A translatable RNAi-driven gene therapy silences PMP22/Pmp22 genes and improves neuropathy in CMT1A mice

Charcot-Marie-Tooth disease type 1A (CMT1A), the most common inherited demyelinating peripheral neuropathy, is caused by PMP22 gene duplication. Overexpression of WT PMP22 in Schwann cells destabilizes the myelin sheath, leading to demyelination and ultimately to secondary axonal loss and disability. No treatments currently exist that modify the disease course. The most direct route to CMT1A therapy will involve reducing PMP22 to normal levels. To accomplish this, we developed a gene therapy strategy to reduce PMP22 using artificial miRNAs targeting human PMP22 and mouse Pmp22 mRNAs. Our lead therapeutic miRNA, miR871, was packaged into an adeno-associated virus 9 (AAV9) vector and delivered by lumbar intrathecal injection into C61-het mice, a model of CMT1A. AAV9-miR871 efficiently transduced Schwann cells in C61-het peripheral nerves and reduced human and mouse PMP22 mRNA and protein levels. Treatment at early and late stages of the disease significantly improved multiple functional outcome measures and nerve conduction velocities. Furthermore, myelin pathology in lumbar roots and femoral motor nerves was ameliorated. The treated mice also showed reductions in circulating biomarkers of CMT1A. Taken together, our data demonstrate that AAV9-miR871–driven silencing of PMP22 rescues a CMT1A model and provides proof of principle for treating CMT1A using a translatable gene therapy approach.


Study design
The goal of this study was first to design new U6 promoter-driven artificial miRs and assess their efficiency to silence human PMP22 and murine Pmp22 in vitro, and then package the lead sequence into AAV9 viral vector for in vivo efficacy studies in C61-het mice, a model of CMT1A. In vitro assessment was performed in HEK293 cells co-transfected with CMV-driven huPMP22 or muPmp22 plasmids along with the candidate miRs, followed by RT-qPCR to quantify gene silencing efficacy.
Early-and late-treated animals were injected at 2 or 6 months of age, respectively, and their motor performance was monitored every two months with behavioral testing until 4 months postinjection. At the final time point of each treatment group, mice were analyzed for PMP22/Pmp22 downregulation and for the expression of other myelin related genes/proteins with RT-qPCR and WB; with nerve electrophysiology testing measuring MNCV and CMAP; axonal degeneration biomarkers including circulating NF-L and Gdf15 levels; PNS tissues morphological evaluation using semithin sectioning analysis; inflammatory response using immunohistochemistry; and finally AAV9 vector biodistribution with VGCN analysis.
Extended early treated animals were injected at 2 months of age and their motor performance was monitored every two months with behavioral testing until 8 months post-injection. At the final time point of this treatment group mice were also analyzed for MNCV and CMAP scores with nerve electrophysiology testing, for axonal degeneration with plasma NF-L levels analysis, for PNS tissues morphological evaluation using semithin sectioning analysis and for AAV9 vector biodistribution with VGCN analysis.

miR sequence generation and cloning
Artificial miRs were designed based on natural human mir-30, maintaining important structural and sequence elements required for normal miRNA biogenesis but replacing mature mir-30 sequences with 22-nt of complementarity with the PMP22 gene. Criteria for miRNA selection were previously described (1, 2). Briefly, a potential 22 nucleotide mature miRNA duplex must meet four criteria (relative to the mature antisense miRNA sequence): (1) the four 3' nucleotides must be at least 75% G:C; (2) the four 5' nucleotides must be 75% A:U; (3) Overall A:U content of at least 40%; (4) lack of RNA pol III termination signal (TTTTTT) in the pre-miRNA sequence. All sequences were cloned as DNA transcription templates in front of the mouse U6 promoter into the U6T6 plasmid as described (1), sequenced verified and used for in vitro screening prior to AAV vector production. Control microRNAs targeting EGFP or LacZ were previously described (3,4).
HEK293 cells were co-transfected using Lipofectamine 2000 (Invitrogen, 11668027) with a plasmid expressing either muPmp22 or huPMP22 gene and a plasmid expressing the U6.miRPMP22 miRs, or controls, at a molar ratio 1:4. Negative controls included empty U6T6 plasmid (no miR) or one expressing a miRNA targeting EGFP (miGFP). Twenty-four hours after the co-transfection, RNA was isolated from the cells using Trizol Reagent (Ambion, 15596018) following manufacturers' protocol. After DNase treatment, RNA was quantified and 1 μg of RNA was used to synthesize cDNA using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, N8080234). Then levels of huPMP22 (Hs00991884_m1), muPmp22 (Mm01333393_m1) and huRPL13A (Hs04194366_g1) were quantified using Taqman gene expression assays (Applied Biosystems). Human RPL13A was used as an endogenous control. Data were collected from 3 independent experiments. All QPCR assays were performed in triplicate.

Production of AAV vectors
U6.miRPMP22 cassettes were subcloned into self-complementary AAV backbones (5) containing separate CMV.eGFP cassettes. Viral vectors were produced by Andelyn Biosciences (Ohio, US) using triple transfection into HEK293 cells of pro-viral plasmids, pHELPER (containing adenoviral helper genes), and a plasmid containing AAV2 rep and AAV9 cap genes. The 3' inverted terminal repeats (ITR) of AAV2 contained a deletion of the terminal resolution site (trs) to enable formation of double-stranded AAV genomes (6). All vectors were purified by iodixanol gradient ultracentrifugation followed by FPLC purification, then titrated by ddPCR using primer/probes to detect the AAV2 ITR.

Experimental animals and lumbar intrathecal injection
All experimental procedures in this study were conducted in accordance with animal care protocols approved by the Cyprus Government's Chief Veterinary Officer (project license CY/EXP/PR.L3/2017) according to national law, which is harmonized with EU guidelines (EC Directive 86/609/EEC). All inoculations were performed under anesthesia and all efforts were made to minimize animal suffering. The protocols were approved by Cyprus Government's Chief Veterinary Officer. In this study, we used adult WT C57BL/6 or C61 Het mice. The C61 Het colony was established from two breeding pairs gifted by Prof R. Martini (Universitäts-Klinikum Würzburg, Germany). C61 Het mice carry 4 copies of human PMP22 along with normal endogenous murine Pmp22 (7). Early, extended-early and late gene therapy trials were conducted using 2-or 6-month-old mice C61 Het mice. Mice were kept in a specific pathogen-free animal facility, housed in open-top system cages. Up to five mice were housed in cages linked with high absorbency wood bedding for laboratory mice, dried by high-temperature treatment, sieved, dedusted, prior to use. Standard mouse diet, certificate for reproduction, weaning, growth, and tap potable water filtered and UV sterilized were administered to the mice. Mice were kept in a 12 h dark/12 h light cycle at a temperature of 22 °C. Both male and female mice were used in our experiments and showed no sex related differences in their behavioral performance or nerve pathology.
Intrathecal injection was performed as described before (8)(9)(10)(11)(12)(13)(14). Briefly, a small skin incision was made along the lower lumbar spine level of anesthetized mice to visualize the spine and the AAV vector was delivered into the L5-L6 intervertebral space. A 50-μL Hamilton syringe (Hamilton, 80530/00) connected to a 26-gauge needle (Hamilton, 7758-02/00) was used to inject 20 μL of AAV stock containing an estimated 5x10 11 vector genomes (vg), at a maximum rate of 1 μL/15sec. A flick of the tail was considered indicative of successful intrathecal administration.
Vector genome copy number (VGCN) determination VGCN was determined as already described in previous studies (13,14). Briefly, genomic DNA was extracted from different PNS and CNS tissues (i.e., lumbar roots, proximal, middle and distal sciatic nerve, femoral nerve, brain, liver, trigeminal and spinal cord) of mice after intrathecal vector delivery using the MagPurix Tissue DNA Extraction Kit (Zinexts Life Science, ZP02004). DNA yield and purity was quantified using a Nanodrop 1000 spectrophotometer. Approximately 20 ng of DNA was used as template for two real-time PCR assays on an Applied Biosystems 7500 Real-Time PCR System involving 45 cycles of 15 s at 95 °C and 60 s at 60 °C. TFRC-specific primers/probe targeting the mouse genome and EGFP-specific primers/probe, which is contained in the transgene, were used. Standard curves were created by serial dilution of quantified mouse genomic DNA, as well as quantified plasmid DNA containing the transgene cassette. The average VGCN per cell was calculated as the total VGCN divided by the total cell number.

Behavioral testing
For rotarod test, animals were trained on an accelerating rotarod apparatus (Ugo Basile, 7650) for three consecutive days by three trials per day with 15-min rest period between trials. The mice were placed on the rod, and the speed was gradually increased from 2.5 to 25 rpm. The trial lasted until the mouse fell from the rod or after the mouse remained on the rod for 600 s and was then removed. Testing was performed on the fourth day using two different speeds, 5 and 17.5 rpm. Latency to fall was calculated for each speed.
To measure hind limb grip strength, a mouse was held by the tail and lowered towards the apparatus (Ugo Basile, 57107) until it grabbed the grid only with the hindlimb paws. Mice were gently pulled back until they released the grid. Each session consisted of six consecutive trials. Measurements of the force in g were indicated on the equipment. Hind limb force was calculated by averaging the scores of each trial for each animal.
For wire hang testing, animals were placed atop a wire, which was then inverted; causing the mice to hang from the paws. Latency to fall was then recorded. This test was performed once a day for three days and then the average performance was calculated and reported.
For hindlimb clasping evaluation, mice were suspended by the base of the tail and three pictures captured every 5 seconds. The average angle of each mouse hindlimb opening was calculated using ImageJ software.

Electrophysiological analysis
For MNCV and CMAP measurements, the bilateral sciatic nerves were stimulated in anesthetized animals at the sciatic notch and distally at the knee via bipolar electrodes with supramaximal square-wave pulses (5 V) of 0.05 ms. MNCV was calculated by dividing the distance between the stimulating and recording electrodes by the result of subtracting the distal latency from the proximal latency. The latencies of CMAP were recorded by a bipolar electrode inserted between digits 2 and 3 of the hind paw and measured from the stimulus artefact to the onset of the negative M-wave deflection. A fixed distance was used between distal stimulation and recording sites for calculating distal latency to avoid errors arising from variations in knee-paw distance in each mouse.

Plasma neurofilament light (NF-L) levels
Blood samples were collected from retro-orbita as previously described and processed within one hour (14). Blood samples were collected in EDTA-containing tubes and centrifuged at 20 °C at 3500 rpm for 10 minutes. Centrifugation separated blood samples in two phases and the top plasma phase was collected and stored at −80 °C until testing. Plasma NF-L concentration was measured at University College London (UCL) using a commercially available NF-Light kit on a Single molecule array (Simoa) HD-X instrument (Quanterix) (15,16).

Serum growth differentiation factor 15 (Gdf15) levels
Blood samples were collected from retro-orbita as previously described (17). Blood samples were collected in serum separation tubes, allowed to clot for 15-30 minutes and centrifuged at 20 °C at 3000 g for 10 minutes. Serum stored at −80 °C until testing. Serum protein levels were determined at the Department of Clinical Neurosciences, University of Cambridge, by ELISA according to manufacturer's instructions for growth differentiation factor 15 (R&D Systems, MGD150).

Morphometric analysis of myelination in lumbar roots and peripheral nerves
Morphometric analysis was performed as described before (9,11,14). Mice were transcardially perfused with 2.5% glutaraldehyde (Agar, R1010) in 0.1M PB buffer. The lumbar spinal cord with multiple spinal roots attached, as well as the femoral and sciatic nerves, were dissected and fixed overnight at 4 °C, then osmicated (SPI, 02601-AB), dehydrated, and embedded in resin (mixture of 17% Araldite resin (Agar,R1040), 25.5% Agar 100 (Agar,R1043), 55.5% dodecenylsuccinic anhydride (Agar, R1051), 2% 2,4,6-tri(dimethylaminomethyl)phenol (Agar,R1064)). Transverse semi-thin sections (1 μm) of the lumbar spinal cord with roots and the middle portion of the femoral motor and sciatic nerves were obtained and stained with alkaline toluidine blue (SPI,02576-AB). Sections were captured using a Nikon Eclipse Ni microscope with a digital camera (DS-Fi3) using NIS Elements software. Images of roots or transverse femoral nerve sections were obtained at 200x and 400x final magnification, respectively. Series of partially overlapping fields covering the entire cross-sectional area of the roots or the nerves were captured at 600× final magnification. Sciatic nerves detailed pictures were obtained at 1000x final magnification.
In brief, all demyelinated, thinly myelinated and normally myelinated axons were counted using the following criteria: axons larger than 1 μm without a myelin sheath were considered demyelinated; axons with myelin sheaths <10% of the axonal diameter were considered thinly myelinated; axons surrounded by circumferentially arranged SC processes and extracellular matrix were considered as "onion bulbs"; all other myelinated axons were considered normally myelinated.      In tissues of AAV9-miRLacZ injected mice EGFP autofluorescence is also seen. Arrowheads are pointing to indicative CD+ cells. Data were compared using One-way ANOVA with Tukey's Multiple Comparison Test followed by Bonferroni correction. Significance level of all comparisons, P<0.05. Scale bars: 20 μm. Immunostaining images of 6month-old non-injected CMT1A mice, as well as quantification data of 6-month-old non-injected WT and CMT1A mice are also used in Figure 5. In tissues of AAV9-miRLacZ injected mice EGFP autofluorescence is also seen. Arrowheads are pointing to indicative CD+ cells. Data were compared using One-way ANOVA with Tukey's Multiple Comparison Test followed by Bonferroni correction. Significance level of all comparisons, P<0.05. Scale bars: 20 μm. Quantification data of 6 months old non-injected WT and CMT1A mice are also used in Figure 5. was calculated in relation to the total cell number (Mean, SD). Nuclear staining with DAPI (blue). In tissues of AAV9-miRLacZ injected mice EGFP autofluorescence is also seen. Arrowheads are pointing to indicative CD+ cells. Data were compared using One-way ANOVA with Tukey's Multiple Comparison Test followed by Bonferroni correction. Significance level of all comparisons, P<0.05. Scale bars: 20 μm.

Fig. S11. Lack of inflammatory reaction in the dorsal root ganglia (DRGs) of AAV9-miRLacZinjected CMT1A mice at 4 months post injection (6 months of age).
Representative images of DRGs sections from CMT1A mice injected with AAV9-miRLacZ at 4 months post injection compared to agematched non-injected CMT1A mice, immunostained for various inflammatory cell markers, as indicated (A-G). Arrowheads are pointing on cells positive for the B-cell marker CD20 (red) (A, B), leukocyte marker CD45 (red) (C, D), macrophage marker CD68 (green) (E, F), or T-cell marker CD3 (red) (G, H). Nuclear staining with DAPI (blue). In tissues of AAV9-miRLacZ injected mice EGFP green autofluorescence is also visible, including several DRG neuron cell bodies.     A-B), as indicated. Tissues were immunostained against B-cell marker CD20 (green or red in tissues of injected animals as indicated), leukocyte marker CD45 (red), macrophage marker CD68 (green) and T-cell marker CD3 (red), and counterstained with DAPI (blue). In tissues of AAV9-miR871 injected mice EGFP autofluorescence is also seen. Arrowheads are pointing to indicative CD+ cells. Scale bar: 20 μm. Immunostaining images of 6-month-old non-injected CMT1A and CMT1A-miR871 mice are also shown in Figure 5.  A-B), as indicated. Tissues were immunostained against B-cell marker CD20 (green or red in tissues of injected animals as indicated), leukocyte marker CD45 (red), macrophage marker CD68 (green) and T-cell marker CD3 (red), and counterstained with DAPI (blue). In tissues of AAV9-miR871 injected mice EGFP autofluorescence is also seen. Arrowheads are pointing to indicative CD+ cells. Scale bar: 20 μm.         A-B), as indicated. Tissues were immunostained against B-cell marker CD20 (green or red in tissues of injected animals as indicated), leukocyte marker CD45 (red), macrophage marker CD68 (green) and Tcell marker CD3 (red), and counterstained with DAPI (blue). Arrowheads are pointing to indicative CD+ cells. In tissues of AAV9-miR871 injected mice EGFP autofluorescence is also seen. Scale bar: 20 μm (Immunostaining images of 10-month-old non-injected CMT1A and CMT1A-miR871 mice are also used in Figure 8).  Cell nuclei were counterstained with DAPI (blue). In tissues of AAV9-miR871 injected mice EGFP autofluorescence in hepatocytes is also seen. Arrowheads are pointing to indicative CD+ cells. Data were compared using One-way ANOVA with Tukey's Multiple Comparison Test. Significance level of all comparisons, P<0.05. Scale bar: 20 μm.

Effects of AAV9-miR871 on WT mice
To assess potential toxic effects of Pmp22 over-silencing, we injected the AAV9-miR871 and -miRLacZ vectors into 2-months-old WT mice, expressing only normal levels of muPmp22, and evaluated effects 6 weeks later with western blot, and then 4 months post-injection with real-time PCR, western blot analysis, behavioral testing, NF-L testing, electrophysiological examination, and morphometric observation (Fig. S28). WT mice injected with AAV9-miR871 showed reduced of Pmp22 protein levels in spinal roots, sciatic and femoral nerves. In addition, MPZ protein levels were reduced in sciatic nerves at 6-weeks and 4 months post-injection, suggesting abnormal myelination (Fig. S29). Pmp22 mRNA showed variable reduction in AAV9-miR871-treated WT mice 4 months after injection that reached significance only in spinal roots. A similar trend was observed for muMpz levels. muGjb1 transcripts increased in AAV9-miR871-treated WT sciatic and femoral nerves ( Fig. S30 and table S4). We assessed motor performance in all mice groups before injection and until the end of the observation period using rotarod at 5 and 17.5 rpm, grip strength, and hang test analyses (Fig. S31). AAV9-miR871-treated WT mice showed transiently impaired rotarod performance (at both speeds) at 2 months post-injection with recovery in function 2 months later (4 months post-injection). They also showed attenuated grip strength performance at all-time points and worsening in hang test performance at 4 months post injection. However, they did not present hindlimb clasping phenotype (Fig. S32) or MNCV deficits 4 months after vector injection (Fig. S33) but CMAP scores were significantly reduced (AAV9-miR871: 4.62±1.22, AAV9-miRLacZ: 7.03±0.72, WT: 6.89±1.76). Interestingly, NF-L plasma levels were not elevated (Fig. S34) and morphological analysis did not reveal any apparent morphological defects (Fig. S35) in AAV9-miR871 injected WT mice.        S1: CMT1A mice mRNA transcripts fold change at 6 weeks post intrathecal injection of AAV9.miR871. Quantitative real-time PCR analysis of huPMP22 and muPmp22 as well as of muMpz, muCnp, mu Gldn and muGjb1 gene expression in lumbar roots, sciatic and femoral nerves at 6 weeks postinjection (n=3/group). Fold relative mRNA expression levels of AAV9-miR871 injected CMT1A mice were calculated compared to AAV9-miRLacZ injected CMT1A (Mean ± SD). All samples were normalized to endogenous control Gapdh. Table S2: Early treated CMT1A mice mRNA transcripts fold change at 4 months post intrathecal injection of AAV9.miR871. Quantitative real-time PCR analysis of huPMP22 and muPmp22 as well as of muMpz, muCnp, mu Gldn and muGjb1 gene expression in lumbar roots, sciatic and femoral nerves at 6 weeks post-injection (n=4/group). Fold relative mRNA expression levels of AAV9-miR871 injected CMT1A mice were calculated compared to AAV9-miRLacZ injected CMT1A (Mean ± SD). All samples were normalized to endogenous control Gapdh. Table S3: Late treated CMT1A mice mRNA transcripts fold change at 4 months post intrathecal injection of AAV9.miR871. Quantitative real-time PCR analysis of huPMP22 and muPmp22 as well as of muMpz, muCnp, mu Gldn and muGjb1 gene expression in lumbar roots, sciatic and femoral nerves at 6 weeks post-injection (n=4/group). Fold relative mRNA expression levels of AAV9-miR871 injected CMT1A mice were calculated compared to AAV9-miRLacZ injected CMT1A (Mean ± SD). All samples were normalized to endogenous control Gapdh. Table S4: WT mice mRNA transcripts fold change at 4 months post intrathecal injection of AAV9.miR871. Quantitative real-time PCR analysis of muPmp22 as well as of muMpz, muCnp, mu Gldn and muGjb1 gene expression in lumbar roots, sciatic and femoral nerves at 6 weeks post-injection (n=3/group). Fold relative mRNA expression levels of AAV9-miR871 injected CMT1A mice were calculated compared to AAV9-miRLacZ injected CMT1A (Mean ± SD). All samples were normalized to endogenous control Gapdh.