Introduction
Neuromyelitis optica (NMO) is a neuroinflammatory demyelinating disease that affects spinal cord and optic nerve, and to a lesser extent brain. Most NMO patients are seropositive for IgG1-class autoantibodies against astrocyte water channel aquaporin-4 (AQP4). It is thought that the anti-AQP4 autoantibodies (called AQP4-IgG) are pathogenic in NMO by a mechanism involving complement- and cell-dependent astrocyte damage and an inflammatory response, which leads to oligodendrocyte injury, demyelination and neurological deficit [
6,
29,
36]. Current NMO therapeutics include immunosuppressants, plasma exchange and B-cell depletion therapy [
42]. Alternative targets under consideration for NMO therapy include the AQP4-IgG antibody and its binding to AQP4 [
40], complement and complement inhibitory proteins [
32,
33] and various immune cells including plasma cells and granulocytes [
7,
11,
16,
28].
Here, we evaluated remyelination as a potential therapeutic approach in NMO, with the goal of reducing axonal degeneration and neuronal loss following demyelination associated with disease exacerbations, which could reduce cumulative neurological deficit. Though the subject of remyelination therapeutics is under active investigation for multiple sclerosis [
12,
17,
26], remyelination has received little attention in NMO, perhaps in part because of theoretical challenges in effecting remyelination in NMO, because:
(i) primary astrocyte damage in NMO could interfere with oligodendrocyte-astrocyte interactions that might be important in oligodendrocyte functions [
5,
19,
45];
(ii) blood–brain barrier disruption in NMO could inhibit oligodendrocyte migration along microvessels [
41]; and
(iii) the inflammatory environment in active NMO lesions could inhibit remyelination and produce irreversible axonal injury. However, limited analysis of early pathology in NMO suggests similar axonal preservation in NMO and multiple sclerosis [
3], which would support the evaluation of remyelinating therapeutics in NMO.
Notwithstanding these challenges, here we investigated the potential efficacy of small molecule remyelinating compounds in NMO. For in vivo studies we modified an established, passive-transfer mouse model of NMO in which intracerebral administration of AQP4-IgG and human complement by stereotaxic infusion produces characteristic NMO pathology with loss of AQP4 and GFAP, complement deposition, inflammation and demyelination, but with minimal axonal damage. We evaluated 14 potential remyelination drugs, as listed in Table
1, based on our review of the literature and selection of those drugs that have a mechanism consistent with use in NMO and for which the data are most clear-cut. All compound have been identified in in vitro drug screens of oligodendrocyte precursor cell (OPC) maturation or function. One compound, the approved drug clobetasol, promoted oligodendrocyte maturation in the primary OPC cultures, and remyelination in AQP4-IgG treated cerebellar slice cultures and mice, providing proof of concept for remyelinating therapy in NMO.
Table 1
Compounds with reported remyelinating activity
Benztropine | EAE, cuprizone mouse model | Muscarinic agonist | |
CDP-choline | EAE, cuprizone mouse model | Protein kinase C-mediated OPC proliferation | |
Clemastine | Lysolecithin mouse model | Antihistamine, anticholinergic | |
Clobetasol | Lysolecithin model, EAE | Glucocorticoid receptor signaling, Hedgehog signaling, OPC differentiation | |
Enprofylline | Kainic acid-induced spinal cord injury ex vivo | Adenosine receptor antagonist | |
Fasudil | OPC culture | Rho-kinase inhibitor, vasodilator | |
GC-1 | OPC culture/P7 mouse model of myelination | Thyroid receptor agonist, OPC differentiation | |
Indazole | EAE | Estrogen receptor beta agonist | |
Miconazole | Lysolecithin model, EAE | ERK1/2 activator, OPC differentiation | |
Olesoxime | Lysolecithin mouse model, cuprizone mouse model | Mitochondrial pore modulator | |
Quercetin | EAE | γ-secretase inhibition interfering with canonical Notch signaling | |
Quetiapine | EAE, cuprizone model, cerebral ischemia | Free radical scavenging, neurotrophic factor stimulation | |
Retinoic acid | Lysolecithin model, ethidium bromide model | Retinoid X receptor γ agonist | |
Y-27632 | Lysolecithin in cerebellar slice cultures | Rho-kinase inhibitor | |
Discussion
Remyelination has received considerable attention as a potential therapeutic strategy in multiple sclerosis, with several drugs in pre-clinical development or clinical trials, including anti-Lingo-1 antibody, antagonists of M1 and/or M3 muscarinic receptors (benztropine), antihistamine and anticholinergic drugs (clemastine), and mitochondrial pore modulators (TRO19622) [
8,
20,
50]. In addition, screens of approved and investigational drugs with oligodendrocyte precursor cells have yielded additional candidate drugs for potential repurposing in demyelinating disorders [
24], some of which are listed in Table
1. The study here was done to investigate the appropriateness of remyelination for NMO therapy. Because NMO pathogenesis involves primary astrocyte cytotoxicity and an inflammatory environment with blood–brain barrier disruption, it was not clear a priori whether drug-induced OPC maturation could promote remyelination in NMO. Our results support this possibility, as clobetasol produced significant remyelination in a mouse model of NMO produced by passive-transfer of AQP4-IgG. However, it is difficult to predict with confidence whether this conclusion from a mouse model will translate to human NMO, though recent findings showing similar early axonal loss in NMO and multiple sclerosis [
3] supports the possibility of a remyelination approach in NMO, at least as an early intervention during a disease exacerbation before gliosis and axonal injury occurs.
The in vivo mouse model of NMO chosen for the studies here involved a single, stereotaxic injection of a recombinant monoclonal AQP4-IgG and human complement into brain under conditions that produce robust, reproducible NMO pathology with loss of AQP4 and GFAP immunoreactivity, inflammation, complement deposition, and, most importantly, demyelination with minimal axonal loss. The single injection model used here, as adapted from the original description [
36] and follow-on applications [
30,
32,
35], was modified with regard to injection details and amounts of AQP4-IgG and complement to give demyelination with minimal axonal injury. A pulled glass pipette with 40-μm tip diameter minimized traumatic brain tissue injury, as did the minimal volume of injected fluid and the slow infusion rate and pipette withdrawal. We did not use continuous AQP4-IgG infusion models, as described [
49], for technical simplicity and to avoid potential pathology caused by chronic needle placement. We did not use rat models, as described [
1], because the pharmacology and efficacy of the drugs tested had been established in mice, as well as practical considerations in creating the model in a sufficient number of animals to yield statistically significant differences. It is acknowledged, however, that available animal models of NMO are imperfect, as NMO pathology requires invasive, passive-transfer of AQP4-IgG rather than spontaneous autoimmunity, and because of differences between rodents and humans in astrocyte:neuron ratios and their biology.
Our approach was to evaluate drug candidates that were reported to induce differentiation and/or proliferation of OPC cultures, and promote remyelination in cerebellar slices and mice following AQP4-IgG-induced demyelination. Following initial evaluation in pure OPC cultures and cerebellar slices, the most efficacious compounds were screened in the mouse model of NMO, from which clobetasol was selected for full analysis. Because the study here was done primarily for proof of concept we were not overly concerned with selection of the very best molecule for clinical development. Though we think it unlikely, it is possible that an NMO-specific remyelination drug screen might yield alternative, more efficacious compounds. It is not practical, however, to screen thousands of molecules for efficacy in a model of NMO demyelination.
Clobetasol was discovered as a potential remyelinating agents in a screen of 727 drugs in an in vitro phenotypic assay using mouse epiblast stem cell (EpiSC)-derived OPCs [
24]. Another screen of 1200 FDA-approved drugs in a mouse immortalized oligodendrocyte cell line identified clobetasol as one of the top ranking compounds in promoting myelin basic protein expression [
34]. Clobetasol is a FDA-approved topical corticosteroid used clinically for the treatment of various skin disorders including eczema, psoriasis, alopecia areata, vitiligo, lichen sclerosus and lichen planus. Unlike benztropine and clemastine, clobetasol does not inhibit muscarinic receptor subtypes (M1-M5), nor does it inhibit various kinase isoforms [
24]. As a corticosteroid, clobetasol modulates glucocorticoid receptor signaling in cytoplasm. Clobetasol has been shown to promote Schwann-cell-mediated myelination in the peripheral nervous system [
23] and functions as a smoothened (Smo) agonist [
43,
44] that activates Hedgehog signaling and stimulates OPC proliferation and differentiation [
10,
27,
34]. Clobetasol is currently approved only for topical administration in humans; however, recent pharmacokinetics data in rodents [
24] showed effective blood–brain barrier penetration following systemic administration. Several lines of evidence suggest that the remyelinating efficacy of clobetasol in our model does not involve a glucocorticoid anti-inflammatory effect, but rather an action on OPCs, as: (i) clobetasol induced OPC maturation in pure OPC cultures; (ii) clobetasol promoted remyelination in cerebellar slices; (iii) clobetasol was effective when given after primary astrocyte injury, inflammation and demyelination; and (iv) high-dose dexamethasone did not promote remyelination. Though dexamethasone has anti-inflammatory actions, it did not have effect here probably because the acute inflammation in our model is largely resolved by 4 days. Like clobetasol, dexamethasone has effects on Smo and Hedgehog signaling pathways [
25,
43,
46], though there is no literature on dexamethasone action on OPC maturation.
Though our results provide proof of concept for the potential utility of remyelinating drugs in NMO, several caveats are noted in extrapolating the data here to predicting the efficacy of remyelinating drugs in human NMO. As mentioned above, there are differences in the proportions of astrocytes vs. neurons in rodents and humans, and there may be differences in the biology of oligodendrocyte maturation and interaction with astrocytes and microvascular endothelia. There may be differences between mice and humans in drug pharmacokinetics and penetration into the central nervous system. With regard to our mouse model of NMO produced by passive, intracerebral transfer of AQP4-IgG, though it recapitulates the major pathological features of human NMO, it is far from the ideal, not yet realized model of NMO in which spontaneous AQP4 autoimmunity produces NMO pathology in spinal cord, optic nerve and brain. Finally, we note that remyelination requires intact axons and neuronal viability, which may be heterogeneous in timing and extent in human NMO.
Conclusion
In summary, our results provide evidence for remyelinating therapy in NMO, which might be most effective when administered early during a disease exacerbation. As remyelination requires intact axons, therapeutics promoting remyelination may be of limited benefit when administered late in the course of NMO. Because of their distinct mechanisms of action and targets, remyelination therapeutics may be efficacious in combination with therapeutics used currently in NMO, as well as therapeutics in the development pipeline. For example, AQP4-IgG targeted approaches, plasma exchange and immunosuppressants target upstream disease-initiating events, inflammation and demyelination, while remyelinating drugs would protect against neuronal injury and reduce the cumulative neurological deficit. Our data support clinical testing of remyelinating drugs in NMO.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
TS and XY carried out cell culture and slice studies. XY carried out mouse studies. XY and ASV analyzed data and wrote the manuscript. All authors read and approved the final manuscript.