A novel small molecule remyelination therapy for multiple sclerosis Authors: Yan Yang, Brain Bai, Lars J. S. Knutsen, Richard J. Hamlyn, Qiao-Ling Cui, Jack Antel, Wendy Macklin, Satish Medicetty, Bruce D. Trapp Published: 2026 ABSTRACT We conducted a phenotypic screen of 20,000 central nervous system (CNS)-biased small molecules for their ability to stimulate differentiation of mouse oligodendrocyte progenitor cells (OPCs) into oligodendrocytes. We identified a lead hit (CN045) with an EC50 of 40 nM in the OPC differentiation assay and a chemical scaffold conducive to modifications. In OPC differentiation assays, CN045 demonstrated higher potency than a known OPC differentiation compound Triiodothyronine (T3). CN045 promoted myelin-like ensheathment of engineered nanofibers by mouse and human OPCs and significantly increased remyelination in white and gray matter regions of mouse brain following cuprizone/rapamycin-induced demyelination. In terms of pharmacokinetics, CN045 is CNS-penetrable with low cytotoxicity. CN045 has a short half-life in vivo, but its chemical scaffold is conducive to structural modifications that can improve its metabolic properties. Collectively, these results demonstrate that CN045 is an attractive lead candidate for enhancing OPC differentiation and remyelination in multiple sclerosis patients. Introduction Multiple sclerosis (MS) is an inflammatory-mediated demyelinating disease of the central nervous system (CNS) and the major cause of atraumatic neurological disability in young adults. Most MS patients have a bi-phasic disease course that starts with episodes of reversible neurological disability termed relapsing-remitting MS (RRMS), which eventually transforms into a secondary-progressive (SPMS) disease course of continuous and irreversible neurological disability with few or no relapses1. Relapses are associated with immune-mediated demyelinating events that can be visualized by brain imaging studies using gadolinium enhancement. There are over 20 FDA-approved therapies that suppress inflammatory-mediated demyelination and reduce the incidence of relapses and demyelinated lesions2. However, current therapies do not stop the irreversible disability progression that eventually afflicts most MS patients. Due to the absence of myelin trophic support, demyelinated axons eventually degenerate and contribute to irreversible neurological disability in SPMS3,4. Loss of myelin is not necessarily permanent, as the adult human brain has the capacity to generate new oligodendrocytes that remyelinate demyelinated axons in MS lesions. By reducing degeneration of demyelinated axons, therapeutic enhancement of remyelination is the best-characterized neuroprotective approach for MS patients5,6. Remyelination requires the generation of new oligodendrocytes from oligodendrocyte progenitor cells (OPCs) that are normally distributed throughout the entire adult human brain in a tile-like pattern7,8. Variable amounts of OPCs are present in MS lesions and OPCs surrounding demyelinating lesions or new OPCs generated in the subventricular zone can migrate into demyelinated lesions9. While recent studies of rodents and humans indicate that mature oligodendrocytes can participate in remyelination10–12, mature oligodendrocytes are unlikely to be a major source of remyelinated internodes in MS. OPCs are the major cellular target for remyelination therapies in MS. Identification of novel remyelinating therapies is a multi-step process. Since a limiting step in remyelination is production of new oligodendrocytes, initial studies often focused on in vitro assays that monitor oligodendrocyte production from OPCs. The source and number of OPCs are both critically important. Since OPC cell lines are not available, most studies utilize primary OPCs from developing rodent brain. The phenotypic markers for oligodendrocyte maturation are well-described and include expression of myelin-specific proteins and dramatic changes in cell shape when compared to the small bipolar shape of OPCs. It is common to measure the affinity of newly produced oligodendrocytes to ensheathe axons or axon-like structures. Approaches include co-cultures with dissociated neurons13, slice cultures14, organoids15, and micropillars16 or nanofibers engineered to mimic axons17. These in vitro studies often provide initial “go/no-go” outcomes for individual therapeutics. An important question is whether promising therapeutics have similar effects on human OPCs. Sources of human OPCs include induced pluripotent stem (iPS) cells18 and OPCs obtained from neurosurgeries of individuals with epilepsy or CNS tumors8. The latter are relatively late lineage cells recognized by A2B5 and O4 antibodies and have been shown to be capable of myelinating axons when transplanted into Shiverer mice19. iPS cells are scalable, while OPCs isolated from adult human brain are limited. While studies of human OPCs are warranted, it appears that the vital mechanisms that govern oligodendrocyte production and remyelination are conserved between rodent and human7. The demonstration of enhanced remyelination in rodent models of demyelination/remyelination is critical for the development of remyelinating therapies. Several models have been developed and have been shown to be effective in testing therapeutic enhancement of remyelination in vivo, with each model having pros and cons. The most commonly used MS animal model, experimental autoimmune encephalomyelitis (EAE), has provided proof-of-principle for many anti-inflammatory MS drugs currently on the market20. However, CNS destruction in this autoimmune model includes both myelin and axons, and large areas of demyelinated axons do not occur in predictable locations. This has limited the use of EAE as an animal model to study CNS remyelination21. Focal injections of lysolecithin and ethidium bromide create small, demyelinated lesions and have been successfully used to test remyelinating therapies immediately following acute demyelination22. When added to mouse chow, the oligodendrocyte toxin cuprizone causes extensive demyelination of the corpus callosum, cerebral cortex, and hippocampus23, but has little effect on the optic nerve or spinal cord. Because cuprizone does not effectively destroy new oligodendrocytes, significant spontaneous remyelination occurs on demyelinated axons during extended cuprizone treatment24. This spontaneous remyelination is dramatically reduced by daily intraperitoneal (i.p.) injections of the mTOR inhibitor rapamycin, which inhibits production of new oligodendrocytes, during the cuprizone treatment stage24,25. Following 12 weeks of cuprizone/rapamycin (cup/rap) treatment, thyroid hormone triiodothyronine (T3) enhanced remyelination in the corpus callosum, cerebral cortex, and hippocampus25. Cuprizone provides a platform for investigating remyelination of chronically demyelinated lesions. Details of molecular, ultrastructural, physiological, and behavioral alterations following cup/rap hippocampal demyelination/remyelination have been described25. Cup/rap experiments are time-consuming (12 weeks of the demyelinating phase with cuprizone chow daily injections of rapamycin followed by 6 weeks of therapeutic treatment). In general, toxin models do not include primary immune-mediated demyelinating events like MS. Identification of remyelination therapies includes target-based screens, phenotypic screens of repurposed FDA-approved drugs, and small molecule libraries. Development of novel remyelination therapies initially focused on a target-based approach using antibodies and thyromimetics. LINGO-1 was identified as an axonal inhibitor of remyelination26. Anti-LINGO antibodies (opicinumab) were developed to mask this inhibitory effect and showed promising results in preclinical studies27. While a large Phase II clinical trial was unsuccessful28, it was informative on remyelination clinical trial design. T3 was found to enhance remyelination in rodent models but did not meet human safety criteria29. T3 is a common positive control in rodent studies. The remyelination therapy field received a major boost by screening libraries of FDA-approved drugs in OPC differentiation assays30. This “repurposing” approach identified multiple OPC differentiation targets including muscarinic receptors, estrogen receptors, thyroid hormone receptors, and NMDA receptors. Clemastine, an antihistamine with antimuscarinic properties, showed functional benefits in a small Phase II clinic trial31. Benzotropine, Miconazole, and Bazedoxifene showed promising results in preclinical and limited clinical studies32. The large number of hits in the repurposing screens raised the possibility that they may share off-target effects33. One significant off-target effect shared by 5% of FDA-approved repurposing drugs is inhibition of cholesterol synthesis34. Therapeutic development of additional hits identified in the repurposing screens is proceeding, as is the development of inhibitors of cholesterol synthesis35. We screened a CNS-biased small molecule library for the ability to stimulate differentiation of primary mouse OPCs into oligodendrocytes. From 20,000 small molecules, we identified a lead hit, CN045, which has a chemical scaffold that is conducive to structural modifications and shows a high potency in driving OPC differentiation into oligodendrocytes in vitro and promoting remyelination in vivo. CN045 is brain-penetrable with low cytotoxicity in vitro as well as in vivo and appears to be an attractive contender that can be improved by further lead optimization to generate a clinical candidate for enhancing remyelination in MS patients. Results Establishing a cell-based high-throughput phenotypic OPC differentiation assay In order to develop a reliable and reproducible high throughput phenotypic OPC differentiation assay for small molecule screening, we used primary OPCs generated from PLP-EGFP transgenic mice, which express EGFP driven by the myelin PLP promoter36. Primary mouse OPCs expressing the PLP-EGFP transgene were generated from neurospheres derived from E14.5 embryonic cortex. Neurospheres were maintained in vitro for 5 days, dissociated into single cells, and plated at a concentration of 6000 cells/well in 96-well plates. Ninety-five percent of these plated cells were positive for the OPC marker NG2 or PDGFα (Supplementary Fig. S1). Since T3 has been shown to promote OPC differentiation into oligodendrocytes in vitro37, we used it as a positive control for assay development. T3 significantly increased the number of differentiated PLP-EGFP+ oligodendrocytes (Fig. 1B) following 5 days of treatment compared to the DMSO-only negative control (Fig. 1A).Fig. 1Development of a phenotypic mouse OPC differentiation assay for high-throughput screening.Representative images of EGFP-positive oligodendrocytes following 5 days of treatment with DMSO (A) and T3 (B, 10 ng/ml). C Z-scores from four separate experiments compared T3 and DMSO. D Example of primary screening of the CNS-biased small molecule library tested at 10 µM. DMSO was used as a negative control (first lane, yellow bar) and T3 was used as a positive control (second lane, green bar). Small molecules with similar or greater differentiation activity as T3 were selected for secondary screening. E Example of secondary screening testing 16 small molecules (shown as numbers on X-axis) were tested at five half-log concentrations (10 µM to 0.1 µM with 10 µM to the left) along with DMSO (first lane) and T3 (second lane). F An example of a dose response EC50 curve of one of the top hits. G The top 18 small molecules with EC50 < 0.2 µM. RB Rotatable Bonds, cLogP Calculated Logarithm of Partition Coefficient, tPSA Topological Polar Surface Area, Hacc Hydrogen-Bond Acceptors, Hdon Hydrogen-Bond Donors. Error bars in (C) and (F) = standard deviation. ImageJ was used to generate Figs. (A) and (B). Graphs were generated using either Microsoft Excel or GraphPad Prism 11.0.0. Development of a phenotypic mouse OPC differentiation assay for high-throughput screening. Representative images of EGFP-positive oligodendrocytes following 5 days of treatment with DMSO (A) and T3 (B, 10 ng/ml). C Z-scores from four separate experiments compared T3 and DMSO. D Example of primary screening of the CNS-biased small molecule library tested at 10 µM. DMSO was used as a negative control (first lane, yellow bar) and T3 was used as a positive control (second lane, green bar). Small molecules with similar or greater differentiation activity as T3 were selected for secondary screening. E Example of secondary screening testing 16 small molecules (shown as numbers on X-axis) were tested at five half-log concentrations (10 µM to 0.1 µM with 10 µM to the left) along with DMSO (first lane) and T3 (second lane). F An example of a dose response EC50 curve of one of the top hits. G The top 18 small molecules with EC50 < 0.2 µM. RB Rotatable Bonds, cLogP Calculated Logarithm of Partition Coefficient, tPSA Topological Polar Surface Area, Hacc Hydrogen-Bond Acceptors, Hdon Hydrogen-Bond Donors. Error bars in (C) and (F) = standard deviation. ImageJ was used to generate Figs. (A) and (B). Graphs were generated using either Microsoft Excel or GraphPad Prism 11.0.0. We calculated the Z prime (Z’) factor, which is a widely-accepted35 assay quality indicator for high throughput screening and reflects the assay’s ability to distinguish a “hit” from noise. A Z’ factor score between 0.5 to 1 indicates suitable assay quality for high-throughput screening38. The Z’ factor score in 4 OPC differentiation assays consisting of 3–6 wells/experiment ranged from 0.58 to 0.69 (Z1 = 0.58; Z2 = 0.6; Z3 = 0.61; Z4 = 0.69), thereby meeting the criteria for a reliable and robust screening assay (Fig. 1C). The Z’ factor was used to develop and standardize the OPC differentiation assay. It was not used during the screening of the small molecule library. Phenotypic screening of 20,000 small molecules Primary screening was conducted using 20,000 CNS-biased small molecules (ChemBridge Library) that were selected based upon multiparameter optimization (MPO) using a weighted scoring approach39. The MPO approach considers 6 physicochemical properties (clogP, clogD, MW, TPSA, HBD, and pKa) for blood-brain barrier penetration, cytochrome P450-mediated metabolism, and inhibition of dofetilide binding. All small molecules were tested in triplicate at a single concentration (10 μM). The primary screening threshold for positive compounds was set at >50% of the efficacy of T3 in inducing OPC differentiation into EGFP+ oligodendrocytes with a “lily pad” appearance. A total of 390 small molecules met the inclusion threshold in primary screening. Figure 1D shows data from one of the 96-well plates where 1 compound had significantly higher efficacy than T3. The 390 hits from the primary screening were subjected to secondary screening with half-log dilutions (0.1 μM to 10 μM). A total of 57 small molecules met the criteria of >50% efficacy of T3 at concentration of 1 μM. Figure 1E shows an example of secondary screening, where multiple compounds show >50% efficacy compared to T3. To determine the potency of each compound, dose response curves were generated with 6–8 half-log dilutions and EC50 was calculated (Fig. 1F shows a representative curve). A total of 43 compounds had EC50s lower than 1 µM and 18 compounds had EC50s lower than 0.2 µM (Fig. 1G). Three small molecules with the lowest EC50s (0.030 µM, Fig. 2F) are analogs of benzylpiperazine, which has several limitations, including IP position, off-target effects, and toxicity (see Discussion for details).Fig. 2CN045 increases mouse OPC differentiation in vitro.A Chemical structure of CN045. B Dose response curve for CN045 (error bars = standard deviation). Representative images of PLP-EGFP+ (green) cells after 5 days of treatment with DMSO (C) T3 (D) or CN045 (E). F CN045 significantly increased OPC differentiation compared to DMSO and T3 (**P < 0.01, error bars = standard deviation). PLP-EGFP+ OLs (G & I, green) express the myelin proteins PLP (H, magenta) and MBP (J, magenta). ImageJ was used to generate all images. Graphs were generated using GraphPad Prism 11.0.0. CN045 increases mouse OPC differentiation in vitro. A Chemical structure of CN045. B Dose response curve for CN045 (error bars = standard deviation). Representative images of PLP-EGFP+ (green) cells after 5 days of treatment with DMSO (C) T3 (D) or CN045 (E). F CN045 significantly increased OPC differentiation compared to DMSO and T3 (**P < 0.01, error bars = standard deviation). PLP-EGFP+ OLs (G & I, green) express the myelin proteins PLP (H, magenta) and MBP (J, magenta). ImageJ was used to generate all images. Graphs were generated using GraphPad Prism 11.0.0. From the 20,000 small molecules initially tested in the OPC differentiation assay, CN045 was chosen as the lead hit for further development as a small molecule therapy for remyelination. The structure of CN045, which includes an attractive scaffold for optimization, is shown in Fig. 2A. CN045 had an EC50 of 0.040 µM (Fig. 2B). Compared to DMSO (Fig. 2C) and T3 (Fig. 2D), CN045 (Fig. 2E) significantly increased the generation of PLP-EGFP+ oligodendrocytes (Fig. 2F) in the OPC differentiation assay. CN045 also generated PLP-EGFP+ cells with a more complex “lily pad” appearance compared to the other treatments (insets in Fig. 2D, E). CN045 generated EGFP+ cells (Fig. 2G, I) expressed the myelin proteins PLP (Fig.2H, magenta) and MBP (Fig. 2J, magenta), indicating translation of endogenous myelin-specific genes in these EGFP+ oligodendrocytes. To address whether CN045-treated cells can mimic morphological changes of process extension and myelination, immunopurified OPCs from PLP-EGFP mice were plated on 24-well culture plates containing PLL-coated 3D-engineered nanofibers and treated for 14 days with DMSO, T3, or CN045. Compared to DMSO (Fig. 3A) and T3 (Fig. 3B), CN045 (Fig. 3C) generated more oligodendrocytes with a greater affinity to associate with nanofibers. DMSO-treated oligodendrocytes rarely associated with multiple nanofibers (Fig. 3A). The number of oligodendrocytes (Fig. 3E) and the total length of myelin-like nanofiber ensheathment (Fig. 3F) generated by CN045 and T3 were significantly increased compared to DMSO. In addition, CN045 treatment showed significantly higher numbers of oligodendrocytes and length of myelin-like nanofiber ensheathment compared to T3 (Fig. 3E, F). MBP was present in the myelin-like processes associated with nanofibers (Fig. 3D, red).Fig. 3CN045 promotes myelin-like ensheathment of engineered nanofibers.Representative images of PLP-EGFP expression after treatment with DMSO (A), T3 (B) or CN045 (C). D EGFP-positive oligodendrocytes (green) extend MBP-positive processes (red) that ensheathe nanofibers. Compared to DMSOand T3, CN045 treatment signifi cantly increased the total number of diff erentiated PLP-EGFP+ OLs/per well (E) and the total length of ensheathed nanofi bers per well (F). (*P < 0.05, paired Student’s t test). Experiments were performed in triplicate for each treatment. Error bars = standard deviation. ImageJ was used to generate all images. Graphs were generated using GraphPad Prism 11.0.0. CN045 promotes myelin-like ensheathment of engineered nanofibers. Representative images of PLP-EGFP expression after treatment with DMSO (A), T3 (B) or CN045 (C). D EGFP-positive oligodendrocytes (green) extend MBP-positive processes (red) that ensheathe nanofibers. Compared to DMSOand T3, CN045 treatment signifi cantly increased the total number of diff erentiated PLP-EGFP+ OLs/per well (E) and the total length of ensheathed nanofi bers per well (F). (*P < 0.05, paired Student’s t test). Experiments were performed in triplicate for each treatment. Error bars = standard deviation. ImageJ was used to generate all images. Graphs were generated using GraphPad Prism 11.0.0. CN045 promotes human OPC differentiation Since adult human OPCs are the target of remyelinating therapies, we tested the ability of CN045 to induce differentiation of OPCs purified from surgically removed human cerebral cortex40. Compared to DMSO treatment (Fig. 4A), 14 days of CN045 treatment significantly increased the percentage of O4+ cells, which extended single or bipolar processes that ensheathed nanofibers (Fig. 4B). CN045 treatment resulted in significant increases in the percentage of O4+ oligodendrocytes (Fig. 4C) and myelin-like ensheathment length (Fig. 4D) compared to DMSO treatment. CN045 therefore promotes human OPC differentiation and nanofiber ensheathment.Fig. 4CN045 increases human OPC differentiation and ensheathment of nanofibers.Human OPCs were isolated from 3 surgically resected brain specimens (ages 2, 13 and 35 years: 1 male, 2 females), cultured on nanofibers, and treated with DMSO (A) or CN045 (B) for 14 days. Compared to DMSO, CN045 increased the percentage of O4-positive oligodendrocytes (C) and the total length of MBP-positive nanofiber ensheathment (D). **P < 0.01, paired Student’s t test. Experiments were performed in triplicate. Error bars = standard deviation. ImageJ was used to generate all images. Graphs were generated using GraphPad Prism 11.0.0. CN045 increases human OPC differentiation and ensheathment of nanofibers. Human OPCs were isolated from 3 surgically resected brain specimens (ages 2, 13 and 35 years: 1 male, 2 females), cultured on nanofibers, and treated with DMSO (A) or CN045 (B) for 14 days. Compared to DMSO, CN045 increased the percentage of O4-positive oligodendrocytes (C) and the total length of MBP-positive nanofiber ensheathment (D). **P < 0.01, paired Student’s t test. Experiments were performed in triplicate. Error bars = standard deviation. ImageJ was used to generate all images. Graphs were generated using GraphPad Prism 11.0.0. CN045 enhances remyelination in the Cuprizone/Rapamycin demyelinating mouse model We previously demonstrated that 12 weeks of cuprizone/rapamycin (cup/rap)-mediated demyelination offers several advantages for studying therapeutic enhancement of remyelination in both white and gray matter25. The experimental design for testing CN045 in the cup/rap-induced demyelination/remyelination mouse model is shown in Fig. 5A. Following 12 weeks of cup/rap-induced demyelination, mice were treated by IP injections of vehicle or CN045 for an additional 6 weeks to assess remyelination. Compared to age-matched WT mice (Fig. 5B, F, J), the cup/rap regimen for 12 weeks demyelinated the corpus callosum (Fig. 5C), hippocampus (Fig. 5G), and cerebral cortex (Fig. 5K). After the cup/rap regimen, we compared the remyelination effects of 6 weeks of vehicle or CN045 treatment. Compared to vehicle treatment, CN045 significantly increased remyelination in the corpus callosum (Fig. 5D, E, N), hippocampus (Fig. 5H, I, O) and cortex (Fig. 5L, M, P). These results demonstrate that CN045 treatment significantly increased remyelination in both white and gray matter in a mouse model of cup/rap-induced demyelination and remyelination.Fig. 5CN045 promotes remyelination in vivo.A Experimental paradigm showing 12-weeks of cup/rap-induced demyelination followed by 6 weeks of CN045 or Vehicle treatment. Compared to the respective regions in the age-matched control group (B, F, J), the cup/rap regimen for 12 weeks induced significant demyelination in corpus callosum (C), hippocampus (G), and cortex (K). PPD+-stained myelinated axons in corpus callosum after Vehicle (D) or CN045 (E) treatment. PLP immunostaining in hippocampus after Vehicle (H) or CN045 (I) treatment and cortex after Vehicle (L) or CN045 (M) treatment. CN045 significantly increased remyelination in corpus callosum (N), hippocampus (O, truncated axis), and cortex (P, truncated axis). N = 10–12 animals/group. P* < 0.05; P** < 0.01, two-tailed Student’s t test. Error bars = standard deviation. ImageJ was used to generate all images. Graphs were generated using GraphPad Prism 11.0.0. CN045 promotes remyelination in vivo. A Experimental paradigm showing 12-weeks of cup/rap-induced demyelination followed by 6 weeks of CN045 or Vehicle treatment. Compared to the respective regions in the age-matched control group (B, F, J), the cup/rap regimen for 12 weeks induced significant demyelination in corpus callosum (C), hippocampus (G), and cortex (K). PPD+-stained myelinated axons in corpus callosum after Vehicle (D) or CN045 (E) treatment. PLP immunostaining in hippocampus after Vehicle (H) or CN045 (I) treatment and cortex after Vehicle (L) or CN045 (M) treatment. CN045 significantly increased remyelination in corpus callosum (N), hippocampus (O, truncated axis), and cortex (P, truncated axis). N = 10–12 animals/group. P* < 0.05; P** < 0.01, two-tailed Student’s t test. Error bars = standard deviation. ImageJ was used to generate all images. Graphs were generated using GraphPad Prism 11.0.0. CN045 emerging structure-activity relationship (SAR) Preliminary assessment of PK showed high intrinsic clearance of CN045 in mouse and human hepatocytes in vitro (Supplementary Fig. S2B), P450 inhibition (Supplementary Fig. S2C), and low half-life in vivo (Supplementary Fig. S3B), and therefore structural modifications are required to develop CN045 into a clinical candidate. We identified 20 CN045 analogs in the ChemBridge small molecule library using a structure-based similarity search and screened these analogs in the OPC differentiation assay, resulting in a range of positive activity at 0.1 µM to >10 µM (data are reported as fold increase compared to DMSO in Fig. 6). These results indicate an early emerging SAR, highlighting key areas of the CN045 scaffold that drive potency. The SAR can be divided into three areas: 1 - aryl groups, 2 - linker (alkyl ether) and 3 - aminoalkyl (Fig. 2A). The current data indicate that basic nitrogen is important for OPC differentiation potency, whereas both the linker, which tolerated various substitutions, and the aryl group, which tolerated variations in the chain length, provide an opportunity for optimizing the drug-like properties of the molecule.Fig. 6The structures of 20 commercial analogs of CN045 were tested in the OPC differentiation assay.Each analog was tested at 7 half-log concentrations (0.01, 0.03, 0.1, 0.3, 1, 3, and 10 µM) to determine the efficacy for promoting OPC differentiation and cell toxicity. The green highlighted columns identify compounds with significantly increased oligodendrocyte differentiation activity compared to DMSO. All analog structures were created using BIOVIA Draw V23. The structures of 20 commercial analogs of CN045 were tested in the OPC differentiation assay. Each analog was tested at 7 half-log concentrations (0.01, 0.03, 0.1, 0.3, 1, 3, and 10 µM) to determine the efficacy for promoting OPC differentiation and cell toxicity. The green highlighted columns identify compounds with significantly increased oligodendrocyte differentiation activity compared to DMSO. All analog structures were created using BIOVIA Draw V23. Discussion Degeneration of chronically demyelinated axons is considered as a major cause of permanent neurological disability in individuals with MS. While remyelination can spontaneously occur in MS brains, this process eventually fails. Development of novel therapeutics that stimulate OPC differentiation into oligodendrocytes is the most attractive approach for myelin repair in MS. Development of remyelinating therapies is a daunting task requiring multiple stages. Using a robust and highly-reproducible assay for differentiating mouse OPCs into oligodendrocytes, we screened 20,000 small molecules, identified 18 compounds that significantly increased OPC differentiation (EC50 < 200 nM), and chose CN045 as a lead compound. CN045 significantly induced the differentiation of mouse and human OPCs into oligodendrocytes, which longitudinally ensheathed nanofibers in vitro and enhanced remyelination of white and gray matter in a rodent model of demyelination/remyelination (cup/rap). These observations support the potential of CN045 as a remyelinating therapy for MS. Our approach was phenotypic screening, which has generated several candidates for remyelination over the last decade35. It is interesting to note that a review of first-in-class new molecular entities approved by the FDA between 1999 and 2008 showed that phenotypic screening has identified more (28 vs.17) small molecule drugs than target-based drug discovery41. Another review of first-in-class drugs approved by the FDA between 1999 and 2013 showed more drugs discovered through a target-based approach (45 vs. 33); however, it is important to note that a majority (25 out of 33) that were identified without a known target were found through a chemocentric approach, where the discovery process was initiated with known compounds42. Therefore, de-risking the process of phenotypic drug discovery by using a known and selective library of small molecules was proposed as an effective process to identify novel therapeutics41. The ChemBridge Library of 20,000 CNS-biased small molecules was chosen because it was designed to be enriched in brain-penetrable small molecules using Multiparameter Optimization (MPO) scoring39. The screening assay we utilized was designed specifically for high-throughput screening, as a single neurosphere experiment can produce enough OPCs for 10 or more 96-well plates. Analyses were computer-based and relied upon detection of EGFP+ oligodendrocytes that have a large “lily pad” appearance. The substantial increase in the size of these EGFP+ cells compared to the small bipolar OPCs indicates initiation of a myelinating program. The detection of PLP and MBP in EGFP+ cells and the myelin-like ensheathment of nanofibers by these cells support the induction of a myelinating genome in EGFP+ cells. There are no immortalized OPC cell lines and several phenotypic screens using primary rodent or human OPCs depend upon immunostaining for myelin proteins as a primary readout, which limits their use for high-throughput screening. We use OPCs generated from embryonic mouse brain. These OPCs contain an oligodendrocyte myelin-specific reporter, PLP, which drives EGFP. Differentiated oligodendrocytes can thus be detected without any immunostaining. The ability to generate large numbers of OPCs and to quantify OPC differentiation without immunostaining makes this a unique and effective phenotypic screening platform. We initially screened 20,000 small molecules at a concentration of 10μM. A positive hit was identified as >50% of the differentiation potency of T3, which was included in all plates. Small molecules were tested in triplicate and the oligodendrocyte differentiation efficacy was averaged. This initial screen included over 60,000 assays and resulted in 390 positive hits, which underwent a subsequent secondary screen of half-log dilutions (0.1 μM to 10 μM) performed in triplicate. The secondary screen identified 18 compounds that significantly increased OPC differentiation into oligodendrocytes with an EC50 < 200 nM. Among the 18 compounds identified, benzylpiperazine analogs and CN045 demonstrated the highest potency (EC50 of 30 nM and 40 nM, respectively) for OPC differentiation. Benzylpiperazine analogs have several limitations, including: 1) IP Position. The benzylpiperazine scaffold has been extensively optimized in the past. A simple substructure search for benzylpiperazines in SciFinder results in over 10,000 hits. CN045 has a clearer path for moving into patentable space; 2) Off-target activity. Benzylpiperazines are known to be active in the serotonergic and dopaminergic systems (benzylpiperazine itself is a designer drug). This can potentially result in stimulatory and hallucinatory effects (similar to as amphetamines), and given that our target is CNS-biased, this could provide additional issues to resolve when compared to CN04543; and 3) Toxicity: hERG blocker pharmacophores are characterized by a basic group flanked by two hydrophobic regions in which the benzylpiperazine scaffold fits quite well. If the position effects in the OPC assay are driven by the same chemical features, then this could be a real challenge to design against hERG with selectivity44. As a result, CN045 presents a better scaffold for optimization and was selected over benzylpiperazine analogs. Among the positive hit series, CN045 stands out because of the following features: high efficacy and potency, low toxicity, good solubility, and promising physiochemical properties. In addition, when compared to the positive control T3, CN045 showed significantly higher efficacy in the OPC differentiation assay in vitro. Furthermore, screening of 20 CN045 analogs obtained from the ChemBridge small molecule library revealed some initial SAR, highlighting key areas that drive OPC differentiation potency (Fig. 6). Therefore, we selected CN045 as our top hit for further investigation. As a potential lead, CN045 had a higher potency than T3 in promoting mouse and human OPC differentiation to oligodendrocytes in vitro. The number of processes per oligodendrocyte that demonstrated myelin-like ensheathment of nanofibers was also increased. These observations supported the utility of testing the remyelination potential of CN045 in vivo. There are several rodent remyelination models, including injection of the myelin toxin lysolecithin45 and the immune-mediated demyelinated model EAE20. However, we chose to use the cuprizone model of demyelination for several reasons. Cuprizone is a myelin toxin that is added to rodent chow, thus minimizing experimental manipulation of study participants. We have shown previously that 12 weeks of cuprizone diet and daily intraperitoneal injections of rapamycin (which reduces spontaneous remyelination) can significantly demyelinate mouse white and gray matter25. Most importantly, compared to six weeks of cuprizone treatment, remyelination following the 12-week cup/rap protocol can be enhanced by 6 weeks of T3 treatment25. Demyelination in the cup/rap model is relatively chronic compared to other models and is almost complete in the corpus callosum, hippocampus, and cerebral cortex, thus maximizing lesional sampling. CN045 significantly increased remyelination in the corpus callosum, hippocampus, and cerebral cortex, thereby demonstrating preclinical proof-of-concept for further development. The increase of CN045-induced remyelination was statistically significant but modest. This is consistently found in in vivo models of remyelination due to the significant rate of spontaneous remyelination in the vehicle group. The inclusion of gray matter in remyelinating studies is clinically relevant, as studies support the concept that cortical demyelination may exceed white matter demyelination in many MS patients46. In addition, spontaneous remyelination in MS brains appears to be greater in the cerebral cortex than in subcortical white matter47. Inclusion of cortex in MS remyelinating clinical trials will require more sensitive imaging readouts of both cortical demyelination and remyelination. CN045 is CNS-penetrable with low toxicity. However, there is room for further optimization of its pharmacokinetic properties, especially its relatively short half-life of 2.7 h (Supplementary Figs. S2, S3). Improvements to these factors will require structural modifications. This compound has a chemical scaffold that is conducive to modifications that can further optimize drug-like properties and SAR to generate novel analogs with better metabolic and therapeutic potential. In addition to the high metabolic turnover, the compound structure may pose additional liabilities, including CYP activity and hERG inhibition. Strategies to mitigate such risks are well-documented48. Assays such as CYP and hERG form part of the screening cascade such that any liability encountered can be addressed during the lead optimization phase. To support the optimization of the ADME (Absorption, Distribution, Metabolism and Excretion) properties in this series, the following data will be required as part of the screening cascade: liver microsomal stability, CYP activity, hERG inhibition, and Caco-2 permeability. We have described a small-molecule approach for identifying a novel remyelinating therapy. Our approach shows the benefits of a phenotypic screen, as the intellectual property path for CN045 appears to be open. A variety of molecular targets that promote the differentiation of OPCs into oligodendrocytes have been identified including Lingo-1, Muscarinic receptor 1, Histamine H3 receptor, and Estrogen receptors35. Our phenotypic OPC differentiation assay is agnostic to specific molecular targets. Therefore, CN045 may be promoting OPC differentiation through a previously-described or novel OPC target. We are currently pursuing the target of CN045 as results of obvious go/no-go studies have been encouraging. Additional lead optimization studies have been performed with novel analogues that have comparable OPC differentiating potency as CN045 but improved physiochemical and ADME properties. These data will be part of a future publication. Methods Animals Mice were housed in the Biomedical Research Unit (BRU) at Cleveland Clinic Research. All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the Cleveland Clinic. Six-week-old C57/BL/6J male mice were purchased from Jackson Laboratory (Stock# 000664, Bar Harbor, ME) and used for the cuprizone/rapamycin-induced demyelination and remyelination model. A colony of homozygous PLP-EGFP transgenic mice36 was maintained to generate embryos at 14.5 days of gestation for in vitro OPC differentiation assays. All animals used in this study were anesthetized by intraperitoneal injections of ketamine/xylazine (100 mg/kg ketamine + 10 mg/kg xylazine) and euthanized by perfusion with 4% paraformaldehyde (PFA, Electron Microscopy Sciences, Cat# 15714) through cardiac puncture. Human brain specimens Human OPC cells (A2B5 + O4 + ) were derived from surgical resections of cortical tissue obtained from 3 individual donors (ages 2, 13, and 35 years: 1 male, 2 females) with non-tumor-related intractable epilepsy, in accordance with the guidelines set by the Biomedical Ethics Unit of McGill University. Detailed methods have been described previously40. Primary mouse OPC preparations Primary mouse OPCs were obtained from neurospheres generated from 14.5-day embryonic brains (E14.5) from PLP-EGFP transgenic mice as described previously36. Briefly, cortices of E14.5 embryos were dissected, mechanically triturated, filtered through a 40μm cell strainer (Corning, Cat# REF 431750), and cultured in 25 cm2 culture flasks (Falcon, Cat# 353109) to form neurospheres. To produce dissociated OPC cell suspensions, neurospheres were mechanically dissociated into single cells by using a mouse NeuroCult dissociation kit (StemCell Technologies, Cat# 05707) according to the manufacturer’s instructions. High-throughput phenotypic OPC differentiation assays Dissociated single-cell suspensions were plated on poly-L-Ornithine (Sigma, P3655)-coated 96-well plates (Falcon, Cat# REF 353072) at a cell density of 6000 cells/well and placed in an incubator maintained at 37 °C and 5% CO2. Cells were maintained in OPC proliferation media containing DMEM/F12 (Corning, Cat# REF15-090-CV), B27 neural supplement with Vitamin A (Gibco, Cat# 17504044), platelet derived growth factor (PDGF, 10 ng/ml, Peprotech, Cat# 100-13 A) and basic fibroblast growth factor (bFGF, 10 ng/ml, Peprotech, Cat# 100-18B) for 2 days. Typical bipolar OPC morphology was observed 48 h after plating. After these 2 days of plating, OPCs were treated with small molecules at various concentrations for 5 days in a differentiation media containing DMEM/F12, B27 media without the growth factors PDGF and bFGF. Dimethyl Sulfoxide (0.1% DMSO, Sigma, Cat# D2650) and T3 (10 ng/ml, Sigma, Cat. #T2877) were included in each plate as negative and positive controls, respectively, to gauge assay performance and for post hoc normalization of inter-plate variability. After five days, cells were fixed with 4% PFA and Hoechst-stained for nuclei (Abcam, Cat# ab228551). The percentage of PLP-EGFP+ cells was determined. Subsets of cultures were immunostained for rat monoclonal antibody of anti-myelin proteolipid protein antibody (PLP, Millipore Sigma, MABN2620) and rat monoclonal antibody of myelin basic protein antibody (MBP, Millipore Sigma, MAB386). The Z prime (Z’ = 1 − 3X (δT3 + δC/µT3-µC) was calculated to assess reliability and reproducibility of cell-based screening on a per-plate basis38. PLP-EGFP+ cells were imaged in a Cellomics VTI ArrayScan reader (Artisan Technology Group) or Cytation 5 Cell Reader (Agilent BioTek). A total of 20–25 images/well were captured at 10× magnification. Computer algorithms were used to count the total number of PLP-EGFP+ oligodendrocytes, as well as the total cell nuclei and pyknotic nuclei. To address any variability in cell numbers, each condition was repeated in three or six replicates and only the 60 center wells of individual 96-well plates were seeded with cells. Screening of small molecules using OPC differentiation assays A CNS-biased library of 20,000 small molecules (ChemBridge Corporation, CA) was screened for the ability to induce OPC differentiation into EGFP+ oligodendrocytes. For primary screening, each small molecule was tested at 10μM in triplicate. A total of 20–25 images were captured/well and total PLP-EGFP+ cells, as well as total cell nuclei and pyknotic cells, were counted using an automated algorithm to calculate the average of cell numbers either from 3 wells (primary screening) or 6 wells (secondary screening). The percentage of PLP-EGFP+ oligodendrocytes was calculated and compared to DMSO and T3 (10 ng/ml). A small molecule was considered to be a positive hit if it showed at least 50% of the effect shown by T3. A total of 390 molecules met this criterion, which were subsequently used for secondary screening. For secondary screening and EC50 experiments, the small molecules were tested at half-log (0.01 μM to 10 μM) or full-log concentrations (0.0003 μM to 10 μM) in the OPC differentiation assay. Immunopurified mouse OPCs grown on engineered nanofibers Immunopurified A2B5+ OPCs were derived from the cortex of postnatal mouse pups (P2-P3) from PLP-EGFP mice49. Cell culture dishes were pre-coated with secondary antibody IgM (10 μg/ml, Millipore, Cat# 55460) and primary monoclonal antibody directed against the OPC marker A2B5 (Abcam, Cat# Ab53521). Dissociated cortical cells were incubated in the pre-coated culture dishes for 30 min. Immunopanned A2B5+ OPCs were collected in DMEM with a purity of ~80–90%. Cells were allowed to proliferate in media containing FGF and PDGF. Tissue culture plates (24-well) were layered with customized 3D-engineered 0.4μm diameter nanofibers made from poly L-lactic acid (Nanofiber Solutions, Dublin, Ohio). Purified OPCs were seeded to each well at a density of 30,000 and cultured at 37°C and 5% CO2 for three days in OPC proliferation medium. After 3 days, cells were treated with CN045 (1 μM), T3 (10 ng/ml), or DMSO in differentiation medium without PDGF, FGF, or EGF and allowed to differentiate for 14 days, with treatment medium changes every 48 h. Each condition was tested in triplicate with a minimum of 6 wells/plate. Cells were then fixed with 4% PFA, stained with MBP antibodies, and imaged in a Cellomics VTI ArrayScan reader. Twenty images were captured from each well. Automated analysis was performed using the Cellomics algorithm and ImageJ to quantify the total number of EGFP+ cells with myelin-like ensheathment and the total length of MBP+ ensheathed nanofibers per well. Adult OPCs generated from human brain Adult human brain tissue was obtained at the Montreal Neurological Institute using the human specimen guidelines of the Biomedical Ethics Unit of McGill University8. Tissue was digested with trypsin and DNase I, followed by cell separation using a 30% Percoll density gradient (Pharmacia LKB, Cat# 17-0891-01, Canada) as described previously40. Dissociated cells were cultured in DMEM/F12 medium containing N1 (Sigma, N6530, Canada), 0.01% bovine serum albumin (BSA dissolved in sterile 1XPBS, Millipore Sigma Cat# A0281), and 1% penicillin–streptomycin (Invitrogen, Cat# 15070063, Canada) overnight. The floating cells, which contain the oligodendrocyte/progenitor cell pool, were collected (the adherent cells were microglia) and incubated with A2B5 antibody-conjugated microbeads (Miltenyi Biotec, Cat# 130-097-864, CA). A2B5 antibody-selected cells were plated (at density of 10,000 cells per microwell) in PLL- and ECM-coated 96-well plates containing engineered nanofibers (Sigma, Cat# Z694541). In an incubator at 37 °C and 5% CO2, cells were cultured in DMEM/F12 + N1 medium supplemented with B27 (Gibco, Cat# 17504044) and T3 (2 ng/ml, Sigma, Cat# T2877), and treated with CN045 (1 μM) or DMSO (negative control) for 14 days. Media were changed every two days. After 2 weeks of treatment, live cells were stained with O4 antibody as a marker for oligodendrocyte cell bodies. Cells were then fixed with 4% PFA and stained for MBP-positive processes ensheathing nanofibers. Images were taken and analyzed for the percentage of O4+ cells and the number and length of MBP+ processes ensheathing nanofibers. The average number of oligodendrocytes and total length of MBP+ processes associated with nanofibers were calculated to determine the amount of myelin-like membrane produced. Immunocytochemical staining Cells were fixed with 4% PFA for 15 min at room temperature (RT) and washed three times with 1X PBS. After blocking with 2% BSA (BSA dissolved in 1xPBS) for 30 min, cells were then incubated with antibodies of MBP (Millipore Sigma, MAB386, anti-rat monoclonal antibody 1:100), PLP/DM20 (Sigma, Cat# MABN2620, clone AA3, 1:50), and O4 (Sigma-Aldrich, Cat# O7139, mouse monoclonal IgM, 1:200) overnight at 4 °C and incubated with one of the following secondary antibodies: anti-rat AlexaFluor 594 (Gibco, A-11007, 1:400); Cy3 Donkey Anti-Mouse IgM, Alexa Fluor® 647 Donkey Anti-Goat IgG, Alexa Fluor® 488 Donkey Anti-Rat IgG, (Jackson Immuno, Cat# 715-165-020; Cat# 705-605-147; Cat# 712-545-150). Hoechst 33342 staining (Abcam, ab228551) was used to visualize cell nuclei. Cuprizone/rapamycin demyelination and remyelination mouse model All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the Cleveland Clinic. The cuprizone/rapamycin demyelination/remyelination mouse model25 was used to test the in vivo remyelination potency of CN045. Eight-week-old C57/BL/6J male mice purchased from Jackson Laboratory (Stock#:000664, Bar Harbor, MR) were used for all experiments. To induce demyelination, mice were fed a pre-mixed chow containing 0.3% cuprizone (bis-cyclohexanone oxaldihydrazone, Sigma-Aldrich, CAS# 370-81-0) purchased from Harlan Teklad (Madison, WI) for 12 weeks ad libitum. During this period, all cuprizone-fed mice received daily i.p. injections of rapamycin (10 mg/kg body weight) reconstituted fresh in vehicle solution (5% PEG-1000, 5% Tween-80 and 5% ethanol). Cuprizone chow was changed every two days, and mouse weights were monitored weekly. At the end of the 12-week demyelination period, mice were either perfused to assess the extent of demyelination or placed into a group treated with CN045 (a soluble small molecule dissolved in 0.9% saline, 10 mg/kg body weight, daily i.p. injection) or vehicle (0.9% saline) that were maintained on normal chow minus rapamycin for 6 weeks to assess the degree of remyelination. A group of age-matched mice that was not exposed to the cup/rap regimen was used as a control to assess demyelination. At the end of the 6-week remyelination period, all mice were anesthetized with ketamine/xylazine and perfused with 4% PFA in 0.08 M phosphate buffer. N = 10–12 mice/per group. Tissue processing PFA-fixed mouse brains were removed, dissected, and processed to quantify remyelination at discrete locations in the corpus callosum, cerebral cortex, and hippocampus as described in detail by Bai and colleagues25. Myelin fiber density in the corpus callosum was quantified in 1μm-thick, epon-embedded sections, while remyelination in the cerebral cortex and hippocampus was determined by the area occupied by PLP-positive myelin staining in 30 μM-thick free-floating sections. For precise comparisons of similar areas of corpus callosum, 1 mm-thick coronal brain slices located approximately at Bregma 0 to −1mm were cut by using a custom-made brain mold25 and placed in 2.5% glutaraldehyde and 4% PFA for 48 h. The corpus callosum was removed from this slice and embedded in Epon. For cortical and hippocampal studies, brain slices (1 mm thick) located at Bregma −1.25 were placed in 4% PFA for an additional 48 h and subsequently sectioned into 30 μm-thick sections using a freezing/sliding microtome (Leica SM2010R). Two 30μm-thick brain sections containing the similar regions of hippocampus and cortex were stained for PLP. Immunohistochemistry of 30 μm-thick free-floating brain sections Free-floating 30 μm-thick brain sections were rinsed in PBS and microwaved in 0.01 M citric buffer (pH 6.0) for 2 min. Sections were incubated in 3% hydrogen peroxide and 0.1% Triton-X100 in PBS for 30 min and then placed in 3% normal goat serum (NGS) blocking solution (3% NGS in 1xPBS). The floating sections were incubated overnight at 4 °C with a rat monoclonal antibody directed against rat monoclonal anti-myelin proteolipid protein (PLP) antibody (Millipore Sigma, MABN2620). Sections were then rinsed in PBS with 0.1% Triton X-100 and stained by avidin-biotin complex (ABC) and diaminobenzidine (DAB) as described previously50. After several washes, sections were mounted on glass slides, dried overnight at RT, and cover slipped. Imaging and quantitative analyses of PLP staining in hippocampus and cerebral cortex PLP-stained sections were scanned in a Mirax high-resolution slide scanner equipped with a 20× objective (Zeiss). Each image tile was acquired from 5 focal levels that were 2 μm apart and projected into one final composite. Images were analyzed using NIH ImageJ software. The regions of interest (ROIs) for hippocampus and cerebral cortex were described previously25. The threshold masks and areas of ROIs were measured to calculate the percentage area occupied by PLP staining. The threshold mask was co-registered with the original image to confirm accuracy of thresholding. Two brain slices were analyzed per mouse. Quantification of myelinated axons in corpus callosum To analyze myelinated axons, brain sections containing the midline of the corpus callosum were embedded in epon resin (Electron Microscopy Sciences). Approximately 50–70 1 μm-thick serial sections were cut and collected. To match or compare similar areas of corpus callosum, every 4th section was stained with toluidine blue to determine the section numbers that were ~0.6 mm lateral to the midline. Two new sections were then selected and stained with 4% paraphenylenediamine (PPD) (Sigma), which is essentially a fat stain based upon the osmication of unsaturated lipids, to visualize myelinated axons. Two images of the corpus callosum were captured from the ROI and a total of 4 images from each brain were obtained. Using ImageJ software, a grid of 300 μm2 was overlaid on top of the images and nine boxes were determined in corpus callosum areas within the grid and myelinated axons were counted manually. The total numbers of myelinated axons from four images/brain were counted. The counts represent the number of myelinated axons in a 180 μm unit length of longitudinal corpus callosum in an ROI. Preliminary assessment of in vitro and in vivo Pharmacokinetics (PK) CN045 was tested by a contract research organization (Xenogesis, Inc.) for intrinsic hepatic clearance in male mouse (C57BL6) and human (mixed gender) liver microsomes to assess the loss of compound with time following incubation with hepatocytes in vitro. In addition, Xenogesis, Inc. tested CN045 for time-dependent P450 inhibition with and without NADPH. We delivered CN045 via i.p. injections to 5 groups of C57BL/6J mice (n = 3 per group) at a dose of 10 mg/kg body weight, and collected plasma and brain tissues at 1, 2, 4, 8, and 24 h post-injection. Tandem mass spectrometry (LC/MS/MS) was used to assess the concentrations of CN045 in plasma and brain. Statistical analyses All data were analyzed in a blinded fashion. For data with normal distribution, a two-tailed Student’s t test or a one-way ANOVA was used, followed by post hoc testing using Holm-Sidak (or Tukey’s) analysis for pair-wise comparisons as recommended by SigmaPlot/SigmaStat (SYSTAT, San Jose, CA) or open-source software RKward (KDE.org). p < 0.05 was deemed as statistically significant. Outliers were identified based upon Chauvenet’s criteria and were excluded from the analysis. Data are presented as mean ± standard deviation (SD) as indicated. Supplementary information Supplementary information Supplementary information