Adaptation of Slow Myofibers: The Effect of Sustained BDNF Treatment of Extraocular Muscles in Infant Nonhuman Primates

Abstract

Purpose.: We evaluated promising new treatment options for strabismus. Neurotrophic factors have emerged as a potential treatment for oculomotor disorders because of diverse roles in signaling to muscles and motor neurons. Unilateral treatment with sustained release brain-derived neurotrophic factor (BDNF) to a single lateral rectus muscle in infant monkeys was performed to test the hypothesis that strabismus would develop in correlation with extraocular muscle (EOM) changes during the critical period for development of binocularity.

Methods.: The lateral rectus muscles of one eye in two infant macaques were treated with sustained delivery of BDNF for 3 months. Eye alignment was assessed using standard photographic methods. Muscle specimens were analyzed to examine the effects of BDNF on the density, morphology, and size of neuromuscular junctions, as well as myofiber size. Counts were compared to age-matched controls.

Results.: No change in eye alignment occurred with BDNF treatment. Compared to control muscle, neuromuscular junctions on myofibers expressing slow myosins had a larger area. Myofibers expressing slow myosin had larger diameters, and the percentage of myofibers expressing slow myosins increased in the proximal end of the muscle. Expression of BDNF was examined in control EOM, and observed to have strongest immunoreactivity outside the endplate zone.

Conclusions.: We hypothesize that the oculomotor system adapted to sustained BDNF treatment to preserve normal alignment. Our results suggest that BDNF treatment preferentially altered myofibers expressing slow myosins. This implicates BDNF signaling as influencing the slow twitch properties of EOM.

Childhood strabismus is a disorder characterized by a misalignment of the eyes. Left untreated, such misalignment can impair development of normal binocular visual sensitivity. Strabismus is a common disorder, affecting 3% to 5% of children worldwide.^1,2 If one considers only nonparalytic, nonanisometropic strabismus, the etiology of the majority of the cases of childhood strabismus is not understood. In addition, it is difficult to produce strabismus in a normally binocular infant nonhuman primate without significant visual or motor perturbations. The most effective methods for producing strabismus in nonhuman primates have used sensory deprivation in the form of prism goggles³ or alternating monocular occlusion⁴ during the first 3 months of life – the critical period for the development of binocularity.^5,6 Additionally resection/recession surgery also has resulted in strabismus in infant-nonhuman primates.^7–10 However, surgery disturbs the normal muscle and connective tissue relationships within the orbit,¹¹ and could make this model of strabismus problematic for studies directed at causal mechanisms and the development of new treatment approaches. 

We have been studying the use of neurotrophic factors and the changes they induce in extraocular muscle (EOM) function and structure. Neurotrophins are signaling molecules that can have significant influences on skeletal muscle and motor neurons, inducing changes in muscle growth, progenitor cell proliferation, and nerve outgrowth.^12,13 Our working hypothesis is that sustained treatment of an EOM with neurotrophic factors would produce functional alterations in the EOM and their innervating motor neurons. These alterations could be manifested by adaptations in the treated muscle as well as adaptations in either yoked and/or antagonist/agonist muscle pairs. One such neurotrophic factor is insulin-like growth factor I (IGF-1), which has direct effects on EOM and is known to be retrogradely transported.¹⁴ Previous work in adult rabbits showed that sustained treatment with IGF-1 of an EOM resulted in significantly increased muscle size and force generation.¹⁵ Similar results were shown after single injections of IGF-1 into the orbits of chicks.¹⁶ These studies support the hypothesis that modulating neurotrophic factor levels in the periphery could have significant functional effects relative to eye position and eye movements. We recently showed that unilateral treatment with sustained release of IGF-1 to one medial rectus muscle in newborn infant nonhuman primates produced a strabismus.¹⁷ Interestingly, beyond changes in myofiber cross-sectional area after the 3 months of sustained IGF-1 treatment, there were no significant changes in neuromuscular junction size or amount of innervating nerve¹⁷ despite the development of strabismus. 

Insulin-like growth factor I is just one of a large number of neurotrophic factors that are expressed in motor neurons and/or skeletal muscle and have the potential to modulate ocular motor function. Another important neurotrophic factor is brain-derived neurotrophic factor (BDNF), which is found in skeletal muscle, Schwann cells, and neuromuscular junctions.^18,19 Brain-derived neurotrophic factor has been shown to promote neurite outgrowth specifically in developing oculomotor and other brainstem neurons.^18,20 Also, BDNF has important effects on neuromuscular junction maturation^19,21 and function.^22,23 In addition, when applied to an axotomized abducens nerve in adult cats, BDNF was shown to be neuroprotective, and it also promoted reinnervation by afferents onto the abducens motor neurons. These effects were associated with restoration of the tonic firing properties of the abducens motor neurons.^24,25 Complementary findings provided further compelling evidence supporting a role for BDNF in signaling to the visual ocular motor system. For example, BDNF was shown to be able to reactivate ocular dominance plasticity in adult visual cortex in cases of monocular deprivation during early development.^26,27 

The present study examined the potential efficacy of 3 months of unilateral sustained BDNF treatment of a lateral rectus muscle in infant monkeys in an attempt to produce a strabismus. We also examined the expression of BDNF in normal EOM. Eye alignment was determined at the end of the study. Myofiber size changes as compared to EOM from normal age-matched control monkeys were determined. Because BDNF has diverse signaling roles at the neuromuscular junction, including promotion of neurite outgrowth,²⁰ maturation of neuromuscular junctions,²⁸ and potentiation of neuromuscular transmission,^22,29 we analyzed the effect of sustained BDNF treatment on innervational density, neuromuscular junction localization, and size. In contrast to leg skeletal muscle, EOM has a continuum of myofiber types that vary between the orbital and global layer, and from the proximal to distal ends of the muscle.^30–34 Within this broad range of properties, myofibers can be coarsely divided into myofibers expressing slow or fast MyHC isoforms, which are related to the type of neuromuscular junctions and myofiber shortening velocity.^30,35–37 Fast myofibers predominately have centrally located en plaque endings, and slow fibers have smaller en grappe endings along their length, although both myofibers types can be multiply innervated and have both types of neuromuscular junctions.^38–40 To simplify the complexity of EOM for analysis, we stained EOM fibers for slow myosin to help differentiate between neuromuscular junctions on these two broad myofiber types. 

Methods

Animals and Surgery for Sustained BDNF Treatment

Infant monkeys were obtained from the breeding colony at the Washington National Primate Research Center at the University of Washington. Normal adult C57BL6 mice were obtained from Harlan (Indianapolis, IN, USA). All experiments received approval from the Animal Care and Use Committee at the University of Washington, and were performed in compliance with guidelines for use of animals in research from both the National Institutes of Health and the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research. Normal human EOM were obtained either at autopsy or by the eye bank during enucleation. All procurement was approved by the IRB, and this research complied with the Declaration of Helsinki. 

Two Macaca nemestrina infant monkeys aged 1 to 2 weeks received unilateral treatment of BDNF to one lateral rectus muscle with a pellet prepared to provide the sustained release of 2 μg BDNF per day for 90 days. Under general anesthesia, an incision was made in the lateral conjunctiva to visualize the lateral rectus muscles. A small muscle hook was used to lift the muscles at their point of insertion to allow a sustained release pellet (Innovative Research of America, Sarasota, FL, USA) to be placed on the surface of one lateral rectus muscle. This dose was selected based upon previous studies of sustained neurotrophic factor treatment to monkey EOM,⁴¹ and on previous studies using anterograde delivery of BDNF to motor neurons.^25,42 After pellet implantation was performed, the conjunctiva was closed using 8-0 ophthalmic suture. 

At 3 months after pellet implantation, the corneal light reflex was used to determine eye alignment.⁴³ After deep sedation with ketamine, the monkeys were euthanized with an IACUC approved overdose of barbiturate anesthesia by a staff veterinarian. At this time, pellet placement was verified to be in location on the treated lateral rectus muscles. All 12 EOMs were dissected from origin to insertion, embedded in tragacanth gum, and frozen in 2-methylbutane chilled to a slurry on liquid nitrogen. Tissue also was collected from two additional age-matched control monkeys with normal alignment, which also were used as controls in a complementary report.¹⁷ The tissue blocks were stored at −30°C until processed. 

Western Blot for Tyrosine Kinase B Receptor (TrkB) Protein

Due to the difficulty of obtaining control nonhuman primate infant EOMs and brainstem, and the need to process the treated tissues for histological analysis, whole tissue lysate samples of mouse EOM and brain were obtained to examine normal protein levels of TrkB, a receptor for BDNF.⁴⁴ The collected mouse tissues were homogenized on ice in N-PER neuronal protein extraction reagent (Thermo Scientific, Waltham, MA, USA) with Complete Protease Inhibitor Cocktail (Hoffmann-La Roche, Basel, Switzerland). Homogenized samples were centrifuged at 8519g for 10 minutes. Supernatants were collected, and protein concentration was determined by bicinchoninic acid (BCA) protein assay (Thermo Scientific). Each lane received 50 μg protein, which was separated on 12% mini-PROTEAN TGX gels (Bio-Rad, Hercules, CA, USA), and transferred to nitrocellulose. Blots subsequently were blocked in Odyssey blocking buffer (LI-COR Biosciences, Lincoln, NE, USA) for 1 hour at room temperature, and then incubated overnight at 4°C in a primary antibody to TrkB (1:100, Santa Cruz Biotechnology, Santa Cruz, CA, USA), and to GAPDH (1:10,000; Meridian Life Sciences, Memphis, TN, USA) for a loading control. The following day, blots were washed in TBST (1× Tris-buffered saline, 0.1% Tween 20), and then incubated for 30 minutes at room temperature in the secondary goat anti-mouse-IR700, and goat anti-rabbit-IR800 (1:1,000; Rockland, Limerick, PA, USA). The blot was imaged with an Odyssey Infrared Imaging System (LI-COR Biosciences). Densitometry values were obtained using Odyssey instrument software. 

Immunohistochemistry and Histological Processing

The EOMs from the infant monkeys were sectioned frozen at a thickness of 30 μm using a Leica cryostat (Leica, Wetzlar, Germany), and the sections were stored at −80°C until processed. Every 30th section was stained with hematoxylin and eosin (H and E) to be used for myofiber area measurements. For visualization of neuromuscular junctions, every 40th section was washed in 0.01 M PBS, pH 7.4 containing 0.1% Triton X-100 (PBS/TX), blocked in 10% goat serum for 1 hour, and incubated overnight at 4°C with an antibody to the slow heavy myosin chain isoform (MyHC; Vector Laboratories, Burlingame, CA, USA) or to synaptophysin (1:300; Abcam, Cambridge, MA, USA). Sections were washed in PBS, followed by incubation in secondary antibody, goat anti-mouse IgG labeled with Cy3 (1:1000; Jackson ImmunoResearch Laboratories, West Grove, PA, USA) for 1 hour at room temperature. To visualize neuromuscular junctions, sections were double labeled by incubation with α-bungarotoxin conjugated to AlexaFluor488 (1:3000; Molecular Probes, Eugene, OR, USA) overnight. A subset of sections were double or triple labeled for nerves by incubation in an antibody to neurofilament (1:1000, smi-31; BioLegend, Dedham, MA, USA) overnight, and visualized with goat anti-mouse DyLight 405 or donkey anti-mouse AlexaFluor488 (1:1000; Jackson ImmunoResearch Laboratories). 

Using the above procedures, human inferior oblique specimens were immunostained for TrkB (1:50 sc-8316; Santa Cruz Biotechnology), a receptor for BDNF, as the antibodies we tested did not work with the monkey tissues. Human tissue from three to four areas throughout each control muscle was immunostained for BDNF (1:100; Abcam; 1:100 sc-546; Santa Cruz Biotechnology; 1:50; Promega, Madison, WI, USA) to assay neurotrophin and receptor distribution through the EOM. Sections were visualized with the secondary antibodies goat anti rabbit Rhodamine Red or donkey anti sheep Cy3 (Jackson ImmunoResearch Laboratories). To further characterize BDNF distribution, BDNF was doubled labeled with either α-bungarotoxin conjugated to AlexaFluor488 as described above, or with a variety of myosin isoforms using the above procedure. Antibodies against myosins from the Developmental Studies Hybridoma Bank (Iowa City, IA, USA) were: type 2X MyHC (1:20), embryonic MyHC (1:40), slow MyHC (1:1000), and neonatal MyHC (1:20). Antibodies obtained from Abcam were α-cardiac MyHC (1:10) and 2a MyHC (1:100). The MyHC isoforms were visualized with labeling by the secondary antibody donkey anti mouse AlexaFluor488 (Jackson ImmunoResearch Laboratories). All sections were washed and mounted on glass slides and coverslipped with Vectashield (Vector Laboratories). 

Image Processing and Analysis

For calculation of mean myofiber cross-sectional areas, the areas were manually entered into our morphometry system (Bioquant, Nashville, TN, USA) using the H and E stained sections. A minimum of 200 myofibers were measured in both the orbital and global layers. 

Immunostaining for BDNF, TrkB, and MyHC expression was visualized using a confocal scanning microscope at ×40 oil immersion when sections were colabeled with neuromuscular junctions or nerve, or was visualized with an epifluorescence microscope at ×20 for all other cases. Representative fields were chosen for visualization. 

For visualization of nerve and neuromuscular junctions, slides were imaged with scanner laser confocal microscopy (FV1000; Olympus Corporation, Tokyo, Japan) to sequentially capture images in the same focal plane with different filters. Neuromuscular junctions were imaged at ×20 oil immersion. Every 40th section was imaged for analysis. With a random starting point, every third field throughout the entire muscle specimen was taken. Additionally, every fifth field was imaged at ×60 oil immersion to capture neuromuscular junction morphology at higher power. 

Z-stacks were collapsed for each field, and the following measurements were taken using the analysis software ImageJ (National Institutes of Health [NIH], Bethesda, MD, USA; available in the public domain at http://imagej.nih.gov/ij): muscle area, mean diameter for myofibers positive or negative for slow MyHC, total number of neuromuscular junctions, and the area of each neuromuscular junction endplate as determined by hand circling. For every fiber with a neuromuscular junction, it also was noted if the fiber was positive or negative for slow myosin. For each field, the muscle layer (orbital, global) and region (proximal, endplate zone, distal) were noted. Neuromuscular junctions were categorized based on slow myosin expression, endplate area, and properties to distinguish between en grappe and en plaque endplates. Additional staining among control and experimental muscle with double-labeling of SNAP-25 (1:1000; Covance, Dedham, MA) or synaptophysin (1:300; Abcam), nerve (smi-31), and α-bungarotoxin revealed that all α-bungarotoxin labeled neuromuscular junctions in the EOM specimens colocalized with presynaptic components. Based on this, we are confident that the majority of α-bungarotoxin positive neuromuscular junctions in our samples were active endplates. For determination of neuromuscular junction area, only fibers with complete neuromuscular junctions were included. Counts were conducted masked to treatment. Data are presented as mean ± SD. 

Results

Analysis of Eye Alignment

Because of the evidence that BDNF is a target-derived neurotrophin that signals to oculomotor neurons,^18,24,25 we hypothesized that sustained unilateral BDNF delivery to infant monkeys EOMs during the critical period for development of the binocular system would perturb oculomotor development and maturation, resulting in strabismus. However, sustained delivery of BDNF for 3 months to one lateral rectus muscle did not result in misalignment of eye position at the end of treatment. Corneal light reflex photographs of monkeys after 3 months of sustained 2-μg/day BDNF treatment demonstrated that one monkey had a possible microstrabismus of 8° (Fig. 1A), while the second treated monkey had normal alignment, <5° (Fig. 1B). 

Figure 1

 

Corneal light reflex photographs of two monkeys after 3 months of unilateral sustained 2 μg/day BDNF treatment. One monkey (A) had a possible microstrabismus of 8°, while the second treated monkey (B) had normal alignment, <5°.

Effect of BDNF on Neuromuscular Junctions

When all myofibers from the BDNF-treated lateral rectus muscles were examined, there was no change in neuromuscular junction density (data not shown). Interestingly, there was a prominent trend towards larger neuromuscular junctions on slow myofibers (Figs. 2A, 2B, 2F, 2G, 3A) with average end palate area 35% larger than untreated controls. Mean areas were 65.6 ± 8.9 in control untreated lateral rectus and 92.6 ± 1.5 μm² in the BDNF-treated lateral rectus muscles. This trend was most pronounced in the endplate and distal regions of the global layer (Fig. 3C), and proximal and distal region of the orbital layer (Fig. 3E). When neuromuscular junction size was normalized to the diameter of the myofibers, the trend persisted, suggesting this effect was not due to changes in neuromuscular junction area that was just in proportion to the increased myofiber size. In contrast, there was no observed difference in the area (Fig. 3B), length, or complexity of en plaque neuromuscular junctions on fast myofibers (Figs. 2C–E, 2H–J). 

Figure 2

 

Confocal images of representative neuromuscular junctions immunostained with an antibody to slow MyHC isoform (red) and α-bungarotoxin (green) to visualize the postsynaptic component of neuromuscular junctions. Compared to control muscle (A, B), neuromuscular junctions on slow myofibers in BDNF-treated muscle had larger endplates (F, G). No change was observed in the size or morphology of en plaque neuromuscular junctions on fast myofibers between control (C–E) and BDNF-treated (H–J) muscles. Scale bar for all photomicrographs (shown in [J]): 25 μm.

Figure 3

 

Brain-derived neurotrophic factor treatment resulted in large endplates on slow myofibers. (A) Histogram of the area of all quantified neuromuscular junctions on slow myofibers. For the two control subjects, 1861 and 2170 neuromuscular junctions were counted. For the two BDNF-treated lateral rectus muscles, 797 and 865 neuromuscular junctions were counted. (B) Histogram of the area for individual en plaque neuromuscular junctions on fast myofibers. Quantification of average neuromuscular junction area (C, E) or the average area normalized to myofiber size (D, F) in the global (C, D) and orbital (E, F) layer for neuromuscular junctions on slow myofibers. Note that even when compensating for differences in myofiber size, the trend of larger neuromuscular junctions persisted.

Effect of BDNF on Myofiber Size

Because of the demonstrated role of BDNF in altering neuromuscular junctions on slow myofibers,²⁸ and the expression of BDNF in satellite cells in diaphragm and limb skeletal muscle,⁴⁵ we also investigated if BDNF treatment had an effect on the size of myofibers. Complementary to the effect we observed on neuromuscular junctions, BDNF treatment resulted in larger diameters in slow myofibers (Figs. 4, 5A) compared to control lateral rectus muscles. This trend was most dramatic in the endplate region of the global layer (Figs. 4B, 4E, 5C), where slow myofibers had a smaller fiber diameter than elsewhere along the length of the muscle. In contrast, no difference between the fiber diameters of fast myofibers was observed between experimental and control lateral rectus EOM (Fig. 5B) in either the global (Fig. 5D) or orbital (Fig. 5F) layer. 

Figure 4

 

Confocal images of representative fields immunostained with an antibody to slow MyHC isoform (red) and α-bungarotoxin (green). (A–F) Myofibers from the global layer, from control (A–C) and BDNF-treated (D–F) muscle. Fields are shown from the proximal (A, D), endplate (B, E), and distal (C, F) regions of the muscle. Note the larger slow myofiber diameter particularly in the endplate region of BDNF-treated muscle. (G–L) Myofibers from the orbital layer, from control (G–I), and BDNF-treated (J–L) muscle. Fields are shown from the proximal (G, J), endplate (H, K), and distal (I, L) regions of the muscle. Scale bar: 25 μm.

Figure 5

 

Mean diameter of myofibers positive for slow MyHC was larger in BDNF-treated EOM. (A). Histogram of the mean slow myofiber diameter from all measured fields. (B) Histogram of the mean fast myofiber diameter from all measured fields. (C, E) Average myofiber diameter of myofibers positive for slow myosins in the global (C) and orbital (E) layer. Myofibers were larger in diameter with BDNF treatment, most prominently in the endplate region of the global layer. (D, F) Average myofiber diameter of myofibers negative for slow myosin in the global (D) and orbital (F) layers. No change in myofiber diameter was seen.

Treatment with BDNF resulted in larger diameters of myofibers expressing the slow MyHC isoform compared to the untreated control lateral rectus muscles. However, we hypothesized that sustained BDNF treatment may cause compensatory adaptation in the functionally paired, but untreated EOMs. For this analysis, myofiber areas were determined on H and E–stained sections of all the medial and lateral rectus muscles from the control and BDNF-treated animals (Fig. 6). Mean myofiber cross-sectional area throughout the entire EOM suggests that the myofibers from the treated lateral rectus muscles were larger compared to control lateral rectus muscles. Myofibers were 60% larger in the orbital layer, going from 123.5 ± 26.2 to 233.1 ± 23.7 μm², and 48% larger in the global layer, from an average myofiber area of 276.3 ± 110.5 to 452.2 ± 23.9 μm² in the treated lateral rectus. However, changes in area were not just seen in the treated lateral rectus muscles, but also in the ipsilateral antagonist medial rectus to the treated lateral rectus (Fig. 6). The untreated antagonist medial rectus myofibers were 31% larger in the orbital layer (143.2 ± 11.5 μm² compared to 197.9 ± 35.1 μm²), and 15% larger in the global layer (329.8 ± 42.2 μm² compared to 382.4 ± 56.4 μm²) as compared to control medial rectus EOM. These results are consistent with our prior findings and suggest that the developing ocular motor system has the ability to direct adaption to perturbation to one of its muscles in a coordinated manner in untreated muscles in order to preserve eye alignment. 

Figure 6

 

Mean myofiber cross-sectional area of the treated medial rectus muscles compared to the antagonist MR, agonist MR muscle, and normal age-matched control muscles.

Effect of BDNF on Slow Myofiber Size and Density

Complementary to the pronounced effect of BDNF on fiber diameter and neuromuscular junction size on slow myofibers, the percentage of myofibers expressing slow myosin in the proximal region of the muscle also was higher in BDNF treated lateral recti in the global and orbital layer (Fig. 7). The percentage of myofibers positive for slow myosin was larger by 35%, from 20.6% ± 0.3% in control lateral rectus to 29.5% ± 4.5% in BDNF-treated lateral rectus. 

Figure 7

 

Quantification of the percentage of myofibers that express slow myosin in the global (A) and orbital (B) layers. Myofibers in the proximal region had a larger percentage of slow myofibers in BDNF-treated muscle.

Expression of BDNF in EOM

Treatment with BDNF caused pronounced changes in the slow myofibers of treated EOM. We hypothesized that BDNF would be preferentially localized to slow myofibers in normal EOM. Immunostaining with three different BDNF antibodies in human and monkey EOM confirmed localization of BDNF to EOM myofibers. Brain-derived neurotrophic factor appeared diffusely throughout the cytoplasm in many myofibers in the proximal and distal regions of the muscles, but was preferentially expressed in some myofibers over others, creating a mosaic pattern in monkey (Figs. 8A, 8B, 8D, 8F) and human (Fig. 9A) muscle specimens. In some regions, BDNF was most intensely localized to the outer membrane of myofibers (Figs. 8C, 8D, 8E, 9C). Brain-derived neurotrophic factor expression was often slightly stronger in the orbital layer as opposed to the global layer (Fig. 9). Toward the midbelly of the muscle, BDNF immunoreactivity was drastically reduced, with less staining in the cytoplasm and outer cell membrane (Figs. 8G–I). 

Figure 8

 

Confocal images of representative fields immunostained with antibody to BDNF in control monkey lateral rectus. (A–C) The proximal region of EOM had pronounced immunolabeling for BDNF. Myofibers often had diffuse immunoreactivity throughout the cytoplasm, with intensity that varied between myofibers. Note the labeling of presumed nerve fibers as well (A, B). In some proximal regions, BDNF was highly expressed at the cell membrane (C). (D–F) The distal region of EOM had intense immunolabeling for BDNF. Myofibers had variable labeling for BDNF within the cytoplasm (E, F). For some distal regions, staining was very prominent at the outer membrane of myofibers (D, E). (G-I) Brain-derived neurotrophic factor immunoreactivity was less positive in the midbelly and endplate zone of the EOM. (G) Brain-derived neurotrophic factor labeling was very light or absent in the endplate zone. (H) Neuromuscular junctions were labeled with α-bungarotoxin (green in [I]) to mark the endplate zone. (I) Merged image to show an absence of BDNF staining. Note that for this field, there also was no BDNF labeling at the neuromuscular junction. Scale bar: 75 μm.

Figure 9

 

Brain-derived neurotrophic factor was highly expressed in the orbital layer in proximal and distal regions in EOM. (A) Representative immunolabeling of BDNF in adult human inferior oblique muscle. More pronounced labeling was visible in the orbital layer (asterisk) compared to global layer. Varied labeling in the cytoplasm of myofibers was present, as well as labeling at the outer cell membrane of a subset of myofibers. (B, C) Representative fields from infant monkey lateral rectus muscle from the orbital layer. (B) Note the more prominent BDNF labeling in the orbital layer (asterisk). Scale bar: 75 μm.

The mosaic pattern of BDNF immunoreactivity in the distal and proximal ends of the EOM suggested that BDNF expression could be related to myofiber type. To test if BDNF expression was related to fiber MyHC composition, lateral rectus muscle was colabeled by immunostaining for six different MyHC isoforms. The EOM express nine different myosin isoforms, which are often coexpressed, and vary in expression along a fiber's length.^30,32–34 In the distal and proximal regions of the muscle, BDNF often was colocalized with the fast MyHC type2x and type2a myofibers (Figs. 10A, 10B). Less colocalization was present with the “developmental” myosins, embryonic (Fig. 10C) or neonatal (Fig. 10D) MyHC. The least overlap was observed with α-cardiac MyHC (Fig. 10F), and surprisingly, there was little colocalization with myofibers expressing slow MyHC (Fig. 10E). 

Figure 10

 

Representative images of coexpression of BDNF (center, red in merged image) with different myosin heavy chain isoforms (left, green in merged image) in control monkey lateral rectus. (A) Brain-derived neurotrophic factor was highly colocalized with 2x MyHC. (B) Brain-derived neurotrophic factor was highly colocalized with 2a MyHC. (C) Some BDNF-positive myofibers also were positive for embryonic MyHC (arrow), but this coexpression pattern was inconsistent. (D) The pattern of BDNF expression was not related to the expression of neonatal MyHC. (E) Few myofibers positive for slow MyHC were positive for BDNF. (F) Few myofibers positive for BDNF were also positive for α-cardiac MyHC. Scale bar: 75 μm.

Brain-derived neurotrophic factor was localized to neuromuscular junctions (Fig. 11) and nerve (Fig. 12) as visualized with double labeling with α-bungarotoxin and an antibody to neurofilament. However, not all endplates had prominent BDNF immunolabeling (Figs. 8I, 11), and BDNF was not expressed in all nerve fibers. No clear trend was observed in BDNF intensity of immunostaining in the neuromuscular junctions and nerve in the proximal to distal extent of the muscles examined (not shown). 

Figure 11

 

Representative confocal images of myofibers colabeled for BDNF (red in merged image) and neuromuscular junctions (green in merged image). (A) Example neuromuscular junction without BDNF staining. (B) Example neuromuscular junction with immunostaining at the endplate. Scale bar: 25 μm.

Figure 12

 

Representative confocal images of myofibers colabeled for BDNF (red in merged image) and neurofilament to label nerve (green in merged image; arrow). Scale bar: 25 μm.

Expression of the BDNF Receptor TrkB

Mature BDNF predominantly signals through the TrkB, although BDNF does signal with lower binding affinity to other Trk receptors and p75^NTR.^46–48 To confirm that the main receptor for BDNF was present in the EOM, and, therefore, present to allow for signaling of the exogenously applied BDNF in the EOM, the presence of TrkB in adult mouse EOM was demonstrated with Western blot (Fig. 13A). Tyrosine kinase B receptor protein expression was localized to EOM with immunolabeling in human inferior oblique specimens. Similar to BDNF, TrkB was present diffusely within individual myofibers, and also expressed in the outer cell membrane (Fig. 13). Colocalization of TrkB with BDNF demonstrated that fibers expressing BDNF were frequently positive for TrkB (Fig. 13). As with BDNF, TrkB was present at neuromuscular junctions (Figs. 14A, 14C) and in nerve (Figs. 15B, 15C). However, as observed with BDNF localization, some neuromuscular junctions (Fig. 14A) and nerve (Fig. 15A) had low to no immunoreactivity for TrkB. 

Figure 13

 

(A) Western Blot demonstrating TrkB protein expression (green) in mouse EOM. Brainstem (BS) served as a positive control, GAPDH as a loading control (red). (B–D) Immunostaining with antibodies to BDNF (red in merged image) and its receptor TrkB (green in merged image) in adult human inferior oblique muscle. Brain-derived neurotrophic factor immunolabeling was generally present in myofibers with TrkB staining, although myofibers were not always equally bright for both. (B) Example of nerves positive for BDNF, but negative for TrkB. (C) Example of the mosaic pattern of BDNF and TrkB expression in EOM. The expression patterns of BDNF and TrkB were similar, but did not always overlap. (D) A nerve bundle and myofiber positive for both BDNF and TrkB. Scale bar: 25 μm.

Figure 14

 

Representative confocal images of myofibers colabeled for neuromuscular junctions (green in merged image) and TrkB (red in merged image) in adult human inferior oblique. (A) Myofiber negative for TrkB in the cytoplasm, but with faint TrkB staining at the endplate. (B) Example neuromuscular junctions with light to no labeling with TrkB, despite one myofiber being lightly positive for TrkB. (C) Higher power image of myofiber positive for TrkB with intense labeling at the cell membrane and positive for TrkB at the endplate. Scale bar: 25 μm.

Figure 15

 

Representative confocal images of myofibers colabeled for neurofilament to label nerve (green in merged image) and TrkB (red in merged image) in human inferior oblique. (A) Nerve bundle negative for TrkB. (B) Example of a nerve presumably innervating a neuromuscular junction that is positive for TrkB. (C) A fiber bundle mostly negative for TrkB. Scale bar: 75 μm.

Discussion

Brain-derived neurotrophic factor is known to have direct effects on EOM and innervation in some animal models. We were interested in evaluating BDNF treatment of macaque EOM because of the utility of this model in studies of strabismus. We treated the lateral rectus muscles of two infant monkeys unilaterally with 2 μg/day of BDNF for 3 months via a sustained-release pellet. One goal was to assess if unilateral BDNF treatment could alter eye position. We found that sustained BDNF treatment during the sensitive period for development of binocularity did not result in clinically significant strabismus, defined as >10°. In contrast, analysis of the treated EOM suggested that BDNF treatment differentially altered the phenotype of the treated lateral rectus muscles. The most pronounced effect was on slow myofibers. On average, slow myofibers had larger diameters, larger neuromuscular junctions, and there was increased percentage of myofibers positive for slow MyHC at the proximal end of the treated muscles. These results strongly suggest that BDNF signaling controls the characteristics of slow myofibers, and may uniquely contribute to the complex phenotype of EOM. In contrast, we found no significant effects of BDNF treatment on fast myofibers. 

Development of the visual system in macaques closely mirrors humans,^6,49 making macaques a useful model to study strabismus and the effects of neurotrophin signaling on the developing binocular system. In a prior study, we treated infant macaque medial rectus muscles bilaterally with 1 μg/day of IGF-1 for 3 months.⁴² The treated monkeys had normal eye alignment, despite larger myofibers after the IGF-1 treatment. Our analysis suggested that binocularity developed because there were coordinated alterations of myofiber size in the treated and nontreated muscles over the 3-month duration of treatment, presumably through communication within the oculomotor system in the brain. Our results complement this study, as normal eye alignment developed in the infant macaques after sustained BDNF treatment despite increases in myofiber area and larger neuromuscular junctions in the treated lateral rectus muscle. Further, the myofiber area measurements suggest that there were coordinated changes in the nontreated muscles that were sufficient to compensate for the increased fiber size in the treated muscles. 

In humans and monkeys, the two eyes move as a functional pair (cyclopean eye) to create sensory fusion and depth perception based on retinal image disparity. The oculomotor system is heavily biased to develop normal eye alignment. This may be due in part to the presence of binocularly sensitive neurons at birth, which would drive precise eye alignment.⁵⁰ Creating animal models of strabismus is difficult and requires persistent sensory deprivation during the early sensitive period of the binocular system.^3,4,44 The oculomotor system has a robust ability to adapt to perturbations to preserve eye alignment. 

In our recent study, we found that an increased dose of 2 μg/day IGF-1 unilaterally delivered for 3 months to the medial rectus muscle of three infant macaques resulted in a clinically significant strabismus at the end of treatment.¹⁷ In all three of our macaque studies, we observed adaptation in a subset of untreated EOM, including the antagonist lateral rectus muscle in the same orbit after unilateral IGF-1 treatment, and the antagonist muscle in the fellow eye in bilateral IGF-1 treatment.^14,17 In the current study, myofiber area was larger in the treated lateral rectus, but also in the ipsilateral antagonist medial rectus and contralateral lateral rectus (Fig. 6). Adaptation between paired EOM also has been observed after strabismus surgery modeled in rabbits.^51,52 It is possible that the dose of the BDNF treatment was not sufficient to functionally change eye movement or alignment. However, we hypothesize that after sustained BDNF treatment to one EOM, specifically, the lateral rectus in this case, adaptations occurred in the paired untreated EOM that were sufficient to allow formation of normal eye alignment in the treated subjects. Collectively, these studies support the view that the central nervous system has the ability to respond to changes in EOM force and shortening velocity produced by the exogenous addition of neurotrophic factor, and adapt over time to maintain balance between antagonist/agonist paired EOM. The mechanisms by which this occurred have yet to be delineated, but could involve signaling of internuclear neurons between the oculomotor and abducens motor nuclei. 

The EOM produce a variety of eye movements, which in part is reflected in the variety of myofiber properties, anatomical location, biochemical adaptations, and physiological properties of myofibers.^34,53,54 Sustained and lower velocity eye movements, such as smooth pursuit or fixation of a visual target, require slow and sustained contractions and a nonfatigable muscle.^55–57 Eye muscles have the fastest shortening velocities among mammalian skeletal muscles,⁵⁸ and they are required for quick eye movements, such as saccades. The organization of the motor neurons provides for a continuum with respect to phasic and tonic activity⁵⁴ that presumably reflects the different trophic backgrounds received from distinct types of myofibers within the EOM. 

Sustained BDNF treatment to EOM uniquely altered myofibers expressing slow myosin, which are implicated in the more tonic or step phase of muscle contraction.^39,59–61 Brain-derived neurotrophic factor did not alter the overall density of neuromuscular junctions, even though in culture BDNF promoted neurite outgrowth of ocular motor neurons and stabilized neuromuscular junctions.^18,28 However, the EOM is richly innervated compared to other skeletal muscles, with multiply innervated fibers and very small motor units.⁶² It is possible that normal neurotrophic signaling modifies EOM properties through changing myofiber type and potentiating neuromuscular junctions rather than by complete repatterning of muscle innervation.^63–65 

The sustained BDNF treatment of the infant lateral rectus muscles resulted in larger neuromuscular junctions on slow myofibers compared to those in control lateral rectus neuromuscular junctions. However, the sustained BDNF treatment did not have an effect on the size or morphology of en plaque neuromuscular junctions localized to fast myofibers (Figs. 2, 3). It is well established that neuromuscular junctions are subject to local, positive feedback loops between various synaptic regulatory factors that translate activity into structural changes at neuromuscular synapses.^66,67 These studies showed the modulatory effect of BDNF on synapse structure and active zone size, for example. In other skeletal muscles, BDNF and its receptor TrkB were shown to influence neuromuscular maturation and maintenance.^{21,28,44,48,68,69} In addition, BDNF treatment of skeletal muscle increased evoked endplate potentials.^22,29 Functional studies are needed to extend our findings to determine the functional result of the increased neuromuscular size seen after the sustained BDNF treatment on enhancing neuromuscular transmission. Further, it is unknown if BDNF molecular signaling is the same for all neuromuscular junctions in the EOM, or only a subset, such as en grappe endplates. 

The sustained BDNF treatment of the infant lateral rectus muscle also was associated with larger slow myofibers (Figs. 3, 4). Slow myofibers are generally smaller in diameter towards the midbelly of the muscle, where large en plaque endplates are concentrated. In the BDNF-treated muscle in the present study, the slow myofibers were slightly larger throughout the EOM, but the most dramatic increase in myofiber size was observed in the endplate zone. In the diaphragm, BDNF was shown to be expressed in satellite cells, and was specifically shown to have a role in maintenance of the myogenic precursor pool.⁴⁵ Although the pattern of immunolabeling in myofibers did not suggest enriched BDNF in satellite cells, it is possible that BDNF treatment encouraged myonuclear addition selectively into slow myofibers, particularly in the endplate zone, where endogenous levels of BDNF appeared to be less than in the proximal and distal regions of the muscles. Alternatively, if BDNF signaling potentiated neuromuscular signaling at slow neuromuscular junctions, which are present along the entire fiber length, it is possible that higher activity level at the synapse may have promoted larger myofibers. Such myofibers are most pronounced in the endplate region, where slow myofibers are proportionally the smallest. 

Based on the changes induced by sustained BDNF treatment on the properties of the slow myofibers, we hypothesized that BDNF expression in control untreated lateral rectus muscles would be localized to this population of fibers. We confirmed protein expression of BDNF in infant monkey and adult human EOM with immunolabeling (Fig. 9). Surprisingly, immunolabeling of BDNF in control monkey lateral rectus muscle revealed that BDNF was seldom colocalized to the slow myofibers (Fig. 10E). Coexpression of BDNF with five other MyHC isoforms was examined in the present study. The EOM express nine different myosin heavy chains, which often overlap in expression.^{32,34,70–72} Brain-derived neurotrophic factor was most often coexpressed with 2x and 2a MyHC, which execute fast muscle contractions. Interestingly, 2a MyHC^32,36 and BDNF expression in myofibers are drastically reduced in the endplate region of the EOM (Figs. 9G–I). The functional sequelae of this coexpression pattern are unknown. 

Recent studies suggest that neurotrophic factors differentially promote specific afferent inputs and firing patterns in neurons.^73–76 In the oculomotor system, axotomized abducens motor neurons underwent synaptic stripping, a process where afferent synaptic endings are lost on the motor neurons.²⁴ However, delivery of BDNF to the lesioned nerve preferentially restored the tonic or step phase of firing of the treated motor neurons, but not the phasic firing.²⁴ Our findings that BDNF immunolabeling was strongest in areas outside the location of the en plaque neuromuscular junctions, and that BDNF influenced properties of slow myofibers preferentially, support a role for BDNF in the periphery and in motor neurons that would promote slow, sustained eye movements. 

In addition to influencing the properties of neurons, neurotrophic factors also influence the properties of muscles.^16,40,77,78 Muscle-derived BDNF is expressed in the developing EOM and retrogradely transported to oculomotor neurons.^18,25 Brain-derived neurotrophic factor mRNA is also present in adult mouse EOM.⁷⁹ We localized BDNF protein to individual myofibers of infant monkeys and also in adult human EOM. This immunolabeling confirmed that BDNF expression was maintained in adult monkey and human EOM (Figs. 8, 9), and highly implicates BDNF as an important signaling molecule even in the mature EOM. In a study of diaphragm muscle, BDNF was differentially localized to satellite cells,⁴⁵ and in studies of other skeletal muscles was localized to the cell cytoplasm, edge of fibers, and at the synapses.^80,81 Similar to these studies, we observed BDNF in the myofiber cytoplasm, around the cell membrane, and in neuromuscular junctions. 

Although BDNF was not uniquely localized to slow myofibers, the highest level of expression was in the orbital layer in the proximal and distal regions of the muscle. This demonstration of a region-specific expression of BDNF is similar to that seen for other neurotrophic factors in the EOM. For example, IGF-1 was expressed in EOM as a gradient along the muscle length, with highest expression in the distal region of the muscle.⁸² The distribution pattern of BDNF along the length of the EOM has interesting functional implications, as BDNF myofiber staining was strongest in areas of the muscle hypothesized to be more involved in the tonic or step phase of muscle contraction.^83,84 Large twitch movements are hypothesized to be executed predominantly by the large en plaque endplates on fast myofibers of the global layer,^38,40 which was the region with the lowest expression level of BDNF. 

In addition to the presence of BDNF within individual myofibers, BDNF was present at neuromuscular junctions, in nerve terminals, and in nerve fiber bundles. It is interesting that not all nerve and neuromuscular junctions were immunolabeled for BDNF expression. In addition, BDNF reactivity was seen in en plaque and en grappe neuromuscular junctions located proximally and distally in the muscles. Interestingly, approximately 20% of abducens motor neurons were shown to lack the TrkB receptor,⁸⁵ which suggests that a subpopulation of motor neurons may not use BDNF as a neurotrophin. A more detailed analysis is needed to more fully delineate if BDNF expression at neuromuscular junctions is related to specific neuromuscular junction morphology or myofiber type. Further studies are needed to determine what controls BDNF expression and the role of BDNF signaling within the EOM. We also confirmed with Western blot and immunolabeling that TrkB, the highest affinity receptor for mature BDNF, is present in adult EOM. This further supports an active role of BDNF in the maintenance of mature EOM properties. Further studies are needed to determine the localization of the receptors TrkB and p75^NTR in the EOM. This information would provide more insight into how the patterning of this neurotrophin and its receptors contributes to the varied properties of the EOM across its length. 

In summary, sustained delivery of BDNF to infant monkey EOM resulted in changes in characteristics of slow myofibers. Compared to control lateral rectus muscles, neuromuscular junctions on slow myofibers had larger areas, the size of slow myofibers was larger, and the percentage of fibers expressing slow myosin was higher in the proximal region of the muscle. Despite these changes, the sustained BDNF treatment throughout the first 3 months of development of the binocular system did not result in strabismus. We hypothesize that the ocular motor system adapted to BDNF treatment to preserve eye alignment. Based on the morphological changes observed in EOM after BDNF treatment, further studies are needed to examine if BDNF changed muscle contractile properties. Larger slow myofibers could result in slower saccades, a longer duration or stability during fixation, or more precise eye movements and better gaze holding. These results support a role for BDNF in promoting the slow sustained contraction properties of EOM, and implicate BDNF as a potential novel treatment for ocular motor disorders where stabilization of eye movements would be desirable. 

Acknowledgments

Supported by Grants EY15313 and EY11375 (LKM) and EY06069 (MJM) from the National Eye Institute; the University of Washington National Primate Research Center (ORIP P510D010425); NIH T32 NS048944 (TJE); the Minnesota Medical Foundation; the Minnesota Lions and Lionesses; and unrestricted grants to the Departments of Ophthalmology (University of Minnesota, University of Washington) from Research to Prevent Blindness, Inc. This work is part of the PhD Thesis of CL Willoughby. The authors alone are responsible for the content and writing of the paper. 

Disclosure: C.L. Willoughby, None; J. Fleuriet, None; M.M. Walton, None; M.J. Mustari, None; L.K. McLoon, None