Voluntary exercise increases axonal regeneration from sensory neurons

Abstract

Recent advances in understanding the role of neurotrophins on activity-dependent plasticity have provided insight into how behavior can affect specific aspects of neuronal biology. We present evidence that voluntary exercise can prime adult dorsal root ganglion neurons for increased axonal regeneration through a neurotrophin-dependent mechanism. Dorsal root ganglion neurons showed an increase in neurite outgrowth when cultured from animals that had undergone 3 or 7 days of exercise compared with sedentary animals. Neurite length over 18-22 h in culture correlated directly with the distance that animals ran. The exercise-conditioned animals also showed enhanced regrowth of axons after an in vivo nerve crush injury. Sensory ganglia from the 3- and 7-day-exercised animals contained higher brain-derived neurotrophic factor, neurotrophin 3, synapsin I, and GAP43 mRNA levels than those from sedentary animals. Consistent with the rise in brain-derived neurotrophic factor and neurotrophin 3 during exercise, the increased growth potential of the exercise-conditioned animals required activation of the neurotrophin signaling in vivo during the exercise period but did not require new mRNA synthesis in culture.

Figures

Fig. 1.

DRG neurons from exercised animals have increased neurite outgrowth in culture. (A–D) Representative images of DRG cultures (20 h duration) from sedentary (A and B) or 3-day-exercised animals (C and D) that were immunostained for tubulin βIII. The increased neurite length in the 3-day-exercised DRGs required a montage to include the entire neuritic arbor of these neurons. Mean neurite length in DRG cultures from 0-, 3-, and 7-day exercise-conditioned animals is shown in E (error bars = 2 × SEM). Student's t test showed that differences in length of 3 and 7 days vs. 0 and 7 days vs. 3 days were statistically significant (P < 0.0001 and P < 0.01, respectively). Mean neurite length in individual cultures vs. the distance run for individual animals (number of wheel revolutions) is shown in F. Matched sedentary and 3-day-exercised animals are displayed as red circles and sedentary and 7-day-exercised animals as blue squares with neurite length expressed as the percent of average sedentary length for each sample set (error bars = 2 × SEM). Linear regression revealed R2 of 0.855. Spearman's rho correlation coefficient calculation showed a highly significant correlation between distance run and neurite length (rho coefficient = 0.626, P < 0.001).

Fig. 2.

Enhanced neurite outgrowth from exercise-conditioned neurons does not require new mRNA synthesis in vitro. L4–5 DRGs were isolated from sedentary and 7-day-exercised animals and dissociated for culture. The RNA polymerase inhibitor DRB (80 μM) was included in the culture medium to evaluate the requirement for new mRNA synthesis by the exercise conditioned DRGs. After 22 h in vitro, cultures were fixed and stained for neurofilament. Results are displayed as mean neurite length (error bars = 2 × SEM). Student's t test revealed P < 0.0001 for exercise plus DRB cultures vs. sedentary cultures. Comparing exercised neurons cultured with vs. without DRB and sedentary neurons cultured with vs. without DRG showed no statistically significant differences (P > 0.05).

Fig. 3.

Exercise-conditioned animals show increased nerve regeneration after crush injury. Sedentary and 7-day-exercised animals were subjected to sciatic nerve crush. Three days after crush, the nerve was transected at 8 mm distal to the crush site and was exposed to FluoroGold for additional 2 days. FluoroGold-positive neurons were more abundant in the L4–5 DRGs from the exercised animals (B) than in those from the sedentary animals (A). Costaining for neurofilament antibody showed similar numbers of neurons in these sections (C and D). Mean number of retrogradedly labeled neurons (FluoroGold-positive, left axis) and total neurons (neurofilament-positive, right axis) are presented of three serial sections from three animals for each condition in E (error bars = SD). (G and H) Representative sections of sciatic nerve distal to the crush site that were stained for neurofilament (green) and counterstained with 4′,6-diamidino-2-phenylindole (blue). Sections of uninjured sciatic nerve taken at a distal thigh level are shown for comparison (F). Only a few axons were discernable in the sections of crushed nerve from the sedentary animals (arrow, H). (Scale bars: 50 μmin A–D and 20 μmin F–H).

Fig. 4.

Exercise increases mRNA levels in the DRG. Levels of BDNF, NT3, synapsin I, and GAP43 mRNAs in the L4–5 DRG were measured by real-time RT-PCR. All values were normalized for the amplification of GAPDH. BDNF, NT3, synapsin I, and GAP43 mRNA levels are expressed as a percentage relative to the 0-day-exercised animals (error bars = standard SD for ≥5 animals per condition). BDNF, NT3, synapsin I, and GAP43 each showed statistically significant differences when comparing the 3- and 7-day with the 0-day samples with P < 0.05 (*)or <0.01 (**) by ANOVA.

Fig. 5.

Inhibition of Trk receptors before exercise blocks exercise-induced growth and synapsin I mRNA increase. (A)L4–5 DRGs were pretreated with K252A (gray columns) or CytC (white columns) before exercise as described in Materials and Methods. After a 3-day exercise period, L5 DRGs were cultured for 20 h. Mean neurite length is presented (error bars = 2 × SEM). The difference between CytC-exercised DRG vs. K252A-exercised DRG is statistically significant (**, P < 0.0001 by Student's t test). Although mean neurite length in the K252A exercised DRG is closer to that of the sedentary conditions, the difference between exercise and sedentary K252A-treated DRGs is statistically significant (*, P < 0.01 by Student's t test). (B and C) RNAs from the L5 DRGs of the sedentary and exercised animals from A were used as a template for real-time RT-PCR to evaluate levels of synapsin I (B) and GAP43 (C) mRNAs. Signals were normalized to GAPDH levels and are expressed as the relative to the CytC-injected, sedentary signals (error bars = SD for six animals per condition). Synapsin I showed statistically significant differences when comparing the exercise CytC with the exercise K252A samples (*, P < 0.01 by ANOVA). Differences in GAP43 mRNA levels between CytC and K252A exercise samples were not statistically significant (P > 0.05).