AAVrh74.tMCK.NT-3 Surrogate Gene Therapy in a Mouse Model of CMT2A

TL;DR

This study tests a gene therapy using NT-3 in a mouse model of CMT2A, showing improvements in muscle function and mitochondrial health.

Contribution

The study demonstrates the disease-modifying potential of AAVrh74.tMCK.NT-3 gene therapy in a CMT2A mouse model.

Findings

NT-3 gene therapy improved grip strength and rotarod performance in Mfn2+/− mice.

The therapy reduced mitochondrial abnormalities and oxidative stress in muscle tissue.

Carbohydrate metabolism in muscle was remodeled by NT-3 treatment.

Abstract

Mutations in the Mitofusin 2 (MFN2) gene cause Charcot–Marie–Tooth type 2A (CMT2A). Neurotrophin 3 (NT-3) is an autocrine factor that supports Schwann cell survival and differentiation, axon regeneration and myelination, neuromuscular junction (NMJ) integrity, and mitochondrial function. In this study, we assessed the efficacy of NT-3 gene therapy using the AAVrh74 serotype in the Mfn2+/− mouse model for CMT2A. Although haploinsufficiency is not reported in CMT2A patients, our model shows some features of CMT2A, including axonal atrophy, muscle atrophy, length-dependent axon loss, and abnormal mitochondria, in muscle in the enzyme histochemistry. Eight-month-old Mfn2+/− mice received a 3 × 1011 vector genome dose of AAVrh74.tMCK.NT-3 intramuscularly, and functional, electrophysiological, and histological outcomes were assessed six months post-treatment. NT-3 gene therapy in Mfn2+/− mice significantly improved grip strength and rotarod performance, and ameliorated electrophysiological abnormalities and NMJ denervation in lumbrical muscles. Additionally, our therapeutic approach improved muscle histopathology with reductions in mitochondrial abnormalities and oxidative stress. NT-3 further remodeled carbohydrate metabolism in muscle. Our study indicated that AAV.NT-3 gene therapy has a disease-modifying effect in the Mfn2+/− model of CMT2A, providing further support for the translational potential of this surrogate gene therapy approach to CMT2A patients.

Figures

Full text

1. Introduction

Charcot–Marie–Tooth (CMT) has been grouped into two main categories depending on pathogenesis and clinical assessment of nerve conduction velocity (NCV) in the median or ulnar nerve: demyelinating/type 1 CMT (CMT1) and axonal/type 2 CMT (CMT2). CMT2A, an axonal form, is the most common form of CMT2 [1,2,3], caused by nuclear-encoded mitofusin 2 (MFN2) gene mutations [4]. This subtype is notable for early onset and rapid progression during childhood [5,6]. CMT2A is predominantly inherited as autosomal dominant, and the majority of patients present with a classic CMT phenotype with involvement of motor and sensory functions in a length-dependent manner, although some patients show additional symptoms of central nervous system involvement [7,8,9], including laryngeal paralysis, loss of vision, sensorineural hearing loss, and upper motor neuron signs [5,10,11,12,13]. The majority of MFN2 mutations are missense, affecting the amino terminal GTPase and the coil-coiled domains, with disease onset in the first 2 years of life and an aggressive clinical course [4,14,15], though some patients have mutations at the extreme MFN2 carboxyl terminus and typically exhibit later onset and milder disease [5,16].

MFN2 and MFN1 proteins, located on outer mitochondrial membranes, presumably mediate the conformational changes required for membrane fusion, characterized by MFN-MFN dimers that extend outward from mitochondria into cytosolic space, and are involved in this process in a rate-dependent manner of GTP hydrolysis, which is more efficient for MFN2, providing high membrane-tethering [17]. Although both MFN proteins are involved in mitochondrial fusion, studies have emphasized that only MFN2 has been associated with metabolic and neurodegenerative diseases, suggesting that MFN2 is needed to maintain mitochondrial metabolism. Loss of MFN2 results in impaired mitochondrial respiration and reduced ATP production, originating from a depletion of mitochondrial coenzyme Q biosynthesis [18]. It was reported that the reduced respiratory chain function in cells lacking MFN2 was partially rescued by coenzyme Q10 supplementation [18]. Beyond its role in mitochondrial fusion, MFN2 has been shown to be essential for axonal mitochondrial transport [19,20], and it is also implicated in endoplasmic reticulum (ER)–mitochondria tethering [21] and in ER stress caused by mitochondrial dysfunction [22]. It has been proposed that MFN2 is important for regulating calcium (Ca^2+^) transfer from the ER to the mitochondria, which is essential for its bioenergetic function [23].

The underlying disease mechanism caused by MFN2 mutations is not well understood and may be variant specific, although the majority are considered to produce a gain-of-function/dominant–negative effect, disrupting the function of normal MFN2 protein, which negatively leads to impairment of mitochondrial dynamics, particularly mitochondrial fusion and transport [19]. However, pathomechanisms cannot be explained solely by toxic gain-of-function, suggested by autosomal recessive and semi-dominant CMT2A cases [13,24,25,26,27]. In studies with in vitro models, a loss-of-function mechanism is also proposed to be the basis of loss of mitochondrial fusion. Impaired mitochondrial fusion was shown to cause disruption of mitochondrial dynamics, crucial for the effective function of mitochondria, ultimately leading to cellular energy deprivation; therefore, MFN2 is important for the maintenance of axonal integrity, which is dependent on mitochondria as the most efficient producers of energy in the form of ATP. Studies with MFN2 deficiency models are important for furthering the understanding of the close association of normal mitochondrial dynamics and function with axonal maintenance, and the morphologic and pathophysiological correlates of impaired mitochondrial dynamics upon axonal properties, particularly on its viability distally. MFN2-deficient human neuronal cell lines were shown to exhibit perikaryal inclusions of phosphorylated neurofilament heavy chain, axonal swellings, impaired anterograde and retrograde mitochondrial transport within axons, axonal degeneration of spinal motor neurons, and defects in mitochondrial morphology and function [28]. Impaired mitochondrial transport associated with MFN2 deficiency was found to downregulate motor proteins KIF1A and KIF5A. In another study, mitochondrial dysfunction resulting from the MFN2-deficiency state was attenuated in mutant MFN2 mouse neurons by intermittent pharmacological activation of the wild-type (WT) MFN2 [29].

In this study, we created a mouse model of CMT2A with an Mfn2-deficiency state, Mfn2^+/−^ mouse, to test the efficacy of a neurotrophin-3 (NT-3) gene therapy approach. From a therapeutic gene therapy viewpoint for CMT2A, gene editing could potentially correct causal MFN2 mutant alleles, although the large number of different causal MFN2 mutations tampers the enthusiasm for such an approach. Our previous pre-clinical studies showed the proven efficacy of NT-3 in several mouse models for CMT subtypes, including those with mitochondrial dysfunction [30], and in accelerated or naturally occurring sarcopenia mouse models [31,32]. In these studies, NT-3 gene therapy increased oxidative phosphorylation markers dramatically, along with reversing abnormalities in oxidative enzyme histochemistry in muscle. Previously, it was also shown that systemic NT-3 could prevent depolarization of the mitochondrial inner-membrane potential in sensory neurons in a diabetic sensory neuropathy model [33]. This small-molecular-weight, versatile peptide has a short serum life; when NTF3 cDNA is delivered intramuscularly (IM) via an adeno-associated virus (AAV) capsid, the transduced muscle produces and secretes NT-3 into the circulation continuously at a therapeutic level, which can be measured in serum with ELISA [30,34,35]. In our current study, we show that NT-3 gene therapy in the Mfn2-deficient mice produced clear treatment efficacy in functional and electrophysiological studies, neuromuscular junction (NMJ) connectivity, and muscle histopathology associated with improved mitochondrial function and decreased oxidative stress in muscle. The results of our AAV.NT-3 surrogate gene therapy approach to CMT2A, which is agnostic to the MFN2 genotype, provide further support for its translational potential, including in this large CMT patient population.

2. Results

2.1. rAAV.NT-3 Vector Production and Potency

Design of the scAAV.tMCK.NT-3 expression cassette was performed at Nationwide Children’s Hospital, Columbus. The rAAVrh74.tMCK.NT-3 vector was provided by Sarepta Therapeutics, Inc. (Cambridge, MA, USA). Blood samples were collected at the endpoint when Mfn2^+/−^ mice were 14 months old, and circulating NT-3 levels were determined using ELISA as described previously [30,34,35]. Serum NT-3 levels were detected by ELISA, confirming the biopotency of the vector in the treated cohort (5.76 ± 0.65 ng/mL). We also noted that females in this model had higher serum NT-3 levels compared to males (females: 7.05 ± 0.82 ng/mL; males: 4.48 ± 0.71 ng/mL; Figure 1).

2.2. Characterization of Mfn2+/− Mouse Model

Mfn2^+/−^ conditional heterozygous mice were obtained using the Cre/Lox system by breeding homozygous Mfn2^Lx/Lx^ (loxP-flanked exon 4 of Mfn2 gene) and homozygous nestin-Cre (Cre^+/+^) mice, in which the Nestin promoter allows Cre recombinase expression in the central and peripheral nervous system. The Mfn2^Lx/−^; Cre^+/−^ (Mfn2^+/−^) mouse is heterozygous for Mfn2 deletion. Deletion was observed not only in the central and peripheral nervous system but also in other tissues, including muscles, leading to a systemic haploinsufficiency (Supplementary Figure S1). This observation also explains the lack of Mfn2^−/−^ pups in our initial attempts to obtain mice homozygous for Mfn2 deletion.

Mfn2^+/−^ mice showed a mild clinical phenotype, noted through a decline in functional performance, which became significant around 7–8 months of age compared to WT. Histopathologic studies revealed no axonal loss within the mid sciatic or distal tibial nerves compared to age-matched WT controls (Supplementary Tables S1 and S2). However, we found an apparent loss of myelinated fibers and swollen axons within the intramuscular nerve bundles of intrinsic foot muscles from the hind paw; such changes were absent in the WT controls (Figure 2A–C). Ultrastructural examination of lumbrical muscle samples from the hind paws showed evidence of axonopathy with tightly packed, irregularly oriented neurofilaments within axons of thinly myelinated fibers (Figure 2D), as reported in sural nerve biopsies from CMT2A patients [36]. Axons showing clusters of mitochondria and branched tubular profiles—presumably derived from smooth endoplasmic reticulum—were present (Figure 2E), similar to the axonal membranous organelle alterations observed in models of neurofilamentous neuropathies [37]. Reminiscent of findings seen in nerve biopsy samples from patients with CMT2A [11,36], we observed enlarged mitochondria with the loss of or distorted cristae—mainly located in the axons of unmyelinated fibers—and clusters of mitochondria within Schwann cells (Figure 2F,G). In addition, partially or fully denervated postsynaptic junctions in the intrinsic foot muscle (Supplementary Figure S2) were commonly encountered as morphologic evidence of reduced compound muscle action potential (CMAP; see Section 2.3). These observations indicate that the phenotypic features of the Mfn2^+/−^ mouse support the use of this model for testing the efficacy of AAv.NT-3 as a therapeutic strategy for CMT2A.

2.3. NT-3 Gene Therapy Improves Function and Attenuates Electrophysiological Abnormalities in Mfn2+/− Mice

We introduced NT-3 gene therapy at the time of onset of clinical phenotype, when Mfn2^+/−^ mice showed a significant decline in functional studies compared to WT, which was found to be at eight months of age. The treatment cohort (NT-3 cohort) received a 3 × 10^11^ vector genome (vg) dose of AAVrh74.tMCK.NT-3 via IM injection to the gastrocnemius muscle. Age- and sex-matched Mfn2^+/−^ mice were used as untreated controls. Grip strength and rotarod performance at baseline were compared to endpoint six months after AAV.NT-3 injection, when mice were fourteen months old. Grip strength analysis showed that NT-3-treated mice performed significantly better than the untreated mice by 24.5% at endpoint (NT-3: 0.109 ± 0.004 vs. UT: 0.088 ± 0.004, p = 0.0063; Figure 3A). In addition, we found that NT-3 treatment preserved grip strength, while performance of the untreated mice significantly declined over the study duration (NT-3, baseline: 0.112 ± 0.005 vs. endpoint: 0.109 ± 0.004, p = 0.74; UT, baseline: 0.115 ± 0.006 vs. endpoint: 0.088 ± 0.004, p = 0.0001). Similarly, NT-3-treated mice maintained their rotarod performance at steady levels, while the untreated mice significantly declined in performance over the period of six months (NT-3, baseline: 0.82 ± 0.06 vs. endpoint: 0.67 ± 0.04, p = 0.2236; UT, baseline: 0.81 ± 0.11 vs. endpoint: 0.54 ± 0.07, p = 0.0381; Figure 3B). At the endpoint, although meager, the NT-3 cohort showed a stronger trend towards better performance than the untreated mice by 23.1% (NT-3: 0.67 ± 0.04 sec/g vs. UT: 0.54 ± 0.07 sec/g, p > 0.05). No significant differences were observed between males and females in the grip-strength or rotarod assays.

We assessed the efficacy of NT-3 gene therapy on electrophysiological parameters at baseline and endpoint and found that the treatment maintained the CMAP values during the six months after treatment, while there was a significant decline in the untreated cohort during the same period. In addition, CMAP was significantly greater with treatment by 16.4% at the endpoint (NT-3: 49.54 ± 1.19 vs. UT: 42.55 ± 1.45 mV, p = 0.0013; Figure 3C). Moreover, nerve conduction velocity (NCV), which was lower than WT at baseline, was normalized in the treated cohort at endpoint. In the untreated mice, however, NCV remained low; 26.1% less compared to the WT (NT-3: 47.75 ± 3.34 m/s; UT: 37.42 ± 2.50 m/s; WT: 50.67 ± 1.36 m/s; UT vs. NT-3, p = 0.0299; UT vs. WT, p = 0.0223; NT-3 vs. WT, p = 0.8077; Figure 3D).

2.4. NT-3 Gene Therapy Improves Neuromuscular Junction Connectivity in Mfn2+/− Mice

Histopathological studies on the sciatic nerves from the untreated Mfn2^+/−^ mice showed the presence of myelin alterations, such as myelin corrugation, inpouching, and outpouching, compatible with axonal atrophy in this model; these myelin alterations were reduced in the treated cohort (Figure 4A,B). Distally, the tibial nerves of untreated mice (Figure 4C) revealed rare Wallerian degeneration/myelin ovoids and a few swollen axons, with disproportionately thin myelin for the axon diameter, which were not encountered in the treated samples (Figure 4D). The treatment also appeared to improve the index of circularity for axons, showing fewer irregularly shaped fibers with myelin alterations. The morphometric analyses of the mid sciatic and tibial nerves revealed that the myelinated fiber density (mean number of myelinated fibers per unit of endoneurial area) was not altered significantly in the Mfn2^+/−^ mutant, treated, or untreated cohorts compared to WT nerves (Supplementary Tables S1 and S2). However, the histograms showing axon size distribution of the myelinated fibers revealed the presence of a shift towards smaller axon size in the sciatic nerves from the Mfn2^+/−^ mice in both cohorts, compatible with axonal atrophy (Figure 4E). In contrast, distally in the tibial nerves, an axon subpopulation with a diameter >6 µm showed a shift to larger diameters compared to WT (Figure 4F), which we interpret as the result of an increase in neurofilament content. In addition, g ratio analysis detected improvements in myelin thickness/smaller g ratio with treatment in the Mfn2^+/−^ mice (Supplementary Figure S3, Supplementary Table S3), correlating with the improvements in NCV in the electrodiagnostic studies.

The investigation of the distal neuropathic process, within the intramuscular nerves in the lumbrical muscles from the hind paw of Mfn2^+/−^ mice, has established the morphologic correlate of decreased CMAP as described in the histopathological phenotype characterization studies. These observations include loss of myelinated axons and denervated Schwann cells within the intramuscular nerve twigs, and vacuolar degeneration of axon terminal bulbs at the NMJ (Figure 2A,B, Supplementary Figure S2). Our quantitative analysis of NMJ connectivity in these hind paw muscles revealed clear evidence for AAV.NT-3 treatment efficacy, showing a significant increase in the percent of innervated NMJs in the treated Mfn2^+/−^ cohort compared to the untreated counterparts (NT-3: 58.42 ± 4.55% vs. UT: 32.68 ± 4.47%, p = 0.0004, Figure 4G,H).

2.5. NT-3 Gene Therapy Improves Muscle Histopathology and Mitochondrial Abnormalities with Remodeling of Carbohydrate Metabolism and Decreases Oxidative Stress in Mfn2+/− Mice

To assess the fiber-size variability and neurogenic changes, we used succinate dehydrogenase (SDH) staining on the freshly frozen sections of tibialis anterior and gastrocnemius muscles from the Mfn2^+/−^ mice. Atrophic angular fibers and fiber-size variability were evident in the untreated samples, which showed an apparent decrease in the treated muscle at endpoint (Figure 5A,B). Morphometric analyses of tibialis anterior and gastrocnemius muscle sections confirmed these observations; fiber-size distribution shifted significantly towards larger-diameter fibers following NT-3 gene therapy, which was more prominent in the tibialis anterior muscle (Figure 5C,D). We also observed a significant increase in the mean total muscle fiber diameter, as well as fiber-type-specific diameter increases in both oxidative and glycolytic fiber subtypes with treatment in this anterior compartment leg muscle (Figure 5E, Supplementary Table S4). There were, however, no fiber-type-specific diameter alterations in the gastrocnemius/posterior compartment leg muscle of Mfn2^+/−^ mice, untreated or NT-3-treated, compared to WT (Supplementary Table S5).

Evaluations of toluidine blue-stained semi-thick sections from soleus, as well as the enzyme histochemistry for SDH and cytochrome C oxidase (COX) activity on the anterior and posterior compartment muscles from the hindlimbs of Mfn2^+/−^ mice, revealed an apparent increase in the number of muscle fibers with altered mitochondrial distribution and content. These alterations included abnormally increased mitochondria as subsarcolemmal accumulation, patchy areas, or complete loss of mitochondrial enzyme activity within fibers (Figure 6A–C), similar to observations in CMT2A patient muscle biopsies [38,39]. Further support for the presence of excessive mitochondria in the Mfn2-deficient muscle were the findings of increased mitochondrial DNA (mtDNA) copy number and impaired mitophagy at the molecular level (Supplementary Figure S4). NT-3 treatment resulted in a decrease in fibers with abnormal mitochondrial content and distribution (Figure 6D,E, Supplementary Figure S4A). Quantifications of these findings in the gastrocnemius and tibialis anterior muscles showed that number of abnormal fibers significantly reduced in the NT-3 cohort compared to untreated counterparts, in both muscles (Figure 6F,G). These abnormalities in the mitochondrial enzyme histochemistry in the Mfn2^+/−^ muscle were associated with marked increases in the expression levels of mtDNA-encoded subunits of complex IV, Cox3, and mitochondrial ATP synthase, Atp5d, compared to WT, suggestive of a compensatory change in mitochondrial function (Figure 6H,I). Unsurprisingly, we found that NT-3 treatment lowered the expression of these markers towards WT levels. As predicted, we also found that NT-3 had a normalizing effect on the oxidative stress levels in the Mfn2^+/−^ muscle (Figure 6J). The oxidative stress levels when using the 3-Nitrotyrosine (3-NT) ELISA assay showed that the MFN2-deficient muscle had significantly higher 3-NT levels than WT (UT: 177.5 ± 7.9 ng/mL vs. WT: 90.0 ± 9.8 ng/mL, p = 0.0002); AAV.NT-3 treatment, however, reversed this abnormality significantly, normalizing the oxidative stress levels towards WT levels (NT-3: 114.17 ± 12.14 ng/mL).

We next investigated expression levels of the glycolytic pathway markers [40] in the Mfn2^+/−^ muscle from the untreated and NT-3-treated cohorts (Figure 6K–N). Pfkm-encoding muscle phosphofructokinase-1, which is an important regulatory enzyme of glycolysis, Ldha, the M isoform of lactic dehydrogenase, involved in glycogen final stage breakdown [41], and Pkm, involved in the final stage of glycolysis for pyruvate production, were significantly low in the MFN2-deficient muscle compared to WT, which all significantly increased with treatment. There was also an increase in the carbohydrate-responsive element-binding protein expression, MLXipl (also known as ChREBP) [42], without reaching statistical significance. As previously illustrated [31], these findings show that in the presence of mitochondrial dysfunctional states, NT-3 is capable of inducing carbohydrate-metabolism remodeling with upregulation of glucose uptake and glycolytic pathways in muscle for energy, and that this anabolic effect is manifested as fiber size increase in muscle.

3. Discussion

Mutations in MFN2 are causal for CMT2A, a prototype for axonal CMT, for which no curative treatment currently exists, although understanding the molecular pathogenesis of disease-causing mutations is thought to offer insights for developing rational therapeutic approaches. Some therapeutic efforts have been focused on developing multifaceted platforms for drug-discovery evaluations based on the studies highlighting the functional importance of conformational changes within the heptad repeat region of MFN2 for its function in facilitating its intermolecular interactions and mitochondrial tethering [43]. These include testing MFN2 agonists or activators that promote mitochondrial fusion by interacting with endogenous MFN1 in in vitro systems, including primary neuronal cultures derived from CMT2A mouse models and patient-derived induced pluripotent stem cell (iPSC)-differentiated motor neurons [44]. Daily oral treatment with a mitofusin activator compound promoting mitochondrial fusion and motility in Mfn2 T105M knock-in mice reportedly normalized the phenotype, including the histopathology, within a short treatment period of 6 weeks [45]. Intermittent mitofusin activation with another small molecule in the same mouse model was also reported to be efficacious by increasing mitochondrial transport within peripheral axons and promoting mitochondrial localization to neuromuscular junctional synapses [29].

As another strategy, a combined RNAi and MFN2 gene therapy in vitro was shown to rescue the CMT2A motor neuron phenotype, stabilizing the altered axonal mitochondrial distribution and correcting abnormal mitophagy [46] that can result from MFN2 mutations [47,48,49,50]. The same approach in vivo reportedly resulted in silencing of endogenous MFN2 and expression of exogenous MFN2 in the brains of MitoCharc1 newborn mice after intracerebroventricular delivery using the AAV9 vector [46]. In addition, SARM1 inhibition has been proposed to be a primary therapeutic target for CMT2A and other types of axonal and demyelinating CMT subtypes [51,52], and histone deacetylase 6 (HDAC6) inhibition was promoted as a treatment strategy for CMT2A as well [53]. Although these studies demonstrate the molecular feasibility of different approaches both in in vitro and in vivo models, further studies are needed to support their translational potential.

In this study, we show that the AAV.NT-3 gene therapy in the Mfn2^+/−^ mouse model mitigated motor dysfunction and CMAP decline, accompanied by morphometric evidence of corresponding histopathological improvements in NMJ connectivity. Although the Mfn2^+/−^ mouse does not recapitulate the length-dependent axonal loss seen in patients’ main peripheral nerves, we want to emphasize that, in our model, the neuropathic process is illustrated with axon loss within the most distal intramuscular nerve bundles from the hind paws, along with NMJ denervation, confirmed by decreased CMAP in the electrophysiological studies. However, we cannot determine whether the process is a distal terminal axonopathy (starting at the NMJ/presynaptic terminal bulb first) or a distal non-terminal axonopathy affecting most distal segments of the tibial nerve, with the intramuscular nerve bundles followed by a loss of NMJ connectivity. Nevertheless, based on the key phenotypic features described here, we believe that the use of the Mfn2^+/−^ mouse model is well-justified and suitable for evaluating the treatment efficacy of NT-3.

An MFN2-deficient state is known to inhibit mitophagy, a process essential for removing damaged mitochondria, and MFN2 deficiency was thought to link age-related sarcopenia and impaired autophagy to activate an adaptive mitophagy pathway [22]. In our model, as predicted, NT-3 had no corrective effect on impaired mitophagy; however, we found a decreased number of fibers with abnormal mitochondria in the treated Mfn2^+/−^ muscle with an mtDNA content and an oxidative stress marker not different from WT. These observations are in support of the less appreciated properties of NT-3, i.e., its ability to protect mitochondria and inhibit oxidative stress by reducing reactive oxygen species and inhibiting lipid peroxidation, which have been reemphasized recently [54]. In previous experimental models, NT-3 has been shown to prevent a dysfunctional state of mitochondria by activating pathways that increase mitochondrial membrane polarization [33], leading to more efficient ATP production, enhancing mitochondrial biogenesis, and preserving mitochondrial function [30,31]. In addition to its well-recognized properties on Schwann cell survival, proliferation, differentiation, and axon regeneration, other biological effects of NT-3 relevant to this study include its role in the maintenance of NMJ integrity [55] and radial growth of muscle fibers through activation of the Akt/mTOR pathway in neurogenic muscle [56]. Our study shows that therapeutic serum NT-3 levels were achieved in the Mfn2^+/−^ mouse model following IM delivery of the AAVrh74.tMCK vector, therefore enabling multiple biological properties of NT-3 to become functional systemically—which are all relevant to the neuropathic process—as well as the mitochondrial cytopathy in muscle that can be part of the CMT2A phenotype [22,57,58,59].

The therapeutic potential of NT-3 should also be considered in the context of prolonged denervation with accompanying Schwann cell atrophy and Schwann cell loss by apoptosis, leading to an overall decreased regeneration capacity of peripheral nerve related to loss of receptors and reduced expression of growth factors, with downstream inability to maintain bands of Bungner [60,61,62]. These features of prolonged denervation are common to all longstanding neuropathic conditions, including CMT neuropathies, and from a translational viewpoint there is a need for targeting denervated Schwann cells to maintain a state of readiness for regeneration.

One of the major challenges of the gene silencing, gene editing, or gene replacement strategies proposed for CMT neuropathies is the presence of a decreased Schwann cell pool harboring a population of atrophic, quiescent Schwann cells, which are likely to be more prominent distally. For efficient regeneration, this atrophic Schwann cell population should be converted to a functional, regeneration-promoting mode, yet it is not known if these quiescent cells could be transduced with AAV vectors effectively and produce an efficient therapeutic effect, even if vector delivery to endoneurium of nerve fascicles in the sciatic nerve is achieved successfully.

Our previous studies have already proven the ability of the NT-3 molecule to stimulate Schwann cell proliferation and differentiation, and to overcome impaired regeneration in mouse models [34,63,64]. Although many other neurotrophic factors—such as nerve growth factor (NGF), brain drive neurotrophic factor (BDNF), glial cell line-derived neurotrophic factor (GDNF), and neurotrophin-4/5 (NT-4/5)—were also demonstrated to have similar functions [65], NT-3 is considered unique due to its specific TrkC (Tropomyosin Receptor Kinase C) receptor binding and distinct expression patterns during development, which are directly relevant to overcoming prolonged denervation. Developing Schwann cells survive independently of axons by establishing an autocrine survival circuit, which is present from embryo day 18 onwards [66]. The key components of this loop include insulin-like growth factor (IGF), platelet-derived growth factor-BB (PDGF-BB), and NT-3 [67,68]. In adult nerves, this autocrine loop, in the absence of axons, is crucial for the survival and differentiation of Schwann cells, which is an absolute prerequisite for nerve regeneration [69,70,71]. This important feature underlines the rationale for testing NT-3 for various CMT subtypes, including CMT2A, in our lab. On the other hand, the benefits of BDNF were most significant with multimodal strategies, including multiple neurotrophins and stem cells [72]. In comparison to other neurotrophic factors, NT-3 has been extensively studied as a therapeutic agent in mouse models of CMT and in other acquired neuropathy animal models [73,74,75,76,77], showing significant success in promoting nerve regeneration, increasing myelination, and improving motor function. In addition, from a translational viewpoint, the safety profile of NT-3 has been documented within comprehensive toxicology and biodistribution studies as part of an Investigational New Drug (IND) application for the AAV.NT-3 gene therapy approach, and the adverse events were considered acceptable in a pilot clinical trial of recombinant methionyl human NT-3 (r-metHuNT-3) given subcutaneously to CMT1A patients [63]. Similar studies on the off-target effects of neurotrophins will likely emerge as the therapeutic potential of other neurotrophins is explored. A recent study demonstrates an interest in exploring the rescuing effect of BDNF on impaired axonal transport in CMT2D mice [78] using a previously described gene delivery approach that we used in the same model [30].

In summary, this report describes the fifth CMT subtype we have studied in our laboratory that shows the efficacy of AAV.NT-3 gene therapy, providing further support for the translational potential of this surrogate gene therapy approach to large groups of CMT patients. Considering that this small peptide secreted from muscle has the ability to cross the blood–brain [79] and blood–nerve barriers [30,34,35,80], and is exerting its efficacy following a simple IM delivery approach using AAV vectors, its translational potential for more complicated neuromyopathic phenotypes [31] can easily be appreciated. It should also be noted that IM delivery of AAV has a proven safety profile [81], which is attractive for potential considerations of NT-3 gene therapy, alone or as part of a combinatorial approach for several disease conditions, genetic or acquired.

Limitations of the Study

In our study, we assessed the efficacy of AAV.NT-3 gene therapy on an Mfn2 haploinsufficiency mouse model. Although haploinsufficiency has not been reported for CMT2A patients to date, autosomal recessive and semidominant CMT2A cases [13,24,25,26,27] have been identified, in addition to heterozygous missense MFN2 mutations observed in the majority of the patients. Moreover, in vitro studies showed that MFN2 knock-down with short hairpin RNA caused axonal degeneration of spinal motor neurons and impairment in mitochondrial morphology, function, and transport [28]. A 50% reduction in functional MFN2 protein may be insufficient to form required amounts of homooligomeric or heterooligomeric (with MFN1) complexes, as well as fine-tune the fusion–fission balance necessary for maintaining mitochondrial quality over the long term, particularly in the context of age-associated cumulative effects. This may explain the late disease-onset observed in our mice, similar to that observed in human cases, with semidominant variants. When the variety in the mode of inheritance is considered, pathomechanisms of CMT2A cannot be explained solely by toxic gain-of-function [82]. The ambiguity in the genotype–phenotype correlation and discrepancies in the mouse models and CMT2A clinical phenotype complicates the pre-clinical studies. A recent article reviewing the CMT2A mouse models [83] stated that models “with genetics similar to human patients, failed to develop a disease phenotype,” suggesting pathomechanisms in mice and humans may follow different routes, and mice may fail to fully recapitulate the CMT2A disease, manifesting only some characteristics of the disease. Similar to the previously reported transgenic mice carrying a single copy of the MFN2 point mutations [83], the Mfn2^+/−^ mice showed a mild clinical phenotype, noted through the significant decline in functional performance around 7–8 months of age compared to WT. We emphasize that our model shows some of the classic CMT2A features: axonal atrophy, muscle atrophy, and length-dependent axon loss, as shown within the intramuscular nerve bundles of lumbrical muscles from the hindlimb foot. These claims are supported by robust quantitative studies of proximal and distal sciatic nerves (tibial branch), NMJ, and the tibialis anterior muscle for histopathological characterization. We found reduced axon size, myelin infoldings, and outfoldings (axonal atrophy) in sciatic nerves, as well as muscle fiber atrophy and muscle fibers with abnormal mitochondrial enzyme histochemistry, similar to observations in nerve and muscle biopsies from CMT2A patients. Moreover, ultrastructural examination of the intramuscular nerve bundles revealed abnormal mitochondria in clusters within axons and Schwann cells and evidence of axonopathy with tightly packed, irregularly oriented neurofilaments within axons, as reported in sural nerve biopsies from CMT2A patients [11,36]. However, in patients’ sural nerve biopsies, there is clear evidence of axonal loss, which we did not find in the sciatic nerve branches in our model. Therefore, further studies in aged mice are needed to determine whether the neuropathic process can be detected in the distal nerves resulting from disease progression in a so called “dying back” fashion. Therefore, we believe that studies with MFN2 deficiency models are useful for furthering our understanding of the close association of normal mitochondrial dynamics and function with axonal maintenance, and its viability distally.

4. Materials and Methods

4.1. Generation of Mfn2+/− Mouse Model

Charles River Laboratories, Inc. (Wilmington, MA, USA) created a mouse model expressing an Mfn2 conditional allele by inserting LoxP sequence flanking exon 4 (NCBI-Gene ID: 170731). These mice were then bred with B6.Cg-Tg(Nes-cre)1Kln/J mice, which were purchased from The Jackson Laboratory (Strain #:003771), to produce Mfn2^+/−^ heterozygous mice. Heterozygous deletion of the Mfn2 gene was confirmed with PCR. All animal experiments were performed according to the guidelines approved by the Research Institute at the Nationwide Children’s Hospital Animal Care and Use Committee that operates in full accordance with the Animal Welfare Act and the Health Research Extension Act (Institutional Animal Care and Use Committee approval number: AR18-00076).

4.2. Treatment Cohorts

Eight-month-old Mfn2^+/−^ mice received a 3 × 10^11^ vg dose of AAVrh74.tMCK.NT-3 via intramuscular (IM) injection to gastrocnemius muscle (NT-3 cohort). Age- and sex-matched Mfn2^+/−^ mice with equal sex distribution were used as untreated (UT) controls (n = 12 for both cohorts). The mice were sacrificed six months after NT-3 gene delivery, at fourteen months of age. Functional assays involving rotarod and grip strength tests and electrophysiology were performed at baseline and endpoint. Mice were then euthanized with a xylazine/ketamine cocktail, and blood, sciatic nerves, and selected muscles from front and hind limbs were collected for detailed quantitative histopathological studies, including an analysis of the innervation status of NMJs.

4.3. rAAVrh74.tMCK.NT-3 Vector Production

The self-complementary recombinant AVV serotype rh74 vector production expressing the human NT-3 transgene under the control of a muscle-specific tMCK promoter and biopotency of the vector was performed by Sarepta Therapeutics, Inc. (Cambridge, MA, USA).

4.4. Functional Tests

Rotarod and grip strength functional tests and electrophysiological measurements were performed at baseline and endpoint.

4.4.1. Grip Strength

A grip strength meter was used to measure the bilateral, simultaneous hindlimb grip strength (Chatillon Digital Meter, Model DFIS-2, Columbus Instruments, Columbus, OH, USA). Bilateral strength was assessed by allowing the animals to grasp a platform simultaneously with their hindlimbs, followed by pulling the animal until it releases the platform. The force measurements were recorded in three trials, and the average of these three measurements was included in the analysis. n = 12 for both treated and untreated cohorts with equal sex distribution; n = 14 at baseline and n = 6 for the endpoint for wild-type cohort (equal sex distribution).

4.4.2. Rotarod

Mice were tested for motor function and balance via an accelerating rotarod instrument (Columbus Instruments, Columbus, OH, USA). At least 24 h before the experimental run, an acclimation run was performed. The protocol run was carried out at 5 rpm with a constant acceleration of 0.5 rpm/s, and the average of three runs was included in the analysis. Run time was divided by the weight of the mice to obtain final data. n = 12 for NT-3-treated, untreated, and wild-type cohorts with equal sex distribution.

4.4.3. Electrophysiological Analysis

Nerve conduction studies were performed on the right sciatic nerves of mice under 2% isoflurane anesthesia, using a Nicolet Viasys Viking Select EMG EP System (Nicolet Biomedical, Madison, WI, USA). Mice were kept on a heating pad during measurements to maintain body temperature and 27 G disposable needle electrodes were used for stimulation, recording, and as reference using a protocol described previously [35]. n = 12 for both treated and untreated cohorts with equal sex distribution; n = 5 and n = 6 for CMAP and NCV, respectively, for the wild-type cohort (equal sex distribution).

4.4.4. Histological Studies

Peripheral nerves, skeletal muscles, and lumbrical intrinsic foot muscles for detection of NMJs were assessed with quantitative histological experiments. Samples were excluded from analysis only if they were not suitable based on criteria, including staining quality, contrast, and artefacts such as wrinkles in the section. Outcomes of behavioral or physiological analyses are not considered in the exclusion/inclusion process.

Nerve Histology

Sciatic and tibial nerves and lumbrical muscles from hindlimbs, removed under a dissecting microscope, were fixed in their in situ length in 3% glutaraldehyde in 0.1 M phosphate buffer for 30 min. Specimens were then cut into small blocks for additional fixation and processed for plastic embedding using our well-established routine protocols in our laboratory. One-micrometer-thick, toluidine blue-stained cross-sections from mid sciatic and distal half of tibial nerves were used for quantitative analyses. Selected tissue blocks from lumbrical muscles were thinly sectioned for electron microscopy.

Myelinated fiber-size distribution analysis in sciatic nerves

Axon diameter measurements of myelinated fibers were done on one-micrometer-thick cross-sections from mid-sciatic nerves, embedded in plastic, stained with toluidine blue. Two randomly selected non-overlapping areas were photographed for each mouse at 90× magnification (Nikon Eclipse Ti2-E, Tokyo, Japan). An average of 1950.7 ± 52.4 and 2225.3 ± 97.1 fibers per mouse were analyzed for treated and untreated mice, respectively (n = 6, equal sex distribution for each cohort; n = 5 for wild-type cohort), for generation of myelinated fiber-size distribution histograms (mean ± SEM, number/0.1 mm^2^).

Myelinated fiber-size distribution and g ratio analysis in tibial nerves

Axon diameter measurements of myelinated fibers for each mouse were derived from one-micrometer-thick, toluidine blue-stained cross-sections from tibial nerves. Two randomly selected, non-overlapping areas were photographed at 100× magnification using an Olympus BX41 microscope (Olympus Corporation, Tokyo, Japan) and SPOT Insight 12Mp sCMOS camera (SPOT Imaging, Sterling Heights, MI, USA). Measurements were obtained using BioQuant TCW14 Life Sciences imaging software version 15.5.60 (BioQuant Image Analysis Corporation; Nashville, TN, USA). The mean total measurements per mouse from the tibial nerves were 793.7 ± 39.4 in the treated (n = 3) and 865.0 ± 62.5 in the untreated cohorts (n = 4). Composites of axon diameter distribution histograms of myelinated fibers and the mean densities of myelinated fibers (mean ± SEM, number/0.1 mm^2^) were generated.

G ratio (calculated by dividing axonal diameter with fiber diameter [84,85]) measurements were done by outlining the myelin interior and exteriors in ImageJ (ImageJ 1.53e) to determine area, which was used to derive diameters to yield g ratio. Data from each mouse were derived from the stored images used for myelinated fiber-size distribution analyses. A total number of 1195 and 1551 myelinated fibers were analyzed for untreated (derived from n = 4 mice) and treated mice (derived from n = 3 mice), respectively, to calculate g ratio and generate percent g ratio distribution histograms.

Muscle Histology

Twelve-micrometer-thick freshly frozen cross-sections from gastrocnemius and tibialis anterior muscles were subjected to succinic dehydrogenase (SDH) enzyme histochemistry to assess metabolic fiber-type differentiation using the standard protocol established in our laboratory. Randomly selected representative images from deep, intermediate, and superficial zones of the muscle sections were photographed at 20× magnification using an Olympus BX41 microscope and SPOT Insight 12 Mp sCMOS camera. Staining quality, contrast, and lack of artifacts were considered during selecting samples for quantitative analysis; functional or physiological outcomes were not regarded. Fiber types were determined based on staining intensity as slow-twitch oxidative (STO), fast-twitch oxidative (FTO), and fast-twitch glycolytic (FTG) fibers. The shortest distance across the muscle fiber was measured as fiber diameter (Zeiss Axiovision LE4 software V4.9.1.0) and mean fiber diameter (mean ± SEM) was calculated for each fiber type (STO, FTO, FTG), as well as the mean total fiber size by combining all fiber types. Data were obtained from a total of 4973 and 3221 fibers of the treated cohort, and 4888 and 4320 fibers of the untreated cohort for gastrocnemius (n = 8 for both cohorts) and tibialis anterior (n = 6 for both cohorts) muscles, respectively.

4.4.5. Statistical Analysis

Statistical analyses were performed using GraphPad Prism 10.6.0 software. Unpaired t-test, one-way ANOVA with Tukey’s multiple comparison test, and two-way ANOVA with Tukey’s multiple comparison test were used based on the data. The tests that best met the assumptions of the data were chosen, and the statistical analyses performed are given in the figure legends, as well as the number for each experiment, provided both in the methods and figure legends. Significance level was set at p < 0.05. Sample size for each experiment was established based on our previous studies with analogous experiments [30,80]. Blinding was used for functional and morphometric data collection.