Leucine-Rich Glioma Inactivated 1 Promotes Oligodendrocyte Differentiation and Myelination via TSC-mTOR Signaling

Leucine-rich glioma inactivated 1 (Lgi1), a putative tumor suppressor, is tightly associated with autosomal dominant lateral temporal lobe epilepsy (ADLTE). It has been shown that Lgi1 regulates the myelination of Schwann cells in the peripheral nervous system (PNS). However, the function and underlying mechanisms for Lgi1 regulation of oligodendrocyte differentiation and myelination in the central nervous system (CNS) remain elusive. In addition, whether Lgi1 is required for myelin maintenance is unknown. Here, we show that Lgi1 is necessary and sufficient for the differentiation of oligodendrocyte precursor cells and is also required for the maintenance of myelinated fibers. The hypomyelination in Lgi1−/− mice attributes to the inhibition of the biosynthesis of lipids and proteins in oligodendrocytes (OLs). Moreover, we found that Lgi1 deficiency leads to a decrease in expression of tuberous sclerosis complex 1 (TSC1) and activates mammalian target of rapamycin signaling. Together, the present work establishes that Lgi1 is a regulator of oligodendrocyte development and myelination in CNS.

Apart from epileptogenesis, it has been shown that Lgi1 plays roles in neuronal and glial development. The ablation of Lgi1 causes a subtle neuronal dyslamination in the cortex (Silva et al., 2015) and reduces the thickness of external granule cell layer in embryonic cerebellum (CB; Su et al., 2015;Xie et al., 2015). Members of Lgi family are yet involved in the myelination of Schwann cells in the peripheral nervous system (PNS). In particular, Lgi4 binds to ADAM22 and facilitates cellular interaction and axon sorting in the developing PNS (Özkaynak et al., 2010;Kegel et al., 2014) and the loss function of Lgi4 contributes to the abnormalities in claw paw (clp) mutant mice (Bermingham et al., 2006). Moreover, EM observations ON the sciatic nerve demonstrate Lgi1 knockout impairs myelination of axons in PNS . These findings raise a question whether Lgi1 coordinates the myelination of oligodendrocytes (OLs) in CNS.
In the present work, we show that Lgi1 is necessary and sufficient to the differentiation of oligodendrocyte precursor cells (OPCs) and myelination and is also required for the maintenance of myelinated fibers facing cuprizone challenge. The hypomyelination caused by Lgi1 deficiency attributes to inhibited biosynthesis of lipids and proteins in OLs. We further show that Lgi1 deficiency decreases the expression of tuberous sclerosis complex 1 (TSC1) and breaks the balance of mTOR signaling in OLs, which might be the cause of hypomyelination.

Animals
All experiments were approved by the Animal Experimentation Ethics Committee of Zhejiang University and were specifically designed to minimize the number of animals used. Original breeding pairs of Lgi1 −/− mice were obtained from John Cowell (Augusta University; Yu et al., 2010;Zhou et al., 2018) and were maintained at the Experimental Animal Center of Zhejiang University. Mice were kept under temperaturecontrolled condition on a 12:12 h light/dark cycle with food and water ad libitum. In vivo experiments were done in a batch of mice of either sex.

OPC Culture
Purified OPCs from SD rats were isolated by shaking off as described previously (Zhou et al., 2014). In brief, OPCs were collected from glial cultures by shaking for 1 h at 200 rpm, incubating in fresh medium for 4 h, and shaking at 250 rpm at 37 • C for 16 h. Collected OPCs in the medium were re-plated onto poly-D-lysine-coated plates and grew in Neurobasal medium supplemented with 2% B27. PDGF-AA (10 nM) was added in the medium to keep OPCs undifferentiated. Alternatively, T3 (40 ng/ml) was added to the medium for 3 days to allow their differentiation (Li et al., 2013;Zhou et al., 2014).

Lentivirus Construction and Transfection
Lentivirus encoding small hairpin RNA (shRNA) for Lgi1 (sequence: 5'-CCT AAG AGG GAA CTC ATT T-3') was prepared by OBIO (Shanghai). Overexpressing Lgi1 was based on the coding sequence of rat Lgi1 gene (GenBank accession number 145769). Lgi1-shRNA and scrambled RNA were driven by U6 promoter, whereas overexpressed Lgi1 was driven by CMV promoter. OPCs were transfected with Lgi1-shRNA for 72 h and Lgi1 overexpression virus for 24 h before experiments. Only when >60% of cultured OPCs were transfected, which was confirmed by GFP fluorescence, experiments were continued.

Immunohistochemistry and Immunocytochemistry
Thirty micrometer sagittal sections were prepared and placed in blocking solution (1% BSA, 0.3% Triton, and 10% goat serum) for 1 h at room temperature (RT). After washing with PBS, sections were incubated sequentially with primary antibodies overnight at 4 • C and secondary antibodies for 1 h at RT. The secondary antibodies were diluted at 1:1000. The sections were mounted using ProLong Gold Antifade Reagent with DAPI (Invitrogen). Cultured cells were fixed with 4% paraformaldehyde for 15 min at RT, washed with PBS and permeabilized in 0.2% Triton X-100 for 10 min, blocked in 10% BSA for 1 h, and labeled with primary antibodies overnight at 4 • C, then cells were incubated with secondary antibodies (1:1000) for 1 h at RT. All antibodies were diluted in PBS containing 1% BSA and 1% normal goat serum. The dilution ratios of primary antibodies for immunohistochemistry and immunocytochemistry were MBP (1:1000), pS6 (1:1000), and FASN (1:100). To calculate the areas of cultured cells, somata and processes were manually tracked by tracing their edges. A process was determined by criteria including the explicit focus, the clear origin point from the soma, and the separation from other processes.

Cell Counts
Three to four animals per genotype were used to examine the cellular marker expression for each time point. In the cortex, three to five nonadjacent sections were counted per animal. In the corpus callosum (CC), images were acquired to include only the CC at the midline.

Eriochrome Cyanine Staining
Cryosections on glass slides were rinsed with PBS and submerged in Eri-C solution for 1 h. The sections were then differentiated in 10% (w/v) iron alum for 30 min until the nuclei became transparent. Slides were then dehydrated and coverslipped. All operations were performed at RT.

Electron Microscopy (EM)
Tissues for the electron microscopy (EM) were prepared as described previously (Pereira et al., 2010;Zou et al., 2011). Ultra-thin sections were obtained using Ultracut UCT (Leica) and stained with 2% uranyl acetate and lead citrate. Electron micrographs were taken with a Philips CM100 microscope (FEI). The ratio of axonal diameter/fiber diameter (g-ratio) was acquired using ImageJ software.

Human Material
The human study was approved by the Medical Ethical Committee of Zhejiang University School of Medicine and was conducted in conformance with policies and principles included in the Federal Policy for Protection of Human Subjects and in the Declaration of Helsinki. Informed written consents were given by all participants. Basic clinic information of these patients is given in Table 1. Brain tissue specimens containing neocortex were obtained as part of planned surgical margin of resection surrounding the tumor core. The para-cancerous tissue, tumor infiltration area and tumor mass area of glioma foci was distinguished by visible appearance and separated under an inverted microscope (Pallud et al., 2014). Aliquots of tissues from each patient were used for protein extraction and western blot, in which the subdivision of each area was confirmed by the different expression of PSD95 and GFAP (Louis et al., 2007).

Statistics
Data analysis was performed using Excel 2003 (Microsoft), Igor Pro 6.0 (Wavemetrics), and SPSS 16.0 statistical program (SPSS). Statistical differences were determined using unpaired two-sided Student's t-test. The accepted level of significance was p < 0.05. ''n'' represents the number of animals or cultures tested. Data in the text and figures are presented as mean ± SEM. The experimental protocols, analytic methods and study material that support this study are available from the corresponding authors upon reasonable request.

Expressions of Lgi1 and MBP Are Correlated in OLs
We measured the expression of MBP and Lgi1 in the postnatal brain and spinal cord (SC). Western blots showed that MBP expression increased in proportion to Lgi1 in both regions ( Figure 1A). Similarly, in cultured OPCs and mature OLs, we found that the expression of Lgi1 was significantly elevated in OLs compared with OPCs ( Figure 1B). Immunocytochemical staining corroborated this result by showing high fluorescence of Lgi1 in OLs, which was not limited to soma but widespread in elaborated processes, a characterization of mature OL ( Figure 1C). In order to further determine the correlation between Lgi1 and myelin proteins, we measured their expressions in glioblastomas. Total 18 fresh brain tissue specimens obtained from six patients (three tissues from each patient) with supratentorial, hemispheric, and diffuse high-grade gliomas ( Table 1) were physically subdivided into control (n = 18), tumor infiltration (n = 18), and tumor mass (n = 18), which were confirmed by levels of GFAP and PSD95 in western blot ( Figure 1D). In consistent with previous report (Besleaga et al., 2003), our results showed Lgi1 expression was significantly attenuated in gliomas ( Figures 1E,F). Meanwhile, the expressions of myelin-associated proteins (MBP, MOG, MAG and CNP) were unanimously reduced in gliomas ( Figures 1E,F). Hence, these results provide evidence showing the correlated expression between Lgi1 and myelin proteins.

Lgi1 Deficiency Causes Hypomyelination in CNS
We next used EM to compare myelin formation in the SC from wild-type (WT) and Lgi1 −/− littermates. Several abnormalities were found in Lgi1 −/− mice at P4: (i) there appeared to be an increased incidence of loose myelin layers (Figure 2A); (ii) myelin thickness was reduced, which was revealed by morphometric quantification showing an increased g-ratio in myelin sheath of both smaller (  Frontiers in Molecular Neuroscience | www.frontiersin.org a hypomyelination phenotype in postnatal Lgi1 −/− mice. Moreover, this hypomyelination persisted in P14 Lgi1 −/− mice, as shown by increased g-ratio and loose myelin layers (Figures 2C,D) and a significant reduction in the number of myelinated axons compared to WT ( Figure 2D). The hypomyelination in Lgi1 −/− mice is confirmed by western blot and immunohistochemistry. First, the expression of myelin proteins, including MBP, MOG and MAG, was prominently attenuated in the CC, ON, CB and SC from Lgi1 −/− mice at P14 (Figure 2E). Second, the immunostaining presented a broad loss of MBP-positive fibers in Lgi1 −/− mice ( Figure 2F). Third, eriochrome cyanine staining showed that Lgi1 −/− mice had a significant reduction in the density of white matter tracts in both CC and SC of P14 Lgi1 −/− mice ( Figure 2G). Taken together, these findings suggest that deletion of Lgi1 leads to myelination defects in CNS.

Lgi1 Ablation Impairs OPC Differentiation in the Brain
To determine how Lgi1 deficiency causes hypomyelination, we analyzed the expression of cellular markers selected for OPC and OL. Our results showed that the total number of OLs indicated by Olig2+ cells in the CC from P14 Lgi1 −/− mice was not affected ( Figure 3A). However, the number of differentiated OLs expressing CC1 and Olig2 simultaneously was reduced by 23% compared with WT ( Figure 3A). These results suggest that Lgi1 contributes to OPC differentiation in the brain. We continued to examine if the proliferating capacity of OPCs was influenced by Lgi1 ablation. PDGFαR staining in CC and cortex showed that there was no significant difference in the number of PDGFαR+ OPCs between WT and Lgi1 −/− littermates at P14 (Figure 3B). Furthermore, immunohistochemical staining with antibodies against Ki67 and Olig2 was used to assess the cell cycle of OPCs. Statistically, the number of proliferating OLs (Ki67+/Olig2+) of Lgi1 −/− mice was normal compared with that of WT in the CC ( Figure 3C). These data suggest that Lgi1 is not required for OPC formation and proliferation.

Lgi1 Is Necessary and Sufficient for OPC Differentiation
In vitro observations were next used to determine the role of Lgi1 in OPC differentiation by changing its expression with lentiviral transfection in purified OPC cultures (Zhou et al., 2014). OPCs were transduced with GFP-tagged Lgi1-shRNA or scrambled RNA prior to T3 stimulation and affected cells were distinguished by GFP fluorescence. The efficiency of transfection was defined by western blot assay ( Figure 4A). We found that T3 promoted the differentiation of OPCs in control group, as shown by characteristic extensive processes of MBP+ cells. In contrast, shRNA-induced Lgi1 knockdown restricted the development of processes and remarkably reduced the area of MBP+ cells, though the number of MBP+ cells remained unchanged ( Figure 4B). In consistent with immunocytochemical observations, the protein expression of MBP and CNP was also reduced in Lgi1-shRNA group (Figure 4A). These results indicate that Lgi1 may be necessary for OPC differentiation.
To investigate whether Lgi1 is sufficient to promote OPC differentiation, we transduced cultured OPCs with either control virus or lentiviral plasmid encoding Lgi1, as shown by the expression of Flag (Figure 4C). The OPC cultures were then treated with T3 for 1 day, a time shorter than normal OPC differentiation. Our results showed that overexpression of Lgi1 significantly increased the area of MBP+ cells compared with vector control (Figure 4D); and MBP and MAG were also elevated (Figure 4C), suggesting that Lgi1 overexpression enhances the differentiation capacity of OPC.

Lgi1 Attenuates Cuprizone-Induced Myelin Degeneration
The degeneration of myelin sheath occurs in many demyelinating diseases, for example multiple sclerosis (Raine et al., 1999) and in gliomas (Underhill et al., 2011;Liu et al., 2014;Mantero et al., 2015). It was of interest to explore whether Lgi1 is involved in myelin loss. To this end, we established a demyelination model by feeding 8-week old adult mice with a diet containing 0.2% cuprizone for 4 weeks. In this scenario, Lgi1 +/− mice was used because homozygous Lgi1 −/− mice died usually after P17 Zhou et al., 2018). In agreement with previous work (Chen et al., 2012), cuprizone administration led to massive demyelination in cortical layers, cingulum bundles, and CC, as shown by greatly attenuated cyanine-stained myelin tracts in the CC and cortex ( Figure 5A). Interestingly, cuprizoneinduced loss of myelin sheath was much severer in Lgi1 +/− mice than WT (Figure 5B), indicating that the haploinsufficiency of Lgi1 exacerbates the degeneration of myelinated fibers. This conclusion was also confirmed by western blot, which showed the reduction in the expression of myelin-associated proteins, including MBP, MAG, and MOG, was more prominent in Lgi1 +/− mice compared with WT littermates when they were both treated with cuprizone ( Figure 5C). Meanwhile, the myelin tracts (Figures 5A,B) and the expression of myelin proteins ( Figure 5B) were not changed in Lgi1 +/− mice compared to WT mice, indicating that myelin maturation is not affected by the haploinsufficiency of Lgi1. Hence, these results demonstrate that Lgi1 prevents the myelin degeneration caused by cuprizone. However, the precise roles of Lgi1 in myelin maintenance shall be determined using adult mice with the deletion of Lgi1 by PLP-ER-Cre.

Lgi1 Deficiency Inhibits Lipid Biosynthesis in OLs
Lipid contents provide the building blocks of multilayered membrane structure of OLs (Woelk and Borri, 1973;Fressinaud et al., 1990;Salles et al., 2002;Saher et al., 2005). Since increased g-ratio and loose myelin layers have been found in Lgi1 −/− mice (Figure 2), we investigated whether Lgi1 ablation changes the lipogenesis in OLs. Western blot revealed that the expression of FASN was significantly decreased in Lgi1 −/− mice ( Figure 6A). Because the contribution from cell types other than OLs might have masked the attenuation of FASN, we performed immunohistochemistry on P14 CC. In Lgi1 −/− mice, Olig2+ OLs showed strongly diminished FASN staining present in the sections, leading to reduced FASN+Olig2+ cells in Lgi1 −/− mice ( Figure 6B). This finding was corroborated and extended by in vitro experiments utilizing GFP-tagged Lgi1-shRNA in cultured OPCs, showing that the fluorescence of FASN was significantly reduced in shRNA-transfected cells compared with those unaffected (Figure 6C). These results suggest that Lgi1 ablation might impair myelin layers via altered lipid biogenesis in OLs.

Lgi1 Deficiency Activates mTOR Signaling in OLs
The results above imply that Lgi1 deficiency changes intracellular signaling in OLs, but the downstream effectors of Lgi1 largely remain unknown. mTOR signaling is intimately linked with TSC1 (Laplante and Sabatini, 2009) and TSC1-mTOR signaling acts as a critical checkpoint for OL homoeostasis and proper CNS myelination (Norrmén and Suter, 2013;Lebrun-Julien et al., 2014;Zou et al., 2014;Jiang et al., 2016). Interestingly, mTOR regulates the lipogenesis in passage cell lines (Laplante and Sabatini, 2010;Soliman, 2011;Lamming and Sabatini, 2013) and in OLs (Lebrun-Julien et al., 2014). Accordingly, we investigated whether TSC1-mTOR signaling is altered and thereby underlies observed phenotypes in Lgi1 −/− mice. The expression of TSC1 and ribosomal protein S6, the substrate of mTORC1, and the phosphorylation of S6 (pS6) was examined in WT and Lgi1 −/− littermates. Indeed, we found that mTOR signaling is activated in Lgi1 −/− mice because pS6 was increased while TSC1 expression was decreased in these mice (Figure 7A). Given that this phenotype might hold partial contributions from other cell types, two experiments argued that TSC1-mTOR signaling was indeed altered in OLs. First, immunohistochemistry on CC sections demonstrated that the intensity of pS6 signal was strongly increased in Olig2+ OLs from Lgi1 −/− littermates compared to WT (P14) ( Figure 7B). Second, pS6 was increased but TSC1 expression was reduced in shRNA-treated rat OPC cultures where Lgi1 expression was suppressed ( Figure 7C). These results indicate that Lgi1 ablation upregulates mTORC1 pathway in OLs, which might further reduce myelin formation and cause demyelination.

DISCUSSION
Lig1 has been documented as an epilepsy-linked gene because Lgi1 mutations result in ADLTE (Kalachikov et al., 2002). Previous EM observations indicate that Lgi1 knockout impairs myelination in the PNS and CNS , however, the mechanisms underlying the function of Lgi1 in oligodendrocyte myelination remain elusive. Our present study demonstrates that Lgi1 promotes the differentiation of OPCs in part through activating the biosynthesis of lipids and proteins in OLs by tipping the balance of TSC1-mTOR signaling. We further  find that Lgi1 is critical for the maintenance of myelinated fibers.
We observed that Lgi1 expression was positively correlated with MBP expression ( Figure 1A) and Lgi1 was significantly elevated during OPC differentiation (Figures 1B,C). Moreover, the difference of g-ratio between WT and Lgi1 −/− mice became smaller at P14 compared to P4 (Figures 2A-D). These data suggest that there may be stage-dependent functions of Lgi1 in oligodendrocyte development, which needs to be confirmed by using conditional Lgi1-knockout mouse, as the global deletion of Lgi1 causes the death of animal  Importantly, the knockdown of Lgi1 in OPCs impaired OPC differentiation and maturation (Figure 4). The present work using isolated OPC hypothesizes a cell-intrinsic function of Lgi1 in OLs development and CNS myelination ( Figure 7D, blue lines), distinct from Lgi4-induced reciprocal cell-neuron communications in PNS myelination (Özkaynak et al., 2010;Kegel et al., 2014). Nonetheless, we could not exclude the possibility of non-cell autonomous effects of Lgi1 loss in other neural cell types on myelinogenesis, because either autocrine or paracrine of Lgi1 was not prevented in our experiments. The mice with the specific deletion of Lgi1 in OL lineage cells will be a better model for a clear conclusion.
The secreted protein Lgi1 can interact with receptors such as ADAM22/23/11 (Kegel et al., 2014) and Nogo receptor 1 (NgR1; Thomas et al., 2010). However, none of these receptors is expressed in OLs. Leucine rich repeat and Immunoglobin-like domain-containing protein 1 (LINGO-1) is a transmembrane protein expressed in OLs and negatively regulates OPC differentiation and myelination (Mi et al., 2005;Jepson et al., 2012). Both Lgi1 and LINGO-1 are LRR-containing proteins, which might enable them to heterodimerize with each other. Alternatively, Lgi1 and LINGO-1 may compete with each other for binding to NgR1 and this competition may play an indirect role in CNS myelination ( Figure 7D, red lines). The potential interaction between Lgi1 and LINGO-1 remains to be determined using transgenic mice with a mutation disrupting their association.
Balanced mTORC1 activity is required for CNS myelination. On one hand, mTORC1 activity is required for myelination (Flores et al., 2008) and in vitro or in vivo inhibition of mTOR signaling in OLs displays impaired OL development and hypomyelination in the CNS (Narayanan et al., 2009;Tyler et al., 2009;Zou et al., 2011Zou et al., , 2014Guardiola-Diaz et al., 2012;Bercury et al., 2014;Wahl et al., 2014). On the other hand, overactivation of mTORC1 or TSC1 mutation causes hypomyelination in part by downregulating Akt signaling and lipogenic pathways as well as enhancing stress responses (Lebrun-Julien et al., 2014;Jiang et al., 2016). In the present work, we found that Lgi1 deficiency decreases the expression of TSC1 and activates mTORC1 in OLs, and Lgi1 −/− mice showed impaired lipogenic biosynthesis and loose myelin layers. These phenotypes corroborate with the importance of balanced mTORC1 activity in CNS myelination (Lebrun-Julien et al., 2014;Jiang et al., 2016). The finding that TSC1 expression was downregulated in Lgi1 −/− mice is unexpected and the function relationship between Lgi1 and TSC1 remains to be determined. Intriguingly, TSC1 mutant mice display a much reduced myelination (Meikle et al., 2007), in consistent with present study. The mutations in both Lgi1 and TSC1 lead to epileptogenesis (Meikle et al., 2007). The loss-of-function mutation of TSC1 in the cortex causes defective neuronal development at postnatal stages (Meikle et al., 2007;Normand et al., 2013;Tee et al., 2016). The ablation of Lgi1 also causes neuronal dyslamination in the cortex (Silva et al., 2015) and reduces the proliferation of granule precursor cells in embryonic CB . These observations suggest a potential correlation of Lgi1 and TSC1 functions not only in OPC differentiation but also myelin maintenance. In addition, it is possible that TSC2 is also involved in Lgi1-controlled myelination, because TSC2 deletion causes the hypomyelination in the brain (Carson et al., 2015).
Previous works show that Lgi1 inhibits the proliferation and causes the apoptosis of neuroblastoma cells (Kunapuli et al., 2003;Gabellini et al., 2006), which might be mediated by altered AKT and ERK (extracellular-signal-regulated kinase) signaling (Kunapuli et al., 2003;Sirerol-Piquer et al., 2006) However, the roles of Lgi1 in oncogenesis still remain quite unclear. We found that Lgi1 is critical for the maintenance of myelin in the brain, suggesting that Lgi1 loss may contribute to gliomarelated dysmyelination. Future study of Lgi1 functions would need to define its biological functions in gliomagenesis and its therapeutic potential.

CONCLUSION
Epileptic gene Lgi1 is necessary and sufficient for OPC differentiation and myelination, which may be associated with the modulation on the activity of TSC1-mTORC1 signaling.

AUTHOR CONTRIBUTIONS
Y-JX, YW and LiangZ designed the research. Y-JX, LinZ, N-WJ, SC, C-YS and X-TW performed the research. X-YL provided unpublished tools and techniques. Y-JX, LinZ and LiangZ analyzed the data. Y-JX, LinZ, YW, YS and LiangZ wrote the article. All authors read and approved the final manuscript.