Neurotrophin-3–enhanced Nerve Regeneration Selectively Improves Recovery of Muscle Fibers Expressing Myosin Heavy Chains 2b

8-wk-old inbred male Lewis rats were anesthetized using 0.3 ml/kg of intramuscular Hypnorm (0.315 mg/ml fentanyl citrate and 10 mg/ml fluanisone; Janssen Pharmaceuticals Ltd.) and 2.5 mg/kg of intraperitoneal diazepam (Phoenix Pharmaceuticals Ltd.).

Using an operating microscope (Wild Heerbrugg Ltd.), the left sciatic nerve was exposed 5 mm distal to the sciatic notch via a gluteal muscle splitting incision. A 5-mm length of nerve was removed using sharp microsurgery scissors to produce a gap of 10 mm after retraction of the nerve ends. This gap was grafted with fibronectin impregnated with 500 ng/ml NT-3 (NT) (Sterne et al., 1997); alternatively plain fibronectin mats (FN) or autologous nerve grafts (NG) were inserted as controls.

During insertion, 14-mm-long fibronectin mats were entubulated, the hydroscopic nature of the fibronectin causing them to expand, obliterating the lumen. The proximal and distal nerve stumps were drawn into either end of the entubulated mat to produce a resultant gap of 10 mm and two 10/0 monofilament nylon (Ethicon Ltd., Edinburgh, UK) interrupted sutures were used at each anastomosis. For autologous nerve grafts, a 1-cm length of sciatic nerve was removed, reversed, and reanastomosed to bridge the resulting defect. Animals received postoperative analgesia for 24 h and were kept warm until fully recovered from the anesthetic before their return to the holding room, where they had free access to food and water. On the first postoperative day, the animals were caged in groups of three.

Samples from six animals from each group were harvested at 5, 10, 15, 30, 120, and 240 d postoperatively. The animals were killed and the calf muscles approached by extending the lateral thigh incision beyond the ankle joint. Gastrocnemius muscles were removed bilaterally from each animal. Using an operating microscope, the entire gastrocnemius muscle was carefully cleaned and dissected out, dividing its two tendinous origins and tendinous insertion flush with the bone. Each muscle was then weighed separately to allow calculation of the percentage reduction in muscle mass (denervated muscle weight vs. contralateral muscle weight) after denervation and reinnervation. For the animals harvested at 240 d only, the effects of reinnervation on the gastrocnemius muscle were investigated immunohistochemically. To ensure biopsy of a constant part of the muscle, while holding the knee joint fully extended and the ankle joint flexed to 90°, the midpoint of the gastrocnemius muscle, lying half way between the knee and ankle, was identified and marked with a blue surgical marking pen. The muscles were then harvested as before.

After weighing, a transverse disc of the whole muscle 5 mm long was cut and carefully orientated on a small piece of cork to allow transverse sections to be made. The muscle specimen, embedded in OCT, was rapidly frozen by rapid inversion and plunging into isopentane cooled in liquid nitrogen for about 30 s.

A panel of mouse monoclonal antisera was used for immunostaining, and it included BA-F8 antibodies raised against rat MHC 1, SC-71 antibodies raised against rat MHC 2a, and BF-F3 antibodies raised against rat MHC 2b. All primary antibodies were used at a 1:200 dilution in PBS, pH 7.4. Antibodies to the MHCs (Schiaffino et al., 1989) were supplied by Regeneron Pharmaceuticals, USA.

Serial 8-μm transverse cryostat sections were cut from each muscle block, collected on Vectabond-coated slides (Vector, UK), and air dried for 30 min at room temperature before immunofluorescent staining. The sections were permeabilized by immersion in 0.2% Triton X-100 in PBS (Merck/BDH, UK) for one hour, rinsed three times in PBS, and incubated in 3% goat serum (TCS Biologicals, UK) in PBS to block any background binding from the secondary labeled antibody. Sections were then incubated with the primary antibodies for 24 h at 4°C in a humid chamber, washed in PBS/Tween (0.02%) to remove nonspecifically bound antibody, rinsed three times in PBS, and incubated with cyanine-3–conjugated goat anti–mouse serum (Affinity, UK, diluted 1:100 in PBS). Nonspecifically bound second layer antibodies were again removed by washing with PBS/ Tween, and the sections were then rinsed in PBS and mounted with PBS/ glycerol containing 2.5% (wt/vol) 1,4-diazabicyclo [2.2.2] octane (Aldrich Chemical, Gillingham, UK) as an antifading agent.

All morphometric measurements were performed on coded sections without knowledge of their source. The sections were viewed using an Olympus (Lake Success, NY) microscope connected to a computerized image analysis system (Seescan Analytical Services, Cambridge, UK) appropriately calibrated for each objective magnification. Six fields (10× objective) were chosen at random from each muscle section providing a minimum of 100 positive fibers from each section. After image capture, background subtraction, image enhancement, thresholding, and final editing, automatic counting and measurement of the diameter of each positive fiber was performed using a dedicated calibrated software program.

An assessment of the degree of fiber type distribution was performed by placing a randomly orientated line from an eyepiece graticule across the entire muscle section, and the number of fibers in each group of positive and negative fibers along this line were recorded. 10 random linear transects were made on each muscle, counting a minimum of 900 fibers. These counts were used as a representative indication of the total number of muscle fibers in each whole muscle section and allowed to quantify the proportion, diameter, and size distribution of each fiber type in each experimental group.

Statistical analysis was carried out using the STATA (STATA Corporation, College Station, TX) and SIGMASTAT (Jandel Scientific, Erkrath, Germany) software packages. A one-way analysis of variance (ANOVA) was used to analyze the results of the muscle masses, fiber type distributions, and diameters. The normality assumption of ANOVA was checked using Shapiro-Francia's W′ test and Bartlett's test was used to check the equal variance assumption. If data violated the assumptions of normality and variance they were compared using a Kruskal-Wallis one-way ANOVA on ranks, and data was then expressed as the median values (25th–75th centiles). Differences between individual groups were subsequently isolated by the use of an appropriate all pair-wise multiple comparison procedure. To statistically analyze the size distribution profiles, the 5th and 95th centiles of each fiber type in normal muscle were calculated and the frequencies and percentages of muscle fibers in each group that were smaller, intermediate, or larger than these centiles were then compared using Fisher's exact test, introducing a Bonferroni adjustment to allow for the number of comparisons made.

The mass of both gastrocnemius muscles in each animal was recorded to allow calculation of the percentage atrophy (mass of denervated muscle/mass of normal contralateral muscle × 100) in each group at each time point (Fig. 1). After denervation, initially there was a rapid loss of muscle mass equally in each group up to day 15. By day 30, the muscles in the groups repaired with fibronectin conduits continued to lose muscle mass at an equal rate; however, in the animals repaired with autologous nerve grafts, the reduction in muscle mass was significantly less (P < 0.05). At subsequent time points, the NT group showed steady recovery of the muscle mass, which reached 47% of the weight of contralateral normal muscle at 8 mo. The degree of recovery of muscle mass was higher after NT-3 administration than in the FN group, which reached only 39% normal muscle weight at the final time point. This difference failed to reach significance at 120 d but was significant at 240 d (P < 0.05). The muscle in the NG group returned to 71% of their original mass by day 120 but remained stable at this level by day 240 (P < 0.001 NG vs. NT and FN).

Fibers staining positive for MHC 1 were confined to the deeper third of the contralateral normal muscle, distributed in a mosaic pattern, and were rarely seen in the peripheral superficial two thirds of the gastrocnemius muscle. After reinnervation, the distribution of these groups was much more scattered throughout the muscle: although MHC 1 staining fibers predominated in the deeper parts of the muscle, there were considerable numbers of groups in more superficial regions. There was no obvious difference between experimental groups in the redistribution patterns of type 1 fibers after reinnervation.

In normal muscles, MHC 2a staining fibers had a similar distribution to that of MHC 1 fibers, namely a mosaic distribution within the deep third of the muscle with few fibers in the superficial two thirds. After reinnervation, this pattern was disrupted with groups present throughout the muscle but with a slight preponderance for the deeper parts remaining; again, no obvious intergroup differences were noted.

Staining for MHC 2b shows a fiber distribution mainly in the superficial two thirds of the gastrocnemius muscle and to a lesser extent in the deep muscle regions. The usual mosaic distribution pattern was evident in all areas of the muscle (Fig. 2,a). Reinnervation resulted in significant disruption of the mosaic pattern (Fig. 2 b), but the general preponderance of MHC 2b stained fibers within the superficial two thirds of the muscle persisted; no difference was evident between experimental groups.

From 10 transects across the entire section of each muscle, a minimum of 900 fibers were counted, allowing a comparison of the estimated number of muscle fibers (Table I) and of the percentage distribution of each fiber type in each group (Table II). From these counts it was evident that there was no statistical difference between mean fiber counts from different groups. An interesting difference in fiber proportions was noted for type 2b fibers (Table II). After reinnervation, the percentage of type 2b fibers in the NT group was similar to that in normal muscle and significantly greater than in either of the NG or FN groups (P < 0.05). In contrast, the type 2b proportion fell significantly in the NG and FN groups compared with normal muscle (P < 0.01).

The proportion of type 1 muscle fibers fell significantly from normal levels following denervation and reinnervation (P < 0.02), with no differences between experimental groups in the reinnervated muscles. The proportion of type 2a muscle fibers in all experimental groups rose significantly compared with normal after reinnervation (P < 0.01), without differences between the three groups.

Consistent with the results of percentage distribution of fiber types, morphometric analysis of muscle fiber diameters showed that type 2b fibers regained normal average size after NT-3–mediated reinnervation (Table III). Conversely, in the two control groups the diameter of type 2b fibers was significantly smaller than in muscle of normal and NT groups (P < 0.01). The diameters of type 2a fibers in reinnervated muscles of all groups were significantly smaller than in normal muscles (P < 0.01). This was also the case for type 1 fibers (P < 0.01), with the exception of the NG group, where they recovered normal size.

Analysis of distribution profiles of individual muscle fiber diameters (Fig. 3, a–c) showed that the distribution profile of type 2b fibers for the reinnervated NT group closely resembled that seen in control muscle (P = 1, NT vs. control). The distribution profiles for the NG and FN groups contained more small fibers and fewer large fibers, with a pattern different from both control and NT groups. The difference was particularly evident after reinnervation without NT-3 administration (FN group), where there was a statistically significant difference from both control and NT groups (P < 0.005).

Reinnervated muscle in all experimental groups demonstrated a shift towards small type 1 fibers compared with normal muscle, though a few large fibers were still present. Instead, type 2a fibers showed a shift toward intermediate size fibers, with fewer large and small diameter fibers than normal muscle. For both type 1 and 2a fibers, statistical analysis confirmed a significant shift toward smaller sizes (P < 0.005 for each group vs. normal).

Here we present evidence that NT-3 improves motor target organ reinnervation in terms of recovery of both normal proportions of type 2b muscle fibers and their return to normal diameter. These results are consistent with our previous report that the local delivery of NT-3 to injured sciatic nerves in rats enhances both the rate and amount of axonal regeneration (Sterne et al., 1997).

After denervation, the gastrocnemius muscle was found to lose weight rapidly, falling to ∼30% of its original weight by day 30, consistent with similar results reported previously (Sunderland, 1978). At day 30, muscles in the NG group were significantly less atrophic than in both fibronectin groups. It is likely from the results of the regeneration rate in the NG group (Sterne, G.D., unpublished results) that by day 30 the muscle was already becoming reinnervated. By day 120, the recovery of muscle mass seen after NT-3 administration was improved compared with control FN group, although it was still less than after nerve grafts. The muscle recovery of the NT-3 group was maintained and actually improved with statistical significance by 8 mo. This correlates well with our previously reported myelinated fiber counts distal to the nerve repair in which NT-3–impregnated mats supported a significant improvement in myelinated fibers at 8 mo after repair (Sterne et al., 1997). By comparison, the muscles in the NG group had regained some 70% of their original mass at 4 mo. However, the benefit did not progress, remaining unchanged up to 8 mo postoperatively. This level of recovery is similar to that reported after nerve transection and muscle reinnervation (Gillespie et al., 1987; Bertelli et al., 1995), though nerve crush injury, with the preservation of Schwann cell basal lamina tubes, results in greater muscle recovery (Bertelli et al., 1995).

The mechanism by which NT-3 exerts its effect on muscle is uncertain. NT-3 may act directly on motoneurons, for example by enhancing survival (Eriksson et al., 1994; Lefebvre et al., 1994) or stimulating axonal branching (Schnell et al., 1994; Isaacson et al., 1996). This might explain the increased number of myelinated axons previously observed in NT-3–treated animals (Sterne et al., 1997), which in turn may lead to reinnervation of greater numbers of muscle fibers. Alternatively, NT-3 may be acting via muscle afferent innervation. Recent studies have shown that in transgenic mice, NT-3 overexpression in muscle influences the survival of proprioceptive sensory neurons (Ozaki et al., 1996; Henderson et al., 1996) and increases the number of muscle spindles in gastrocnemius muscle without an overall increase in motoneuron numbers (Wright et al., 1996). Consistently, deletion of NT-3 during development results in a decreased survival of muscle sensory afferents during development (Ernfors et al., 1994; Kucera et al., 1995a,b; Oakley et al., 1995). It may be possible that after nerve injury, the survival or enhanced regeneration effect of NT-3 on this subpopulation of neurons leads to a greater reinnervation of muscle spindles and Golgi tendon organs, consequently resulting in improved sensory feedback from the muscle. This in turn would result in greater use of the muscle and hence lead to improvement in muscle mass.

Our data show that there are specific and highly significant effects of both nerve graft and NT-3 treatment upon type 1 and 2b muscle fibers, respectively. Nerve graft, by the nature of it, offers the regenerating nerve an almost ideal environment because of the presence of activated Schwann cells and a scaffolding of basal lamina to facilitate the progression of regenerating axons. However, treatment by nerve graft normalized type 1 fiber size but did not bring their proportion back to normal, while NT-3 treatment normalized both size and proportion. It should be remembered that we were unable to analyze MHCx/d, and therefore we cannot say if the increase in muscle mass seen in the NG group was also the result of the presence of large and numerous IIx/d fibers.

After reinnervation, morphometric analysis showed that the diameter distribution of type 2b fibers showed a pattern similar to normal only in the NT group. Furthermore, the general trend for the other fiber types in all experimental groups showed a shift to smaller diameter sizes. Analysis of the proportion of different muscle fiber types in each group at 8 mo postoperatively revealed a generalized reduction of type 1 fibers and a significant increase of type 2a fibers. By contrast, the proportion of type 2b muscle fibers was restored to normal only in the NT group. This result is consistent with the enhanced diameters of type 2b fibers in NT-3–treated animals, which uniquely return to the same size as in normal muscle. This result is strongly indicative that NT-3 applied to the cut nerve may influence the regeneration of a physiologically distinct subset of motor neurons supplying type 2b muscle fibers. While functional assessment of NT-3–enhanced reinnervated muscles has yet to be performed, it is known that for reinnervated muscle, as in normal muscle, force generated is proportional to cross-sectional area of muscle fibers (Schantz et al., 1983; Davis et al., 1988; Totosy de Zepetnek et al., 1992) and increase in diameter and proportion of 2b fibers seen in the NT group is likely to result in increased force generation.

In conclusion, we have demonstrated that NT-3 enhancement of peripheral nerve regeneration does have a beneficial biochemical and morphological effect on the reinnervation of motor target organs, supporting the hypothesis of a specific phenotypic effect of growth factor–enhanced nerve regeneration. This is the first indication of a differential effect of NT-3 on neuromuscular interaction in normal adult animals in vivo. It seems doubtful that muscle reinnervation alone may be able to provoke a selective response from the muscle fibers, and it is likely that a more complex signaling system, possibly involving neurotrophic factors, may be involved. This still awaits clarification, but these results also suggest the remarkable possibility that other subpopulations of motoneurons, as classified by the phenotype they confer on muscle fibers, may be differentially responsive to other neurotrophic factors, and this leads us to believe that single or combined application of trophic factors might prove of future therapeutic value.

The authors are grateful to The Geoffrey Burton Trust, Smith's Charity, and Glaxo Holdings for financial support.