Gene Therapy (2021) 28:659–675

https://doi.org/10.1038/s41434-021-00250-0

ARTICLE

AAV9-mediated Schwann cell-targeted gene therapy rescues a

model of demyelinating neuropathy
Alexia Kagiava 1 Christos Karaiskos1 Jan Richter2 Christina Tryfonos2 Matthew J. Jennings3
● ● ● ● ●

Amanda J. Heslegrave4 Irene Sargiannidou1 Marina Stavrou1 Henrik Zetterberg5,6,7,8 Mary M. Reilly4
● ● ● ● ●

Christina Christodoulou2 Rita Horvath3 Kleopas A. Kleopa 1,9

● ●

Received: 17 December 2020 / Revised: 15 February 2021 / Accepted: 19 February 2021 / Published online: 10 March 2021
© The Author(s) 2021. This article is published with open access

Abstract
Mutations in the GJB1 gene, encoding the gap junction (GJ) protein connexin32 (Cx32), cause X-linked Charcot-Marie-
Tooth disease (CMT1X), an inherited demyelinating neuropathy. We developed a gene therapy approach for CMT1X using
an AAV9 vector to deliver the GJB1/Cx32 gene under the myelin protein zero (Mpz) promoter for targeted expression in
Schwann cells. Lumbar intrathecal injection of the AAV9-Mpz.GJB1 resulted in widespread biodistribution in the peripheral
1234567890();,:
1234567890();,:

nervous system including lumbar roots, sciatic and femoral nerves, as well as in Cx32 expression in the paranodal non-
compact myelin areas of myelinated fibers. A pre-, as well as post-onset treatment trial in Gjb1-null mice, demonstrated
improved motor performance and sciatic nerve conduction velocities along with improved myelination and reduced
inflammation in peripheral nerve tissues. Blood biomarker levels were also significantly ameliorated in treated mice. This
study provides evidence that a clinically translatable AAV9-mediated gene therapy approach targeting Schwann cells could
potentially treat CMT1X.

Introduction velocities [2–5]. CMT1X affects mostly men with onset at

the age of 5–20 years [5–7], while affected women may
X-linked Charcot-Marie-Tooth (CMT1X) is the second present at a later age [8, 9] with milder symptoms [10].
most common inherited demyelinating neuropathy [1] CMT1X is caused by mutations in the GJB1 gene
characterized by slowly progressive muscle weakness and encoding the gap junction (GJ) protein connexin 32 (Cx32).
atrophy, loss of reflexes, and reduced nerve conduction Cx32 is a myelin-related protein that forms intracellular GJ
channels through the non-compact myelin layers at para-
nodal loops and Schmidt–Lantermann incisures of myeli-
nating Schwann cells [11–13] serving important
Supplementary information The online version contains
homeostatic and axon-glial signaling functions. There are
supplementary material available at https://doi.org/10.1038/s41434-
021-00250-0. over 400 different GJB1 mutations reported so far

* Kleopas A. Kleopa 5
Department of Neurodegenerative Disease, UCL Institute of
kleopa@cing.ac.cy Neurology, London, United Kingdom
6
1 UK Dementia Research Institute at UCL, London, United
Neuroscience Department, The Cyprus Institute of Neurology and
Kingdom
Genetics and Cyprus School of Molecular Medicine,
7
Nicosia, Cyprus Department of Psychiatry and Neurochemistry, Institute of
2 Neuroscience and Physiology, the Sahlgrenska Academy at the
Department of Molecular Virology, The Cyprus Institute of
University of Gothenburg, Mölndal, Sweden
Neurology and Genetics and Cyprus School of Molecular
8
Medicine, Nicosia, Cyprus Clinical Neurochemistry Laboratory, Sahlgrenska University
3 Hospital, Mölndal, Sweden
Department of Clinical Neurosciences, University of Cambridge,
9
Cambridge, United Kingdom Center for Neuromuscular Disorders, The Cyprus Institute of
4 Neurology and Genetics and Cyprus School of Molecular
Department of Neuromuscular Diseases, UCL Queen Square
Medicine, Nicosia, Cyprus
Institute of Neurology, London, United Kingdom
660 A. Kagiava et al.

(http://hihg.med.miami.edu/code/http/cmt/public_html/ ubiquitous promoters and was not targeted to Schwann

index.html#/) affecting all domains of Cx32, as well as the cells. Thus, the improvement observed could result from
non-coding gene regions [14, 15]. Frameshift, premature cross-correction through the transfer of the protein from
stop, and non-coding mutations are predicted to cause axons and other cells surrounding the PNS. The Mpz pro-
complete loss of protein function through failed synthesis or moter has been used to target Schwannoma cells with
rapid degradation. Several missense and inframe mutations AAV1 vectors resulting in the regression of the tumors
are retained intracellularly [16–18] in the ER and/or Golgi following intraneural injection [40, 41].
[18–22] with the inability to form functional channels, In order to develop translatable gene therapy for
while other mutants form membrane channels with altered CMT1X, we used here an AAV9 vector driven by the
biophysical characteristics [22]. Schwann cell-specific Mpz promoter in order to achieve
Despite the large number and different mechanisms of targeted GJB1 gene expression in Schwann cells throughout
CMT1X-associated mutations, clinical studies have the PNS. Following a single lumbar intrathecal injection, we
demonstrated that males suffering from the disease have a demonstrate widespread vector biodistribution to the lumbar
similar phenotype and severity that correlates with age and roots, as well as to sciatic and femoral nerves, leading to
is independent of the specific mutation, comparable with the high expression rates of Cx32 specifically in myelinating
phenotype of patients with a complete GJB1 deletion Schwann cells. Using this approach, we demonstrate a
[23, 24]. Furthermore, Gjb1-null mice with deletion of the significant therapeutic benefit in the CMT1X mouse model
Gjb1/Cx32 gene develop a progressive, predominantly when treated before as well as after the onset of peripheral
motor demyelinating peripheral neuropathy beginning at neuropathy. This study provides the proof of principle for a
about three months of age with reduced sciatic motor nerve gene therapy approach to treat CMT1X patients.
conduction velocity (MNCV) and motor amplitude [25, 26].
Expression of wild type (WT) human Cx32 protein driven
by the rat myelin protein zero promoter (Mpz/P0) promoter Materials and methods
prevented demyelination in Gjb1-null mice [27], confirming
that loss of Schwann cell autonomous expression of Cx32 is Cloning and production of AAV vectors
sufficient to cause CMT1X pathology. Transgenic mice
expressing CMT1X mutations showed no detectable Cx32 We generated novel constructs for AAV vector gene
protein in the 175 fs mutant line [28], while R142W, T55I, delivery designed to provide Schwann cell-specific
R75W and N175D transgenic mice showed retention of the expression of either the reporter gene EGFP (pAAV-Mpz.
mutant protein in the perinuclear region and developed a Egfp, mock vector; Fig. 1A) or Cx32 (pAAV-Mpz.GJB1,
demyelinating neuropathy similar to Cx32 KO mice full, therapeutic vector; Fig. 1B), both under the rat 1.2 kB
[29, 30]. Taken together, these findings suggest that most Mpz promoter shown to drive expression specifically in
GJB1 mutations cause Schwann cell-autonomous loss of Schwann cells [27, 32]. The Mpz/Cx32 ORF was PCR
Cx32 function, indicating that gene replacement therapy is a amplified from a lentiviral construct previously made in
valid approach to treat the disease. the lab [42]. The primers used for the amplification were
We previously showed that a single lumbar intrathecal P0-Cx32-F 5′–AGGGGTACCCTTCCTGTTCAGACT–3′
injection of a lentiviral vector driven by the Mpz promoter (SEQ ID No. 13) and P0-Cx32- R 5′–CCGCTCGAGGGA
carrying the GJB1 gene can partially improve the demye- TCCTC AGCAG–3′. The PCR product (2030 bp) was gel
linating neuropathy in Gjb1-null mice both at early and at purified and digested using KpnI and XhoI. The AAV
later stages of the disease [31, 32]. However, low levels of vector was also digested with the same restriction enzymes.
lentiviral derived expression may limit the potential for The entire expression cassette was confirmed by direct
translating this approach for a subset of CMT1X patients sequencing of the ORFs.
with Golgi-retained interfering mutations [33, 34]. In The pAAV-Mpz.Egfp and pAAV-Mpz.GJB1 plasmids
addition, the mutagenic risk of lentiviral vectors through were cross-packaged into AAV9 capsid. For AAV9 vector
genome integration [35] limits their potential to be used in production HEK293T, cells are grown in 10 cm tissue
the treatment of patients. Therefore, we considered as an culture dishes in IMDM supplemented with 10% FCS. Two
alternative approach the use of an adeno-associated viral hours prior to transfection, the medium is changed. When
vector (AAV). AAV vectors have been extensively used in cells have reached ~70–80% confluency, they are trans-
the last decades for gene delivery mainly to the central fected using polyethylenimine (PEI; branched, MW 25,000;
nervous system (CNS) and in some applications also to the Sigma) with a three-plasmid mix of pAAV, pRepCap, and
peripheral nervous system (PNS) [36–39]. Although pre- pXX6 plasmid provided by Dr. Jude Samulsky, the Uni-
vious studies showed that the genes carried by AAV vectors versity of North Carolina at Chapel Hill at a ratio of 1:2:1
can be expressed in the PNS, expression was driven by and the medium is changed 16 h after transfection. As it was
AAV9-mediated Schwann cell-targeted gene therapy rescues a model of demyelinating neuropathy 661

shown previously [43] that the cell media contains sig- Study design
nificant amounts of virions, both the cells and the cell
supernatant are collected 48 h later for vector purification. For the analysis of EGFP expression, we used 6 WT mice
Cells are pelleted by centrifugation at 500 g for 10 min and injected with the pAAV-Mpz.Egfp (mock) vector and for
are subsequently re-dissolved in lysis buffer (0.5% sodium Cx32 expression analysis, we used ten Gjb1-null mice
deoxycholate, 150 mM NaCl, 20 mM Tris, 50 U/ml Ben- injected with the pAAV-Mpz.GJB1 (full) vector. For the
zonase; Sigma-Aldrich, Munich, Germany) and incubated at early treatment study, we used ten Gjb1-null mice injected
37 °C for 1 h. The lysate is then clarified by centrifugation with the mock vector and 10 injected with the full vector.
at 3000 g for 10 min and filtered in parallel with the Finally, for the late treatment 20 Gjb1-null mice were used
supernatant through 0.45 µm Millipore filters (Millipore, in the mock group and 20 in the full group. Untreated Gjb1-
San Salvador, El Salvador). Next, RNAseA and a Protease null mice (n = 7), as well as WT mice (n = 4) of the same
inhibitor cocktail are added to both the lysate and the age, were also assessed for the same outcome measures. We
supernatant and incubated for 2 h at 37 °C followed by used 3–4 mice in each group for the expression analysis in
clarification at 3000 g for 15 min. The preparations are then order to obtain reliable data for statistical analysis (t-test).
combined with a precipitation mix (PEG/NaCl) at a ratio of For the treatment trials, we used 8–12 mice for both elec-
3:1, which is incubated at 4 °C overnight. The mixture is trophysiological and morphological analysis in order to
centrifuged at 3000 g for 30 min and the aqueous super- overcome the variability between mice and obtain reliable
natant discarded. The virion-containing pellet is then data for statistical analysis. We did not exclude outliers in
resuspended in 3 ml Pellet suspension buffer (250 mM NaCl case we had. Foot grip tests were performed three times and
solution), which is clarified by centrifugation at 10,000 g for the mean value was used while muscle contraction experi-
10 min at 4 °C. The virus-containing aqueous layer is then ments were performed twice.
transferred to the Beckmann UltraClear SW41 tube and is The aim of this study was to examine whether a gene
centrifuged at 149,000 g for 3 h. The aqueous layer is then addition therapy can treat peripheral neuropathy or prevent
discarded and the viral pellet resuspended in 200 µl pellet the development of peripheral neuropathy in the mouse
suspension buffer. The vector genome copy number model of CMT1X neuropathy both at early and late stages,
(VGCN) is determined by qPCR targeting the WPRE before and after the onset of the neuropathy. The gene
sequence [44]. therapy study was conducted using four groups of Gjb1-null
mice. A minimum of 8–12 mice per treatment group for each
Experimental animals outcome measure was considered adequate for assessing
statistically significant differences based on our previous
In this study, we used adult WT C57BL/6 or Gjb1-null/Cx32 studies using similar models [32, 42]. Mice were treated at
KO (C57BL/6_129) mice weighing 20–25 g. Early and late the age of 2 months for the early treatment and at 6 months
gene therapy trials were conducted using 2- and 6-month-old of age for the late treatment (Fig. 2A). Littermate mice were
mice Gjb1-null/Cx32 KO mice weighing 20–25 g. Gjb1-null/ randomized to receive either AAV9-Mpz.GJB1 (full) treat-
Cx32 KO mice were obtained from the European Mouse ment or AAV9-Mpz.Egfp (mock treatment, control group)
Mutant Archive, originally generated by Prof. Klaus Willecke and were assigned a coding number for further identification.
(University of Bonn). In these mice, the neor gene was inserted 2-month-old mice were evaluated before the treatment, and
in-frame into the exon 2 of Gjb1 gene which contains the ORF again at the age of 4 and 6 months, while 6-month-old mice
[45]. Mice were kept in a specific pathogen-free animal facility, were evaluated before the treatment, and at the age of 8 and
housed in open-top system cages. Wood bedding for laboratory 10 months. Mice were evaluated by behavioral testing by an
mice, dried by high-temperature treatment, sieved, de-dusted, examiner blinded to the treatment condition and used at the
of high absorbency, was used and mice were housed up to five age of 6 months for the early treatment and 10 months for
in each cage. Standard mouse diet, certificate, for reproduction, the late treatment for electrophysiology or for quantitative
weaning, growth, and tap potable water, filtered and UV ster- morphometric analysis of semi-thin sections. Animals sub-
ilized were administered to the mice. Mice were kept in a 12 h jected to muscle contraction experiments were not used for
dark/12 h light cycle at a temperature of 22 °C. Both male and morphological analysis. Analysis of physiological and
female mice were used in our experiments and showed no (sex- morphological results was also performed blinded to the
related) differences in their behavioral performance or nerve treatment condition.
pathology. All experimental procedures in this study were
conducted in accordance with animal care protocols approved Intrathecal vector delivery
by the Cyprus Government’s Chief Veterinary Officer (project
license CY/EXP/PR.L3/2017) according to national law, which We delivered the AAV vectors by a single lumbar intrathecal
is harmonized with EU guidelines (EC Directive 86/609/EEC). injection as previously described [32, 46]. Briefly, a small
662 A. Kagiava et al.

skin incision was made along the lower lumbar spine level of Waltham, MA USA), rat CD68 (1:50; Biorad, California,
anesthetized mice to visualize the spine and the AAV vector USA), CD45 (1:100; Abcam, Cambridge, UK) and goat CD20
was delivered into the L5-L6 intervertebral space. A 50-μL (1:100; Santa Cruz, Texas, USA) all diluted in blocking solu-
Hamilton syringe (Hamilton, Giarmata, Romania) connected tion and incubated overnight at 4 °C. Slides were then washed
to a 26-gauge needle was used to inject 20 µL of AAV stock in PBS and incubated with mouse cross-affinity fluorescein-
containing an estimated 2 × 1010 vector genomes (vg)/mL, at conjugated (1:1000; Invitrogen, Waltham, MA USA), rat
a maximum rate of 5 µL/min. A flick of the tail was con- cross-affinity purified rhodamine-conjugated (1:2000; Invitro-
sidered indicative of successful intrathecal administration. gen, Waltham, MA USA), goat cross-affinity fluorescein-con-
jugated (1:700; Abcam, Cambridge, UK) and rabbit cross-
VGCN determination affinity purified rhodamine-conjugated (1:500; Jackson
ImmunoResearch, West Grove, USA) secondary antibodies for
Genomic DNA was extracted from different PNS tissues 1 h at RT. Cell nuclei were visualized with DAPI (1 µg/ml;
(i.e., lumbar roots, proximal and distal sciatic nerves, and Sigma, Munich, Germany). Slides were mounted with fluor-
femoral motor nerves) of mice 4 or 8 weeks after intrathecal escent mounting medium and images photographed under a
vector delivery using the Invitrogen iPrep PureLink gDNA fluorescence microscope with a digital camera using a fluor-
Kit (Thermo Fisher Scientific, Waltham, MA USA). The escence microscope (Nikon Eclipse Nἱ; Tokyo, Japan) with a
extracted DNA was analyzed for yield and purity using a digital camera (DS-Qi2) using NIS-Elements software.
Nanodrop 1000 spectrophotometer. Approximately 20 ng of
DNA was used as template for two real-time PCR assays on Immunoblot analysis
an Applied Biosystems 7500 Real-Time PCR System
involving 45 cycles of 15 s at 95 °C and 60 s at 60 °C. Fresh sciatic and femoral nerves and lumbar spinal roots
TFRC-specific primers/probe targeting the mouse genome were collected at 8 weeks post-injection and lysed in ice-
and WPRE-specific primers/probe, which is contained in cold RIPA buffer (10 mM sodium phosphate, pH 7.0, 150
the transgene, were used. Standard curves were created by mM NaCl, 2 mM EDTA, 50 mM sodium fluoride, 1%
serial dilution of quantified mouse genomic DNA, as well Nonidet P-40, 1% sodium deoxycholate, and 0.1% SDS all
as quantified plasmid DNA containing the transgene cas- from Sigma-Aldrich, Munich, Germany) containing a
sette. The average VCN per cell was calculated as the total mixture of protease inhibitors (Roche, Sigma-Aldrich,
VCN divided by the total cell number. Munich, Germany). Proteins (150 μg) from the lysates were
fractionated by 12% SDS/PAGE and then transferred to a
Immunofluorescence staining Hybond-C Extra membrane (GE Healthcare Life Sciences,
Logan, USA) using a semidry transfer unit. Nonspecific
For immunostaining, mice were anesthetized with avertin sites on the membrane were blocked with 5% non-fat milk
according to institutionally approved protocols, and then in PBS with Tween 20 (PBST) for 1 h at room temperature.
transcardially perfused with normal saline followed by fresh Immunoblots were incubated with rabbit antisera against
4% paraformaldehyde in 0.1 M PB buffer. The lumbar-sacral EGFP (1:1000; Abcam, Cambridge, UK) or Cx32 (clone
spinal cords with spinal roots attached, as well as the bilateral 918, 1:3000) [47] and mouse β-tubulin (1:4000; Develop-
sciatic and femoral motor nerves were dissected and post-fixed mental Studies Hybridoma Bank, Iowa, USA) at 4 °C
in 4% PFA, the spinal cord for 2 h while sciatic and femoral overnight. After washing, the immunoblots were incubated
nerves for 30 min. Spinal roots were frozen for cryosections with an anti-mouse or anti-rabbit HRP-conjugated second-
while sciatic and femoral nerves were isolated and teased into ary antiserum (Jackson ImmunoResearch, diluted 1:3000,
fibers under a stereoscope. Teased fibers or sections were West Grove, USA) in 5% milk–PBST for 1 h. The bound
permeabilized in cold acetone and incubated at RT with a antibody was visualized by an enhanced chemilumines-
blocking solution of 5% BSA (Sigma-Aldrich, Munich, Ger- cence system (GE Healthcare Life Sciences, Logan, USA).
many) containing 0.5% Triton-X (Sigma-Aldrich, Munich,
Germany) for 1 h. Primary antibodies used were: mouse Behavioral analysis
monoclonal antibody against the contactin-associated protein
(Caspr, 1:50; gift of Dr. Elior Peles, Weizmann Institute of Grip strength testing
Science), NeuN (1:400 Chemicon, San Salvador, El Salvador),
CC1 (1:50, Calbiochem, San Salvador, El Salvador), GFAP To measure grip strength, mice were held by the tail and
(1:400, Sigma-Aldrich, Munich, Germany), rabbit antisera lowered towards the apparatus (Ugo Basile, Varese, Italy) until
against EGFP (1:1,000; Invitrogen, Waltham, MA USA), they grabbed the grid with the hind paws. Mice were gently
Caspr2 (1:200, Alomone Labs, Jerusalem, Israel), CD3 (1:100; pulled back until they released the grid. Measurements of the
Abcam, Cambridge, UK) and Cx32 (1:50; Invitrogen, force in g were indicated on the equipment. Each session
AAV9-mediated Schwann cell-targeted gene therapy rescues a model of demyelinating neuropathy 663

consisted of three consecutive trials and measurements were was collected prior to sacrificing the animals using standard
averaged. Hind limb force was compared between AAV9.Mpz- methods [49]. We measured the levels of neurofilament
GJB1 and AAV9.Mpz-Egfp treated mice. light (NF-L) and neural cell adhesion molecule (NCAM-1).

Electrophysiological analysis Measurement of NF-L concentration

Motor nerve conduction velocity (MNCV) Blood samples were processed within one hour. Blood was
collected into EDTA-containing tubes and centrifuged at 20 °C
MNCV was measured in vivo using published methods [48] at 3500 rpm for 10 min. Plasma was aliquoted and stored at
from bilateral sciatic nerves following stimulation in anes- −80 °C until testing. Plasma NF-L concentration was mea-
thetized animals using two stimulation sites, one near the sured at University College London (UCL) using a commer-
sciatic notch and one distally at the ankle via bipolar elec- cially available NF-Light kit on a Single molecule array
trodes with supramaximal square-wave pulses (5 V) of (Simoa) HD-1 instrument (Quanterix, Billerica, MA) [50, 51].
0.05 ms. The latencies of the compound muscle action
potentials (CMAP) were recorded by a bipolar electrode Measurement of NCAM-1 concentration by ELISA
inserted between digits 2 and 3 of the hind paw and mea-
sured from the stimulus artifact to the onset of the negative Serum was separated using 0.5 ml Minicollect Z serum
M-wave deflection. MNCV was calculated by dividing the separation tubes. Blood samples were taken and centrifuged
distance between the stimulating and recording electrodes 20 °C at 3000 g for 10 min. Serum was aliquoted and stored
by the result of subtracting distal from proximal latency. at −80 °C until testing. Serum protein levels were deter-
mined at the Department of Clinical Neurosciences, Uni-
Quadriceps muscle contractility study versity of Cambridge, by ELISA according to
manufacturers’ instructions for neural cellular adhesion
Quadriceps muscle contractility study was performed only in molecule (NCAM1; Rockland KOA0716).
the late treatment group in order to assess in situ the function
of the lumbar root and femoral motor axons. Mice subjected to Morphometric analysis of myelination in lumbar
muscle contraction study were not used for morphological roots and peripheral nerves
assessment. We measured the contraction properties of the
quadriceps muscle innervated by the femoral nerve in an Mice were transcardially perfused with 2.5% glutar-
anesthetized mouse as previously described [32]. After expo- aldehyde in 0.1 M PB buffer. The lumbar spinal cord
sure of the motor part of the femoral nerve a stimulating hook with multiple spinal roots attached, as well as the
electrode was used to stimulate the motor branch of the femoral and sciatic nerves, were dissected and fixed
femoral nerve at 1 Hz using a constant current stimulator overnight at 4 °C, then osmicated, dehydrated, and
(DS3; Digitimer, Welwyn Garden City, UK) with 5–6 mA and embedded in araldite resin (all purchased from Agar
200 μs duration pulse. The muscle contraction of the partially Scientific, Essex, UK). Transverse semi-thin sections
exposed quadriceps muscle was recorded with a force dis- (1 μm) of the lumbar spinal cord with roots and the
placement transducer (FT03; Grass Technologies), which was middle portion of the femoral motor and sciatic nerves
attached to the muscle with a silk suture. The transducer was were obtained and stained with alkaline toluidine blue
connected to a micromanipulator and for the experiment the (Sigma-Aldrich, Munich, Germany). Sections were
muscle was extended 1 mm each time until the muscle con- visualized with 10×, 20×, and 40× objective lenses and
traction reached the maximum value. The average amplitude captured using a Nikon Eclipse Ni microscope (Tokyo,
and duration of the force generated by the quadriceps muscle Japan) with a digital camera (DS-Fi3) using NIS-
contraction was compared between treatment groups. Elements software. Images of whole root or transverse
nerve sections were obtained at 100–200× final magni-
Biomarkers determination in blood samples of fication, and a series of partially overlapping fields
Gjb1-null treated mice covering the entire cross-sectional area of the roots or the
nerves were captured at 400× final magnification.
In order to further access the potential value of clinically These images were used to examine the degree of
relevant blood biomarkers and their treatment responsive- abnormal myelination in all treatment groups as described
ness, we performed biomarker analysis from blood samples previously [30, 42, 52]. In brief, all demyelinated, remye-
of WT compared to untreated Gjb1-null mice, as well as linated, and normally myelinated axons were counted in the
from Gjb1-null mice treated either with the therapeutic or entire root or nerve cross-section using the following cri-
with the mock vector, all at the age of 10 months. Blood teria: axons larger than 1 μm without a myelin sheath were
664 A. Kagiava et al.

considered demyelinated; axons with myelin sheaths <10% of the sciatic nerve (n = 4 mice; Fig. 1F). EGFP expression
of the axonal diameter and/or axons surrounded by “onion rates in immunostained tissue sections reached 35.0 ± 3.23 %
bulbs” (i.e., circumferentially arranged Schwann cell pro- in anterior lumbar roots and 39.9 ± 2.49 % in sciatic nerves (n
cesses and extracellular matrix) were considered remyeli- = 3 mice; Fig. 1G). EGFP was also detected by immunoblot
nated; all other myelinated axons were considered normally analysis of lumbar roots, sciatic and femoral nerve samples of
myelinated. injected mice with a proximal to distal expression level gra-
In addition, we counted the number of foamy macro- dient, while it was not detected in non-injected WT mouse
phages present within the entire cross section of each root samples run as negative controls (Fig. S1A–C).
or nerve, as an indication of inflammation. Macrophages In order to confirm the cell-specificity of the Mpz pro-
were identified in semi-thin sections at 400× magnifica- moter we performed double immunostaining for EGFP
tion as cells laden with myelin debris, devoid of a base- with cell markers in lumbar spinal cord sections and
ment membrane, and extending small, microvilli-like confirmed that EGFP expression was not detected in oli-
processes, as described previously [53, 54]. The macro- godendrocytes, neurons, or astrocytes (Fig. S2A–C).
phage count was calculated as the ratio per 1000 myeli- Likewise, staining of sciatic nerve sections with vimentin,
nated fibers, to account for size differences between a fibroblast marker, and Glut-1, a perineurial cell marker,
different spinal roots and nerves. All pathological ana- showed no EGFP expression either in fibroblasts or in
lyses were performed blinded to the treatment condition perineurial cells (Fig. S2D, E). Furthermore, in order to
of each mouse. exclude an inflammatory response that could occur after
intrathecal injection of the vector we immunostained sec-
Statistical analysis tions of spinal cord with anterior lumbar roots attached
using different inflammatory cell markers. We did not
The percentages of EGFP-positive Schwann cells or Cx32- detect any CD68+ macrophages, CD3+ T lymphocytes,
expressing paranodal myelin areas in immunostained spinal CD45+ leukocytes or CD20+ B-lymphocytes in CNS or
roots and sciatic nerves of WT and Gjb1-null mice injected PNS tissues of injected mice (Fig. S3).
with the mock or full vector, respectively, were compared
with Student’s t-test. Behavioral testing results, electro- Cx32 expression following intrathecal injection of
physiological results, and blood biomarker levels were AAV9-Mpz.GJB1
compared using one-way ANOVA with Bonferroni post-
test. Morphological analysis data obtained from mock- AAV9-Mpz.GJB1 (full, therapeutic vector; Fig. 1B) driving
treated and fully treated groups were compared using the expression of the human GJB1 open reading frame was
Mann–Whitney U test (significance level for all compar- delivered by lumbar intrathecal injection (estimated 2 × 1010 vg
isons, P < 0.05). All tests were performed using GraphPad injected) into both 2- and 6-month-old Gjb1-null mice in order
Instat3 software (GraphPad, San Diego, USA). to study the expression of Cx32 throughout the PNS in the
absence of endogenous Cx32 at the age groups relevant for the
treatment trials. Cx32 expression and VGCNs were determined
Results 4 weeks post-injection in anterior lumbar roots and sciatic
nerves. Cx32 was detected as expected at the paranodal areas
EGFP expression following intrathecal injection of of myelinating Schwann cells in both lumbar roots and sciatic
AAV9-Mpz.Egfp nerves, similar to the WT tissues (Fig. 1H–M). At 4 weeks
post-injection, VGCNs reached 0.2 ± 0.05 (n = 4 mice) in
AAV9 vectors (both mock and full) were produced to titers of anterior lumbar roots and 0.18 ± 0.07 in the sciatic nerve (n = 4
1 × 1012 vg/ml. AAV9-Mpz.Egfp (mock vector; Fig. 1A) was mice; Fig. 1N). We also measured the VGCNs in other tissues
delivered by lumbar intrathecal injection (estimated 2 × 1010 vg including the quadriceps muscles (0.05 ± 0.03), the spinal cord
injected) into WT mice in order to study the biodistribution of (0.17 ± 0.08), and the liver (17.32 ± 3.30). The percentage of
the vector throughout the PNS. VGCNs and EGFP expression Cx32-immunoreactive paranodal areas among all paranodal
were examined 8-weeks post-injection in anterior lumbar roots areas visualized in each tissue based on Caspr paranodal
and sciatic nerves. EGFP was detected by immunofluorescent labeling was quantified reaching 62.7 ± 3.75 % in anterior
labeling (Fig. 1C, E) as well as by auto-fluorescence without lumbar roots (n = 5 mice) and 69.9 ± 6.90 % in sciatic nerves
the use of an antibody (Fig. 1D) in the perinuclear cytoplasm of (n = 5 mice) (Fig. 1O). Immunoblot analysis confirmed high
a subset of Schwann cells in both the lumbar roots and sciatic expression levels of Cx32 in anterior lumbar root and sciatic
nerves. VGCNs in DNA extracted from PNS tissues reached nerve lysates from injected animals similar to animals trans-
0.24 ± 0.09 in anterior lumbar roots, 0.82 ± 0.46 in the prox- genically expressing Cx32, while Cx32 was absent in non-
imal part of the sciatic nerve and 0.18 ± 0.08 in the distal part injected Gjb1-null mice tissues (Fig. S1D, C).
AAV9-mediated Schwann cell-targeted gene therapy rescues a model of demyelinating neuropathy 665

Fig. 1 Biodistribution and expression of AAV9-Mpz.Egfp and PNS tissues examined at 8 weeks post-injection (G). Following
AAV9-Mpz.GJB1 vectors. Diagrams showing the expression cas- injection of AAV9-Mpz.GJB1 in Gjb1-null mice immunostaining of
sette of the AAV9-Mpz.Egfp (mock) (A) and AAV9-Mpz.GJB1 lumbar root sections (H)–(J) and sciatic nerve teased fibers (K)–(M)
(full) (B) vectors, in which the expression of either the EGFP for Cx32 (red) and paranodal axonal marker Caspr (white arrows)
reporter gene or of the human GJB1 open reading frame, respec- shows that virally delivered Cx32 (red arrows) is correctly expressed
tively, is driven by the rat myelin protein zero (Mpz) promoter. at some of the paranodal myelin areas (white arrows) only in
Immunostaining for EGFP (red) of lumbar root sections (C) and injected Gjb1-null mice (J) and (M) similar to WT mice (H) and (K)
sciatic nerve teased fibers (E) reveals perinuclear EGFP immunor- but not in their non-injected Gjb1-null littermates (I) and (L). Cx32
eactivity (arrows) in a subset of Schwann cells 8 weeks after intra- is also detected in the Schmidt-Lanterman incisures (yellow arrows)
thecal injection of AAV9-Mpz.Egfp vector in wild type (WT) mice. in both WT (A) and AAV9-Mpz.GJB1 injected mice (J) and (M).
Auto-fluorescent EGFP signal (green) is also evident in sciatic nerve N VGCN determination demonstrates biodistribution of AAV9-Mpz.
sections without antibody detection (D). Cell nuclei are stained with GJB1 to both lumbar roots and sciatic nerves in addition to the
DAPI (blue). Determination of the vector genome copy numbers spinal cord (SC) and to a lesser degree to the quadriceps muscle.
(VGCN) confirms AAV9 biodistribution to lumbar roots (LR) as O Quantification of the percentage of Cx32-positive paranodal areas
well as both proximal (prox) and distal (dis) sciatic nerves (SN) (F). in lumbar roots and sciatic nerves shows comparable results in both
Quantification of the percentage of EGFP-positive cells in lumbar tissues. VGCNs and expression rates values represent Mean ± SEM.
roots and sciatic nerves shows no significant differences between Scale bars: (C) and (D): 20 μm; (E)–(M): 10 μm.
666 A. Kagiava et al.

Fig. 2 Analysis of functional outcomes of AAV9-Mpz.GJB1 injec- to controls and was not significantly different from the performance
ted Gjb1-null mice compared to AAV9-Mpz.Egfp (mock) treated of wild type (WT) mice of the same age (D) and (H). Longitudinal
littermates. A Diagram showing the pre- and post-onset treatment analysis within each group showed improved motor performance
trial time course with a timing of functional and morphological over time in both pre- and post-onset treated Gjb1-null mice while
analysis. Results of foot grip analysis in pre- (B)–(E) and post-onset mock groups did not improve (E) and (I). Sciatic nerve motor
treatment groups (F)–(I) comparing AAV9-Mpz.GJB1 treated conduction velocity (MNCV) also showed significant improvement
(GJB1) and mock-treated Gjb1-null mice, as indicated. There were 4 months after treatment compared to the mock groups in both pre-
no differences between the treatment groups before the injection (B) (J) and post-onset (K) treated mice (n = 10 for all groups). MNCVs
and (F), while 2 months post-injection grip strength improved in in early treated mice did not differ significantly from those of age-
both pre- and post-onset treated Gjb1-null mice compared to the matched WT mice (J) while in post-onset treated mice they
mock groups (n = 10 per group) (C) and (G). The improved per- remained below those of age-matched WT mice (K) (One-
formance continued at 4 months post-injection in both pre- (6- way ANOVA with Bonferroni post-test; *P < 0.05, **P < 0.01 and
month old) and post-onset (10-months old) treated mice compared ***P < 0.001).
AAV9-mediated Schwann cell-targeted gene therapy rescues a model of demyelinating neuropathy 667

Improvement of motor performance in pre- and Longitudinal comparison of the motor performance at dif-
post-onset treated Gjb1-null mice ferent time points within each treatment group showed that
fully treated mice improved significantly 2 months post-
Following confirmation of therapeutic vector expression in injection, from 6 to 8 months of age (p = 0.0441) and even
both 2- and 6-month-old Gjb1-null mice, we proceeded further from 8 to 10 months of age (p = 0.0045), while
with pre- and post-onset treatment trials in which littermate mock-treated littermates did not show any significant dif-
mice were randomized into two groups, one receiving the ferences over time (p > 0.05). Overall, our motor perfor-
mock (AAV9-Mpz.Egfp) and the other the full (AAV9-Mpz. mance findings at different stages of the treatment trials
GJB1) vector. Pre-onset treatment groups were injected at indicate a continuous improvement of motor performance in
2 months of age and monitored until 6 months of age when both pre- and post-onset treatment groups following the
outcome was assessed by behavioral, electrophysiological delivery of the AAV9-Mpz.GJB1 vector.
and morphological analysis. Post-onset treatment groups
were injected at 6 months of age and outcome was similarly Improvement of electrophysiological properties of
assessed at 10 months of age, with additional biomarker pre- and post-onset AAV9-Mpz.GJB1 treated
analysis (Fig. 2A). Gjb1-null mice
Motor performance was assessed before the injection and
until the end of the observation period by foot grip analysis In the pre-onset treatment groups, we performed sciatic
in all groups (Fig. 2B–E). Foot grip strength in pre-onset MNCV studies at the age of 6 months, 4 months after vector
groups at the age of 2 months, at baseline before the injection. MNCV was improved in the fully treated mice
initiation of the treatment, showed as expected no difference compared to the mock group (n = 10 mice per group) as
between the two groups or the WT (p > 0.05) with values indicated by the values that reached 37.5 ± 1.81 m/s in the
reaching 66.6 ± 3.59 g for the full vector group (n = 10), fully treated group and 28.3 ± 1.92 m/s in the mock group
71.0 ± 6.69 g for the mock vector group (n = 10) and 76.6 ± (p < 0.01) (Fig. 2J) without significant difference to the WT
5.87 g for the WT group (n = 12). At the age of 4 months, values at the same age (41.7 ± 1.09; n = 11; p > 0.05).
2 months post-injection, muscle strength was increased in Although MNCV was improved in the fully treated mice the
fully treated mice reaching 74.3 ± 3.57 g (n = 10) compared amplitude of the CMAP was similar between the two
to the mock group (53.4 ± 5.0; n = 10; p > 0.05) but without groups reaching 4.6 ± 0.56 mV (n = 10) in the mock group
reaching the WT values (116.0 ± 8.06 g; n = 12; p < and 4.1 ± 0.49 mV (n = 10) in the full group (p > 0.05;
0.0001). At 4 months post-injection the force generated by Fig. S4A).
the hindlimbs of the treated mice was increased reaching In the post-onset treated groups, we performed both
120.0 ± 7.87 g (n = 10) compared to the mock group (69.3 MNCV and quadriceps muscle contraction experiments that
± 3.59 g; n = 10; p < 0.001) and was similar to the WT reflect chronic axonal loss and muscle denervation. Mea-
values (106.6 ± 2.89; n = 12; p > 0.05). Longitudinal com- surements were performed at 10 months of age, 4 months
parison at different age groups of fully treated mice showed post-injection. MNCV was improved in the fully treated
that the force generated by the hindlimbs increased sig- mice (34.8 ± 1.44 m/s; n = 11) compared to the mock group
nificantly over time (p < 0.0001), while mock treated mice (30.4 ± 0.87 m/s; n = 10) (p < 0.05; Fig. 2K) but remained
did not show any differences (p > 0.05). significantly below the values of age-matched WT mice
Foot grip analysis in post-onset treatment groups (43.7 ± 1.05; n = 11; p < 0.001). As in the early treated
(Fig. 2F, I) showed no differences at baseline with the groups, we also measured the CMAP amplitudes and found
values of the fully treated group reaching 74.5 ± 5.46 g (n = no significant differences between the two groups (p > 0.05)
21) and the mock 76.0 ± 9.14 g (n = 20), while both per- with the values reaching 4.1 ± 0.42 mV (n = 11; full group)
formed worse than WT mice of the same age (106.6 ± 2.89; and 3.2 ± 0.35 mV (n = 10; mock group; Fig. S4B). We
n = 12; p < 0.01). At the age of 8 months, 2 months post- further examined the muscle contractility properties of the
injection, we observed an increase in the force generated by quadriceps muscle after stimulation of the femoral motor
the hindlimbs in the fully treated compared to the mock nerve in situ (Fig. S4C, D) as previously described [32].
groups reaching 93.5 ± 5.18 g (n = 21) and 76.2 ± 3.41 g Both force and duration were improved in the AAV9-Mpz.
(n = 20; p < 0.05), respectively with the fully treated group GJB1 treated (n = 11) compared to the mock-treated (n =
reaching values similar to those of the WT mice (103.8 ± 10) group. Amplitudes reached 0.22 ± 0.010 N in the treated
5.26; n = 12; p > 0.05). At the final stage, at 10 months of group, similar to WT [32], and 0.18 ± 0.008 N in the mock
age, we observed further improvement of the force in fully group (p = 0.0059). Furthermore, the duration of contrac-
treated mice (103.4 ± 6.01 g; n = 21), compared to the mock tion was significantly increased in the treated group reach-
group (73.2 ± 5.15 g; n = 20; p < 0.01) with no significant ing 0.098 ± 0.004 sec compared to 0.084 ± 0.002 s in the
difference to WT values (109.6 ± 7.46 g; n = 16; p > 0.05). mock group (p = 0.0015). Overall, our electrophysiological
668 A. Kagiava et al.

Fig. 3 Improvement of blood biomarkers after treatment in Gjb1- post-onset treated mice compared to mock treated mice (n = 6 per
null mice. Analysis of blood levels of molecules relevant to nerve group). B: Neural cell adhesion molecule-1 (NCAM-1) serum levels
pathology in 10-month-old AAV9-Mpz.GJB1 treated (GJB1) Gjb1- were similarly increased in untreated Gjb1-null (n = 4) compared to
null mice compared to AAV9-Mpz.Egfp (mock) treated littermates, as WT (n = 3) mice and showed significant improvement in treated
well as age-matched WT and untreated Gjb1-null mice: A Neurofila- compared to mock-treated group (n = 7 per group) (one-way ANOVA
ment light (NF-L) plasma concentration is increased in Gjb1-null (n = with Bonferroni post-test; *P < 0.05, **P < 0.01 and ***P < 0.001).
5) compared to WT (n = 8) mice and shows significant amelioration in

studies with emphasis on motor fibers that are pre- In addition to NF-L, we evaluated NCAM-1, a novel
dominantly affected in this CMT1X model [26, 30] showed blood biomarker with potential relevance to peripheral nerve
significant improvement following both pre- as well as post- disease. Similar to NF-L, analysis of NCAM-1 levels showed
onset gene addition therapy. an elevation in untreated Gjb1-null mice (179787.4 ± 6953.2
nmol/μl; n = 4) compared to WT mice (135494.5 ± 10649.9
Establishment of treatment-responsive blood nmol/μl; n = 3; p > 0.05; Fig. 3B). Importantly, there was
biomarkers in the CMT1X model a significant amelioration of NCAM-1 elevation in the
treated group (140651.6 ± 6798.4 nmol/μl; n = 7) compared
Based on previous data showing that biomarkers can be used to the mock-treated mice (196598 ± 9424.6 nmol/μl; n = 7;
in order to access treatment outcome in CMT neuropathy p < 0.001).
models [31, 74] we performed further analysis of blood
samples for disease-relevant biomarkers including Neurofi- Improved pathology in pre-onset AAV9-Mpz.GJB1
lament light (NF-L) and neuronal cell adhesion molecule-1 treated Gjb1-null mice
(NCAM-1). Clinical studies have shown elevation of plasma
NF-L levels in patients with different CMT types including We performed a morphological analysis in transverse
CMT1X compared to controls, as well as a correlation semithin sections of anterior lumbar roots, mid-sciatic and
between elevated concentrations and disease severity, sug- femoral motor nerves of 6-month-old Gjb1-null mice
gesting that this may be a useful biomarker for assessing injected at the age of 2 months with either the full (AAV9.
disease progression and potential response to treatment in Mpz-GJB1) or the mock (AAV9.Mpz-Egfp) vector. Multiple
future clinical trials [51]. In keeping with our previous roots, bilateral sciatic and femoral motor nerves from each
observations [31], we found that plasma NF-L levels were mouse were examined and the ratio of abnormally myeli-
increased in untreated 10-month old Gjb1-null mice (246.84 nated fibers was counted and their proportion to the total
± 20.18 pg/ml) compared to age-matched WT mice (111.48 number of fibers was calculated [32, 33, 42]. We also
± 26.77 pg/ml; p < 0.001). Moreover, comparison of treat- counted the number of foamy macrophages in each section
ment groups (n = 6 mice per group) showed that NF-L [32], and calculated their numbers per 1,000 myelinated
levels were significantly ameliorated in Gjb1-null mice fibers (to account for variations in root and nerve size). In
treated post-onset at the age of 6 months (152.4 ± 9.7 pg/ml; the anterior lumbar roots, a reduction of the abnormally
range: 124.9–191.2) compared to mock vector treated lit- myelinated fibers and foamy macrophages was observed in
termates (245.5 ± 13.4 pg/ml; range: 197.3–280.7; Fig. 3A; early AAV9-Mpz.GJB1 treated mice compared to their
p < 0.05). This reduction of NF-L levels in the AAV9-Mpz. mock-treated littermates (Fig. 4A–D). The ratio of abnor-
GJB1 treatment group by 1.6-fold compared to the mock mally myelinated fibers reached 0.12 ± 0.02 in the fully
group is in accordance with their improved motor function treated group compared to 0.18 ± 0.02 in the mock-treated
indicating that NF-L levels can be used as a treatment- mice (n = 10 mice per group; p = 0.0232; Fig. 4E and
responsive and clinically relevant biomarker for future gene Table S1). Reduction was also observed in the number of
therapy. foamy macrophages that reached 5.2 ± 0.97/1000 fibers in
AAV9-mediated Schwann cell-targeted gene therapy rescues a model of demyelinating neuropathy 669

Fig. 4 Morphological analysis of anterior lumbar roots and AAV9-Mpz.GJB1 injected mice show improved myelination in all
femoral motor nerves of Gjb1-null mice following pre-onset tissues compared to mock-treated littermates with fewer demyelinated
intrathecal delivery of the AAV9-Mpz.GJB1 vector. These are (*) and remyelinated (r) fibers. Quantification of the ratios of abnor-
representative images of semithin sections of anterior lumbar spinal mally myelinated fibers in multiple roots and femoral nerves (n = 10
roots attached to the spinal cord at low (A) and (B) and higher (C) and mice per group) confirms significant improvement in the numbers of
(D) magnification, with morphometric analysis results (E) and (F), as abnormally myelinated fibers (E) and (K), as well as reduction in the
well as of semithin sections of femoral motor nerves at low (G) and numbers of foamy macrophages (F) and (L) in the treated compared
(H) and higher (I) and (J) magnification and corresponding morpho- with mock treated littermates (see also Tables S1 and S3). Scale bars:
metric analysis results (K) and (L), from mock and full vector treated (A, B), (G, H) 50 μm, (C, D), (I, J) 10 μm.
mice as indicated, at 6 months of age (4 months after treatment).

the treated group and 12.7 ± 1.60 in the mock group (p = group and 0.38 ± 0.02 in the mock group (n = 10 per group;
0.0009; Fig. 4F and Table S1). p < 0.0001; Fig. 4K and Table S3), while foamy macrophages
Likewise, in femoral motor nerves, the ratio of abnormally were reduced to 6.1 ± 0.30/1000 fibers in the treated group
myelinated fibers and the numbers of foamy macrophages compared to 13.4 ± 0.54 in the mock group (p < 0.0001;
were reduced after gene therapy (Fig. 4G–J). The ratio of Fig. 4L and Table S3). The sciatic nerves showed similar
abnormally myelinated fibers was 0.20 ± 0.01 in the treated results (Supplementary results, Table S2 and Fig. S5A–F).
670 A. Kagiava et al.

Improved pathology in post-onset AAV9-Mpz.GJB1- Mpz promoter with known specificity for myelinating
treated Gjb1-null mice Schwann cells [27, 32] we delivered by a single lumbar
intrathecal injection the AAV9 vector into 2- and 6-month-
As with early, pre-onset treatment groups, we performed a old Gjb1-null mice and achieved widespread expression of
morphological analysis in transverse semithin sections of the virally delivered human Cx32. This treatment resulted in
anterior lumbar roots, mid-sciatic, and femoral motor nerves significant improvements in all outcome measures in this
of 10-month-old Gjb1-null mice injected either with the full neuropathy model both when treated at early as well as at
or mock vector at the age of 6 months. Tissues of 10-month later stages of the disease.
old WT and untreated Gjb1-null mice were also examined AAV vectors are known to provide high expression
for assessment of the magnitude of the therapeutic effect. In levels in transduced cells [55–57] offering the potential to
anterior lumbar roots, similar to the early treated mice, post- treat patients with different CMT1X mutations, including
onset treated mice showed improvement in morphological those with potential dominant-negative effects [33]. More-
properties with a reduced ratio of abnormally myelinated over, the use of AAV vector provides a more translatable
fibers and numbers of foamy macrophages (Fig. 5A–D). approach since these vectors remain mostly episomal
The ratio of abnormally myelinated fibers decreased to 0.22 minimizing the risk for insertional mutagenesis [58]. In
± 0.03 in the treatment group (n = 10) compared to 0.32 ± contrast, lentiviral vectors integrate into the host genome
0.02 in the mock group (n = 10; p = 0.0147) (Fig. 5E and [35] increasing the risk for mutagenesis and therefore have
Table S4). Furthermore, the number of foamy macrophages limited potential for systemic in vivo delivery. Importantly,
was 9.31 ± 1.2/1000 fibers in the full group (n = 10) com- the AAV9 vector is already used in clinical trials with an
pared to 14.85 ± 1.38 in the mock group (n = 10; p = established safety profile and recently FDA-approved
0.0068) (Fig. 5F and Table S4). The sciatic nerves showed application for the treatment of spinal muscular atrophy.
similar results (Supplementary results, Table S5 and Intrathecal injection of the AAV9 vector resulted in
Fig. S5G–L). widespread expression of the reporter gene EGFP in WT
Finally, as in lumbar motor roots and sciatic nerves, we mice as well as of Cx32 in both 2- and 6-month-old Gjb1-
observed improved morphological properties also in the null mice. Analysis of vector genomes in PNS tissues
femoral motor nerves of post-onset treated mice indicates a gradient of biodistribution from proximal to
(Fig. 5G–J), where the ratio of abnormal fibers reached distal PNS tissues, although there was considerable varia-
0.21 ± 0.009 in the treated group compared to 0.33 ± 0.012 bility between animals. As expected, muscle and spinal cord
in the mock group (n = 10 per group; p < 0.0001) (Fig. 5K transduction by the AAV vectors were also evident, but
and Table S6). Moreover, foamy macrophages were EGFP or Cx32 expression was restricted to myelinating
reduced to 4.9 ± 0.73/1000 fibers in the treated group (n = Schwann cells under the control of Mpz promoter. More-
10) compared to 10.0 ± 0.47 in the mock group (n = 10; p < over, we demonstrate that virally delivered Cx32 is cor-
0.0001) (Fig. 5L and Table S6). Thus, our morphological rectly localized in the paranodal areas of myelinating
analysis of the PNS tissues revealed a significant Schwann cells as previously achieved with lentiviral gene
improvement of the pathology including de- and remyeli- delivery [31, 32]. However, AAV9 appears to provide
nation and inflammatory changes that are characteristic of higher expression rates and levels as evident by the
this model of CMT1X both in pre- as well as in post-onset immunoblot analysis in PNS tissues of injected mice. In this
treated mice. However, morphological improvement in study, we injected AAV amounts of 2 × 1010 vg. Higher
treated mice was only partial, as demonstrated by the vector titers allowing the injection of more viral particles
comparison to WT tissues (Tables S4–S6). On the other exceeding >1011 vg may improve further the biodistribution
hand, comparison of 10-month old mock treated with into the PNS and expression rates. Furthermore, it remains
untreated Gjb1-null mice of the same age showed similar to be shown whether the biodistribution achieved by intra-
degree of pathology (Supplementary results, Fig. S6, and thecal injection can be also obtained in a larger animal to
Tables S4–6), indicating that mock vector injection and support the clinical application.
EGFP expression had no detrimental effects in this model. Although AAV9 provided higher Cx32 expression levels
compared to our previous studies, only a subset of Schwann
cells expressed the gene of interest resulting only in partial
Discussion phenotype correction. Interestingly, it appears that restoring
Cx32 function in at least 50% of myelinating Schwann cells
In this study, we used an AAV9 vector for targeted gene in PNS tissues may be sufficient to provide significant
delivery to Schwann cells in order to treat the peripheral functional and morphological improvement in this model
neuropathy in the Gjb1-null mouse model of CMT1X. and potentially also in CMT1X patients. This partial cor-
Using the GJB1 gene coding sequence driven by the rat rection may reflect the situation caused by the random X
AAV9-mediated Schwann cell-targeted gene therapy rescues a model of demyelinating neuropathy 671

Fig. 5 Morphological analysis of anterior lumbar roots and AAV9-Mpz.GJB1 injected mice show improved myelination com-
femoral motor nerves of Gjb1-null mice following post-onset pared with mock-treated littermates with fewer demyelinated (*) and
intrathecal delivery of the AAV9-Mpz.GJB1 vector. These are remyelinated (r) fibers in both tissues as well as fewer foamy macro-
representative images of semithin sections of anterior lumbar spinal phages (arrows in C). Quantification of the ratios of abnormally
roots attached to the spinal cord at low (A) and (B) and higher (C) and myelinated fibers in multiple roots and femoral nerves (n = 10 mice
(D) magnification, with morphometric analysis results (E) and (F), as per group) confirms significant improvement in the numbers of
well as of semithin sections of femoral motor nerves at low (G) and abnormally myelinated fibers (E) and (K) as well as significant
(H) and higher (I) and (J) magnification with related morphometric reduction in the numbers of foamy macrophages (F, L) in the treated
analysis results (K) and (L), from mock and full (GJB1) vector treated compared with mock treated littermates (see also Tables S4 and S6).
mice as indicated, at 10 months of age (4 months after treatment). Scale bars: (A, B), (G, H) 50 μm, (C, D), (I, J) 10 μm.

chromosome inactivation in female patients affected by axon, there may be a threshold of functioning Schwann
CMT1X, who typically present with a milder and non- cells ratio for maintaining axon survival explaining why the
progressive phenotype [59]. As thousands of myelinating correction of a subset of Schwann cells was sufficient to
Schwann cells are needed to myelinate the length of a single improve the outcome in the disease model.
672 A. Kagiava et al.

In this study, we demonstrate that targeted transduction likely to have advanced pathological changes in their nerves
of Schwann cell of the peripheral nervous system can be including demyelination and progressive axonal loss
achieved after a single lumbar intrathecal injection using the [5, 23, 24, 72, 73]. Although the Gjb1-null mouse model
AAV9. AAV9 has been previously used mostly for CNS does not fully reproduce the degree of pathology in patients,
transduction [60] but has not been studied specifically for it develops a significant degree of demyelination and axonal
PNS targeting [61–63]. However, some studies showed that pathology especially in motor fibers at the ages of 6 to
AAV9 can transduce peripheral nerves after intrathecal 10 months [25, 26, 30, 31, 52] in addition to inflammatory
injection [64] with detection of EGFP expression in sciatic changes with macrophage infiltration [53, 54]. Thus,
nerves after intravenous or intrathecal injection in mice demonstration of post-onset efficacy of gene therapy in this
[65, 66]. AAV9 has been mainly used in experimental model indicates that viral expression and restoration of
models of giant axonal neuropathy [61], spinal muscular Cx32 function can be effective even in an intraneural
atrophy [67], and amyotrophic lateral sclerosis [68, 69]. environment of ongoing chronic de- and re-myelination,
These studies demonstrated that AAV9 can transduce effi- axonal degeneration, and inflammation, and is therefore
ciently the CNS resulting in phenotype rescue of these highly relevant for the potential of clinical application.
models. Furthermore, the AAV9 vector has been shown to The therapeutic benefit in this neuropathy model is further
provide gene expression in non-human primates either by demonstrated by the amelioration of NF-L plasma con-
intravenous [70] or by intrathecal [71] and recently by centrations in AAV9-Mpz.GJB1 treated compared to the
subpial delivery [68]. However, there are limited data mock-treated mice. NF-L is a biomarker that correlates with
concerning the efficacy of AAV9 to transduce the PNS. A peripheral nerve degeneration in patients with CMT1X and
recent study showed a gradient of AAV9 biodistribution other CMT types [51]. Since NF-L is associated with axonal
into the PNS following lumbar intrathecal injection or damage, we conclude that treatment with AAV9-Mpz.GJB1
injection in the cisterna magna, with reporter gene expres- either prevents or reverses axonal damage caused by the loss
sion also in the sciatic nerves [64]. In contrast to our study, of Cx32 in Schwann cells. Similar treatment responsiveness
cell specificity of expression could not be demonstrated, of this biomarker was shown in our previous studies using
since ubiquitous promoters were used. The lower efficacy of lentiviral gene therapy in CMT1X and CMT4C demyelinat-
Schwann cell transduction after intrathecal injection in other ing neuropathy models [31, 74]. Interestingly, in this
studies [64] could be in part attributed to the fourfold lower study AAV9-treated Gjb1-null mice showed an even stronger
volumes injected compared to our approach. amelioration of NF-L levels approaching those of WT mice,
Improvement of functional properties in AAV9-Mpz. indicating that the therapeutic benefit and axonal protection
GJB1 treated groups, including behavioral and electro- provided by AAV are stronger. These findings support the
physiological outcomes, indicate that treatment with this use of the AAV9 vector for the treatment of demyelinating
vector not only results in expression of human Cx32 but also neuropathies and confirm NF-L as a relevant biomarker
in restoration of Cx32 function in myelinated fibers. that can be used to measure treatment response in CMT1X
Although pre- and post-onset treatment groups were studied patients.
at different time periods and may not be directly comparable, A further blood biomarker reflecting nerve pathology,
almost similar improvements in functional properties were NCAM1, also showed significant treatment responsiveness.
observed at both time points, indicating that gene replace- Elevated NCAM1 serum levels have been previously asso-
ment can be beneficial either before or after the onset of the ciated with various forms of inflammatory neuropathies and
neuropathy, reversing some of the pathological changes that with CMT1A [75] as well as with Alzheimer’s disease [76],
are already present in older mice. Early treated mice showed indicating that it may be a relatively broad marker of CNS
a trend for higher increase of hind limb force and more and PNS neurodegeneration. Furthermore, NCAM-1 was
improvement of MNCVs approaching those of age-matched shown to regulate synaptic reorganization after peripheral
WT mice, while in post-onset treated mice they remained nerve injury suggesting an important role during regenera-
significantly below those of WT mice. Thus, early inter- tion [77]. We demonstrate here for the first time a significant
vention may provide a bigger therapeutic benefit, although increase of NCAM-1 levels in 10-month-old Gjb1-null mice
not clearly demonstrated in all outcome measures, in compared to WT controls, highlighting its relevance for
accordance with our previous studies [31, 33]. The relatively CMT1X pathology. Importantly, treated Gjb1-null mice
slow progression of the pathological and functional changes showed a significant amelioration in serum NCAM-1 levels
beyond 6 months of age in the Gjb1-null model at baseline compared to the mock group at 10 months of age, suggesting
[25, 31, 33] may account for these observations. the potential value of this biomarker in CMT1X. However,
By the time CMT1X patients become symptomatic, further confirmatory studies in the serum of CMTX1 patients
typically by late childhood to young adulthood, they are are required to validate this finding.
AAV9-mediated Schwann cell-targeted gene therapy rescues a model of demyelinating neuropathy 673

Our morphological analysis revealed that treatment with Compliance with ethical standards
the AAV9 vector results in improvement of the pathological
properties of all the PNS tissues examined. Both the ratios Conflict of interest The study is part of a PCT/EP2020/065312
application in which AK, IS and KAK are co-inventors.
of abnormally myelinated fibers and the numbers of foamy
macrophages were reduced in fully treated mice compared
to the mock group in both pre- and post-onset studies. The Publisher’s note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
percentage of improvement reached 60% for the roots and
the femoral nerve and 50% for the sciatic nerves, although Open Access This article is licensed under a Creative Commons
baseline values were higher in late treated mice as expected Attribution 4.0 International License, which permits use, sharing,
based on the pathology progression in the Gjb1-null mouse adaptation, distribution and reproduction in any medium or format, as
model [30]. Stronger improvement in tissues with motor long as you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons license, and indicate if
fibers including the anterior lumbar roots and femoral motor changes were made. The images or other third party material in this
branches likely reflects the fact that neuropathy in this article are included in the article’s Creative Commons license, unless
model affects predominantly the motor fibers [25, 78]. indicated otherwise in a credit line to the material. If material is not
Overall, our data indicate that AAV9-mediated delivery of included in the article’s Creative Commons license and your intended
use is not permitted by statutory regulation or exceeds the permitted
Cx32 can improve PNS pathology both before and after the use, you will need to obtain permission directly from the copyright
onset of the CMT1X neuropathy. holder. To view a copy of this license, visit http://creativecommons.
In conclusion, this study provides the proof of principle org/licenses/by/4.0/.
that intrathecal injection of AAV9, a well-established vector
for clinical applications, provides widespread biodistribu- References
tion in the PNS and efficient transduction of myelinating
Schwann cells. Moreover, Schwann cell-targeted expression 1. Fridman V, Bundy B, Reilly MM, Pareyson D, Bacon C, Burns J,
et al. CMT subtypes and disease burden in patients enrolled in the
of Cx32 driven by a myelin-specific promoter can prevent Inherited Neuropathies Consortium natural history study: a cross-
or even reverse the demyelination and axonal degeneration sectional analysis. J Neurol Neurosurg Psychiatry. 2015;86:873–8.
occurring in the CMT1X mouse model providing a long- 2. Birouk N, LeGuern E, Maisonobe T, Rouger H, Gouider R,
lasting therapeutic benefit even after the onset of the neu- Tardieu S, et al. X-linked Charcot-Marie-Tooth disease with
connexin 32 mutations: clinical and electrophysiologic study.
ropathy. Evaluation of this approach in larger animal Neurology. 1998;50:1074–82.
models would be valuable in order to confirm adequate 3. Dubourg O, Tardieu S, Birouk N, Gouider R, Leger JM, Maiso-
biodistribution and safety facilitating the path towards nobe T, et al. Clinical, electrophysiological and molecular genetic
clinical translation for treating CMT1X patients. characteristics of 93 patients with X-linked Charcot-Marie-Tooth
disease. Brain. 2001;124:1958–67.
4. Hahn AF, Brown WF, Koopman WJ, Feasby TE. X-linked
Acknowledgements We would like to thank Dr. Alex Rossor for his dominant hereditary motor and sensory neuropathy. Brain.
help with NF-L analysis. 1990;113:1511–25.
5. Kleopa KA, Scherer SS. Molecular genetics of X-linked Charcot-
Author contributions AK conducted the experiments, acquired data, Marie-Tooth disease. Neuromolecular Med. 2006;8:107–22.
analyzed data, and wrote the manuscript. JR, CT, and CC produced the 6. Lu YY, Lyu H, Jin SQ, Zuo YH, Liu J, Wang ZX, et al. Clinical
viral vector. CK conducted behavioral tests. IS conducted cloning and and genetic features of Chinese X-linked Charcot-Marie-Tooth
mice PCR screening. MS performed morphological studies. AH, MR type 1 disease. Chin Med J. 2017;130:1049–54.
and HZ performed an analysis of plasma neurofilament light levels. MJ 7. Saporta AS, Sottile SL, Miller LJ, Feely SM, Siskind CE, Shy
and RH performed and analyzed ELISA analysis of blood biomarkers. ME. Charcot-Marie-Tooth disease subtypes and genetic testing
KAK designed research studies, performed morphometric analysis, strategies. Ann Neurol. 2011;69:22–33.
analyzed data, and wrote the manuscript. All authors critically 8. Liang C, Howells J, Kennerson M, Nicholson GA, Burke D, Ng
reviewed and approved the final manuscript. K. Axonal excitability in X-linked dominant Charcot Marie Tooth
disease. Clin Neurophysiol. 2014;125:1261–9.
Funding This study was co-funded by the Muscular Dystrophy 9. Martikainen MH, Majamaa K. Novel GJB1 mutation causing
Association and Charcot-Marie-Tooth Association (Grants 480030 and adult-onset Charcot-Marie-Tooth disease in a female patient.
603003 to KAK). This work was funded by the Republic of Cyprus Neuromuscul Disord. 2013;23:899–901.
through the Research and Innovation Foundation (Project: CULTURE/ 10. Jerath NU, Gutmann L, Reddy CG, Shy ME. Charcot-Marie-tooth
BR-NE/0416/07). HZ is a Wallenberg Scholar and the fluid biomarker disease type 1X in women: electrodiagnostic findings. Muscle
measurements in the lab of HZ and AJH were supported by the UK Nerve. 2016;54:728–32.
Dementia Research Institute at UCL. MMR acknowledges funding 11. Scherer SS, Deschenes SM, Xu YT, Grinspan JB, Fischbeck KH,
support from the Medical Research Council (MRC MR/S005021/1), Paul DL. Connexin32 is a myelin-related protein in the PNS and
the National Institutes of Neurological Diseases and Stroke, and the CNS. J Neurosci. 1995;15:8281–94.
Office of Rare Diseases (U54NS065712), and Muscular Dystrophy 12. Bruzzone R, White TW, Paul DL. Connections with connexins:
Association (MDA510281). RH is supported by the Wellcome the molecular basis of direct intercellular signaling. Eur J Bio-
Investigator Award [109915/Z/15/Z], the Medical Research Council chem. 1996;238:1–27.
(UK) [MR/N025431/1], the Newton Fund [UK/Turkey, MR/N027302/ 13. Balice-Gordon RJ, Bone LJ, Scherer SS. Functional gap junctions in
1], the Lily Foundation, and the Evelyn Trust. the Schwann cell myelin sheath. J Cell Biol. 1998;142:1095–104.
674 A. Kagiava et al.

14. Murphy SM, Polke J, Manji H, Blake J, Reiniger L, Sweeney M, demyelinating peripheral neuropathy. Proc Natl Acad Sci USA.
et al. A novel mutation in the nerve-specific 5′UTR of the GJB1 2016;113:E2421–9.
gene causes X-linked Charcot-Marie-Tooth disease. J Peripher 33. Kagiava A, Karaiskos C, Richter J, Tryfonos C, Lapathitis G, Sar-
Nerv Syst. 2011;16:65–70. giannidou I, et al. Intrathecal gene therapy in mouse models
15. Tomaselli PJ, Rossor AM, Horga A, Jaunmuktane Z, Carr A, expressing CMT1X mutations. Hum Mol Genet. 2018;27:1460–73.
Saveri P, et al. Mutations in noncoding regions of GJB1 are a 34. Kyriakoudi S, Sargiannidou I, Kagiava A, Olympiou M, Kleopa
major cause of X-linked CMT. Neurology. 2017;88:1445–53. KA. Golgi-retained Cx32 mutants interfere with gene addition
16. Omori Y, Mesnil M, Yamasaki H. Connexin 32 mutations from therapy for CMT1X. Hum Mol Genet. 2017;26:1622–33.
X-linked Charcot-Marie-Tooth disease patients: functional defects 35. Hargrove PW, Kepes S, Hanawa H, Obenauer JC, Pei D, Cheng
and dominant negative effects. Mol Biol Cell. 1996;7:907–16. C, et al. Globin lentiviral vector insertions can perturb the
17. Yoshimura T, Satake M, Ohnishi A, Tsutsumi Y, Fujikura Y. expression of endogenous genes in beta-thalassemic hemato-
Mutations of connexin32 in Charcot-Marie-Tooth disease type X poietic cells. Mol Ther. 2008;16:525–33.
interfere with cell-to-cell communication but not cell proliferation and 36. Bradbury AM, Rafi MA, Bagel JH, Brisson BK, Marshall MS,
myelin-specific gene expression. J Neurosci Res. 1998;51:154–61. Pesayco Salvador J, et al. AAVrh10 gene therapy ameliorates
18. Yum SW, Kleopa KA, Shumas S, Scherer SS. Diverse trafficking central and peripheral nervous system disease in canine globoid
abnormalities of connexin32 mutants causing CMTX. Neurobiol cell leukodystrophy (Krabbe disease). Hum Gene Ther.
Dis. 2002;11:43–52. 2018;29:785–801.
19. Deschenes SM, Walcott JL, Wexler TL, Scherer SS, Fischbeck 37. Karumuthil-Melethil S, Marshall MS, Heindel C, Jakubauskas B,
KH. Altered trafficking of mutant connexin32. J Neurosci. Bongarzone ER, Gray SJ. Intrathecal administration of AAV/
1997;17:9077–84. GALC vectors in 10-11-day-old twitcher mice improves survival
20. Kleopa KA, Yum SW, Scherer SS. Cellular mechanisms of con- and is enhanced by bone marrow transplant. J Neurosci Res.
nexin32 mutations associated with CNS manifestations. J Neu- 2016;94:1138–51.
rosci Res. 2002;68:522–34. 38. Gurda BL, Vite CH. Large animal models contribute to the
21. Martin PE, Mambetisaeva ET, Archer DA, George CH, Evans development of therapies for central and peripheral nervous sys-
WH. Analysis of gap junction assembly using mutated connexins tem dysfunction in patients with lysosomal storage diseases. Hum
detected in Charcot-Marie-Tooth X-linked disease. J Neurochem. Mol Genet. 2019;28:R119–31.
2000;74:711–20. 39. Homs J, Pages G, Ariza L, Casas C, Chillon M, Navarro X, et al.
22. Oh S, Ri Y, Bennett MV, Trexler EB, Verselis VK, Bargiello TA. Intrathecal administration of IGF-I by AAVrh10 improves sensory
Changes in permeability caused by connexin 32 mutations underlie and motor deficits in a mouse model of diabetic neuropathy. Mol
X-linked Charcot-Marie-Tooth disease. Neuron. 1997;19:927–38. Ther Methods Clin Dev. 2014;1:7.
23. Panosyan FB, Laura M, Rossor AM, Pisciotta C, Piscosquito G, 40. Ahmed SG, Abdelanabi A, Doha M, Brenner GJ. Schwannoma
Burns J, et al. Cross-sectional analysis of a large cohort with X- gene therapy by adeno-associated virus delivery of the pore-forming
linked Charcot-Marie-Tooth disease (CMTX1). Neurology. protein Gasdermin-D. Cancer Gene Ther. 2019;26:259–67.
2017;89:927–35. 41. Prabhakar S, Taherian M, Gianni D, Conlon TJ, Fulci G,
24. Shy ME, Siskind C, Swan ER, Krajewski KM, Doherty T, Fuerst Brockmann J, et al. Regression of schwannomas induced by
DR, et al. CMT1X phenotypes represent loss of GJB1 gene adeno-associated virus-mediated delivery of caspase-1. Hum Gene
function. Neurology. 2007;68:849–55. Ther. 2013;24:152–62.
25. Anzini P, Neuberg DH, Schachner M, Nelles E, Willecke K, 42. Sargiannidou I, Kagiava A, Bashiardes S, Richter J, Christodou-
Zielasek J, et al. Structural abnormalities and deficient main- lou C, Scherer SS, et al. Intraneural GJB1 gene delivery improves
tenance of peripheral nerve myelin in mice lacking the gap nerve pathology in a model of X-linked Charcot-Marie-Tooth
junction protein connexin 32. J Neurosci. 1997;17:4545–51. disease. Ann Neurol. 2015;78:303–16.
26. Scherer SS, Xu YT, Nelles E, Fischbeck K, Willecke K, Bone LJ. 43. Arden E, Metzger JMInexpensive. serotype-independent protocol
Connexin32-null mice develop demyelinating peripheral neuro- for native and bioengineered recombinant adeno-associated virus
pathy. Glia. 1998;24:8–20. purification. J Biol Methods. 2016;3:2.
27. Scherer SS, Xu YT, Messing A, Willecke K, Fischbeck KH, Jeng 44. Geraerts M, Willems S, Baekelandt V, Debyser Z, Gijsbers R.
LJ. Transgenic expression of human connexin32 in myelinating Comparison of lentiviral vector titration methods. BMC Bio-
Schwann cells prevents demyelination in connexin32-null mice. J technol. 2006;6:34.
Neurosci. 2005;25:1550–9. 45. Nelles E, Butzler C, Jung D, Temme A, Gabriel HD, Dahl U, et al.
28. Abel A, Bone LJ, Messing A, Scherer SS, Fischbeck KH. Studies Defective propagation of signals generated by sympathetic nerve
in transgenic mice indicate a loss of connexin32 function in X- stimulation in the liver of connexin32-deficient mice. Proc Natl
linked Charcot-Marie-Tooth disease. J Neuropathol Exp Neurol. Acad Sci USA. 1996;93:9565–70.
1999;58:702–10. 46. Kagiava A, Kleopa KA. Intrathecal delivery of viral vectors for
29. Jeng LJ, Balice-Gordon RJ, Messing A, Fischbeck KH, Scherer gene therapy. Methods Mol Biol. 2018;1791:277–85.
SS. The effects of a dominant connexin32 mutant in myelinating 47. Ahn M, Lee J, Gustafsson A, Enriquez A, Lancaster E, Sul JY,
Schwann cells. Mol Cell Neurosci. 2006;32:283–98. et al. Cx29 and Cx32, two connexins expressed by myelinating
30. Sargiannidou I, Vavlitou N, Aristodemou S, Hadjisavvas A, glia, do not interact and are functionally distinct. J Neurosci Res.
Kyriacou K, Scherer SS, et al. Connexin32 mutations cause loss 2008;86:992–1006.
of function in Schwann cells and oligodendrocytes leading to PNS 48. Zielasek J, Martini R, Toyka KV. Functional abnormalities in P0-
and CNS myelination defects. J Neurosci. 2009;29:4736–49. deficient mice resemble human hereditary neuropathies linked to
31. Kagiava A, Richter J, Tryfonos C, Karaiskos C, Heslegrave AJ, P0 gene mutations. Muscle Nerve. 1996;19:946–52.
Sargiannidou I, et al. Gene replacement therapy after neuropathy 49. Parasuraman S, Raveendran R, Kesavan R. Blood sample col-
onset provides therapeutic benefit in a model of CMT1X. Hum lection in small laboratory animals. J Pharmacol Pharmacother.
Mol Genet. 2019;28:3528–42. 2010;1:87–93.
32. Kagiava A, Sargiannidou I, Theophilidis G, Karaiskos C, Richter 50. Rohrer JD, Woollacott IO, Dick KM, Brotherhood E, Gordon E,
J, Bashiardes S, et al. Intrathecal gene therapy rescues a model of Fellows A, et al. Serum neurofilament light chain protein is a
AAV9-mediated Schwann cell-targeted gene therapy rescues a model of demyelinating neuropathy 675

measure of disease intensity in frontotemporal dementia. Neurol- 65. Schuster DJ, Dykstra JA, Riedl MS, Kitto KF, Belur LR, McIvor
ogy. 2016;87:1329–36. RS, et al. Biodistribution of adeno-associated virus serotype 9
51. Sandelius A, Zetterberg H, Blennow K, Adiutori R, Malaspina A, (AAV9) vector after intrathecal and intravenous delivery in
Laura M, et al. Plasma neurofilament light chain concentration in the mouse. Front Neuroanat. 2014;8:42.
inherited peripheral neuropathies. Neurology. 2018;90:e518–24. 66. Tanguy Y, Biferi MG, Besse A, Astord S, Cohen-Tannoudji M,
52. Vavlitou N, Sargiannidou I, Markoullis K, Kyriacou K, Scherer Marais T, et al. Systemic AAVrh10 provides higher transgene
SS, Kleopa KA. Axonal pathology precedes demyelination in a expression than AAV9 in the brain and the spinal cord of neonatal
mouse model of X-linked demyelinating/type I Charcot-Marie mice. Front Mol Neurosci. 2015;8:36.
Tooth neuropathy. J Neuropathol Exp Neurol. 2010;69:945–58. 67. Duque SI, Arnold WD, Odermatt P, Li X, Porensky PN, Schmelzer
53. Groh J, Heinl K, Kohl B, Wessig C, Greeske J, Fischer S, et al. L, et al. A large animal model of spinal muscular atrophy and
Attenuation of MCP-1/CCL2 expression ameliorates neuropathy correction of phenotype. Ann Neurol. 2015;77:399–414.
in a mouse model for Charcot-Marie-Tooth 1X. Hum Mol Genet. 68. Bravo-Hernandez M, Tadokoro T, Navarro MR, Platoshyn O,
2010;19:3530–43. Kobayashi Y, Marsala S, et al. Spinal subpial delivery of AAV9
54. Kobsar I, Berghoff M, Samsam M, Wessig C, Maurer M, Toyka enables widespread gene silencing and blocks motoneuron
KV, et al. Preserved myelin integrity and reduced axonopathy in degeneration in ALS. Nat Med. 2020;26:118–30.
connexin32-deficient mice lacking the recombination activating 69. Foust KD, Salazar DL, Likhite S, Ferraiuolo L, Ditsworth D,
gene-1. Brain. 2003;126:804–13. Ilieva H, et al. Therapeutic AAV9-mediated suppression of mutant
55. Blits B, Derks S, Twisk J, Ehlert E, Prins J, Verhaagen J. Adeno- SOD1 slows disease progression and extends survival in models
associated viral vector (AAV)-mediated gene transfer in the red of inherited ALS. Mol Ther. 2013;21:2148–59.
nucleus of the adult rat brain: comparative analysis of the trans- 70. Gray SJ, Matagne V, Bachaboina L, Yadav S, Ojeda SR,
duction properties of seven AAV serotypes and lentiviral vectors. Samulski RJ. Preclinical differences of intravascular AAV9
J Neurosci Methods. 2010;185:257–63. delivery to neurons and glia: a comparative study of adult mice
56. Hutson TH, Verhaagen J, Yanez-Munoz RJ, Moon LD. Corti- and nonhuman primates. Mol Ther. 2011;19:1058–69.
cospinal tract transduction: a comparison of seven adeno- 71. Gray SJ, Nagabhushan Kalburgi S, McCown TJ, Jude, Samulski
associated viral vector serotypes and a non-integrating lentiviral R. Global CNS gene delivery and evasion of anti-AAV-
vector. Gene Ther. 2012;19:49–60. neutralizing antibodies by intrathecal AAV administration in
57. Mason MR, Ehlert EM, Eggers R, Pool CW, Hermening S, Husei- non-human primates. Gene Ther. 2013;20:450–9.
novic A, et al. Comparison of AAV serotypes for gene delivery to 72. Hahn AF, Ainsworth PJ, Naus CC, Mao J, Bolton CF. Clinical
dorsal root ganglion neurons. Mol Ther. 2010;18:715–24. and pathological observations in men lacking the gap junction
58. Lentz TB, Gray SJ, Samulski RJ. Viral vectors for gene delivery protein connexin 32. Muscle Nerve Suppl. 2000;9:S39–48.
to the central nervous system. Neurobiol Dis. 2012;48:179–88. 73. Hattori N, Yamamoto M, Yoshihara T, Koike H, Nakagawa M,
59. Siskind CE, Murphy SM, Ovens R, Polke J, Reilly MM, Shy ME. Yoshikawa H, et al. Demyelinating and axonal features of
Phenotype expression in women with CMT1X. J Peripher Nerv Charcot-Marie-Tooth disease with mutations of myelin-related
Syst. 2011;16:102–7. proteins (PMP22, MPZ and Cx32): a clinicopathological study of
60. Lykken EA, Shyng C, Edwards RJ, Rozenberg A, Gray SJ. Recent 205 Japanese patients. Brain. 2003;126:134–51.
progress and considerations for AAV gene therapies targeting the 74. Schiza N, Georgiou E, Kagiava A, Medard JJ, Richter J, Tryfonos
central nervous system. J Neurodev Disord. 2018;10:16. C, et al. Gene replacement therapy in a model of Charcot-Marie-
61. Bailey RM, Armao D, Nagabhushan Kalburgi S, Gray SJ. Tooth 4C neuropathy. Brain. 2019;142:1227–41.
Development of intrathecal AAV9 gene therapy for giant axonal 75. Kim YH, Kim YH, Shin YK, Jo YR, Park DK, Song MY, et al.
neuropathy. Mol Ther Methods Clin Dev. 2018;9:160–71. p75 and neural cell adhesion molecule 1 can identify pathologic
62. Hinderer C, Bell P, Katz N, Vite CH, Louboutin JP, Bote E, et al. Schwann cells in peripheral neuropathies. Ann Clin Transl Neurol.
Evaluation of intrathecal routes of administration for adeno- 2019;6:1292–301.
associated viral vectors in large animals. Hum Gene Ther. 76. Todaro L, Puricelli L, Gioseffi H, Guadalupe Pallotta M, Lastiri J,
2018;29:15–24. Bal de Kier Joffe E, et al. Neural cell adhesion molecule in human
63. Ramsingh AI, Gray SJ, Reilly A, Koday M, Bratt D, Koday MT, serum. Increased levels in dementia of the Alzheimer type. Neu-
et al. Sustained AAV9-mediated expression of a non-self protein robiol Dis. 2004;15:387–93.
in the CNS of non-human primates after immunomodulation. 77. Ko HG, Choi JH, Park DI, Kang SJ, Lim CS, Sim SE, et al. Rapid
PLoS One. 2018;13:e0198154. turnover of cortical NCAM1 regulates synaptic reorganization
64. Bailey RM, Rozenberg A, Gray SJ. Comparison of high-dose after peripheral nerve injury. Cell Rep. 2018;22:748–59.
intracisterna magna and lumbar puncture intrathecal delivery of 78. Scherer SS, Wrabetz L. Molecular mechanisms of inherited
AAV9 in mice to treat neuropathies. Brain Res. 2020;1739:146832. demyelinating neuropathies. Glia. 2008;56:1578–89.