Abstract
X-linked Charcot-Marie-Tooth disease (CMT1X), one of the commonest forms of inherited demyelinating neuropathy, results from GJB1 gene mutations causing loss of function of the gap junction protein connexin32 (Cx32). The aim of this study was to examine whether delayed gene replacement therapy after the onset of peripheral neuropathy can provide a therapeutic benefit in the Gjb1-null/Cx32 knockout model of CMT1X. After delivery of the LV-Mpz.GJB1 lentiviral vector by a single lumbar intrathecal injection into 6-month-old Gjb1-null mice, we confirmed expression of Cx32 in lumbar roots and sciatic nerves correctly localized at the paranodal myelin areas. Gjb1-null mice treated with LV-Mpz.GJB1 compared with LV-Mpz.Egfp (mock) vector at the age of 6 months showed improved motor performance at 8 and 10 months. Furthermore, treated mice showed increased sciatic nerve conduction velocities, improvement of myelination and reduced inflammation in lumbar roots and peripheral nerves at 10 months of age, along with enhanced quadriceps muscle innervation. Plasma neurofilament light (NEFL) levels, a clinically relevant biomarker, were also ameliorated in fully treated mice. Intrathecal gene delivery after the onset of peripheral neuropathy offers a significant therapeutic benefit in this disease model, providing a proof of principle for treating patients with CMT1X at different ages.
Introduction
X-linked Charcot-Marie Tooth (CMT1X) disease is one of the commonest inherited neuropathies, characterized by progressive weakness and atrophy of distal limb muscles, loss of reflexes, sensory loss and reduced nerve conduction velocities. Disability increases with age mainly resulting from motor unit loss (1). CMT1X results from mutations in GJB1, the gene encoding the gap junction (GJ) protein connexin32 (Cx32). In addition to peripheral neuropathy, a number of CMT1X mutations also result in central nervous system phenotypes characterized by spasticity, hyperactive reflexes, extensor plantar responses, ataxia, or acute reversible encephalopathy (2–6). CMT1X affects mostly men with onset at the age of 5 to 20 years (3, 7, 8), while affected women may present at a later stage (9, 10) with milder symptoms (11).
Cx32 forms intracellular GJ channels through the non-compact myelin layers at paranodal loops and Schmidt–Lanterman incisures of Schwann cells (12, 13). These channels serve important homeostatic and axon-glial signaling functions in peripheral myelinated fibers. There are over 400 different GJB1 mutations reported so far (http://hihg.med.miami.edu/code/http/cmt/public_html/index.html#/; date last accessed December 2018) affecting all domains of Cx32, as well as the non-coding gene regions (14, 15). Clinical series of CMT1X families with different mutations including complete deletion of the GJB1 open reading frame show that patients have a similar phenotype and clinical course, indicating a loss of function mechanism underlying the disease (1, 16). Therefore, gene replacement therapy may potentially treat the neuropathy.
We have previously developed a gene delivery method by a single lumbar intrathecal injection for targeted expression in Schwann cells. Using a lentiviral vector carrying the GJB1 gene driven by the myelin protein zero (Mpz) promoter, we have shown that gene addition at the age of 2 months can rescue the demyelinating neuropathy in the Gjb1-null/Cx32 knockout (KO) model of CMT1X, either by direct intraneural (17) or by intrathecal injection (18). These results have been replicated in 2-month-old Gjb1-null mice expressing the endoplasmic reticulum-retained T55I mutant associated with CMT1X (19). However, in these studies, we treated Gjb1-null mice before the onset of disease which occurs after 3 months of age (20–22). Given the clinical course of the disease, therapeutic benefit from gene therapy needs to be also demonstrated after the onset of the neuropathy, as this would be highly relevant for treating patients with CMT1X with long established peripheral neuropathy.
Thus, in order to provide a clinically relevant proof of concept for the effectiveness of gene therapy for CMT1X, even after the onset of the disease, we delivered intrathecally a lentiviral vector carrying the human GJB1 gene into 6-month-old Gjb1-null mice. This late intervention resulted in a widespread expression of Cx32 in lumbar roots and sciatic nerves of older mice with correct localization at the paranodal areas. Treatment trial initiated at 6 months of age resulted in increased motor performance that was maintained up to 10 months of age along with electrophysiological and morphological improvement. Our results confirm the efficacy of gene addition to provide therapeutic benefit in the model of CMT1X even after the onset of the neuropathy, supporting further the potential for this approach to treat CMT1X patients.
Results
Virally induced expression of GJB1 gene in 6-month-old Gjb1-null mice
We have previously shown that LV-Mpz.GJB1 (full) lentiviral vector delivery of the human GJB1 gene into 2 month-old Gjb1-null mice, before the onset of peripheral neuropathy, results in stable expression of Cx32 (18). In this study, we examined the expression levels of the same vector following delivery to 6-month-old mice, after the onset of the disease. Expression levels of enhanced green fluorescent protein (EGFP) and Cx32 were determined 4 and 8 weeks post-injection in lumbar roots and sciatic nerves. EGFP was detected in the perinuclear cytoplasm in a subset of Schwann cells in both the lumbar root sections (Fig. 1A and B) and in sciatic nerve teased fibers (Fig. 1C and D). Quantification of EGFP expression (Fig. 1E) 4 weeks post-injection (at 7 months of age, in n = 4 mice examined) showed that EGFP expression rates (% of EGFP-positive Schwann cells) reached 29.6 ± 5.81% in lumbar roots and 42.7 ± 2.61% in sciatic nerves, while at 8 weeks (8 months of age, n = 3 mice examined), expression rates reached 21.0 ± 2.02% in lumbar roots and 33.3 ± 3.93% in sciatic nerves (data not shown). Differences between the two time points were not significant (P > 0.05).
Analysis of reporter gene and Cx32 expression following LV-Mpz.GJB1 injection in 6-month-old Gjb1-null mice. Immunostaining for EGFP (red) of lumbar root sections (A-B) or sciatic nerve teased fibers (C-D) reveals perinuclear EGFP immunoreactivity (asterisks) in a subset of Schwann cells in tissues from injected mice (B, D) that are not detected in non-injected mice (A, C). Cell nuclei are stained with DAPI (blue). Quantification of the percentage of EGFP-positive cells in lumbar roots and sciatic nerves 4 weeks post-injection is shown in E (n = 4). Values represent mean ± SEM. F-G: Double staining of lumbar root sections for Cx32 (red) and paranodal axonal marker Caspr (green) shows that virally delivered Cx32 (white arrows) is correctly expressed at some of the paranodal myelin areas (green arrowheads) only in injected mice (G) but not in non-injected littermates (F). H-I: Double staining of sciatic nerve teased fibers for Cx32 (green) and juxtaparanodal axonal marker Caspr2 (red, arrowheads) reveals expression of Cx32 in paranodal areas adjacent to Caspr2-positive juxtaparanodes in injected (I) but not in non-injected mice (H). Quantification of the percentage of Cx32-positive paranodal areas in lumbar roots and sciatic nerves 4 weeks post injection is shown in J (n = 4). Values represent mean ± SEM. Scale bars: 10 μm.
Expression of Cx32 in peripheral nervous system (PNS) tissues of 6-month-old Gjb1-null mice was also detected both at 4 and 8 weeks after full vector delivery, correctly localized to the paranodal areas of lumbar roots (Fig. 1G) and sciatic nerves (Fig. 1I), but was absent in non-injected littermates (Fig. 1F and H). Quantification of Cx32 expression rates 4 weeks post-injection (n = 4 mice examined) showed that 37.2 ± 3.4% of paranodal areas identified by labeling with paranodal marker Caspr were Cx32-positive in lumbar roots and 43.5 ± 6.18% in sciatic nerves (Fig. 1J). Cx32 expression rates remained stable at 8 weeks post-injection (n = 3 mice examined), reaching 40.1 ± 7.40% in lumbar roots and 42.8 ± 0.62% in sciatic nerves (data no shown).
Improvement of motor performance in treated 6-month-old Gjb1-null mice
After confirming adequate vector expression in 6-month-old Gjb1-null mice with already advanced peripheral nerve pathology, we proceeded to the treatment trial. For this purpose, mice were randomized into two groups, one receiving the mock (LV-Mpz.Egfp) and the other the full (LV-Mpz.GJB1) vector. Motor performance was assessed before the injection at the age of 6 months and again at the age of 8 (2 months post-injection) and 10 months (4 months post-injection) by rotarod (at 20 and 32 RPM) and foot grip tests. Rotarod analysis at the age of 6 months, at baseline before initiating the treatment with the lentiviral vector, showed, as expected, no differences between the two groups at either 20 or at 32 RPM (Supplementary Material, Table S1 and Fig. 2A and B). Likewise, foot grip test showed no significant difference between the two groups (Fig. 2C).
Behavioral analysis of LV-Mpz.GJB1 injected Gjb1-null mice compared to LV-Mpz.Egfp (mock) treated littermates. These are the results of rotarod (A–B, D-E, G-H, J-K) and foot grip (C, F, I, L) testing of motor performance in LV-Mpz.GJB1 treated (GJB1) compared to mock treated Gjb1-null mice, as indicated. Rotarod and foot grip analysis did not show any differences between the two groups at 6 months of age, before the injection (A-C). At the age of 8 months rotarod analysis showed improved motor performance at both speeds examined in fully treated Gjb1-null mice compared to the mock vector treated (n = 35 full and n = 32 mock) (D-E). Foot grip analysis showed also improvement of the fully treated compared with the mock-treated Gjb1-null mice at 8 months of age (F) and this improvement was sustained at 10 months of age at the higher speed (H) and at the foot grip analysis (I), but not at the lower speed (G). Time course analysis of each group showed improved motor performance of fully treated Gjb1-null mice in rotarod and foot grip analysis 2 months post-injection (8 months of age) and then motor performance remained stable up to 10 months of age (J-L). In contrast, mock treated mice did not improve over time as indicated by both behavioral tests (J-L). Values represent mean ± SEM. Data were compared using the Mann-Whitney U test.
At 2 months post-injection, at the age of 8 months, there was a significant improvement in the fully treated group (n = 35) compared with the mock vector-treated mice (n = 32) as indicated by the results of both rotarod and foot grip analysis. Fully treated mice remained on the rotarod significantly longer both at 20 and at 32 RPM (Supplementary Material, Table S1 and Fig. 2D and E). Fully treated mice also generated a higher force in the foot grip test compared to mock-treated mice (Supplementary Material, Table S1 and Fig. 2F). The significant improvement of motor performance in the treatment group was maintained up to 4 months after treatment, at the age of 10 months. Although at the speed of 20 RPM, the improvement in the full treatment group did not reach statistical significance, at 32 RPM, they performed better than mock-treated littermates (Supplementary Material, Table S1 and Fig. 2G and H). Moreover, 10-month-old fully treated Gjb1-null mice showed significantly higher-force generation in the foot grip test compared with mock-treated littermates (Supplementary Material, Table S1 and Fig. 2I).
Longitudinal comparison of the motor performance at different time points within each treatment group showed that fully treated mice improved significantly 2 months post-injection, from 6 to 8 months of age, and this improvement was maintained until 10 months of age (Fig. 2J–L and Supplementary Material, Table S1). Although comparison between the baseline (6 month of age) and either 8 or 10 months of age showed improvement in all behavioral tests in treated mice, there was no significant change from 8 to 10 months of age, indicating that improvement occurred during the first months after treatment while motor performance remained stable thereafter without further improvement. In contrast, in the mock vector-treated group, there was no significant change in any of the behavioral tests between the time points examined (Fig. 2J–L and Supplementary Material, Table S1).
Improvement of sciatic nerve motor conduction velocity and quadriceps muscle contractility in late-treated Gjb1-null mice
Motor nerve conduction velocity (MNCV) of sciatic nerve was measured 4 months after treatment in order to assess functional properties in both groups. MNCV values were improved in the fully treated group reaching 37.0 ± 1.29 m/s (n = 18), while velocities in the mock group were 25.5 ± 1.16 m/s (n = 12) (Fig. 3A; P < 0.0001). MNCV values of the fully treated group were close to those of the wild type (WT) mice which reached 41.7 ± 1.62 (n = 8; P > 0.05). Further to the MNCV, we also measured the compound muscle action potential (CMAP) amplitude in both groups and found a trend for higher amplitudes in treated mice which reached 2.4 ± 0.22 mV (n = 18) compared to the mock group that reached 2.0 ± 0.22 mV (n = 12), although not statistically significant (Fig. 3B; P > 0.05). Sciatic CMAP amplitudes of treated mice remained lower than those of WT mice which reached 3.3 ± 0.29 mV (n = 8; P = 0.0259).
Physiological improvement in LV-Mpz.GJB1 treated Gjb1-null mice. Motor nerve conduction velocity (MNCV) was improved in the 10-month-old fully treated Gjb1-null mice (n = 18) compared to the mock vector injected littermates (n = 12) and approached the values of WT (n = 8) (A), whereas the amplitude of compound muscle action potential (CMAP) was not significantly increased and remained lower than the WT amplitude (B). Both the force (C), as well as the duration (D) of quadriceps muscle contraction were increased in fully treated mice (GJB1; n = 10) at 10 months of age compared to mock-treated mice (mock; n = 9). Values represent mean ± SEM. Data were compared using the Mann-Whitney U test (A, B) and the unpaired two-sided Student’s t-test (C, D).
Furthermore, we analyzed the quadriceps muscle contractility after stimulation of the femoral motor nerve in situ as previously described (18). Both force and duration of contraction were used to assess functional improvement. This analysis showed that force and duration were improved in the fully treated mice compared to the mock group (P < 0.05). The highest mean values were observed at a 6 mm extension in all cases with the highest value for mock-treated mice at 0.27 ± 0.02 N (n = 9 mice) and for fully treated at 0.34 ± 0.02 N (n = 10) (Fig. 3C). The mean duration of quadriceps contraction (Fig. 3D) reached 73.3 ± 3.07 ms in mock-treated mice (n = 9) while the duration of the fully treated mice reached 85.3 ± 4.61 ms (n = 10; P < 0.05). Thus, electrophysiological studies with emphasis on motor fibers that are predominantly affected in this CMT1X model showed significant improvement following post-onset gene therapy.
Improved neuromuscular junction innervation in treated mice
Neuromuscular junction (NMJ) innervation was measured in sections of 10-month-old WT (Fig. 4A) as well as of mock (Fig. 4B) compared to full vector-treated (Fig. 4C) Gjb1-null mice. Partial or completely lost innervation was frequently observed in Gjb1-null quadriceps muscle with bungarotoxin-labeled endplates partially or completely devoid of axon terminals, but these abnormalities were improved in treated mice. Quantification confirmed improved NMJ innervation in fully treated compared with mock-treated mice although the values of fully treated mice did not reach the WT levels. The percentage of fully innervated NMJs showed a trend of increase in the fully treated mice (34.9 ± 6.73%; n = 471 NMJs from 6 mice) compared to the mock group (19.4 ± 5.54%; n = 320 NMJs from five mice) but this improvement was not statistically significant (P > 0.05) and did not reach the NMJ innervation rates of WT mice (97.3 ± 1.10%, P < 0.0001; n = 206 NMJs from 4 mice; Fig. 4D). However, the rate of the partially innervated NMJs was increased in fully treated mice (37.3 ± 1.21%) compared to the mock group (24.9 ± 2.62%; P = 0.0014; Fig. 4E) while completely denervated NMJs were more frequent in the mock group (53.2 ± 7.67%) compared to fully treated mice (28 ± 6.23%, P = 0.0299; Fig. 4F).
Improved neuromuscular junction (NMJ) morphology in LV-Mpz.GJB1 treated Gjb1-null mice. Representative images of immunostained quadriceps muscle sections of 10-month-old WT (A), as well as mock (B) or full vector treated (C) Gjb1-null mice for neurofilament marker SMI31 to label axons (red) and bungarotoxin that binds to post-synaptic acetylcholine receptors (green), shows partial or complete denervation of NMJs in the mock treated muscle, with improved innervation in the fully treated muscle. Cell nuclei are stained with DAPI (blue). Quantification of innervation of NMJs in different groups (D-F) showed nearly normal innervation of NMJs in the WT muscle (n = 4 mice), while mock treated muscles (n = 5 mice) showed mostly partial or complete loss of innervation, and fully treated mice (n = 6 mice) showed fewer completely denervated and more partially innervated NMJs compared to the mock group (values in text). G-I: Morphometric analysis of NMJs by measuring bungarotoxin immunofluorescence revealed an improvement in the area (G) and diameter (H) in the fully treated mice compared to mock group but no difference in NMJ perimeter (I). Values represent mean ± SEM. Data were compared using the Mann-Whitney U test. Scale bars: 30 μm.
Furthermore, measurement of area, diameter and perimeter of the NMJs in all groups showed that both area and diameter were increased in the fully treated compared with the mock group. Average NMJ area in the mock group was 385.6 ± 12.11 μm2 compared 416 ± 10.00 μm2 in the treated group (P = 0.0239; Fig. 4G). Likewise, we observed a small increase in the diameter of the NMJs (mock group: 20.3 ± 0.29 μm; treated group: 21.3 ± 0.22 μm; P = 0.0036; Fig. 4H). In contrast, the NMJ perimeter did not significantly differ between the two groups (119.7 ± 3.34 μm in mock and 110.6 ± 2.11 μm in treated group, P > 0.05; Fig. 4I). Overall, these findings indicate improved although not completely restored muscle innervation at 10 months of age following gene therapy after onset in Gjb1-null mice.
Reduced plasma neurofilament light concentration in treated mice
Recent studies have shown an elevation of plasma neurofilament light (NEFL) levels in patients with different CMT types including CMT1X compared with controls, as well as a correlation between elevated concentrations and disease severity, suggesting that this may be a useful biomarker for assessing disease progression and potential response to treatment in future clinical trials (23). In order to clarify whether the Gjb1-null model of CMT1X shows an elevation of NEFL levels, we first examined blood samples obtained from 10-month-old untreated Gjb1-null and WT littermates. This comparison confirmed a significant elevation of NEFL levels in Gjb1-null mice (n = 10) reaching 328.0 ± 24.94 pg/ml (range: 222 to 481.2) compared to the levels of WT mice (n = 8) reaching 111.5 ± 26.77 pg/ml (range: 34.8 to 266.4), a 2.9-fold elevation (Fig. 5A; P < 0.00001). Having confirmed that plasma NEFL is a biomarker in this neuropathy model, we then compared NEFL levels in 10-month-old Gjb1-null mice that were treated at the age of 6 months with mock vector-treated littermates. This analysis showed a significant reduction of NEFL levels in the full treatment group (n = 5) by 1.5-fold (268.9 ± 6.47 pg/ml; range, 251.5 to 281.6) compared with the mock treatment group (407.2 ± 21.47 pg/ml; n = 8; range: 334.7 to 536.4) (Fig. 5B; P = 0.0016). Thus, plasma NEFL concentration appears to be also a treatment responsive biomarker for CMT1X and supports the findings of improved motor function and muscle innervation following gene therapy.
Reduced plasma neurofilament light (NEFL) concentrations in LV-Mpz.GJB1 treated 10-month-old Gjb1-null mice. A. Comparison of baseline plasma NEFL levels in 10-month-old wild type (WT; n = 8) and untreated Gjb1-null mice (n = 10) confirms a significant elevation of this biomarker in the CMT1X model. B. Fully treated Gjb1-null mice (n = 5) showed an improvement in the plasma NEFL concentrations compared to mock-treated mice (n = 8) at the age of 10 months, 4 months after treatment, without reaching the levels of the WT mice (values in text). Values represent mean ± SEM. Data were compared using the Mann-Whitney U test.
Improved pathology in lumbar roots and peripheral nerves after delayed treatment in Gjb1-null mice
Morphological analysis was performed in transverse sections of anterior lumbar roots, mid-sciatic and femoral motor nerves of 10-month-old Gjb1-null mice injected either with the LV-Mpz-GJB1 or the mock vector. Multiple roots, as well as bilateral nerves when available, were examined, and results were averaged per mouse. The number of abnormally myelinated fibers, including demyelinated and remyelinated fibers, was counted, and their proportion to the total number of fibers was calculated (17–19). Foamy macrophages were also counted (18), and their numbers per 1000 myelinated fibers were compared (to account for variations in root and nerve size).
In the anterior lumbar roots of the fully treated mice the ratio of abnormally myelinated fibers was reduced compared to the mock group (Fig. 6; Supplementary Material, Table S2). The ratio of abnormal fibers was 0.156 ± 0.01 in the fully treated mice (n = 10), compared with 0.259 ± 0.03 in the mock-treated mice (n = 10; P = 0.0011, Mann–Whitney U test). Likewise, macrophage numbers were lower in the anterior roots of fully treated (6.42 ± 0.85/1000 fibers) compared with mock-treated mice (9.75 ± 0.64/1000 fibers; P = 0.0052).
Morphological analysis of anterior spinal roots of Gjb1-null mice following intrathecal delivery of the LV-Mpz.GJB1 vector. Representative images of semithin sections of anterior lumbar spinal roots attached to the spinal cord at low (A-B) and higher (C-D) magnification, as well as morphometric analysis results (E-F) from mock or full (GJB1) vector treated mice as indicated, at 10 months of age (4 months after treatment). LV-Mpz.GJB1 injected mouse roots (B, D) show improved myelination compared with roots of a mock-treated littermate (A, C) with fewer demyelinated (*) and remyelinated (r) fibers. Quantification of the ratios of abnormally myelinated fibers in multiple roots (n = 10 mice per group) confirms significant improvement in the numbers of abnormally myelinated fibers (E) as well as significant reduction in the numbers of foamy macrophages (F) in fully treated compared to mock vector treated littermates. Values represent mean ± SEM. Data were compared using the Mann-Whitney U test. Scale bars: in A-B:100 μm in C-D: 10 μm.
Likewise, in mid-sciatic nerves, the ratios of abnormally myelinated fibers were reduced in fully treated mice compared with the mock group (Fig. 7; Supplementary Material, Table S3). The ratio of abnormal sciatic nerve fibers was 0.064 ± 0.003 in fully treated (n = 10) compared with 0.086 ± 0.006 in mock-treated mice (n = 10; P = 0.0089, Mann–Whitney U test). Improvement was also evident in the reduction of foamy macrophages, the numbers of which reached 3.0 ± 0.25/1000 fibers in the fully treated mice compared with 4.5 ± 0.46/1000 fibers in the mock group (P = 0.0232).
Morphological analysis of Gjb1-null mice sciatic nerves following intrathecal delivery of the LV-Mpz.GJB1 vector. Representative images of semithin sections of sciatic nerves at low (A-B) and higher (C-D) magnification from mock or full (GJB1) vector treated mice as indicated, at 10 months of age (4 months after treatment). LV-Mpz.GJB1 injected mouse roots (B, D) show improved myelination compared with mock-treated littermates (A, C) with fewer demyelinated (*) and remyelinated (r) fibers. Quantification of the ratios of abnormally myelinated fibers in multiple nerves (n = 10 mice per group) confirms significant improvement in the numbers of abnormally myelinated fibers (E), as well as significant reduction in the numbers of foamy macrophages (F) in the fully treated compared with mock vector treated littermates. Values represent mean ± SEM. Data were compared using the Mann-Whitney U test. Scale bars: in A-B: 50 μm; in C-D: 10 μm.
Finally, improvement of myelination was observed also in femoral motor nerves (Fig. 8; Supplementary Material, Table S4). The ratio of abnormally myelinated fibers was 0.239 ± 0.02 in the fully treated group compared to 0.332 ± 0.02 in the mock group (P = 0.0052). However, the numbers of macrophages were not significantly reduced in the fully treated mice (4.3 ± 0.88/1000 fibers) compared with the mock group (6.1 ± 0.88/1000 fibers; P > 0.05). In addition, we used semithin sections of femoral motor nerves to determine axonal profiles of full and mock-treated mice and to measure total axons. Although there was a trend for bigger proportion of large diameter fibers (>10 μm) in the treatment group, this was not statistically significant as indicated by the axonal profile analysis (Supplementary Material, Fig. S1A). Total axon numbers did not significantly differ either between the two groups (mock group: 604.0 ± 9.51; n = 10 nerves; 5 mice; and fully treated mice: 642.0 ± 16.90; n = 12 nerves; 6 mice, P > 0.05) (Supplementary Material, Fig. S1B).
Morphological analysis of Gjb1-null mice femoral nerves following intrathecal delivery of the LV-Mpz.GJB1 vector. Representative images of semithin sections of femoral nerves at low (A-B) and higher (C-D) magnification, from mock and full (GJB1) vector treated mice as indicated, at 10 months of age (4 months after treatment). LV-Mpz.GJB1 injected mice (B, D) show improved myelination compared to mock-treated littermates (A, C) with fewer demyelinated (*) and remyelinated (r) fibers as well as foamy macrophages (arrow in C). Quantification of the ratios of abnormally myelinated fibers in multiple nerves (n = 10 mice per group) confirms significant improvement in the numbers of abnormally myelinated fibers in the fully treated compared with mock treated littermates (E). However, the reduction of the numbers of foamy macrophages did not reach statistical significance (F). Values represent mean ± SEM. Data were compared using the Mann-Whitney U test. Scale bars: in A-B: 50 μm; in C-D: 10 μm.
Discussion
In the current study, we used a lentiviral vector-mediated gene replacement therapy targeting Schwann cells to treat for the first time post-onset Gjb1-null mice, the mouse model for CMT1X. Intrathecal delivery of the viral vector carrying the GJB1 gene in 6-month-old Cx32-deficient mice resulted in widespread expression of the virally delivered Cx32 in peripheral nervous system tissues, and provided a significant therapeutic benefit in this model of demyelinating neuropathy.
In our previous studies using the Gjb1-null model of CMT1X, we have treated animals before the onset of the demyelinating neuropathy which occurs around 3 months of age (20–22) and may be preceded by axonal dysregulation as early as at 2 months of age (24). This pre-onset treatment resulted in widespread vector distribution in the PNS and GJB1 gene expression, with rescue of the demyelinating neuropathy (18, 19). In this study, although we treated 6-month-old mice, well after the onset of peripheral neuropathy, we found a similar widespread vector distribution and Cx32 expression rates as that observed in 2-month-old mice (18). This is an indication that gene replacement targeting myelinating Schwann cells can be efficient even after the onset of the neuropathy regardless of age, at least in the mouse model of CMT1X. Gene expression was also restricted to myelinating Schwann cells as in the 2-month-old animals. This is an important aspect of CMT1X gene therapy strategy, since in the PNS, Cx32 is specifically expressed in myelinating Schwann cells where it forms GJ channels through the non-compact myelin layers (25).
This late treatment study in the CMT1X model provided also further insights into the course of the disease. Behavioral analysis showed that mock-treated Gjb1-null mice remained rather stable over time after 6 months of age with no further progression, while fully treated mice showed significant improvement initially, that was maintained until the end of the study period. This indicates that in Gjb1-null mice functional deficits occur before 6 months of age and remain stable over time, perhaps as a result of an equilibrium between de-myelination and re-myelination and associated axonal degeneration. This may be a limitation of this neuropathy model as it does not fully reproduce the steady progression of the disease in patients with CMT1X, characterized by progressive neuropathy that worsens over time (1, 3, 26) mainly due to accumulating axonal loss (27, 28). Nevertheless, post-onset response in the CMT1X mouse model with functional improvement in treated mice indicates that expression of Cx32 can reverse some of the pathological changes already present, leading to an improvement of motor performance within the first 2 months after gene delivery that is maintained up to 10 months of age. However, in particular, the rotarod test showed marked variations in the results even within groups, which has also been observed in WT and in untreated Gjb1-null mice (data not shown). Thus, it may be a less useful test compared with the food grip analysis for the purpose of measuring therapeutic response in this model.
Improved motor performance in late-treated Gjb1-null mice correlated with improvement in electrophysiological properties. MNCVs were increased in LV-Mpz.GJB1 compared with LV-Mpz.Egfp (mock)-treated mice at 10 months of age, similar to our observations in Gjb1-null mice treated prior to the onset of the neuropathy (18). These results are also in accordance with the findings of Duque et al. (2016) (29), showing similar improvement of the CMAP amplitude in pre- and post-symptomatic animals of spinal muscular atrophy (SMA) treated with the adeno-associated vector (AAV)9 viral vector. We found that mock-treated 10-month-old Cx32-deficient mice had lower MNCVs than previously studied untreated Gjb1-null mice at about the same age (20, 30, 31). This may be a side effect of the injection itself. Although we also found a trend towards improvement of the CMAP amplitude in treated mice, this did not reach statistical significance. This may be related to the fact that decreased CMAP amplitudes in initially characterized Gjb1-null mice were only detectable at the age of 6 months and actually reached almost WT levels by 1 year of age (20), perhaps secondary to regenerating units during the course of the disease. Finally, quadriceps muscle contractility was significantly improved following gene therapy in this model with predominantly motor neuropathy (20, 22).
NMJ evaluation was in accordance with the results of electrophysiological analysis, since morphological analysis of NMJ innervation in the quadriceps muscle, which was also used in electrophysiological analysis, showed that fully treated mice presented higher rates of NMJ innervation although the innervation was mainly partial in contrast to the mock group where most of the NMJs were denervated. Our results of NMJ analysis at the age of 10 months are similar to Groh et al. (2010) (30) showing low percentages of denervation in Gjb1-null mice and almost no denervation in WT mice. A more detailed analysis performed in our study demonstrated also the percentage of fully and partially innervated NMJs in this model. Comparison of the Gjb1-null mouse NMJ results and a recently published analysis of NMJ pathology in the CMT4C mouse model, shows similarities in NMJ area, diameter and perimeter but the CMT4C model shows an increase in post-synaptic fragmentation (32). This aspect was not evaluated in our study, but may be indirectly reflected in the tendency for larger NMJ diameter in the mock compared to the fully treated Gjb1-null or the WT group.
Our morphological analysis showed reduction in the abnormally myelinated fibers in all PNS tissues examined, as well as in the numbers of foamy macrophages although the latter change did not reach statistical significance in femoral motor nerves. Although our results indicate morphological improvement and are similar to the animals treated before the onset of the neuropathy (18), the mean values of both abnormally myelinated fibers and foamy macrophages were increased by about 60% in both late-treated groups compared with early-treated groups in most PNS tissues examined, including the anterior lumbar roots and femoral motor nerves. Only the sciatic nerves showed similar ratios of abnormal fibers in both late and early-treated groups, likely reflecting the fact that neuropathy in this model affects predominantly the motor fibers (20, 33). Since the sciatic nerve is mixed, including both motor and sensory fibers, it shows minimal progression, making it less indicative of a therapeutic response. Thus, overall, there was an improvement of pathology in peripheral nerves and lumbar roots of late-treated Gjb1-null mice, although not to the degree we observed with pre-onset treatment and morphometric analysis at 8 months of age (18, 19). Similar observations have been reported in other disease models including the SMA gene therapy study in which both pre-symptomatic and post-symptomatic groups showed an improvement in the morphological properties, but post-symptomatic treatment was less effective compared to pre-symptomatic treatment (29). Early intervention was also more effective in diabetic neuropathy models compared with late intervention (34).
Overall, our functional and morphological analyses suggest a partial rescue of the demyelinating neuropathy. Behavioral tests, MNCV, percentage of innervated NMJs and morphological characteristics did not reach WT levels, indicating a partial improvement in all aspects. This may be explained by the relatively low (around 50%) expression rates of Cx32 in PNS tissues of treated mice. The use of alternative vectors, such as AAVs providing higher expression rates and levels could lead to a higher rate of improvement even reaching WT levels. Although there is no evidence of the minimal Cx32 amount needed for a significant therapeutic response, this is likely to be low.
Further to the demonstration of functional and morphological improvements following post-onset treatment, we also studied plasma NEFL concentration in our treatment groups as a clinically relevant biomarker that correlates with peripheral nerve degeneration and was found to be increased in patients with CMT1X and other CMT types (23). We demonstrate for the first time that plasma NEFL levels are elevated in 10-month-old Gjb1-null mice compared to WT controls, confirming that this is a useful biomarker in this model of CMT1X. Furthermore, we found a significant therapeutic response with decreased NEFL concentrations in treated compared with mock-treated mice, indicating that gene replacement therapy can ameliorate pathological changes affecting the axon itself reducing axonal degeneration, even with post-onset treatment. This biomarker becomes even more relevant for CMT1X because studies both in patient biopsies (27), as well as in the Gjb1-null model (24, 35) indicate a direct negative effect of Cx32 loss on axonal cytoskeleton and neurofilament phosphorylation and function, that likely drives axonal degeneration along with demyelination. Thus, plasma NEFL levels can serve both as a diagnostic as well as treatment-response biomarker in pre-clinical and clinical studies focusing on CMT1X therapies.
It is difficult to correlate precisely the timing of our intervention in the Gjb1-null mouse model with the stages and clinical course of the neuropathy in CMT1X patients. There is a well-described age-dependent loss of myelinated fibers in biopsied nerves of CMT1X patients (26, 36–38), leading to a progressive loss of motor units that correlates with clinical disability (28), while slowing of NCV may be present already in pre-symptomatic children and is less indicative of neurological deficits (39, 40). Thus, post-onset response in the mouse may not be entirely predictive of a late treatment response in patients, for example, after childhood and adolescence, when axonal loss progresses. On the other hand, the ongoing formation of axonal sprouts along with remyelination as a mechanism of regeneration have been recognized as major features both in CMT1X patient biopsies as well as in the mouse model, indicating that by restoring Cx32 expression in viable Schwann cells, one would expect a therapeutic benefit in patients even in adulthood.
Another limitation of our study is that biodistribution and expression rates of the viral vectors obtained in the mouse model may not be fully reproduced in larger animals or in humans, and this remains to be shown. Although lentiviral vectors have a well demonstrated tropism for myelinating Schwann cells (17, 18), including human (41), their direct in vivo application may be limited by their ability to integrate into the host genome, creating possible insertional mutagenesis risks. Alternative vectors, such as AAV vectors that remain episomal, may offer an advantage if serotypes with low immunogenicity and adequate tropism for human myelinating Schwann cells are identified.
In conclusion, this study provides evidence that late, post-onset gene replacement therapy targeting myelinating Schwann cells can partly reverse the combination of demyelinating neuropathy and axonal degeneration in the CMT1X mouse model resulting in significant functional and morphological improvements. Although late treatment provides a clear therapeutic benefit, our results indicate that treatment should be pursued as early in the disease course as possible, in order to achieve the highest possible therapeutic response.
Materials and Methods
Cloning and production of lentiviral vectors
Cloning of the lentiviral vectors and vector production and titration methods have been described in our previous studies (17, 18). Briefly, for the generation of Schwann cell-targeted lentiviral vectors the expression cassette was cloned into the lentivirus transfer vector pCCLsin.PPT.hPGK.GFP.pre by replacing the PGK promoter with the rat myelin protein zero (Mpz) promoter, which drives strongly gene expression in Schwann cells (42), and by downstream insertion of the human GJB1 open reading frame (ORF), along with the IRES.EGFP. A mock vector was also generated by cloning the Mpz-IRES.EGFP expression cassette without the GJB1 ORF into the same lentiviral vector. Correct assembly of the expression cassettes was confirmed by restriction digest mapping and by sequencing the ORFs.
Recombinant lentiviruses were produced by transient co-transfection of HEK 293 T cells with the transfer vector and the helper plasmids (CMVΔR8.74 and pMD2-VSVG for pseudotyping) using the calcium phosphate co-precipitation method. A total of 5 × 106 293 T cells were seeded in 10 m plates 24 h prior to transfection in Iscove-modified Dulbecco culture medium with 10% fetal bovine serum, penicillin (100 IU/ml), and streptomycin (100 mg/ml) in a 5% CO2 incubator. One hour prior to transfection, the culture medium was changed. A total of 64 μg of plasmid DNA was used for the transfection per dish: 16 μg of the envelope plasmid pMD2-VSVG, 16 μg of the packaging plasmid CMVΔR8.74 and 32 μg of the transfer vector plasmid. The precipitate was formed by adding the plasmids to a final volume of 540 and 60 μL of 2.5 M CaCl2 and then adding drop wise 600 mL of 2× HEPES-buffered saline. The precipitate was added immediately to the cultures. The medium was then replaced after 6 h with fresh medium containing 1 mM Na butyrate. The conditioned medium was collected 60 h after transfection, cleared by low-speed centrifugation and filtered through 0.22 μm pore size filters. The media was concentrated down from 6 mL to 250 μL volume by using the Lenti-X columns (CloneTech). The lentivirus titer was calculated using HIV-1 Gag p24 ELISA and one-step RT-qPCR for EGFP.
Experimental Animals
All intrathecal gene delivery experiments were conducted using 6-month-old mice Gjb1-null/Cx32 KO (C57BL/6_129) mice weighing 20–25 g, obtained from the European Mouse Mutant Archive, originally generated by Prof. Klaus Willecke (University of Bonn). In these mice, the neor gene was inserted in frame into the exon 2 of Gjb1 gene which contains the ORF (31). Both male and female mice were used in our experiments and showed no (sex-related) differences in their behavioral performance or nerve pathology. 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).
Intrathecal Vector Delivery
We delivered the lentiviral vector as previously described (18, 43). Briefly, following a small skin incision along the lower lumbar spine level to visualize the spine, the lentiviral vector was delivered into the L5 to L6 intervertebral space of anesthetized mice. A 50 μL Hamilton syringe (Hamilton, Giarmata, Romania) connected to a 26 gauge needle was used to inject 30 μL of lentiviral stock containing an estimated 4.0 × 1011 to 5.3 × 1012 (full vector) and 7.5 × 1011 to 9.9 × 1012 (mock vector) viral particles/mL. A flick of the tail was considered indicative of successful intrathecal administration.
Immunofluorescence staining
For immunostaining, mice were anesthetized with avertin according to institutionally approved protocols, and then transcardially perfused with normal saline followed by fresh 4% paraformaldehyde in 0.1 M PB buffer. The lumbar-sacral spinal cords with spinal roots attached, as well as the bilateral sciatic and femoral motor nerves were dissected. All tissues were frozen for cryosections, while sciatic and femoral nerves were isolated and teased into fibers under a stereoscope. Teased fibers or sections were permeabilized in cold acetone and incubated at RT with a blocking solution of 5% BSA (Sigma-Aldrich, Munich, Germany) containing 0.5% Triton-X (Sigma-Aldrich) for 1 h. Primary antibodies used were: mouse monoclonal antibody against contactin-associated protein (Caspr, 1:50; gift of Dr Elior Peles, Weizmann Institute of Science), rabbit antisera against EGFP (1:1000; Invitrogen, USA), Capr2 (1:200, Alomone Labs, Israel) and Cx32 (1:50; Sigma-Aldrich) all diluted in blocking solution and incubated overnight at 4°C. Slides were then washed in PBS and incubated with fluorescein- and rhodamine-conjugated mouse and rabbit cross-affinity purified secondary antibodies (1:500; Jackson ImmunoResearch, USA) for 1 h at RT. Cell nuclei were visualized with DAPI (1 μg/ml; Sigma-Aldrich). Slides were mounted with fluorescent mounting medium and images photographed under a fluorescence microscope with a digital camera using Axiovision software (Carl Zeiss MicroImaging; Oberkochen, Germany).
Gene therapy study design
The aim of this study was to examine whether a gene addition method can prevent the development of peripheral neuropathy in the CMT1X mouse model at a later stage after the onset of the neuropathy. The gene therapy trial was conducted using two groups of Gjb1-null mice. A minimum of 8 to 12 mice per treatment group for each outcome measure was considered adequate for assessing statistically significant differences based on our previous studies using similar models (17, 18). Animals were treated at the age of 6 months, after the onset of the pathology. Littermate mice were randomized to receive either LV-Mpz.GJB1 (full) treatment or LV-Mpz.Egfp (mock treatment, as a control group) and were assigned a coding number for further identification. Mice were evaluated by behavioral testing before treatment, and again at the ages of 8 and 10 months, by an examiner blinded to the treatment condition, and used at the age of 10 months for electrophysiology or for quantitative morphometric analysis of semithin sections. Analysis of physiological and morphological results was also performed blinded to the treatment condition. WT mice of the same age were also evaluated by nerve conduction velocities and NMJ morphological analysis in order to compared with therapeutic benefit in treatment groups.
Behavioral analysis
Rotarod testing: Motor balance and coordination was determined as described previously (44) using an accelerating rotarod apparatus (Ugo Basile, Varese, Italy). Training of animals consisted of three trials per day with 15 min rest period between trials, for three consecutive days. The mice were placed on the rod and the speed was gradually increased from 4 to 40 rotations per minute (rpm). Testing was performed on the fourth day using two different speeds, 20 and 32 rpm. Latency to fall was calculated for each speed. The test lasted until the mouse fell from the rod or after the mouse remained on the rod for 600 s and was then removed. Each mouse was placed on the rotarod three times at each speed used, and three different values were obtained for each speed. Mean values were used for each mouse at the two different speeds.
Grip strength testing: To measure grip strength, mice were held by the tail and lowered towards the apparatus (Ugo Basile, Varese, Italy) until they grabbed the grid with the hind paws. Mice were gently pulled back until they released the grid. Measurements of the force in g were indicated on the equipment. Each session consisted of three consecutive trials and measurements were averaged. Hindlimb force was compared between LV.Mpz-GJB1- and LV.Mpz-Egfp-treated mice.
Electrophysiological analysis
Motor nerve conduction velocity: MNCV was measured in vivo using published methods (45) from bilateral sciatic nerves following stimulation in anesthetized animals at the sciatic notch and distally at the ankle via bipolar electrodes with supramaximal square-wave pulses (5 V) of 0.05 ms. The latencies of the compound muscle action potentials (CMAP) were recorded by a bipolar electrode inserted between digits 2 and 3 of the hind paw and measured from the stimulus artifact to the onset of the negative M-wave deflection. MNCV was calculated by dividing the distance between the stimulating and recording electrodes by the result of subtracting distal from proximal latency.
Quadriceps muscle contractility study: In order to assess in situ the function of the lumbar root and femoral motor axons, we measured the contraction properties of the quadriceps muscle innervated by the femoral nerve in an anesthetized mouse as recently described (18). After exposure of the motor part of the femoral nerve, a stimulating hook electrode was used to stimulate the motor branch of the femoral nerve at 1 Hz using a constant current stimulator (DS3; Digitimer) with 5–6 mA and 200 μs duration pulse. The muscle contraction of the partially exposed quadriceps muscle was recorded with a force displacement transducer (FT03; Grass Technologies), which was attached to the muscle with a silk suture. The transducer was connected to a micromanipulator, and for the experiment, the muscle was extended 1 mm each time until the muscle contraction reached the maximum value. The average amplitude and duration of the force generated by the quadriceps muscle contraction were compared between treatment groups.
Measurement of NEFL concentration
Blood samples were obtained from 10-month-old untreated Cx32KO, WT, fully treated and mock-treated mice prior to sacrificing using standard methods (46). Blood samples were taken and processed within 1 h. Blood was collected into EDTA-containing tubes and centrifuged at 20°C at 3500 rpm for 10 min. Plasma was aliquoted and stored at −80°C until testing. Plasma NEFL concentration was determined at UCL using a commercially available NF-Light kit on a Single molecule array (Simoa) HD-1 instrument (Quanterix, Lexington, MA) (23, 47).
Assessment of NMJ denervation
To further examine the results of gene therapy on distal nerve endings and muscle innervation, we evaluated NMJs of quadriceps muscles of WT, mock and fully treated mice. Muscles were isolated from mice that were subjected to electrophysiological analysis and maintained in 4% PFA overnight. PFA was removed and muscles were washed three times with PBS and then maintained in 30% sucrose overnight. Muscles were frozen and cut into 25 μm sections. Sections were permeabilized in cold acetone and incubated at RT with a blocking solution of 5% BSA (Sigma-Aldrich) containing 0.5% Triton-X (Sigma-Aldrich) for 1 h and incubated overnight at 4°C with a mouse monoclonal antibody against neurofilament (SMI31; 1:1000; Abcam, Cambridge, MA) diluted in blocking solution. Slides were then washed in PBS and incubated with fluorescein-conjugated mouse cross-affinity purified secondary antibodies (1:500; Jackson ImmunoResearch, USA) along with bungarotoxin that binds to acetylcholine receptors on the postsynaptic membrane (1:100, Molecular Probes, USA) for 1 h at RT. Cell nuclei were visualized with DAPI (1 μg/ml; Sigma-Aldrich) and slides were mounted with fluorescent mounting medium. Z-stack images were obtained using a fluorescence microscope (Nikon Eclipse Nί; Tokyo, Japan) with digital camera (DS-Qi2) using NIS-Elements software. Area, diameter and perimeter of the NMJs were measured using the Image Pro Plus software (version 6.0; Media Cybernetics) using a custom made macro. In addition, the ratios of NMJs that were devoid of innervation or partly innervated (missing completely or partly SMI31-immunoreactive axons entering the NMJ area) were measured in at least 50 NMJs per individual mouse.
Morphometric analysis of myelination in lumbar roots and peripheral nerves
Mice were transcardially perfused with 2.5% glutaraldehyde in 0.1 M 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, dehydrated, and embedded in araldite resin (all purchased from Agar Scientific, Essex, UK). Transverse semithin 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 (Sigma-Aldrich). Sections were visualized with 10×, 20×, and 40× objective lenses and captured with a Nikon DS-L3 camera (Nikon Eclipse-Nί; Tokyo, Japan). Images of whole root or transverse nerve sections were obtained at 100 to 200× final magnification, and a series of partially overlapping fields covering the cross-sectional area of the roots or the nerves were captured at 400× final magnification. These images were used to examine the degree of abnormal myelination in both groups as described previously (17, 21, 24). In brief, all demyelinated, remyelinated 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 and/or axons surrounded by ‘onion bulbs’ (i.e. circumferentially arranged Schwann cell processes and extracellular matrix) were considered remyelinated, and other myelinated axons were considered normally myelinated.
In addition, we counted the number of foamy macrophages present within the entire cross section of each root or nerve, as an indication of inflammation. Macrophages were identified in semithin sections at 400× magnification as cells laden with myelin debris, devoid of a basement membrane, and extending small, microvilli-like processes, as described previously (30, 48). The macrophage count was calculated as the ratio per 1000 myelinated fibers, to account for size differences between different spinal roots and nerves. All pathological analyses were performed blinded to the treatment condition of each mouse. Finally, using semithin sections of femoral motor nerves from both full and mock treatment groups, total axons and the axonal diameter for axon profiling analysis were measured using the Image Pro Plus software (version 6.0; Media Cybernetics) using a custom made macro.
Statistical analysis
The percentages of Cx32-expressing paranodal myelin areas in immunostained spinal roots and sciatic nerves of different mutant lines injected with the full vector were compared with Student's t-test. Behavioral testing results as well as morphological analysis data including the proportion of abnormally myelinated fibers and the number of macrophages in lumbar spinal roots and peripheral nerves obtained from mock-treated and fully treated mice during the treatment trial were compared using the unpaired two-sided Student's t-test as well as the Mann-Whitney U test (significance level for all comparisons, P < 0.05) using GraphPad Instat3 software (GraphPad, USA).
Funding
Muscular Dystrophy Association and Charcot-Marie-Tooth Association (480030 and 603003 to K.A.K.); Republic of Cyprus through the Research Promotion Foundation (CULTURE/BR-NE/0416/07). H.Z. is a Wallenberg Academy Fellow, and the fluid biomarker measurements in the laboratory of H.Z. and A.J.H. were supported by the UK Dementia Research Institute at UCL. Wellcome Trust Postdoctoral Fellowship for Clinicians (110043/Z/15/Z to A.M.R.).
Author contributions
A.K. conducted the experiments, acquired data, analyzed data and wrote the manuscript. J.R., C.T. and C.C. produced the viral vector. C.K. conducted behavioral tests. I.S. conducted cloning and mice PCR screening. A.H., A.R., M.R. and H.Z. performed analysis of plasma neurofilament light levels. K.A.K. designed research studies, performed morphometric analysis, analyzed data and wrote the manuscript. All authors critically reviewed and approved the final manuscript.
Conflict of interest
None declared.
ACKNOWLEDGEMENTS
We thank Dr Elior Peles (Weizmann Institute of Science) for the Caspr antibody and Dr Kyriacos Kyriacou for advice on processing semithin sections. We also thank Martha Foiani for help with measuring NEFL samples.
