Small compounds mimicking the adhesion molecule L1 improve recovery in a zebrafish demyelination model

Demyelination leads to a loss of neurons, which results in, among other consequences, a severe reduction in locomotor function, and underlies several diseases in humans including multiple sclerosis and polyneuropathies. Considerable clinical progress has been made in counteracting demyelination. However, there remains a need for novel methods that reduce demyelination while concomitantly achieving remyelination, thus complementing the currently available tools to ameliorate demyelinating diseases. In this study, we used an established zebrafish demyelination model to test selected compounds, following a screening in cell culture experiments and in a mouse model of spinal cord injury that was aimed at identifying beneficial functions of the neural cell adhesion molecule L1. In comparison to mammalian nervous system disease models, the zebrafish allows testing of potentially promotive compounds more easily than what is possible in mammals. We found that our selected compounds tacrine and duloxetine significantly improved remyelination in the peripheral and central nervous system of transgenic zebrafish following pharmacologically induced demyelination. Given that both molecules are known to positively affect functions other than those related to L1 and in other disease contexts, we propose that their combined beneficial function raises hope for the use of these compounds in clinical settings.

www.nature.com/scientificreports/ straightforward maintenance in the laboratory, is favorable for the screening and testing of bioactive compounds.
To generate a zebrafish model useful for demyelination research, we developed Tg(mbpa:gal4-vp16;uas:NTR-mCherry) transgenic zebrafish, in which mature oligodendrocytes are conditionally ablated under the control of the mbpa promoter 11 . Expression of nitroreductase (NTR), which is encoded by the bacterial cytotoxin gene nsfB, is not cytotoxic; however, upon treatment with the prodrug metronidazole (MTZ), this enzyme converts MTZ into cytotoxins, thus enabling targeted cell ablation of oligodendrocytes and Schwann cells. Therefore, exposure of our transgenic zebrafish to an MTZ-containing medium results in rapid ablation of oligodendrocytes and Schwann cells, followed by subsequent demyelination within the nervous system. L1 cell adhesion molecule (L1CAM; hereafter abbreviated L1) is a cell adhesion molecule with multiple functions in nervous system development, synaptic plasticity, and regeneration after injury 12 . It has been shown to exert beneficial effects in ameliorating the consequences of spinal cord injury and of Huntington's, Parkinson's, and Alzheimer's diseases in animal models. However, the use of L1 in the treatment of these diseases is hampered by the large size of this protein, which is difficult to synthesize recombinantly or to isolate from tissues in larger quantities. To circumvent this issue, small organic compounds were isolated from an NIH library and found to elicit L1 functions by binding to the active site of L1 in the N-terminal part of the third fibronectin type III homologous repeat 12,13 . These compounds yielded beneficial effects in mouse and zebrafish spinal cord injury [14][15][16] . Out of these L1-mimicking compounds, we chose duloxetine and tacrine to investigate their effects in a zebrafish model of demyelination. It should be pointed out that duloxetine is a therapeutic drug for mood disorders and neuropathic pain, acting as a selective serotonin and norepinephrine reuptake inhibitor 17 . Tacrine is known to be a reversible acetylcholinesterase inhibitor 18 . Since L1 has been shown to contribute to myelination, we deemed it important to investigate whether these mimetic compounds are suitable for enhancing nervous system remyelination.
Here, we used the NTR-MTZ model zebrafish to investigate whether agonistic L1 mimetics would influence remyelination after experimentally induced demyelination. Our findings demonstrate that application of tacrine and duloxetine improved oligodendrocyte survival and enhanced remyelination, both in the central and peripheral nervous system. In addition, tacrine, but interestingly not duloxetine, restored the locomotive functions of demyelinated zebrafish larvae, which is in agreement with the observed neuroprotection at the cellular level.

Results
Tacrine and duloxetine promote the regeneration of oligodendrocytes in the central nervous system after oligodendrocyte ablation. To examine the effect of the two selected compounds on remyelination, we took advantage of the transgenic zebrafish model of demyelination using the NTR-MTZ system. Tg (mbpa:gal4-vp16;uas:NTR-mCherry) larvae express mCherry in oligodendrocytes ( Fig. 1a and Supplementary  Fig. 1). To induce oligodendrocyte ablation, 10 mM MTZ was used on larvae for 3 days, from days post-fertilization (dpf) 5-8 ( Supplementary Fig. 2). Following MTZ treatment, larvae robustly showed vacuolated cells with fuzzy red fluorescence, cell death, and reduced myelin sheaths in the spinal cord (Fig. 1b). The number of oligodendrocytes was significantly lower in MTZ-treated larvae than in control transgenic larvae, which were left untreated. Following MTZ treatment for 2 days, larvae were washed to remove MTZ and incubated for another day in an embryonic medium containing 1% DMSO (hereafter referred to as the recovery group) or L1 mimetic compounds thereafter. The number of GFP-expressing oligodendrocytes in compound-treated larvae was determined in comparison to the recovery group, which showed a slight, non-significant restoration of oligodendrocyte numbers (Fig. 1c). Tacrine hydrochloride (hereafter referred to as tacrine) was applied at concentrations of 100, 250, and 500 nM ( Fig. 1d-f). The group treated with 250 nM tacrine showed significantly improved oligodendrocyte regeneration compared to the MTZ-treated group and the recovery group (Fig. 1j). Additionally, the number of vacuolated cells was significantly reduced by tacrine at this concentration (Fig. 1e,k). (S)-duloxetine hydrochloride (hereafter referred to as duloxetine) was used at concentrations of 5, 10, and 20 μM (Fig. 1g-i). Treatment with duloxetine at 10 μM significantly increased the number of oligodendrocytes as compared to the MTZ-treated and recovery groups (Fig. 1j), and duloxetine significantly reduced the number of vacuolated cells compared to the control group (Fig. 1h,k). These results suggest that tacrine and duloxetine enhance the regeneration of oligodendrocytes in the spinal cord of zebrafish following demyelination.
Given that tacrine and duloxetine restore the oligodendrocyte cell population after cell ablation, we assessed whether these two L1 mimetic compounds enhanced oligodendrocyte differentiation during the developmental process. To this end, we treated zebrafish larvae with each compound during oligodendrocyte differentiation. In the normal state, tacrine and duloxetine did not affect the differentiation of oligodendrocytes (Fig. 2a-c). The number of olig2 + dorsally migrated oligodendrocytes and mature MBP + oligodendrocytes was not different in either tacrine-or duloxetine-treated larvae compared to control larvae (Fig. 2d,e). However, unlike the normal state, L1 mimetic compounds, especially tacrine, increase cell proliferation in the spinal cord after oligodendrocyte ablation ( Supplementary Fig. 3). These results indicate that tacrine and duloxetine effectively restore the oligodendrocyte population during demyelination and have no effect on normal development.  (Fig. 3b,k,l). After the washout of MTZ, L1 mimetic compounds, similar to their effect on the CNS, enhanced the regeneration of Schwann cells, compared with the recovery group ( Fig. 3c- (Fig. 3f). After MTZ treatment, these axons were completely demyelinated following Schwann cell ablation (Fig. 3g). To compare the degree of Schwann cell regeneration, we examined the ratio of GFP + myelinated axons to GFP-demyelinated axons (Fig. 3m). Tacrine and duloxetine significantly improved the cytological appearance of these axons compared with the MTZ-treated and recovery groups at concentrations of 250 nM and 5 μM, respectively ( Fig. 3h-j,m). These findings indicate that both tacrine and duloxetine promote the regeneration of Schwann cells in the PLL axons after Schwann cell ablation.

Tacrine and duloxetine induce remyelination in the central nervous system after demyelination.
To determine whether tacrine and duloxetine can induce remyelination in the demyelinated nervous    www.nature.com/scientificreports/ system as well, transmission electron microscopy (TEM) was performed on transverse sections of the spinal cord and PLL (Fig. 4). Axons in the spinal cord and PLL of untreated control larvae were surrounded by a compact and thick myelin sheath ( Supplementary Fig. 4a,c). Upon MTZ treatment, larvae showed robust demyelination compared to control larvae ( Supplementary Fig. 4b,d). The percentage of myelinated axons per ventral hemi-spinal cord was decreased in MTZ-treated larvae and demyelination was accompanied by substantial degeneration of axons associated with myelin debris (Supplementary Fig. 4e-h). In contrast to the recovery group, tacrineand duloxetine-treated larvae showed higher numbers of remyelinated axons (Fig. 4a-c). Tacrine and duloxetine treatment increased the percentage of myelinated axons and concomitantly decreased the number of degenerated axons (Fig. 4d,e). No significant differences were observed in the number of large axons between the different groups (Fig. 4f). To quantitatively evaluate myelin thickness, the g-ratio (the ratio of the inner diameter to the outer diameter of the myelin sheath) was determined. The tacrine-and duloxetine-treated groups exhibited a reduced g-ratio compared to the recovery group, indicating the promotion of remyelination (Fig. 4g,h). Next, we examined whether tacrine and duloxetine promoted remyelination in the peripheral nervous system after demyelination. In the PLL, TEM analysis showed that only the duloxetine-treated group displayed enhanced remyelination compared to the recovery group (Fig. 4i,j,l). However, unlike the CNS, tacrine showed little effect on remyelination in the PLL (Fig. 4i,k). Together, these results demonstrate that tacrine and duloxetine enhance regeneration of myelinating glia and remyelination. Demyelination induces the activation and migration of macrophages/microglia resulting in the rapid removal of immune cells which is necessary for the remyelination step 19 . In order to verify whether L1 mimetic compounds enhance remyelination by modulating the immune response induced by oligodendrocyte ablation, we examined Tg(mpeg:gfp;mbpa:gal4-vp16;uas:NTR-mCherry) larvae, which express GFP in the microglia, and found that mpeg:GFP + microglia were recruited in the spinal cord after the ablation of oligodendrocytes (Supplementary Fig. 5a,d). Interestingly, duloxetine-treated larvae exhibited a significant reduction in mpeg:GFP + cells in the spinal cord ( Supplementary Fig. 5c,d), but tacrine treatment had no effect on the immune response ( Supplementary Fig. 5b,d). These results indicate that duloxetine, but not tacrine, improves the remyelination process by reducing excessive neuroinflammation.

Tacrine restores locomotor activity after demyelination. MTZ-induced demyelination has been
shown to lead to locomotor deficits in zebrafish larvae 20 . We therefore investigated whether the observed remyelination was sufficient to permit functional improvements. To this end, we measured parameters of locomotor activity, that is, distance moved (mm), velocity (mm/s), movement time (s), and resting time (s) over a time period of 20 min. These parameters were robustly reduced in MTZ-treated tg (mbpa:gal4-vp16;uas:NTR-mCherry) larvae compared to control larvae (Fig. 5). Since MTZ-treated wild type larvae showed normal locomotor activity ( Supplementary Fig. 6), we conclude that these defects in motor activity were due to demyelination. In addition, both the MTZ-treated and recovery groups showed similar locomotor activity, despite our finding that the recovery group showed a tendency to partially restore oligodendrocytes (Fig. 1c). These observations suggest that due to the short recovery time, functional recovery was not achieved. In contrast, treatment with 250 nM tacrine significantly increased locomotor activity compared to the recovery group (Fig. 5). Interestingly, locomotor activity was decreased in the duloxetine-treated group, in comparison with not only the recovery group but also the MTZ-treated transgenic zebrafish, which may be explained by the sedative effects of this compound. To investigate its sedative effects, we tested whether duloxetine treatment also affects the behavior of the wild type. Duloxetine-treated wild type zebrafish showed very low locomotor activity ( Supplementary Fig. 6). In addition, the actin cytoskeleton of the skeletal muscles was not affected by duloxetine ( Supplementary Fig. 7), indicating no noticeable effect on the viability of other cell types. We next tested whether functional recovery was achieved after washout of duloxetine. Locomotor behavior was examined for another 2 days after washout. The sedative effects of duloxetine weakened over time, returning to the level of the recovery group within 1 day (Supplementary Fig. 8). Taken together, these observations provide evidence that the decrease in locomotor activity observed with duloxetine was due to a sedative effect. Therefore, tacrine, in contrast to duloxetine, effectively promotes remyelination following demyelination and thereby boosts functional recovery.

Discussion
In this study, we used a robust and reliable transgenic zebrafish demyelination model to evaluate the effects of the L1 mimetic agonist compounds tacrine and duloxetine, which have previously been shown to enhance regeneration in mammalian cell cultures and the spinal cord [14][15][16] . Here, we found that these compounds displayed beneficial effects in both the peripheral and central nervous system of the demyelinated zebrafish, as evidenced by the increased survival and restoration of oligodendrocytes, protection of axons, improved myelin compaction and thickness, and, importantly, recovery of locomotive functions as indicated by several parameters of locomotor activity. These results show that our selected agonistic L1 mimetic compounds improved recovery after demyelination and demonstrate the feasibility and significance of the zebrafish demyelination model for screening and validating drug candidates for the improvement of remyelination.
It should be emphasized that the two evaluated L1 mimetics are known to exert other functions in addition to mimicking L1. Several acetylcholinesterase inhibitors including tacrine, rivastigmine, and donepezil are used in the treatment of individuals with moderate Alzheimer's disease by inhibiting the breakdown of acetylcholine 21 . However, tacrine has been removed from the list of FDA-approved drugs since adverse effects were observed at the prescribed concentrations 22 . Oligodendrocyte lineage cells express several nicotinic acetylcholine receptors, depending on their developmental stage 23 . Cholinergic signaling is important for oligodendrocyte development and myelination under normal physiological conditions and in disease. Thus, acetylcholinesterase inhibitors were reported to induce myelin gene expression, improve myelin integrity, and reduce inflammation 24   www.nature.com/scientificreports/ acetylcholinesterase inhibitor, tacrine may therefore not only act as an L1 agonist but also affect the differentiation of oligodendrocytes, which may well have played a role in the effects on remyelination observed here. Recent research has shown that donepezil, an acetylcholinesterase inhibitor, promotes the differentiation of oligodendrocyte precursor cells to oligodendrocytes in in vitro cell culture systems, while tacrine does not 25 . However, the current study shows that tacrine has the capacity to restore oligodendrocyte cell populations in a pathological state, such as demyelination. Therefore, although tacrine has been taken off the market due to adverse effects in relation to Alzheimer's disease, and despite the fact that findings in zebrafish are not directly translatable to a human context 26 , we would still like to suggest that lower tacrine concentrations may be used in humans since we observed regenerative effects of this drug in the nanomolar range in our experiments.
Here we show that in zebrafish, similar to tacrine, treatment with duloxetine protects oligodendrocytes from degeneration while boosting their regeneration, thereby enhancing remyelination. Duloxetine is known to affect neuroinflammation by regulating cytokine responses 27 , namely by inhibiting the release of pro-inflammatory cytokines while stimulating the production of anti-inflammatory cytokines. Also, in a mouse model of neuropathic pain, duloxetine inhibits immune responses via downregulation of the tumor necrosis factor-α (TNF-α)/ NF-κB signaling pathway 28 . Since multiple pro-inflammatory cytokines including TNF-α are thought to contribute to inflammation and demyelination in MS 29 , it is possible that in the current study, duloxetine improved remyelination by reducing inflammation and thus creating a protective and pro-regenerative environment. Contrary to tacrine-treated larvae, duloxetine-treated larvae did not show improved functional locomotive recovery. Although duloxetine has been reported to induce hyperactivity in several animal models, other studies have observed deviating effects of this compound [30][31][32] . Duloxetine treatment may affect the behavior of zebrafish larvae via 5-HT4 and α 2 -adrenergic receptors in the central and peripheral nervous system and has been reported to cause a decrease in swimming speed and an increase in resting time, which is consistent with our results 33 . Since it has been shown that the effect of duloxetine on locomotor activity varies depending on concentration www.nature.com/scientificreports/ and treatment time 34 , further studies are required to elucidate whether this L1 mimetic may improve functional recovery following demyelination at other doses.
Altogether, the present study shows that tacrine and duloxetine enhance remyelination in a zebrafish demyelination model. These beneficial effects are likely due to a combination of L1′s ability to affect myelination and the compounds' indirect effects on the animal's immune system. The fact that L1 promotes myelination in mammalian cell cultures in the absence of immune system cells 13 is indicative of its direct influence on myelination. The overall effects of these two L1 mimetics on enhancing remyelination raises the hope that tacrine and duloxetine may be repurposed for the treatment of demyelinating diseases.
All fluorescent images were taken in the 2-somite area above the end of the yolk extension using an A1Si laser-scanning confocal microscope (Nikon, Tokyo, Japan). Confocal images (1-μm z-stacking) were processed using NIS-Elements AR Analysis 4.30 software (Nikon).
Transmission electron microscopy. Tissues were prepared using standard procedures as previously described 38 . Zebrafish larvae were anesthetized with tricaine (Cat #A5040, Sigma) and fixed in 10% formaldehyde/2.5% glutaraldehyde/0.1 M phosphate buffer, with a pH of 7.4, at 4 °C overnight. Then, larvae were post-fixed in 1% osmium tetroxide, sequentially dehydrated, and embedded in eponate-12 resin (Cat #18,006, Ted Pella). Sections of 1-μm thickness were obtained using a Reichert-Jung Ultracut E Ultramicrotome (Leica Microsystems, Wetzlar, Germany) and stained with toluidine blue. Images were taken with a Zeiss Axio Observer microscope (Carl Zeiss Microscopy, Jena, Germany). Sections of 60-nm thickness were collected on formvarcoated slot grids, stained with uranyl acetate/lead citrate, and imaged on a H-7500 transmission electron microscope (80 kV, Hitachi, Tokyo, Japan).
Behavior. All behavior tests were performed between 1 and 4 pm in light conditions. At 8 dpf, larvae from each group were individually transferred onto 48-well plates, with each well containing 500 μl of embryo medium. For the behavioral analyses, larvae were acclimated for 1 h before their swimming performance was recorded. Then, locomotion was recorded for 15 min under light conditions and analyzed using the EthoVision XT 12 system (Noldus, Wageningen, Netherlands). The total distance moved was recorded in mm, velocity in mm/s, and total time moved in seconds, each over the course of 15 min. Each test was independently repeated three times.
Statistics. Statistical analyses were conducted using GraphPad Prism 7.0c software (GraphPad Software, San Diego, CA, USA). A one-way analysis of variance (ANOVA) followed by Tukey's test for multiple comparisons was used to analyze differences between the groups. Comparisons of g-ratios were performed using linear regression analysis. Data are expressed as the mean ± standard deviation and p < 0.05 was considered to indicate a significant difference. Statistical significances are indicated as follows: *p < 0.05; **p < 0.01; and n.s. (not significant).
Ethics approval. All experimental procedures were approved by the Korea University Institutional Animal Care and Use Committee and carried out in accordance with the animal experimental guidelines of the Korea National Veterinary Research and Quarantine Service. The study was carried out in compliance with the ARRIVE guidelines.

Data availability
Data supporting the findings of this study are available in the article and the Supplementary Information Files, or from the corresponding authors on reasonable request.