom ImageJ (NIH, USA) script (found in supplemental materials; MuscleFiberTyping.txt).

Dose-response pilot experiment.
There was a trend that 5mg/kg of Resv reduced the basal RER of rats in stroked rats (p=0.055; see supplemental Figure 1). PGC1a protein content was increased in rats that had a stroke and received 5mg/kg of Resv compared to sham (p<0.05; see supplemental Figure 1C).
Similarly, there was a trend that PGC1a protein content was increased in rats that had a stroke and received 5mg/kg of Resv compared to vehicle-treated rats (p=0.053).