AAV-mediated BDNF and GAS6 muscle delivery delays disease onset in SOD1^G93A ALS mice

Vector plasmid construction

pAAV plasmids were generated using the pAAV-CMV-teLuc backbone which was constructed by substituting the eGFP gene from the pAAV-CMV-eGFP (Addgene Cat. 67634) with a teLuc gene PCR fragment, utilizing pcDNA3-teLuc c-myc (Addgene Cat. 100026) as the template. The pAAV-P_mus3-teLuc construct contains a sequence -744 to -958 relative to the first exon of human DES gene (hu DES enhancer) and combined with one 2R5S (murine Mck) enhancer [33] and the Mck core promoter [34]. Mature HuBDNF sequence [35],tagged with the Hemagglutinin (HA) sequence, was ligated downstream of the promoter, followed by a wildtype IRES sequence and reporter gene teLuc [36] to make pAAV-P_mus3-HuBDNF-teLuc. pAAV-P_mus7-HuBDNF-teLuc was constructed by replacing the promoter region of pAAV-P_mus3-HuBDNF-teLuc construct with a sk-CRM4 enhancer element and a murine Des promoter [37]. In pAAV-P_mus7-GAS6, human GAS6 cDNA (MN_000820.3) tagged with the Flag tag was ligated in place of the BDNF sequence.

Cell culture and Western blot

HEK293T cells (Takara Bio, 632180) were cultured in growth medium composed of DMEM (4500 mg/L Glucose, Gibco) supplemented with 10% FBS (Gibco), 1x MEM non-essential amino acid (MEM NEAA, Gibco), 1 mM Sodium Pyruvate, 100 U/mL penicillin-streptomycin (Gibco), and 1.5 g/L Sodium Bicarbonate. C2C12 muscle cells (ATCC, CRL-1772) were maintained in DMEM (4500 mg/L Glucose, Gibco) supplemented with 10% FBS, and differentiated in DMEM (4500 mg/L Glucose, Gibco) supplemented with 2% horse serum. For western blot, C2C12 cells were infected with AAV vectors at MOI 10,000 on differentiation day 4 in a 100 mm culture dish. 4 days after infection, cell pellet and culture medium were collected and tested for expressed and secreted recombinant protein. Cell pellets were lysed in NP-40 (50 mM Tris-HCl, 150 mM NaCl, and 1% NP-40) with 1x proteinase inhibitor (Sigma P8340). Cell lysates were then centrifuged at 18,000 × g for 20 min and supernatants were collected and boiled with 5% of β-mecaptoethanol and 4x Laemmli buffer. Protein concentrations were determined with Pierce BCA Protein Assay Kit (Thermo Scientific 23225). Secreted recombinant proteins were concentrated by immunoprecipitation following the manufactures’ instruction (Anti-DYKDDDDK G1 Affinity Resin, GenScript L00432; Pierce Anti-HA Magnetic Beads, Thermo Scientific 88836). Recombinant huBDNF was revealed with anti HA-Tag (C29F4) Rabbit mAb (Cell Signaling 3724) and Goat-anti-rabbit secondary (Thermo Scientific #32460). Recombinant huGAS6 was revealed with anti-Flag Mouse mAb (Sigma F1804) and Goat-anti-mouse secondary antibody, HRP (Thermo Fisher Scientific #62-6520).

Virus production and purification

AAV.PHP.eB was produced using Gibco AAV-Max Helper-Free AAV production System (A51217). In brief, Viral Production Cells (VPC) were cultured in the Viral Production medium. When reaching 3 million/mL concentration, VPCs were transfected with recombinant pAAV, pHelper (Takara Bio #6234), and pUCmini-iCap-PHP.eB [38] using the AAV-Max transfection kit. 70–72 h post transfection, cells were lysed by adding AAV-Max Lysis buffer, and cell debris were removed by centrifuging at 3,500 × g for 30 min at 4 °C. Virus particles in the supernatants were precipitated with polyethylene glycol (PEG) 8000 solution (8% PEG and 0.5 M NaCl final concentration) and centrifuged at 5000 × g for 1 h at 4 °C. Concentrated AAVs were purified by two rounds of CsCl gradient ultracentrifugation (38,000 × g, 18 h, 20 °C). Virus titers were determined by qPCR with the following primers:

CRM4+: 5’-AGCCGTCTCCCTAGCAGCCA-3’.

DES-: 5’-TCCTCAGGACATCGGGTCCTAGA-3’.

teLuc+: 5’-CCGGCTACAACTTGAGTCAAGTCC-3’.

teLuc-: 5’-CTTCATACGGGATGATGACATGGATGTC-3’.

Immunostaining

Mouse spinal cords were obtained by the hydraulic extrusion protocol [39]. In brief, segment L1-L6 lumbar spinal cord was dissected; Isolated lumbar spinal cord tissues were soaked in 4% paraformaldehyde (PFA) (Thermo Fisher AAJ19943K2) for 24 h followed by immersion in 15% and 30% sucrose solution until they settle down to the bottom of the tube. A total of 100 sections were collected per animal across 10 slides. Slides containing comparable anatomical regions were selected for staining, and motor neurons were quantified. The average motor neuron count per ventral horn was calculated and presented. Sections were treated with prewarmed (60 °C) antigen retrieval buffer (10 mM Sodium Citrate, pH 6.0) for 30 min before primary goat ChAT antibody (Sigma Ab144p, 1:100) incubation for 18 h followed by peroxidase conjugated secondary antibody (Abcam AB214881, 1:1000). DAB (Abcam ab64238) was used as the chromogenic reagent. ChAT positive cells in the ventral horn were counted.

ELISA

Heart, triceps branchii (TB), quadriceps (Quad) muscles, and brain tissues were dissected from both male and female animals and lysed in NP-40 buffer (1% NP-40, 150 mM NaCl, and 50 mM Tris-Cl, pH8) supplemented with 1% (v/v) Proteinase Inhibitor Cocktail (Sigma P8340) for 30 min on ice, supernatants were collected after centrifuge at 1800 g for 20 min. The supernatants were used for recombinant protein detection by the Human BDNF Elisa Kit (Sigma RAB0026) and DYKDDDDDK-tag Detection Elisa kit (Cayman No. 501560).

Transgenic mice

Transgenic mice with the G93A human SOD1 mutation (B6SJL.SOD1-G93A) were originally obtained from the Jackson Laboratory (Bar Harbor, ME, USA). Colonies were expanded by crossing with C57BL/6 J females. The offspring were genotyped by PCR of DNA extracted from tail tissue. Animal numbers of each experimental group were determined by resource equation approach (group comparison, one-way ANOVA). The following experimental groups of mice were used: male SOD1^G93A mice injected with AAV-P_mus7-teLuc (male control, n = 11), AAV-P_mus7-BDNF-teLuc (male BDNF, n = 11) or AAV-P_mus7-GAS6-teLuc (male GAS6, n = 11); female SOD1^G93A mice injected with AAV-P_mus7-teLuc (female control, n = 13), AAV-P_mus7-BDNF-teLuc (female BDNF, n = 10) or AAV-P_mus7-GAS6-teLuc (female GAS6, n = 11), each group of animals were allocated based on weight and litter to ensure similar group composition. Transgenic mice were injected at 6 weeks of age with a total of 2 × 10¹² virus genomes (vg) both intravenously (1 × 10¹² vg, retro-orbital) and intramuscularly (5 × 10¹¹ vg) into the quadriceps of each hindlimb. Investigators were not blinded to the group allocation and when assessing the outcomes.

Behavioral studies

Recombinant AAV vector construction and neurotrophic factor expression

To identify which neurotrophic factors might promote survival of the spinal motors neurons, we analyzed receptor tyrosine kinase (RTK) expression in adult mouse spinal cord using RNA-seq profiling data of spinal motor neuronss in ALS mice [10]. Our results showed that Ret, Ntrk2, and Tyro 3 are among the top RTKs which have higher expression levels in these cells (Table 1). The Ret receptor ligand glial-derived neurotrophic factor (GDNF) has been shown to have neuroprotection and extend life span in mouse ALS models; however, there is no published study on AAV delivery of Ntrk2 receptor ligand BDNF in ALS [41]. The role in ALS of GAS6 which is the Tyro3 receptor ligand has not previously been reported in ALS, either. Although their RPKM values did not significantly differ between groups, we highlighted these genes due to their established neuroprotective roles and their potential to counteract ALS-related neurodegeneration.

In this study, we aimed to express trophic factors BDNF and GAS6 in muscle cells to support innervated motor neuron survival in ALS model mice. Our earlier study showed that the AAV variant PHP.eB had higher expression level in muscle cells and tissues than AAV2-MTP and AAV6 [42], so this AAV capsid was used in this study.

We designed two constructs based on previously published muscle specific promoters and enhancers, as well as using regulatory elements from genes that are highly expressed in the muscle tissue. The P_mus3 promoter was a combination of a human DES enhancer, a 2R5S enhancer [33] and a murine Mck promoter (Fig. 1A). The Pmus7 promoter was adopted from Sarcar et.al. which contains a skCRM4 element and a murine Desmin promoter (Fig. 1A). BDNF-HA were cloned into the AAV-P_mus3 and AAV-P_mus7 vectors together with te-luciferase (teLuc) as a reporter gene [36] to test protein expression in vitro.

Differentiated C2C12 myotubes were infected with AAV-P_mus7-teLuc (control virus), AAV-P_mus3-BDNF-teLuc, and AAV-P_mus7-BDNF-teLuc (MOI 10,000). At 4 days post infection, cell lysates and culture medium were collected for the detection of BDNF expression. BDNF proteins in the medium were first concentrated using anti- HA magnetic beads. Western blots showed that both AAV-P_mus3-HuBDNF-teLuc and AAV-P_mus7-HuBDNF-teLuc vector can express the recombinant protein in C2C12 myotubes, however, AAV-P_mus7-HuBDNF-teLuc vector had higher expression levels of BDNF, which is consistent with the reporter gene assay (Fig. 1B). Thus, we chose to use the AAV-P_mus7-HuBDNF-teLuc as the potential therapeutic vector.

To further elucidate the muscle tissue specificity of these vectors, recombinant AAV viruses with the reporter gene te-Luciferase under the transcriptional control of the pan-cell type promoter CMV or muscle- specific promoter P_mus7 were packaged with AAV2 which infects HEK 293 T cells well or PHP.eB, respectively. After transduction of HEK 293 T cells, robust luciferase activity from the AAV2-CMV-teLuc infected cells were observed, whereas the cells infected with AAV2-P_mus7-teLuc showed limited expression compared to the CMV group (Fig. S1A). In contrast, with PHP.eB capsid, P_mus7 directed higher te-Luciferase expression than CMV in C2C12 muscle cells (Fig. S1B), which indicated the promoter specificity of this vector.

GAS6-FLAG was then cloned into the pAAV-P_mus7 vector. The protein expression and secretion were also validated by transducing the differentiated C2C12 myotubes followed by Western Blot. The secreted GAS6 in the culture medium was first concentrated by anti-FLAG affinity resin. AAV-P_mus7-HuGAS6 also showed protein expression both in the cell lysates and the culture medium (Fig. 1B).

The effects of BDNF and GAS6 expression in skeletal muscle on disease onset

After we confirmed that our AAV vectors were able to successfully express the trophic factors in the muscle cells in vitro, we tested their efficacy in vivo in the ALS mouse model. Directly injecting the virus into skeletal muscles was considered as potentially more efficient for muscle specific expression. In our earlier studies, a re-inducible muscle-specific expression vector, packaged in the same capsid, was administered intravenously to mice, and the expression level of the reporter gene (te-luc) was tracked following repeated inductions. After 18 weeks, reporter gene expression remained sustained following the third induction, indicating long-term expression after systemic injection [42]. In contrast, the same vector was administered intramuscularly to a separate group of mice. After the second round of induction, expression levels decreased by 42%, and by 17% after the third induction, compared to the initial induction (data not shown). To achieve the benefit of both local and systemic routes, we injected 1 × 10¹² vg intravenously (retro-orbital), and 0.5 × 10¹² vg intramuscularly into the skeletal muscle of both hindlimb at 6 weeks of age.

After virus injection, animal bodyweight and neurological score changes were monitored and recorded. Increases in NS score was significantly delayed in BDNF female mice compared with controls (Fig. 2B; 129.3 versus 103.2 days, respectively; p < 0.001). Male BDNF mice also showed significantly delayed NS changes compared to the control group (Fig. 2D; 120.5 versus 110.1 days, respectively; p < 0.01), although the improvement was less than that observed in female mice. Weight loss is another sign of disease onset in transgenic mice. We found that BDNF treated female mice experienced delayed body weight loss compared to control (Fig. 2A; 127 versus 114.5 days, respectively; p < 0.001). Again, in male BDNF treated mice, the onset of weight loss was delayed (Fig. 2C;123.1 versus 116.4 days, respectively), but less so than in female mice.

Expression of GAS6 also delayed disease onset in a similar manner. In female mice, the average NS change was delayed by GAS6 treatment (Fig. 2F; 127 versus 114.5 days, respectively; p < 0.0001). Again, the average NS change in male mice was delayed (Fig. 2H; 117.7 Versus 110.1 days, respectively; p < 0.05) but less so than observed in female mice. The average body weight change in female GAS6 treated mice was also significantly delayed (Fig. 2E; 131.8 versus 114.5, respectively; p < 0.0001); similarly, the time of body weight change was also delayed in male GAS6 group compared to control (Fig. 2G; 121.5 versus 116.4 days, respectively).

Based on these observations, both BDNF and GAS6 significantly delayed disease onset, as measured with both NS changes and bodyweight loss in female mice. Male mice showed significantly delayed NS change, but not the delay in weight loss compared to the control group.

The effects of BDNF and GAS6 expression in skeletal muscle on disease progression

Our data showed that muscle-specific expression of both BDNF and GAS6 treatment delayed disease onset in ALS transgenic mice. To examine the effect of these trophic factors on disease progression, the rate of deterioration in neurological score was monitored through the course of disease, from onset until termination. We used scales from 0 (pre-symptomatic) to 4 (considered moribund and requiring euthanasia) [40]. To evaluate the locomotor performance, rotarod tests were also conducted.

Both male and female BDNF mice showed significant reduction in disease progression as measured by NS (Fig. 3A, B). In female mice, BDNF treatment also significantly improved locomotor performance as measured by rotarod (Fig. 3C); however, no significant improvement was observed in male mice (Fig. 3D). Similarly, GAS6 treated mice exhibited a significant reduction in NS in both male and female mice (Fig. 3E, F), but improvements in rotarod performances were not significant (Fig. 3G, H).

Effect of BDNF and GAS6 treatment in life span of transgenic mice

To evaluate if BDNF and GAS6 treatment had effects on life span, we conducted Kaplan–Meier survival analysis of the percentage of surviving animals after AAV-BDNF/GAS6 administration versus control. However, neither BDNF nor GAS6 (Fig. 4) treated mice showed statistically significant extension of life.

Effects of BDNF and GAS6 on spinal motor neuron survival

To assess whether expression of BDNF and GAS6 delays lower motor neuron loss in ALS, mice at 75 days of age (pre-symptomatic), at 115 days of age (initiation of symptoms) and end-stage animals in control, BDNF and GAS6 treated groups (n = 3–5, mixed sexes) were euthanized and lumbar spinal cord were cryo-sectioned into 20 μm slices. ChAT staining was used as a motor neuron marker, and the number of ChAT positive cells in the ventral horn were quantified.

At 75 days of age (Fig. 5A–D), there were no differences in motor neuron counts in each ventral horn between the wild-type, control, BDNF and GAS6 groups (8.2, 8.9, and 8.3 motor neurons, respectively). At 115 days of age, animals started to show lumbar motor neuron loss in mice injected with the control vectors; the BDNF treated 115-days old group animals had significantly more motor neurons remaining compared to the control group (7.8 versus 6.3 motor neurons, respectively; p < 0.05; Fig. 5E–H). When the animals reached end stage (Fig. 5I–L), an average of 2.6 motor neurons remained in the control group, indicating continued motor neuron death. BDNF and GAS6 treated animals had significantly more ChAT positive cells in the ventral horn (4.3 and 5.3 MN, respectively; p < 0.01, p < 0.0001, respectively versus control).

Expression of BDNF and GAS6 over time in SOD1^G93A mice

Our data show that BDNF and GAS6 expression mediated by recombinant AAV vectors in ALS mice significantly delays the disease onset, restrains disease progression, and prevents motor neuron loss in the lumbar spinal cord. However, time to reach end-stage disease and overall lifespan were not improved significantly.

To assess the level of expression of the trophic factors in targeted tissues of treated animals, and how their levels were maintained over time, control and treated ALS mice were euthanized at 1 month or 2 months after virus injection, and at the end stage disease. Extracts of dissected heart, triceps brachii (TB), quadriceps (Quad), and brain tissues were prepared and lysed for determination of BDNF and GAS6 levels by ELISA.

The ELISA kit for human BDNF also detected endogenous mouse BDNF, because of high homology between the two species. Thus, untreated, control ALS mice tissue extracts exhibited reactivity. In BDNF-treated animals, total BDNF levels (mouse and human) were higher than control in every sample tested, reflecting expression derived from the vector. However, the virus derived protein (in treated mouse sample) appeared to decrease as the disease progressed (Fig. 6A).

Virus-coded GAS6 protein levels were specifically quantified by ELISA measurement of its FLAG tag. Again, significant amounts of protein were detected in all samples, which generally fell as disease progressed (Fig. 6B). To further validate the tissue-specific expression of this vector, GAS6-FLAG mRNA levels were measured by RT-PCR. Consistent with the protein data, mRNA expression was higher in skeletal muscles compared to heart and brain tissues (Fig. S2).