Structure-Based Designed Nano-Dysferlin Significantly Improves Dysferlinopathy in BLA/J Mice

Abstract

Dysferlinopathy is an autosomal recessive muscular dystrophy characterized by the progressive loss of motility that is caused by mutations throughout the DYSF gene. There are currently no approved therapies that ameliorate or reverse dysferlinopathy. Gene delivery using adeno-associated vectors (AAVs) is a leading therapeutic strategy for genetic diseases; however, the large size of dysferlin cDNA (6.2 kB) precludes packaging into a single AAV capsid. Therefore, using 3D structural modeling and hypothesizing dysferlin C2 domain redundancy, a 30% smaller, dysferlin-like molecule amenable to single AAV vector packaging was engineered (termed Nano-Dysferlin). The intracellular distribution of Nano-Dysferlin was similar to wild-type dysferlin and neither demonstrated toxicity when overexpressed in dysferlin-deficient patient myoblasts. Intramuscular injection of AAV-Nano-Dysferlin in young dysferlin-deficient mice significantly improved muscle integrity and decreased muscle turnover 3 weeks after treatment, as determined by Evans blue dye uptake and central nucleated fibers, respectively. Systemically administered AAV-Nano-Dysferlin to young adult dysferlin-deficient mice restored motor function and improved muscle integrity nearly 8 months after a single injection. These preclinical data are the first report of a smaller dysferlin variant tailored for AAV single particle delivery that restores motor function and, therefore, represents an attractive candidate for the treatment of dysferlinopathy.

Single vector adeno-associated viral (AAV) gene therapy packaging limit (∼5 kb) is unable to deliver dysferlin (6.2 kb) to treat dysferlinopathy. Hirsch and colleagues overcome this with Nano-Dysferlin (4.3 kb), improving muscle histology and motor phenotype almost 8 months after systemic injection in dysferlin-null mice.

Introduction

Dysferlinopathy is an autosomal recessive muscular dystrophy caused by the lack of a functional dysferlin protein. Previously asymptomatic patients experience disease onset between 14 and 28 years of age, and muscle atrophy can begin in proximal (LGMD-2B phenotype) or distal lower limb muscles (Miyoshi myopathy).¹ The disease typically progresses to the upper limb and paravertebral muscles while leaving the neck and head muscles unaffected. There is currently no treatment for dysferlinopathy, and accepted disease management consists of improving the patient’s deteriorating mobility with mechanical aids, preventing future contractures through physical therapy, and alleviating arising musculoskeletal trauma through orthopedic surgery.²

There are several therapies under investigation to treat dysferlinopathy encompassing drug, cell, and gene therapies. In 2013, a clinical trial of 25 dysferlinopathy patients randomly treated with the steroid drug Deflazacort found a considerable decrease in blood creatine kinase levels, yet, paradoxically, the patients’ muscle strength rapidly declined compared to control subjects.³ Cell therapy results have been more encouraging because transposase-mediated targeted integration of dysferlin cDNA into mouse muscle satellite cells resulted in dysferlin expression; however, following successful tibialis anterior (TA) muscle engraftment in a dysferlin-deficient mouse model, no clear histological improvement was shown. Furthermore, issues such as gene silencing, immune response, and the potential for “off-target” chromosomal mutations are problems cell-based therapies are working to overcome.⁴

Therapeutic approaches employing adeno-associated viral (AAV) vectors currently lead the gene therapy field in clinical applications due to their safety profile and ability for widespread human transduction. However, for treatment of dysferlinopathy, excitement for a simple gene addition strategy is tempered because the 6.2-kb dysferlin cDNA is too large to be packaged within a single AAV capsid. Approaches for AAV “oversized” transduction, to make up for the limited capsid packaging capacity, rely on multiple vector delivery of portions of a larger gene to a single cell, followed by host-mediated large gene reconstruction.5, 6, 7 These systems are generally termed fragment or dual AAV vectors.8, 9 Although genetically elegant from a basic science standpoint, AAV “oversized” gene transduction is inefficient compared to AAV vectors with an “intact” genome, with decreased performance 5- to 100-fold, with an inverse relationship between the size of the transduced compartment and the efficiency of this strategy.10, 11 Despite these limitations, preclinical success has been noted using fragment and dual AAV vectors for dysferlin transduction in mice, leading Nationwide Children’s Hospital to initiate the first phase I clinical trial that intentionally employs dual AAV vectors for a dysferlin gene addition strategy in the vestigial extensor digitorum brevis muscle (Clinicaltrials.gov). Although intramuscular (IM) injection of dual AAV vectors in a small muscle, such as the one targeted in this trial, in theory could transduce at high levels to facilitate intracellular transgene reconstruction, the desired systemic muscle therapy for dysferlinopathy presents a more difficult challenge that the dual vector technology needs to overcome.

Keeping the efficient single AAV vector strategy in mind and hypothesizing C2 domain redundancy, we engineered and validated a smaller, more compact dysferlin gene variant, termed Nano-Dysferlin, as a potential therapeutic for the treatment of dysferlinopathy.

Results

Design of Nano-Dysferlin Protein

Previous attempts at constructing smaller dysferlin genes have discounted the fact that partially folded protein domains, as a result of inappropriate truncation, could mask any therapeutic value of the smaller gene. To alleviate this issue, careful attention was given to the structural characteristics of C2 domains in order to rationally define each domain of dysferlin. Each of the seven C2 domains in dysferlin was defined by eight predicted β strands, C2 domain topology, integrity of the Ca²⁺-binding site, if applicable, and continuity of the hydrophobic packing in the core of the domain (Table 1). The overall philosophy to construct Nano-Dysferlin is based on three rules. First, the central features of the ferlin family members, FerA and DysF, were maintained intact in all constructs. Second, the first C2A domain and the C2 domain next to the transmembrane span, C2G, were preserved in all constructs. Third, multiple tandem C2 domains contribute individually to the overall membrane avidity. Given these tenants, C2 domains were subsequently excised with knowledge of the folded domain and the flexible linker that joined it to other potentially folded domains. These three rules led to the construction of a compact, potentially therapeutic dysferlin variant (Nano-Dysferlin), which was predicted to be efficiently packaged within a single AAV capsid (open reading frame [ORF] 4,356 nt).

Table 1.

Rationally Defined Dysferlin Domains

Domains	Amino Acid Range
C2A	1–124
C2B	219–352
C2C	366–515
FerA	670–782
DysF	864–1,097
C2D	1,137–1,281
C2E	1,314–1,465
C2F	1,579–1,696
C2G	1,789–1,994
Transmembrane span	2,045–2,067

Rationally defined amino-acid cutoffs that maintain hypothesized biochemical features of each C2 domain based on dysferlin isoform 8 primary sequence.

Expression of Nano-Dysferlin in Mammalian Cells

Nano-Dysferlin was based on the wild-type (WT) dysferlin isoform 8 cDNA (6,240 nt), which contains domains C2A-C2B-C2C-FerA-DysF-C2D-C2E-C2F-C2G-TM (Figure 1A). Initially, western blotting of membrane-associated, or soluble, protein lysates was performed to determine Nano-dysferlin localization following transfection in C2C12 myoblasts. For these experiments, full-length dysferlin and GFP expression cassettes served as the positive and negative controls, respectively. The results demonstrate that Nano-Dysferlin is produced as a single band at its expected size (160 kDa), and, like its parent molecule dysferlin, Nano-Dysferlin is a membrane and membrane-vesicle-associated protein (Figure 1B). Immunofluorescence of Nano-Dysferlin in transfected human HeLa cells demonstrated protein localization and abundance like wild-type dysferlin, with both distributed throughout the cell, likely in membrane vesicles, as it has been previously reported12, 13 (Figure 1C). In vitro toxicity experiments in dysferlin patient myoblasts showed no toxicity by alamarBlue following Nano-Dysferlin or dysferlin overexpression at increasing transfection doses of plasmid DNA (Figure 1D).

Figure 1.

Nano-Dysferlin Design and Expression in Mammalian Cells

(A) Human dysferlin isoform 8, the parent cDNA from which Nano-Dysferlin was derived, contains C2A, C2B, C2C, FerA, Dysf, C2D, C2E, C2F, C2G, and a transmembrane domain at a size of 6,240 nt. Nano-Dysferlin lacks C2D, CDE, and C2F domains, bringing the cDNA size down to 4,356 nt. (B) Western blot analysis of transfected c2c12 mouse myoblasts revealed that soluble protein lysate did not contain Nano-Dysferlin or full-length dysferlin. Contrastingly, membrane-associated protein lysate contains both dysferlin and Nano-Dysferlin. (C) Immunofluorescence imaging in HeLa cells revealed a similar cellular distribution of dysferlin and Nano-Dysferlin. Scale bar, 20 um. (D) Nano-Dysferlin did not display significant toxicity in dysferlin-deficient patient cells, as measured by alamarBlue absorbance at low (0.5 μg), medium (1 μg), or high (1.5 μg) plasmid doses. 0.5% sodium hypochlorite was used as the positive killing positive control. Mean + SD is shown.

Intact AAV Transduction Using a Weak Promoter Is More Efficient Than Fragment AAV Using a Strong Promoter

To find the most efficient therapy for in vivo studies, we evaluated fragment AAV with a strong CMV promoter against intact AAV with a small weak JET promoter for Nano-Dysferlin protein production. CMV-Nano-Dysferlin has a cassette size of 5,597 nt, whereas JeT-Nano-Dysferlin is theoretically within the AAV capsid packaging capacity at 4,849 nt (Figure 2A).¹⁴ To determine Nano-Dysferlin protein production from each cassette in a plasmid context, western blotting was performed following HEK293 cell transfection. As expected, the larger CMV promoter produced approximately 15-fold more Nano-Dysferlin compared to the small JeT promoter (Figure 2B).¹⁴ Given the existing AAV packaging dogma, it was hypothesized that the CMV-Nano-Dysferlin cassette at 5.6 kb would produce fragment AAV, whereas the smaller JeT cassette could be packaged as an intact genome at 4.85 kb in the AAV2 capsid. To investigate this, the capsid-packaged DNA species were separated by alkaline gel electrophoresis and then stained with SYBR gold. A single DNA species of the intended size was observed for the smaller JeT driven cassette, whereas Nano-Dysferlin expressed from the larger CMV promoter resulted in the packaging of heterogeneous DNA species within the size range of approximately 3.5–4.5 kb, much smaller than the intended 5.6-kb genome (Figure 2C). As noted in the introduction, the efficiency of fragment AAV transduction compared to intact AAV is dramatically decreased between 5- and 100-fold.10, 11 To determine if the CMV promoter, which is much stronger that the JeT promoter (Figure 2B), can overcome the decreased efficiency of fragment AAV, Nano-Dysferlin abundance was determined by western blot following transduction at increasing doses. Despite the differences in promoter strength favoring CMV, intact AAV2-JeT-Nano-Dysferlin vector transduction showed superior protein production compared to fragment AAV2-CMV-Nano-Dysferlin when administered at increasing doses (Figure 2D). Given the defined nature of the packaged transgenic DNA (Figure 2C), increased efficiency of intact AAV vector transduction (Figure 2D), envisioned systemic clinical intravenous (IV) administration, and potential for an unwanted immunological response to the vector at high doses in the clinic, we opted for the use of a higher efficiency intact JeT-Nano-Dysferlin-based vector for the remaining in vivo studies.

Figure 2.

Intact AAV Transduction Using a Weak Promoter Is More Efficient Than Fragment AAV Using a Strong Promoter

(A) Two Nano-Dysferlin AAV-ITR cassettes were designed that differ in size based on promoter and poly adenylation (polyA) sequences. JeT-Nano-Dysferlin is 4,849 nt in size, whereas CMV-Nano-Dysferlin is 5,597 nt. (B) Western blot following transfection of constructs depicted in (A) (along with dysferlin and GFP controls) in 293 cells and stained with the indicated antibodies. (C) AAV viral packaging was analyzed by alkaline gel electrophoresis and SYBR gold staining. Intact packaging was observed for the Jet-Nano-Dysferlin cassette, whereas fragmented packaging was seen for the CMV-Nano-Dysferlin cassette. The numbers in the horizontal axis indicate the cesium gradient fraction tested, while vertical numbers indicate the packaged DNA size. (D) Western blot analysis of 293 cells treated with the indicated AAV vectors at the indicated amounts per cell.

AAV-Nano-Dysferlin Improves Muscle Integrity following Intramuscular Injection

Next, the safety and efficacy of AAV-Nano-Dysferlin was investigated in blinded experiments following intramuscular injections using the AAV1 capsid due to its ability for widespread muscle transduction (data not shown). The TAs of 6-week-old dysferlin-deficient (BLA/J) mice were injected with AAV1-JeT-Nano-Dysferlin, with the contralateral leg receiving AAV1-CMV-GFP as a control. 40 hr before sacrifice, at 9 weeks, mice were injected intraperitoneally with Evans blue dye, a muscle damage marker that binds intra-fiber albumin, helping detect breaches in the sarcolemma of damaged muscle fibers.¹⁵ Upon counting positive fibers normalized to total fibers in cross-sections, variability in Evans blue dye-positive fibers in the AAV1-GFP control muscles was observed between individual BLA/J mice, suggesting different disease severities in genetically identical mice (Figure 3A, “GFP”). This is consistent with early disease variability in human dysferlinopathy patients, as previously reported.¹⁶ Despite baseline variations between TAs treated with control vector between the mice, within each mouse, every muscle treated with AAV1-JeT-Nano-Dysferlin demonstrated fewer Evans blue dye-positive fibers compared to the respective contralateral GFP control (Figure 3A). Collectively, the mouse cohort showed a significant difference between treated and control muscles by a paired two-tailed t test, p = 0.005 (Figure 3A). Central nucleation, a marker for muscle regeneration and thus indirectly muscle fiber turnover, was quantitated upon H&E staining of sections. The data indicate a decrease in central nucleation in all but one TA muscle injected with AAV1-Jet-Nano-Dysferlin compared to the internal AAV1-GFP control (Figure 3B; two-tailed t test, p = 0.0125). AAV-treated muscles also showed visibly improved histology (Figure 3C). Immunofluorescence detected Nano-Dysferlin in approximately 30% of muscle fibers; however, its localization in each muscle fiber was more distributed compared to the sarcolemma predominance observed for endogenous dysferlin. This is a common, yet puzzling, observation consistently reported for dysferlin gene addition studies in dysferlin-deficient mice (Figure 3D).5, 6, 17

Figure 3.

AAV-Nano-Dysferlin Significantly Improves Muscle Histology following Intramuscular Injection

(A) The TA muscles of BLA/J dysferlin-deficient mice were contralaterally injected with either AAV1-CMV-GFP or AAV1-JeT-Nano-Dysferlin. Evans blue dye was intraperitoneally administered 40 hr prior to sacrifice. Evans blue dye-positive fibers were normalized to total fibers. Matched pairs statistical analysis revealed a significant reduction of Evans blue dye-positive fibers in AAV1-JeT-Nano-Dysferlin-treated TA compared to contralateral controls. (B) Central nucleated fibers, a marker for muscle turnover, was reduced in all but one muscle treated with AAV1-JeT-Nano-Dysferlin, and statistical analysis showed a significant decrease in central nucleation of Nano-Dysferlin-treated muscles (p = 0.0125). (C) Representative images show improved muscle histology in AAV1-JeT-Nano-Dysferlin-injected muscle, which resembles WT muscle more closely than BLA/J dysferlin-deficient muscle. Scale bar, 40 um. (D) Romeo dysferlin antibody IF staining revealed a different distribution pattern between endogenous dysferlin and Nano-Dysferlin. Approximately 30% of fibers stained positive for Nano-Dysferlin (total fiber n = 455). Scale bar, 40 um. Mean + SD is shown.

AAV-Nano-Dysferlin Improves Motor Function following Systemic Injection

The BLA/J mouse model of dysferlinopathy varies from the human condition, with only mild motor deficits that significantly manifest, depending on the motor challenge and sensitivity of acquisition, starting at 10 weeks of age, considered young adulthood in mice.¹⁸ Consistently, human dysferlinopathy also becomes evident during young adulthood with normal, or even enhanced, athleticism earlier in life. In attempts to mimic the timing of diagnosis and the subsequent human therapeutic window of treatment, BLA/J mice were treated systemically with AAV9-JeT-Nano-Dysferlin (n = 6) or an AAV9-CMV-GFP control vector (n = 4), with a dose of 1e11 viral genomes. Blood creatine kinase activity, a marker often elevated in muscular dystrophies,¹⁹ was measured at 39 weeks, with the AAV9-Nano-Dysferlin cohort, showing a non-significant, yet trending, decrease by an unpaired t test with Welch’s correction (p = 0.13) (Figure 4A). Based on our previous findings of reduced rearing, the ability to stand on the two hind legs with arms/head in the air is reduced over time in older BLA/J mice, this cohort’s rearing activity was observed at 43 weeks of age, roughly 5 and a half months post injection.¹⁸ The data demonstrate a significant increase in total rears, on average >200 more times within an hour, only in mice that received AAV9-JeT-Nano-Dysferlin by a t test with Welch’s correction (p = 0.037) (Figure 4B). Furthermore, analysis of rearing performance over time suggested AAV9-Jet-Nano-Dysferlin-injected mice were not fatigued and maintained rearing at a constant level, whereas the performance of AAV9-CMV-GFP-injected mice decreased over time when analyzed by an ANOVA with repeated measures (p = 0.039) (Figure 4C). Horizontal activity showed no differences over the first 30 min (p = 0.58); however, over the last 30 min of evaluation, a non-significant (p = 0.13), yet trending, higher horizontal activity was observed in Nano-Dysferlin-treated mice by t test. This propensity to early “fatigue” has been observed in a BLA/J dysferlinopathy mouse model when compared to C56B7 mice.¹⁸

Figure 4.

AAV-Nano-Dysferlin Improves Motor Function following Systemic Injection

(A) Creatine kinase activity was found to be higher in AAV9-CMV-GFP-treated mice compared to AAV9-JeT-Nano-Dysferlin-treated mice. (B) Rearing performance was significantly improved over an hour evaluation in BLA/J mice injected with AAV9-JeT-Nano-Dysferlin compared to AAV9-CMV-GFP-treated mice. (C) Analysis of rearing over time demonstrated AAV9-JeT-Nano-Dysferlin-treated mice had increased stamina, indicated by consistent rearing over an hour compared to AAV9-CMV-GFP control mice. Mean + SD is shown.

AAV-Nano-Dysferlin Improves Muscle Integrity following Systemic Injection

The systemically treated cohort described above for motor function was sacrificed at 54 weeks, roughly 8 months following a single injection at 4.5 months of age. Evans blue dye was administered prior to euthanasia, and dye uptake, indicative of damaged muscle, was analyzed in a whole muscle assay and separately in a fiber-by-fiber manner following histology.¹⁵ The whole muscle Evans blue dye assay was performed using the gluteal and psoas muscles, which were determined in our previous work to be the most affected in the BLA/J mouse.¹⁸ In this assay, a higher absorbance indicates increased dye uptake and more muscle damage.¹⁵ The gluteal muscles, thought to be most affected in the BLA/J mouse model by our previous studies,¹⁸ showed significantly lower Evans blue dye uptake in mice treated with AAV9-JeT-Nano-Dysferlin compared to controls by a t test with Welch’s correction (p = 0.037) (Figure 5B). Meanwhile, analysis of the psoas muscle showed a non-significant trend of reduced Evans blue dye whole muscle uptake in AAV9-JeT-Nano-Dysferlin-treated mice (n = 6) compared to controls (n = 4) by a t test with Welch’s correction (p = 0.11) (Figure S1). To confirm the Evans blue dye whole muscle analysis, Evans blue dye-positive fibers were directly counted following histology and normalized to total fibers, with AAV9-Jet-Nano-Dysferlin-treated muscles showing an almost significant (p = 0.056) reduction of Evans blue dye-positive fibers (Figure 5C). Central nucleated fibers, indicative of muscular regeneration and turnover, also revealed a non-significant, yet strong, trend of reduction (p = 0.0835) in AAV9-Jet-Nano-Dysferlin-treated gluteal muscles (Figure 5A). Total central nuclei/total fibers were also evaluated and we found non-significant differences between treatments (Figure S1). As an additional measure of muscle fiber health, gluteal muscle fiber size was measured by the minimal Feret’s diameter from wheat germ agglutinin (WGA) lectin-stained muscle sections,²⁰ and analyzed with ImageJ (see Materials and Methods). Past studies have found increased variability and decreased mean fiber size in dysferlin-null mice muscles when compared to wild-type genetic background mice.¹³ Our results found muscle fibers from systemically treated AAV9-Jet-Nano-Dysferlin-treated mice were significantly larger than GFP-mouse-treated muscle fibers (p < 0.0001) (Figure 5D), with fiber size distribution graphs showing a right-shifted bell curve in the AAV9-Jet-Nano-dysferlin treated cohort (Figure S2). Given the significant fatty infiltration observed in the gluteal muscles in our previous study,¹⁸ we performed oil red staining of lipids in gluteus muscle sections, observing a drastic decrease of staining, which suggested lower lipid accumulation in AAV9-Jet-Nano-Dysferlin-treated mice gluteal muscles. To determine the extent of Nano-Dysferlin production in the gluteal muscles resulting in improved integrity, western blots were performed; however, Nano-Dysferlin was below the limit of detection by this assay and these blots were negative. This was followed by immunofluorescent staining performed on muscle sections, and wheat germ agglutinin lectin was used to stain the muscle sarcolemma. Expression was evident in approximately 10% of muscle fibers (Nano-Dysferlin total fiber, n = 256; no treatment total fiber, n = 185) (Figure 6). Nano-Dysferlin presence was also confirmed in the gluteal muscles of treated mice by RT-qPCR (Figure S3). Nano-Dysferlin appeared to have a preference for sarcolemma localization, with some protein apparently localized throughout the cytosol, similar to the IM injections (Figure 3) and several prior reports.5, 6, 17

Figure 5.

Effect of AAV-Nano-Dysferlin on Muscle Histology following Systemic Injection

(A) Central nucleated fibers, whose presence indicates regeneration and turnover were reduced non-significantly, yet trending (p = 0.0835) in Nano-Dysferlin-treated muscles compared to GFP-treated muscles. (B) Evans blue dye whole muscle absorbance assay, a measure of muscle damage, was significantly decreased in the gluteal muscles of AAV9-JeT-Nano-Dysferlin-injected mice (p = 0.037). (C) Representative image of Evans blue dye-positive fiber histology shows a marked decrease in muscle damage of AAV9-JeT-Nano-Dysferlin-treated muscles compared to the AAV9-CMV-GFP-treated controls. Statistical analysis showed an almost significant reduction (p = 0.056) of Evans blue dye-positive fibers in the gluteal muscles of AAV9-JeT-Nano-Dysferlin-treated mice. Scale bar, 100 um. (D) Minimal Feret diameter, a measure of fiber size, was obtained from gluteal muscle WGA lectin-stained muscle sections, with a significant difference between treatments by unpaired t test (p < 0.0001). (E) Oil-Red-O staining for hydrophobic and negatively charged lipids (red); this representative image showed a marked difference between treatments. Scale bar, 300 um. Mean + SD is shown.

Figure 6.

Nano-Dysferlin Detection by Immunofluorescence

Immunofluorescence staining of gluteal muscles from the indicated mice with a dysferlin antibody (red), Hoeschts nuclear stain (blue), and wheat germ agglutinin membrane stain (green). Nano-Dysferlin localization throughout the membrane and cytoplasm was noted while endogenous dysferlin is uniquely localized to the membrane. Approximately 10% of muscle fibers stained positive for Nano-Dysferlin (total fiber n = 441). Scale bar, 100 um.

Discussion

AAV-mediated gene therapy is currently considered a promising method to treat diseases such as Duchenne’s muscular dystrophy (DMD) and dysferlinopathy.5, 6, 10, 17 However, both these muscle-wasting diseases highlight a primary deficiency of AAV vectors: the viral capsid is too small to package the full-length cDNA for a simple gene addition strategy.⁷ To overcome this limitation, we and others have investigated the ability of multiple AAV capsids to deliver portions of a large gene to the nucleus, wherein the host’s DNA damage response mediates the possibility for large gene reconstruction.8, 10, 21, 22 Although intriguing, these DNA-repair-dependent multiple vector formats for AAV large gene delivery⁹ suffer from dramatically reduced transduction efficiency compared to a single AAV particle with an intact transgenic genome.10, 11 Unlike a single particle AAV gene addition strategy, which theoretically relies on one particle infecting a single cell, AAV oversized gene transduction is highly inefficient, especially when delivered systemically.10, 11 This is due primarily to (1) the requirement for several different vector genomes to be uncoated within a single nucleus, and (2) inefficient homology-directed repair in non-dividing cells, such as muscle fibers that are biased toward non-homologous end joining, thereby generating aberrant non-functional, and potentially immunogenic, transgene products. Due to the decreased efficiency of oversized AAV transduction approaches, higher effective doses are required (compared to single particle AAV transduction).¹⁰ In many cases, increasing the dosage of virus exasperates the problem by producing undesired immunological complications and resulting in therapeutic failure. Additionally, the current production titers of clinical grade AAV vector preparations for other muscular diseases that require only single AAV vector transduction are a serious limitation restricting the number of patients able to be treated. Despite these two major concerns with AAV large gene transduction, preclinical data in a dysferlin-deficient mouse have led to recruitment of dysferlinopathy patients for a phase 1 clinical trial proposing the use of AAV-oversized transduction for the treatment of dysferlinopathy.6, 17 Notably, this will be the first AAV trial intentionally relying on multiple vector transduction of single cells and the capacity of the patients’ DNA damage response for homology-directed repair in muscle fibers for clinical success. To provide an alternative treatment strategy to patients with dysferlinopathy, we have followed suit with the DMD community and rationally designed Nano-Dysferlin, a compact dysferlin-like open reading frame that is amenable to single AAV vector genome packaging and transduction.

In general, C2 domains are modular protein domains that can bind to the inner leaflet of phospholipid membranes.²³ Most C2 domains bind to membranes in a Ca²⁺-dependent manner, but there are some that do not. Wild-type dysferlin possesses seven tandem C2 domains, each separated by long linkers.²⁴ Our central hypothesis in constructing more compact dysferlin proteins is that multiple tandem C2 domains contribute individually to the membrane-binding avidity of the entire protein. Therefore, there must be a point where fewer domains still bind membrane and still provide their function, but can provide therapeutic benefit by being amenable to intact AAV packaging. This strategy implies a knowledge of what makes up a C2 domain. There have been other attempts at minimizing the overall size of dysferlin;²⁵ however, these experiments were conducted without an in-depth understanding of the structure of C2 domains. Without a clear domain definition, the folded inadvertent truncation of even a single folded domain could misfold the entire protein, thereby leading to degradation, loss of function, or even aggregation. After testing several constructs, we discovered that retaining the amino-terminal C2 domains, C2A, C2B, and C2C, with their inter-domain linkers, in addition to the FerA, DysF, C2G, and transmembrane domain results in a molecule correcting for the absence of dysferlin function in a dysferlin-deficient mouse model.

The transgenic DNA packaging limitation of AAV (<5 kB) not only precludes packaging of full-length dysferlin cDNA, but also restricted our promoter size for Nano-Dysferlin expression. Examination of packaged AAV genomes clearly demonstrated that Nano-Dysferlin expressed from the JeT promoter (4,849 nt) is packaged as a single species; in contrast, when using CMV (5,597 nt), heterogeneous DNA species were encapsidated, which ranged in size from 3 to 5 kb (Figure 2C).¹⁴ This fragment AAV vector was less efficient than AAV single vector transduction, even despite the >10-fold increased expression of the CMV promoter when compared to the JeT promoter (Figures 2B and 2D).

In previous experiments, we have demonstrated that fragment AAV-oversized gene transduction is better than or similar to the other approaches of AAV large gene transduction, which in general are referred to “dual vector” approaches (reviewed by Pyradkina et al.,⁷ Hirsch et al.,⁹ and Hirsch et al.¹⁰). In our published work investigating fragment AAV and dual AAV transduction efficiencies, intact AAV remained 5- to 100-fold more efficient than an AAV capsid packaged with an intact genome in vitro and in mouse liver, muscle, and eye.10, 11 Although slight variations of these approaches have been published to modestly increase these oversized AAV transduction approaches,²⁵ their inherent dependency on multiple vector transduction and host-mediated transgene reconstruction places them at a major disadvantage to an approach that relies on single AAV vector transduction. Therefore, our focus for in vivo analysis relied on the JeT-Nano-Dysferlin cassette for single AAV vector transduction. A limitation of our efforts herein is that the JeT promoter is small, as required for intact genome packaging, yet relatively weak and ubiquitous in nature, which is not ideal for a skeletal muscle therapy delivered IV (Figure 2).¹⁴ Currently, the small muscle-specific promoters C2-27 and C5-12 are under investigation, which are hypothesized to allow intact genome packaging when combined with Nano-Dysferlin in an AAV context while likely having significantly enhanced transcriptional activity in muscle.²⁶

Contralateral administration of AAV1-JeT-Nano-Dysferlin directly to dysferlin-deficient skeletal muscle resulted in increased muscle integrity in every mouse tested, as determined by decreased Evans blue dye fiber staining, and all but one mouse tested by central nucleated fibers (Figure 3). This contralateral intra-mouse comparison is important because the dysferlin phenotype between animals (Figure 3, black bars) was variable, perhaps due to environmental contexts (i.e., increased individual activity for particular mice). Despite this inter-mouse variability in disease severity, the results clearly demonstrated increased integrity and significantly improved muscle phenotype as a result of Nano-Dysferlin, evident by immunofluorescence (IF) in approximately 30% of treated fibers (Figure 3C). Interestingly, we note that Nano-Dysferlin localization following gene delivery is not primarily restricted to the sarcolemma, as observed for native dysferlin in WT mice (Figure 3D). This result is puzzling yet not specific to Nano-Dysferlin because restoration of WT dysferlin via a multiple vector approach also results in abnormal intracellular distribution, as evidenced by previous reports.5, 17 The reason for this aberrant localization is speculated to result from restoration of dysferlin (or Nano-Dysferlin) to terminally differentiated myofibers because dysferlin has been suggested to be regulated during differentiation; however, other theories, such as altered abundance per fiber, are also entertained.

Curiously, the onset of dysferlinopathy in human patients generally begins during the teenage years in previously asymptomatic, and often athletic, individuals. Reports have suggested the reason for this may be related to the metabolic switch in cellular respiration from oxidative to glycolytic predominance during this time.27, 28, 29, 30, 31 This is consistent with the emergence of muscular dystrophy phenotype in BLA/J dysferlin-deficient mice starting at 15 weeks of age.¹⁸ In fact, studies have found both dysferlin-deficient BLA/J mice and primary human myoblasts have an impaired glucose and lipid uptake/metabolism.³² Furthermore, prior reports have shown lipid accumulation is a feature observed in human and BLA/J mouse dysferlinopathy yet has not been reported in other muscular dystrophies, such as calpainopathy, DMD, and myotonic dystrophy.³³ Consistent with this line of thought, our previous study found an increase in extramyocellular lipids (EMCLs) in gluteal and psoas BLA/J mouse muscles, the most affected muscles in the BLA/J mouse model, with visible fatty infiltration in MRI images of gluteal muscles.¹⁸ After analysis of muscle sections stained by H&E in the present study, differences in potential fatty infiltrates became apparent between treatments. To confirm this, we performed oil red O staining for lipids, which revealed a drastic reduction of fat infiltrates in AAV9-Jet-Nano-Dysferlin-treated mice (Figure 5E).

The experiments designed herein attempted to imitate a potential clinical situation by systemically treating 4.5 month-old animals already demonstrating progressive muscular disease, with a single dose of AAV9-JeT-Nano-Dysferlin. The results of blinded experiments demonstrate that BLA/J mice treated with AAV9-Jet-Nano-Dysferlin reared on average 200 more times during a 1-hr evaluation, totaling nearly twice the activity of control treated mice. In previous work, we observed that the rearing deficit compared to WT mice increased over time, suggesting earlier onset of fatigue in BLA/J mice.¹⁸ The work herein is consistent with a therapeutic effect of AAV9-Jet-Nano-Dysferlin because, when analyzed over time, treated mice performance strongly suggested fatigue correction and demonstrated rearing levels similar to those of WT mice, as observed in our previous study.¹⁸ One additional take away from this study for future locomotor evaluation of therapeutics, and given the observed “fatigue” of BLA/J mice, is that appropriately designing locomotor experiments that extend the time of activity testing beyond 60 min may reveal stronger, more drastic deficiencies present in this model for dysferlinopathy. This remains to be tested in future studies (Figure 4C).

Post-mortem analysis of Evans blue dye uptake using a whole muscle assay (Figure 5B) by conventional Evans blue dye histology (Figure 5C), central nucleated fibers (Figure 5A), and semi-automated fiber size analysis by Feret diameter (Figures 5D and S2) agreed that BLA/J mice treated with AAV9-Jet-Nano-Dysferlin were increased for muscle integrity in the most affected BLA/J muscle group, the gluteal muscles,¹⁸ where approximately 10% of muscle fibers stained positive for Nano-Dysferlin by immunofluorescence (Figure 6), consistent with the notion that a little dysferlin (or in this case Nano-Dysferlin) goes a long way in maintaining muscle integrity.⁵ In no cases herein, whether dysferlin-deficient patient myoblasts or unrestricted production in the BLA/J model, did we see toxicity for Nano-Dysferlin or AAV vector transduction. However, again, we note that the ubiquitous JeT promoter is relatively weak, resulting in detectable (Figures 2, 3, and 5) but low levels of Nano-Dysferlin.

Further experimentation with stronger and muscle-restricted promoters is needed to confirm this result. In addition, we note that a single new epitope was generated by deletion of the C2D, E, and F domains, which raises the potential of a Nano-Dysferlin-specific cellular-mediated immune response, depending on the nature of the patient’s mutation. This is a similar scenario to the application of micro- or mini-dystrophin to DMD patients or even full-length dysferlin administration to dysferlinopathy patients due to the myriad of possible mutations. Despite these standard therapeutic concerns, Nano-Dysferlin represents the only single AAV-vector-amenable dysferlin variant that restores motor function in dysferlin-deficient mice and represents an attractive candidate for the treatment of dysferlinopathy in the clinic.

Materials and Methods

Study Design

Our study was designed to generate an AAV therapeutic for dysferlinopathy. To test this, Nano-Dysferlin, an abridged dysferlin-like molecule, was created and tested functionally in vivo using AAV technology. We chose to use the currently best animal model of dysferlinopathy, BLA/J mice, due to its clinically relevant phenotypic characteristics. All mouse experiments were blinded to the handler in terms of the type of treatment, and the results were un-blinded only after statistical analysis. The experimental endpoints and time of initial treatment were based on our earlier characterization of the BLA/J model.¹⁸ The in vitro experiments were repeated on at least 2 separate days, with a minimal replicate number of 3 for each occasion. The animal experiments were performed once with the indicated replicate number and duration. For our intramuscular experiment, littermates were administered randomly assigned treatments with contralateral controls. For our systemic experiment, mice were randomly assigned treatments. Investigators performing all animal interaction and data collection were blinded. Alpha was set at the traditional 0.05 for significance. Post hoc power analysis of our rearing behavioral performance assay was done in G-Power 3.1.9.2 software, an effect size of 1.77 was obtained using group means, and standard deviation within each group was estimated by the pooled standard deviation equation, with a power (1-β error probability) of 0.76. One mouse was eliminated from the intramuscular experiment due to a missed Nano-Dysferlin injection, as evidenced by lack of India ink in the targeted TA muscle. In the systemic experiment, one Nano-Dysferlin-treated mouse was eliminated as an outlier because it had less than half the rearing performance of the median for all other Nano-Dysferlin-treated mice. No major changes in p value throughout the performed experiments arose from this exclusion.

Designing Nano-Dysferlin

The Nano-Dysferlin gene was based on the wild-type dysferlin isoform eight cDNA (6,240 nt), which contains domains C2A-C2B-C2C-FerA-DysF-C2D-C2E-C2F-C2G-TM. Wild-type domains were defined in terms of the available primary sequence as follows.

Each C2 domain range in Table 1 was analyzed for predicted β strand content, potential Ca²⁺-binding residues, C2 domain topology, overall C2 domain length, and continuity of hydrophobic packing of the domain’s core. Once this was completed, dysferlin could be edited in silico by defining excision sites that extended from the N-terminal linker to the C-terminal linker of each C2 domain. All abbreviated protein constructs retained the C2A domain, FerA domain, DysF domain, C2G domain, and transmembrane helix in addition to the short extra-cellular portion of the protein. All other C2 domains were dispensable. Finally, genes corresponding to the new proteins were assembled by GenScript, with codon optimization for human synthesis. Nano-Dysferlin itself possesses domains C2A-C2B-C2C-FerA-DysF-C2G-TM at a total length of 4,356 nt.

Cell Lines and Culture Media

HeLa cells were used for immunofluorescence and grown in DMEM supplemented with 10% Sigma fetal bovine serum (FBS) (F7524) and 1% Pen/Strep antibiotic. Immortalized human patient “ER” myoblasts bearing dysferlin exon 44: c.4882G > A HMZ, p.G1628R homozygous mutation were obtained from Dr. E. Gallardo and grown in Promocell Skeletal Muscle Cell Growth Medium Kit (C-23060) supplemented with 15% Sigma FBS (F7524), 2 mM Glutamax by Life Technologies (35050), and 100 μg/mL Primocin by Invivogen (ant-pm-1). C2C12 myoblasts were obtained from ATCC (CRL-1772) and grown in DMEM supplemented with 10% Sigma FBS (F7524) and 1% Pen/Strep antibiotic. HEK293 cells, used for western blots and AAV vector production, were obtained from ATCC (CRL-1573) and cultured in DMEM supplemented with 10% Sigma FBS (F7524) and 1% Pen/Strep antibiotic.

Plasmids and Viral Production

The Nano-Dysferlin nucleotide sequence (available upon request) was generated by GenScript based on our amino acid sequence submission and their human codon optimization algorithm. PCR sub-cloning added a 3X FLAG tag to the 3′ ORF and moved the Nano-Dysferlin sequence into pSJG-JeT-GFP-synpolyA self-complementary plasmid (kind gift of Dr. S. Gray at University of North Carolina [UNC]) at the NcoI and XhoI sites. This cassette was then excised using KpnI and MluI, and the ends were blunted and then cloned into blunted KpnI/SphI sites of pTReGFP (a single-strand AAV plasmid).³⁴ The region from between the AAV2-inverted terminal repeats on this resultant plasmid was then confirmed by sequencing. For these experiments, phpaTRSK-CMV-GFP was used to generate the GFP control AAV vector.³⁵

Virus was produced by triple transfection protocol in HEK293 cells.³⁶ This method used the pXR1, pXR2, and pXR9 plasmids, along with the pXX680 helper (kind gifts of Dr. R.J. Samulski). The titer of all vector preps was determined by southern dot blot and confirmed by qPCR. When applicable, the packaged genome species were confirmed by alkaline gel electrophoresis and SYBR gold staining.³⁶

Nano-Dysferlin Intramuscular and Systemic Administration

For our intramuscular experiment, data shown in Figure 3, AAV1-Nano-Dysferlin or AAV1-CMV-GFP was injected intramuscularly into contralateral TA muscles a single time at 6 weeks of age. Isoflurane-sedated mice were injected with a BD 8-mm 31-gauge needle in 50 ul of total volume (5e10 total viral genomes) administered per TA containing 2% India ink (America Master Tech Cat: STIIN25). For our systemic experiment, AAV9-JeT-Nano-Dysferlin (n = 6) or AAV9-CMV-GFP (n = 4) was administered by a tail-vein injection a single time at 4 and a half months of age with a BD 8-mm 31-gauge needle in a total volume of 200 ul (2e11 total viral genomes).

Western Blots

CMV Nano-Dysferlin plasmid was first tested by western blot alongside CMV wild-type dysferlin 48 hr post-transfections of C2C12 mouse myoblasts using Lipofectamine 3000 (ThermoFisher Cat: L3000001), as described in the product protocol. Mammalian protein extraction reagent (MPER; Thermo Scientific Cat: 78501) was used to extract protein for total protein lysate western blots. Isolated cytoplasm and membrane-associated protein lysates were obtained via the Mem-PER Plus Membrane Protein Extraction kit (ThermoFisher Cat: 89842). For intramuscular and intravenous experiments, muscle was harvested and followed the mammalian protein extraction reagent protocol (ThermoFisher Cat: 78501). All protein lysates were subsequently denatured, added to 4X NuPage solution (ThermoFisher Cat: NP0008) with a final concentration of 5% β-mercaptoethanol, and ran on a precast 4%–12% BIS-TRIS gradient gel (ThermoFisher Cat: NP0321). All dysferlin and Nano-Dysferlin detection experiments employed the Romeo primary antibody (Abcam Cat: 124684) at a 1:2,000 concentration, followed by a secondary anti Rabbit HRP antibody (Abcam Cat: ab6721) at a 1:10,000 concentration. Sirius chemiluminescence kit (Advansta Cat: K-12043-D20) was used for all blots, and blots were imaged by the Amersham A600 imager.

Toxicity Assay

Dysferlin-deficient (ER) human patient cells, courtesy of the Jain Foundation, were plated in a 24-well plate (see Cell Lines and Culture Media). Cells were approximately 70% confluent when Lipofectamine 3000 was used for transfection using the product information-recommended protocol. Low, medium, and high doses consisted of 0.5 μg, 1 μg, and 1.5 μg of pCMV-GFP, pCMV-Nano-Dysferlin, or pCMV-dysferlin DNA plasmids. The cell’s medium was replaced 24 hr after transfection, and 50 ul of alamarBlue cell viability reagent (DAL1100) was added to each well 48 hr after transfection; readouts were followed per product protocol. 100 ul of medium was taken from each well 72 hr after transfection for analysis in a fluorescent plate reader.

Animals and Animal Care

Subjects for all in vivo experiments were a total of 15 BLA/J mice on a C57BL/67 background bred from mice originally obtained from Jackson Laboratory. Intramuscular experiments used an equal number of male and female littermates. Intravenous experiments used three females for both groups, two males for the Nano-Dysferlin group, and one male for the control group. Subjects were group housed in ventilated cages, with free access to water and mouse chow. The housing room was maintained on a 12L:12D circadian schedule, with lights on at 7 AM. All testing procedures were conducted in strict compliance with the “Guide for the Care and Use of Laboratory Animals” (Institute of Laboratory Animal Resources, National Research Council, 1996) and approved by the Institutional Animal Care and Use Committee of UNC.

Evans Blue Dye Assays

Mice were injected intraperitoneally 40 hr prior to sacrifice with Evans blue dye (10 mg/mL) at 5 μL/g of body weight. Mice were housed in a new environment on the last day prior to sacrifice to exacerbate the relatively mild dysferlin-deficient phenotype. For the positive fiber count assay, muscles were cross-sectioned at a 10-μm thickness over seven locations at least 500 μm apart throughout the muscle. Utilizing fluorescent microscopy, total fibers were counted and compared against positive fibers. For the Evans blue dye absorbance assay, muscle pieces were normalized by weight and placed in Eppendorf tubes. 1 mL of formamide was added and incubated at 55° for 2 hr. Samples were centrifuged at 12,000 rpm for 2 min to remove debris, and supernatants were added to a 96-well plate in triplicate for each muscle. Absorbance was measured at 620 nm in a plate reader. One intravenous mouse did not receive Evans blue dye and was used to quantitate immunofluorescence staining.

H&E Central Nucleation

Muscle cross-sections, as described above, were stained for H&E by the UNC Histology Core. Central nucleated fibers were counted against area in mm², as previously evaluated in the literature.³⁷ Additionally, an alternate measure of central nucleation comparing total intact fibers counted against total central nuclei was also evaluated.³⁸

Oil Red O Staining

Muscle cross-sections, as described above, were stained for Oil Red O by the UNC Histology Core.

Fiber Size Analysis

Muscle sections were stained with WGA lectin and analyzed on ImageJ by first splitting RGB channels and using the find edges function with the green channel. This was followed by applying an auto Huang threshold and using the binary options open function set at a “4” count over ten iterations (black background). This was followed by the binary options fill holes function, and remaining open fiber edges were closed manually. This was followed by the analyze particles function, and the minimal Feret diameter measurement was converted to microns.

Immunofluorescence

Muscle tissue from the intramuscular and intravenous cohort were flash frozen in Sakura TissueTek Cryomolds (REF4557) using optimal cutting temperature solution (OCT) by dipping into isopentane cooled by liquid nitrogen. Tissue was then sliced at 10 μm using a Leica CM3050-S cryostat and stored at −80°C. Tissue was then thawed in a humidity chamber at room temperature. Thawed tissue was fixed for 15 min in 4% paraformaldehyde/4% sucrose solution. Muscle was then stained with WGA-Alexa 488 conjugate at a concentration of 50 μg/mL for 10 min at room temperature. 10% BSA was used to block the tissue, and Abcam ab124684 anti-dysferlin antibody was used at a 1:200 dilution for 2 hr at 37°. Secondary antibody goat anti Rabbit 594 Life Technologies (A11037) was used at a 1:1,000 dilution. Hoeschts stain (H3569) was used at a 1:10,000 dilution for 5 min at room temperature. Coverslips were mounted and imaged in an Olympus IX-83 fluorescence microscope.

Immunofluorescence Fiber Counts

For intramuscular experiments, fibers staining above background for Nano-Dysferlin were counted manually against total fibers based on fiber outlines employing the ImageJ cell counter and multi-point analysis tool. This procedure was carried out in both Nano-Dysferlin and its contralateral GFP controls. GFP control “false positives” were then also subtracted to estimate the approximate Nano-dysferlin expression. It is worth mentioning vector systemic shedding is a common occurrence with AAV, which may account for transduction of the contralateral leg. For systemic experiments, due to expected weaker staining, one treated mouse was not injected with Evans blue dye. In this case, WGA-stained outlines were used to determine total fibers, which were used to normalize the total positive fibers observed. Positive fibers observed in a no-treatment control mouse were used to subtract “false positives.”

Creatine Kinase Assay

Blood drawn from the submandibular vein, approximately 200 uL, was placed in EDTA tubes and centrifuged at 1,500 rpm for 10 min to separate blood solids. Plasma was processed using the creatine kinase activity colorimetric assay kit (Abcam Cat: 155-901) following protocol instructions. Samples were measured in a Perkins colorimetric plate reader.

Rearing Behavioral Assay

The number of times the mice stood on two legs (termed rearing) was quantitated over 60 min at 5-min intervals. Rearing in a novel environment was assessed in a photocell-equipped open field automatic (41 cm × 41 cm × 30 cm; Versamax system, Accuscan Instruments). Activity chambers were themselves placed in sound-attenuating containers equipped with fans and houselights.

Author Contributions

T.L., N.N., C.D., L.C., and M.L.H. performed experiments. R.B.S. and M.L.H. conceived experiments and oversaw experimental execution and result analysis. T.L., C.D., L.C., R.B.S., and M.L.H. wrote the manuscript.

Conflicts of Interest

R.B.S. and M.L.H. are co-inventors on an intellectual property disclosure to UNC and Texas Tech that encompasses Nano-Dysferlin variants.