Gene therapy, CMT1X, and the inherited neuropathies

In PNAS, Kagiava et al. (1) report their rescue of the X-linked form of Charcot–Marie–Tooth disease (CMT1X) in mice, through the use of a single intrathecal injection with a lentiviral vector. The vector expresses the gap junction beta 1 (GJB1) gene and its encoded protein connexin32 (Cx32), expressed from a myelin-specific promoter from the myelin protein zero (MPZ) gene. The mice are Cx32-null animals (2), and therefore lack Cx32. Injecting the vector into the subarachnoid space permitted the virus access to the endoneurium of multiple spinal nerves because the dura mater surrounding the spinal cord directly merges with epineurial connective tissue layers from the proximal end of peripheral nerves. Therefore, there is continuity between the subarachnoid space and endoneurial space. The MPZ promoter is only expressed by myelinating Schwann cells, providing tissue specificity to the study. Lentiviruses are incorporated into target cell genomes, where they can provide prolonged expression of the transduced gene. Taken together, these features certainly contribute to the encouraging results reported by Kagiava et al. (1). Specifically, the authors demonstrated robust expression of the vector in Schwann cells with improvement of the neuropathy at all levels in the mice and with little evidence of toxicity or inflammation. Using a similar vector expressing the reporter EGFP, they demonstrated transduction of over 50% of all Schwann cells in sciatic nerve and similar levels in other sites, including the femoral nerve, dorsal root ganglia (DRG), and ventral roots. Transduced Schwann cells were detected as far away as the trigeminal nerve. Moreover, expression rates increased significantly between 2 and 16 wk postinjection. Results with the GJB1 vector were just as impressive. Cx32 was correctly targeted to noncompact areas of myelin, such as paranodes and Schmidt–Lanterman incisures. Treated mice improved by behavioral and physiological criteria. Morphological analysis demonstrated improved myelination in roots and peripheral nerves. Prior studies have suggested a role for macrophages and inflammation in the pathogenesis of neuropathy in Cx32-null mice (3). Interestingly, numbers of foamy macrophages were also reduced in roots and nerves of treated animals. Moreover, no evidence of an immune response was detected. In summary, the results were a remarkable improvement over earlier attempts to repair demyelinating forms of Charcot–Marie–Tooth disease (CMT) by gene therapy.

CMT is caused by mutations in more than 80 known genes that are divided into several major groups: dominantly inherited demyelinating neuropathies (CMT1), dominantly inherited axonal neuropathies (CMT2), X-linked CMT (CMTX), and recessively inherited neuropathies (CMT4). CMT1X is the second most common form of CMT, causing ∼10% of all cases (4). No medical treatments currently slow progression or reverse any form of CMT despite extensive knowledge about pathogenic mechanisms. Attempts to repair various forms of CMT by gene therapy have been attempted as far back as 1993, when the cause of CMT1X was first identified (5). Some of these approaches in mice are cited in Fig. 1. Challenges have hindered the advancement of gene therapy in the peripheral nervous system, and many still remain. Some of these challenges include the following. (i) Schwann cells and neurons are postmitotic. Therefore, vectors that require proliferating cells will not introduce transgenes into these cells. (ii) Mutated Schwann cells continue to ensheath axons even when the myelin they form is abnormal. Therefore, mutant Schwann cells cannot simply be replaced; they need to be repaired. (iii) Cell bodies of motor and sensory neurons originate in different sites (spinal cord and DRG), and their axons extend to different targets (muscle or sensory endings). It has been a challenge to target both simultaneously. (iv) Targeting adequate numbers of Schwann cells or neurons has been difficult. (v) Transgenes expressed from viral vectors have had limited expression and/or introduced a toxic immune response. Attempts to overcome these problems in Schwann cells have typically involved injections directly into nerve, as recently attempted by the authors of the present paper (6), or into lesioned nerve (7, 8). To target neurons, investigators have also used i.m. or s.c. injections of vectors to take advantage of the secretory capabilities of muscle (9) or of retrograde transport pathways in motor and sensory neurons (10, 11), as well as intrathecal injections, although typically not with cell-specific promoters (reviewed in ref. 12). An evolution of various viral vectors, including retroviruses, adenoviruses, adeno-associated viruses, herpes simplex viruses, and lentiviruses, has also occurred in attempts to increase transduction and specificity and to reduce toxicity (reviewed in ref. 12).

Fig. 1.

Gene therapy approaches in mouse models of CMT. Investigators have used various approaches to introduce gene therapy constructs into mouse models of CMT. These approaches include (1) injecting vectors (19) or allowing axons to regenerate into transplanted Schwann cells (8) in lesioned nerve, (2) injecting vectors directly into nerve (6), (3) injecting vectors into muscle for secretion (9, 12) or to facilitate retrograde transport of the gene product (10), and (4) injecting vectors intrathecally as in the present study (1).

It is not clear how well the approach of Kagiava et al. (1) for CMT1X can be extended to other forms of CMT. As the authors point out, autosomal recessively inherited forms of CMT are probably caused by loss of function of the missing protein, whereas dominant forms are thought to cause neuropathy by toxic gain of function or haploinsufficiency. Gene replacement, as is the strategy in the study by Kagiava et al. (1), appears attractive for recessive disorders because the missing protein could simply be replaced. However, this approach would not be the case for types of CMT caused by toxic gain of function. For example, some mutations in MPZ cause CMT1B because the mutant protein is misfolded, retained in the endoplasmic reticulum (ER), and activates a process known as the unfolded protein response (UPR) (13, 14). Simply providing additional MPZ by a lentiviral vector would not repair the UPR. Gene therapy approaches that allow for specific gene editing will likely be necessary to treat these disorders by eliminating the causal mutation. In this vein, it is particularly interesting that CMT1X is considered to be a “loss-of-function” disease and that the Cx32 KO mouse is considered to be an appropriate model for the disease. Hundreds of different GJB1 mutations have been reported, which disrupt the normal function of Cx32 in a variety of ways, including ER retention, trafficking to its normal location in noncompact myelin, or disrupting the transport of ions or small molecules. One might predict that these various mutations would cause a variety of clinical phenotypes in CMT1X, as occurs with forms of CMT like CMT1B or CMT1E (4). However, this prediction does not seem to be the case. We evaluated the clinical impairment of 73 men with CMT1X with 28 different GJB1 mutations and found that when we corrected for age, all had similar phenotypes to individuals completely lacking GJB1 and Cx32 (15). Why males with different mutations do not differ clinically from males with no Cx32 remains unknown. It will be interesting to determine whether models of CMT1X caused by missense mutations, for example, do as well with replacement therapy as the null mice in this study.

A final point concerning the approach of Kagiava et al. (1) is toxicity. The authors detected no evidence of an immune response, and macrophage numbers were decreased, not increased, following treatment with their lentiviral vector. However, there have been concerns in the literature with the use of these vectors that randomly integrate into the genome. Most notably, several patients treated for severe combined immunodeficiency (SCID) subsequently developed leukemia even after their immunodeficiency was corrected (16). Particular care to minimize such complications is especially important for disorders like CMT1X that do not shorten a patient’s lifetime. An ideal vector may be one that does not integrate into the genome but leads to permanent alteration of the targeted genetic defect. In this regard, the recent reports of treating mouse models of Duchenne muscular dystrophy with adeno-associated viral vectors, which are not incorporated into the genome, is exciting. These vectors used the clustered regularly interspaced short palindromic repeats (CRISPR)-Cas9 system specifically to target and eliminate mutant exons of dystrophin in the mice (17, 18). Perhaps similar approaches will prove feasible for Schwann cells and neurons in CMT. It is exciting to see the progress occurring at many levels with gene therapy, and the paper by Kagiava et al. (1) clearly demonstrates the capability of targeting large numbers of Schwann cells specifically in mutant mice, suggesting that gene therapy may become a realistic possibility for patients with these disorders.