Is Essential for the Differentiation of Radial Glial Cells to Oligodendrocytes in the Dorsal Cortex

Home , Oligodendrocyte, Radial glial cell

ORIGINAL ARTICLE Liver X receptor β is essential for the differentiation of radial glial cells to oligodendrocytes in the dorsal cortex

PXu1,6,HXu2,6, X Tang1,LXu1, Y Wang1, L Guo1, Z Yang1, Y Xing3,YWu3, M Warner4, J-A Gustafsson4,5 and X Fan1

Several psychiatric disorders are associated with aberrant white matter development, suggesting oligodendrocyte and myelin dysfunction in these diseases. There are indications that radial glial cells (RGCs) are involved in initiating myelination, and may contribute to the production of oligodendrocyte progenitor cells (OPCs) in the dorsal cortex. Liver X receptors (LXRs) are involved in maintaining normal myelin in the central nervous system (CNS), however, their function in oligodendrogenesis and myelination is not well understood. Here, we demonstrate that loss of LXRβ function leads to abnormality in locomotor activity and exploratory behavior, signs of anxiety and hypomyelination in the corpus callosum and optic nerve, providing in vivo evidence that LXRβ deletion delays both oligodendrocyte differentiation and maturation. Remarkably, along the germinal ventricular zone-subventricular zone and corpus callosum there is reduced OPC production from RGCs in LXRβ− / − mice. Conversely, in cultured RGC an LXR agonist led to increased differentiation into OPCs. Collectively, these results suggest that LXRβ, by driving RGCs to become OPCs in the dorsal cortex, is critical for white matter development and CNS myelination, and point to the involvement of LXRβ in psychiatric disorders.

Molecular Psychiatry (2014) 19, 947–957; doi:10.1038/mp.2014.60; published online 17 June 2014

INTRODUCTION guiding migration of later-born neurons during corticogenesis. Deterioration of white matter structures including axons, myelin Loss of LXRβ in mice causes significant reduction of RGC long 16 sheaths and oligodendrocytes (OLs) contribute to the patho- fibers in the cerebral cortex during later embryonic stages. In the physiology of central nervous system (CNS) diseases.1,2 cerebellum, LXR agonist treatment of neonatal mice promotes the Mature OLs derived from oligodendrocyte precursor cells migration of granular neurons by inhibition of RGC transformation 17 (OPCs) are the myelin-forming cells of the CNS, which surround into astrocytes during development. Consistent with these axons to form myelin and mediate the fast conduction of neuronal findings, LXRβ deletion induced astrogliosis in the spinal cord 18,19 − / − information.3,4 The earliest differentiated OPCs come from the and substantia nigra. Early studies with LXRβ mice first ventral cortex of the embryo and are gradually degraded and revealed the importance of LXRs in maintaining the intregrity of 20 disappear after birth,5,6 whereas the OPCs produced from the myelin sheaths. Later loss of LXRβ was found to exacerbate dorsal cortex continue to differentiate into OLs and are involved in demyelination in mouse experimental autoimmune encephalo- 21 myelination after birth.7,8 However, the pathway of generation of myelitis. Accordingly, as an important regulatory factor involved OPCs in the dorsal cortex has not yet been clarified. In the in the maintenance of RGCs, LXRβ may determine the choice developing CNS, radial glial cells (RGCs) have two functions: of differentiation of RGCs to astrocytes or OPCs, respectively, (1) they act as neural precursor cells to form new neurons in the and thereby affect myelination in the adult CNS. cortex; and (2) they are migrational guides for later-born neurons, Inthepresentstudy,weexaminedtheroleofLXRβ in the myeli- and thus are critical for morphogenesis of the cerebral cortex.9–11 nation and oligodendrogenesis in the corpus callosum. We report − / − During late neurogenesis, RGCs primarily transform into glia here that LXRβ animals exhibit behavioral abnormalities, delayed cells.10 Recently, it has been suggested that RGCs may also contri- myelination and abnormal myelin thickness. Furthermore, we show bute to the production of OPCs in the dorsal cortex,12 although that ablation of LXRβ negatively regulates OL production and the underlying mechanism remains unclear. typically inhibits RGC differentiation into OPCs. Thus, our study Liver X receptors (LXRs) α and β are ligand-activated nuclear reveals a previously uncharacterized role for LXRβ in myelination. receptors, active in the regulation of lipid metabolism.13,14 Analysis of the expression patterns of these receptor subtypes has shown that although LXRα is mainly expressed in tissues with MATERIALS AND METHODS high lipid metabolism such as liver, intestine, adipose tissue and Animals 15 macrophages, LXRβ is expressed ubiquitously. Furthermore, Generation of LXRβ− / − mice was described previously.22 LXRβ+/− female LXRβ has an important role in lamination of the cerebral cortex by mice were mated overnight with LXRβ+/ − males and inspected at

1Department of Histology and Embryology, Third Military Medical University, Chongqing, China; 2Southwest Eye Hospital, Southwest Hospital, Third Military Medical University, Chongqing, China; 3Institute of Immunology, PLA, Third Military Medical University, Chongqing, China; 4Department of Biology and Biochemistry, Center for Nuclear Receptors and Cell Signaling, University of Houston, Houston, TX, USA and 5Department of Biosciences and Nutrition, Karolinska Institute, Novum, Sweden. Correspondence: Professor J-A Gustafsson, Department of Biology and Biochemistry, Center for Nuclear Receptors and Cell Signaling, University of Houston, 3605 Cullen Boulevard, Science & Engineering Research Center Building 545, Houston 77054, TX, USA or Professor X Fan, Department of Histology and Embryology, Third Military Medical University, No. 30, Gaotanyan Street, Chongqing 400038, China. E-mail: [email protected] or [email protected] 6These authors contributed equally to this work. Received 24 March 2014; revised 23 April 2014; accepted 24 April 2014; published online 17 June 2014 Differentiation of radial glial cells PXuet al 948 0900 hours on the following day for the presence of vaginal plugs. Noon of Electron microscopy and morphometric analysis this day was assumed to correspond to E0.5. All animals were housed in Six-week- and three-month-old littermate mice were processed for elec- the animal facility of the Third Military Medical University in a controlled tron microscopy (three LXRβ− / − and three WT. Male mice were perfused environment on a 12-h light/12-h dark illumination schedule and were fed intracardially with 4% paraformaldehyde/2.5% glutaraldehyde in 0.1 M a standard pellet diet with water provided ad libitum. All experimental phosphate-buffered saline. Optic nerves and corpus callosum were cut procedures were performed in accordance with approved principles of into small cubes (1 mm) under a dissecting microscope, then postfixed laboratory animal care and ethical approval by the Third Military Medical overnight at 4 °C. Tissues were rinsed in phosphate-buffered saline, University. postfixed in 1% osmium tetroxide for 2 h, dehydrated in a graded series of ethyl alcohol, infiltrated with propylene oxide and embedded in Epon. Immunohistochemistry and Immunofluorescence Ultrathin sections (~60 nm thick) were generated by an ultramicrotome Brains were dissected and fixed in 4% paraformaldehyde for 24 h at 4 °C. (LKB-V, LKB Produkter AB, Bromma, Sweden) and counterstained with For paraffin sections, tissues were processed for paraffin embedding, and uranyl acetate and lead citrate. Sections were viewed with a transmission coronal sections (5 μm thick) were collected. For cryostat sections, brain electron microscope (TECNAI10, Philips, Eindhoven, The Netherlands). was postfixed in 30% sucrose solution with 4% paraformaldehyde at 4 °C, Digital images were acquired with an AMT XR-60 CCD Digital Camera and coronal cryosections (40 or 15 μm thick) were collected. Paraffin System (Advanced Microscopy Techniques, Danvers, MA, USA) and sections were used for LXRβ immunohistochemistry according to Fan compiled and analyzed using Adobe Photoshop and ImageJ (NIH). The et al.6 In brief, brain sections were deparaffinized in xylene and rehydrated g-ratio of myelinated axons was determined by dividing the axon diameter by the myelin diameter. A minimum of 344 axons was measured for the through graded ethanol. Antigens were retrieved by boiling in 10 mM citrate buffer (pH 6.0). Endogenous peroxidase was quenched with 0.3% corpus callosum of each genotype at each age. hydrogen peroxide in 50% methanol and nonspecific binding blocked with 1% bovine serum albumin. For LXRβ staining, slides were incubated with Behavioral tests − 1 β − − 0.15 units ml of -galactosidase for 2 h at room temperature. Thereafter, Male LXRβ / mice and their WT littermates (6 weeks old, 3, 7 and β sections were incubated with goat anti-LXR antibody (1:1000; made in 10 months old) were housed in a controlled environment (20–23 °C) with the Jan-Ake Gustafsson's laboratory, Karolinska Institute, Novum, Sweden) free access to food and water and maintained on a 12 h/12 h day/night at 4 °C overnight. After thorough washing, sections were incubated for cycle, light on at 0600 hours. Behavioral experiments were performed 2 h with biotinylated secondary antibody (Vector Laboratories, Burlingame, between 1000 and 1400 hours in an arrangement of 24 h in between CA, USA) at a 1:200 dilution in phosphate-buffered saline. The sections behavioral tests. The following assays were performed: open-field and were rinsed and incubated with avidin-biotinylated horseradish peroxidase plus-maze tests to measure motor exploration and anxiety, respectively. complex (Vectastain Elite ABC kit; Vector Laboratories) for 1 h. The For all behavioral experiments, investigators were blinded for LXRβ− / − or – ′ peroxidase substrate reaction was visualized by 3,3 -diaminobenzidine WT mice. After each experiment all the apparatuses were wiped clean to (DAKO, Carpenteria, CA, USA). The sections were slightly counterstained remove traces of the previous assay. with hematoxylin. Open field was measured using an open-field activity system (Biowill, μ Coronal cryosections, 40 m thick, were incubated overnight at 4 °C Shanghai, China) and activity software (Biowill). Mice were placed in the with the primary antibodies in 1% bovine serum albumin: goat polyclonal center of the open-field box, and activity was recorded for a period of α anti-MBP (1:200) and rabbit anti-PDGFR (1:200) (Santa Cruz Biotechnol- 10 min. The total and center-area distances were measured and the time in ogy, Santa Cruz, CA, USA), rabbit anti-Olig2 (1:1000) (Millipore, Temecula, the central area was recorded. CA, USA), and mouse anti-APC/CC-1 (1:100, Calbiochem, San Diego, CA, Elevated plus-maze test was used to measure anxiety levels of mice as – USA). Sections were then incubated with the avidin biotin complex, they avoid the open arms of the plus maze. Briefly, a mouse was initially followed by the biotinylated secondary antibodies (1:200; 2 h, 37 °C) and placed in the center area facing an open arm. The mouse’s entry into any fi ′ nally staining was visualized using the 3,3 -diaminobenzidine substrate of the four arms was counted when all four paws crossed from the central fl μ kit. For immuno uorescence, sections (15 m thick) were incubated with region into an arm. The number of total arm entries and the amount of α rabbit anti-PDGFR 1:200 and mouse anti-Olig2 (1:1000, Millipore), mouse time spent in the open arms during a 10-min testing period were recorded. anti-APC/CC-1 (1:100) and rabbit anti-Olig2 (1:1000), rabbit anti-BLBP (1:500, Millipore) and rat anti-PDGFRα (1:300, Millipore) or mouse anti-Olig2 (1:1000) overnight at 4 °C. Bovine serum albumin alone served as the Quantification of cholesterol in the cerebral cortex negative control. The secondary antibodies, Cy3- or 488-conjugated Cholesterol in the cerebral cortex was extracted by homogenization (both at 1:500, 3 h; Jackson ImmunoResearch, West Grove, PA, USA) with chloroform–Triton X-100 (1% Triton X-100 in pure chloroform) were then added (3 h, at room temperature), respectively. Sections in a microhomogenizer. The mixture was vortexed and centrifuged at were counterstained with 4',6-diamidino-2-phenylindole (Sigma-Aldrich, 16,000 g for 10 min at 4 °C. The lower organic phase was collected and air- St. Louis, MO, USA) and then viewed using the Leica TCS SP2 spectral dried at 50 °C, and chloroform traces were removed with N2 flow. The chole- confocal laser-scanning microscope (Leica, Wetzlar, Germany). sterol content was then determined with a cholesterol quantification kit (BioVision, Milpitas, CA, USA) according to the manufacturer's instructions. Western blot analysis The cerebral cortex was harvested from wild-type (WT) and LXRβ − / − mice at Cell culture, differentiation and factor treatments P2, P7, P10, P14 and P6W. The protein content was measured using a Cultures of radial glial clone L2.3 (a gift from Professor HD Li) have been bicinchoninic acid protein assay with bovine serum albumin as a standard. described previously.23 Briefly, culture medium contained DMEM/F12 Protein was mixed with 5 × loading buffer and then submitted for (Invitrogen, Carlsbad, CA, USA) supplemented with 25 mM glucose (Sigma), electrophoresis in a 12% SDS-polyacrylamide gel electrophoresis gel and 2mM glutamine (Invitrogen), penicillin/streptomycin (Invitrogen), electrically transferred onto a polyvinylidene difluoride transfer membrane. 10 ng ml− 1 FGF2 (BD Biosciences, San Jose, CA, USA), 2 μgml− 1 heparin Membranes were blocked in 5% fat-free milk and then incubated overnight (Sigma), and 1 × B27 (Invitrogen). Cells were propagated as neurospheres at 4 °C with the primary antibodies followed by peroxidase-conjugated and passaged by mild trypsinization (0.025% for 5 min) every 3 days. For secondary antibody labeling for 1 h at room temperature. The following differentiation,24 cells were cultured on laminin-coated coverslips in FGF2 primary antibodies were used: anti-β-actin (1:1000, Santa Cruz Biotechnol- containing serum-free medium for 1 day, then the medium was replaced ogy), anti-MBP (1:1000, Santa Cruz Biotechnology), anti-PDGFRα (1:1000, with DMEM/F12 plus N2 in the presence of FGF2, PDGF and forskolin for Santa Cruz Biotechnology), anti-Olig2 (1:1000, Chemicon, Temecula, 4 days, and the growth factors withdrawn in the presence of 3,3,5-tri- CA, USA) and anti-β-catenin (1:1000, Sigma, St Louis, MO, USA). Final iodothyronine hormone (T3) and ascorbic acid for another 4 days. visualization was achieved using an enhanced chemiluminescence western Treatments with or without the LXR agonist T0901317 (0.1 μM or 1 μM) blotting analysis system (Pierce, Rockford, IL, USA), and the signals were started on day 1 until the end of differentiation. exposed to X-ray films (Kodak, Rochester, NY, USA) and analyzed by the Cultured cells were fixed with 4% paraformaldehyde for 15 min at room Gel-Pro analyzer (Quantity One 4.0; Bio-Rad Laboratories, Hercules, CA, USA). temperature, washed three times with phosphate-buffered saline, followed Western blots were obtained from the cerebral cortex from three animals by incubation for 1 h at room temperature with the primary mouse anti-NG2 of each genotype and age. Data were averaged and represented as antibody (1/200; Chemicon), and rabbit anti-BLBP antibody (1/400). Next, means ± s.e.m. the cells were washed and incubated for 1 h with Cy3-conjugated goat

Molecular Psychiatry (2014), 947 – 957 © 2014 Macmillan Publishers Limited Differentiation of radial glial cells PXuet al 949 anti-mouse or 488-conjugated goat anti-rabbit antibody. Cells were counter- (Figures 1a, c, e, g and i) and the SVZ (Figures 1b, d, f, h and j) from stained with 4',6-diamidino-2-phenylindole, and coverslips were viewed P2 to adulthood. LXRβ expression was higher in the SVZ than that with the Carl Zeiss Axioplan microscope (Oberkochen, Germany). in the corpus callosum from P2 to P10 (Figures 1a–f). There was a similar level of LXRβ expression in the corpus callosum Statistical analysis and SVZ at P14 (Figures 1g and h) and P6W (Figures 1i and j), of β All data are expressed as the mean ± s.e. of the mean, and were analyzed which LXR expression was less at 6W compared with P14. This β by the Student’s t-test, or a one-way analysis of variance followed by expression pattern indicates that LXR might be related to Fisher’s protected least-significant difference post hoc test or a least- postnatal white matter development and myelination. significant difference multiple-comparison t-test. Significance was reached at values of Po0.05. Statistical analysis was performed using the Statistical Abnormality in the total locomotor activity of LXRβ KO mice Product and Service Solutions software V13.0 (SPSS, Chicago, IL, USA). We examined the pattern of free movement of mice at 6 weeks and 3 months of age using open-field equipment. In the open- fi β RESULTS eld test (Figure 2), the LXR knockout (KO) mice at 6 weeks of age showed significantly decreased total locomotor activity β LXR expression in the corpus callosum and ventricular zone (Po0.01), and a similar trend was confirmed for mice aged (VZ) /subventricular zone (SVZ) of mice from P2 to P6W 3 months (Po0.05). However, at 6 weeks and 3 months of age Immunohistochemistry with an anti-LXRβ antibody was used to loss of LXRβ in mice did not significantly affect the time or map the spatiotemporal expression pattern of LXRβ protein in the distance in the central area, indicating that the KO mice had no corpus callosum and SVZ in postnatal mouse brain, at P2, P7, P14 altered anxiety level. Abnormal exploratory behavior remained in and P6W. LXRβ expression was detected in both corpus callosum the KO mice at 7 and 10 months of age (Supplementary Figure S1).

Figure 1. Liver X receptor (LXR)β expression in the corpus callosum and subventricular zone (SVZ) of mice from P2 to P 6W. LXRβ expression was detected in both corpus callosum and the SVZ from P2 to P 6W. LXRβ expression was higher in the SVZ from P2 to P10 (a–f), which was of similar level in the corpus callosum and SVZ at P14 (g and h) and P6W (i and j). Scale bar: a–j,50μm. CC, corpus callosum; LV, lateral ventricle; P, postnatal day; SVZ, subventricular zone; VZ, ventricular zone; W, week.

Figure 2. An illustrative example of travel pathway in the open-ﬁeld test of a control and an liver X receptor (LXR)β knockout (KO) mouse at 6 weeks of age (a and b) and 3 months of age (f and g). LXRβ KO mice at 6 weeks of age traveled less distance overall (c) compared with wild- type (WT) controls (n = 11, **Po0.01), whereas there was no change either in the time (d) or distance traveled (e) in the center. Similarly, traveled distance overall (h) was decreased by loss of LXRβ in the mice of 3 months of age (n = 16, *Po0.05). Meanwhile, LXRβ KO mice and WT controls spent similar time (i) and traveled similar distances (j) in the center. NS, not signiﬁcant.

Figure 3. An illustrative example of the travel pathway on the elevated plus-maze assay of a control and an liver X receptor (LXR)β knockout (KO) mouse at 6 weeks of age (a and b) and 3 months of age (e and f). There is no alteration in the percentages of open-arm entries (c)or open-arm time (d) between wild-type (WT) and LXRβ KO mice at 6 weeks of age. By 3 months of age, percentage of open-arm entries of the LXRβ KO mice was decreased compared with WT mice (g,*Po0.05), although no alteration was observed in the percentage open-arm time (h). NS, not signiﬁcant.

Figure 4. Myelinated axons in 6-week-old and 3-month-old liver X receptor (LXR)β-deficient corpus callosum. (a–d) Corpus callosum prepared from 6-week-old wild-type (WT) (a and b) and LXRβ-deficient (c and d) mice were examined by electron microscopy. (e–h) Corpus callosum prepared from 3-month-old WT (e and f) and LXRβ-deficient (g and h) mice were examined by electron microscopy. Scale bar: a, c, e and g, 1 μm; b, d, f and h,1μm. Higher magnification of images demonstrate proper axonal myelination (b and f), and loosely wrapped or lack of myelin ensheathment (d and h) (examples denoted by asterisk). (i) The number of myelinated axons is significantly decreased in LXRβ knockout (KO) mice compared with WT controls (n=15 fields from at least 3 animals per genotype at each age were analyzed). *Po0.05, **Po0.01 by analysis of variance statistical analysis, followed by Tukey’s test. Error bars indicate +/ − s.e.m. (j) Graph shows that loss of LXRβ induced changes in G ratios compared with WT littermates at P6W and P3M (n = 3 brains for each age and each genotype; 344 axons analyzed (**Po0.01). M, month; W, week.

Figure 5. The expression of myelin basic protein (MBP) is reduced during postnatal development in liver X receptor (LXR)β knockout (KO) mice compared with littermate controls. At P10 (a–d), P14 (e–h) and P6W (i–l), MBP immunohistochemistry reveals a reduction in MBP expression in the corpus callosum and overlying cortex of LXRβ KO mice compared with their WT littermates. Scale bar: a–h, 100 μm; i–l, 200 μm. CC, corpus callosum; CTX, cortex; P, postnatal day; W, week. (m) Western blots reveal reductions in MBP in LXRβ KO mice compared with controls. (n) Graph represents changes in expressions of MBP in the cerebral cortical lysates indicating a reduction induced by loss of LXRβ (n = 3; brains for each age and each genotype; **Po0.01).

Figure 6. Maturation of oligodendrocytes is delayed during postnatal development in liver X receptor (LXR)β knockout (KO) mice compared with littermate controls. (a–d) At P6W, CC1 immunohistochemistry reveals a reduction of CC1-positive cells in the corpus callosum and overlying cortex of LXRβ KO mice (b and d) compared with their wild-typw (WT) littermates (a and c). (e) The mean cell density of CC1 per volume (mm3) of corpus callosum was determined for WT (white bar) and LXRβ KO (black bar) animals at P6W. (f–k) At P14, Olig2 and CC1 double-positive cells were decreased by loss of LXRβ.(l) The mean cell densities co-expressing CC1 and Olig2 per volume (mm3) of corpus callosum were determined for WT (white bar) and LXRβ KO (black bar) animals at P14. *Po0.05, **Po0.01 compared with WT. Scale bar: 100 μm(a and b); 50 μm(c, d and f–k). CC, corpus callosum; P, postnatal day; W, week.

Figure 7. The expressions of PDGFRα and Olig2 were reduced during postnatal development in liver X receptor (LXR)β knockout (KO) mice compared with littermate controls. (a–l) Immunohistochemistry on corpus callosum sections from P7, P10 and P14, with antibodies against PDGFRα (a–f) and Olig2 (g–l) as indicated. Oligodendrocyte progenitor cells (PDGFRα+)(m) and (Olig2+)(n) per mm3 in the corpus callosum in wild-type (WT) (white bars) and LXRβ KO (black bars) littermates at ages of P7, P10 and P14. Graphs are mean ( ± s.e.m.) counts obtained from four different areas of the corpus callosum (n = 3; brains for each age and each genotype; *Po0.05, **Po0.01). Western blot analysis showed a marked decrease in PDGFRα (o) and Olig2 (p) expression in the cerebral cortical lysates from LXRβ KO mice compared with WT from P7 to P14. Graphs represent changes in PDGFRα (q) and Olig2 (r) expression from P7 to P14 indicating a reduction in their level by loss of LXRβ in brains for each age and each genotype; *Po0.05, **Po0.01). Scale bar: 50 μm(a–f and g–l). CC, corpus callosum; P, postnatal day.

In the elevated plus-maze assay (Figure 3), a two-way analysis of ratio between the diameter of the axon proper and the outer variance revealed no difference between WT and LXRβ KO mice at diameter of the myelinated ﬁber) in the corpus callosum. At 3 months 6 weeks of age for either percentage open-arm entries or percentage of age, the demyelination of LXRβ KO mice (Figures 4g and h) in the open-arm time. By 3 months of age, the percentage of open-arm corpus callosum was more severe and the thickness of myelin entries for LXRβ KOmicewaslessthanthatinWTmice(Po0.05), surrounding axons was reduced compared with WT (Figures 4e and although no alteration was recorded for the percentage of open-arm f) resulting from a decrease in the number of wraps (Figures 4i and j). time. But, in mice 7 and 10 months of age, the percentage open-arm We also observed slight axonal swelling or degeneration in the LXRβ entries or percentage open-arm time was not different from those of KO mouse (Figures 4d and h). These changes were not limited to the WT mice (Supplementary Figure S1). These results suggest that the corpus callosum, as hypomyelination was also detected in the optic anxiety in LXRβ KOmiceat3monthsistemporary. nerve (Supplementary Figure S2). Hypomyelination in the optic nerve (Supplementary Figure S3) was conﬁrmed by the prolonged latency Electron micrographs reveal extensive hypomyelination in the of the P2 component of the Flash visual evoked potentials in the corpus callosum and optic nerve in LXRβ KO mice at ages of LXRβ KO mice. 6 weeks and 3 months To examine myelin sheaths in LXRβ KO mice, we analyzed the corpus Delay in the initiation of myelination and hypomyelination in LXRβ callosum and optic nerve by electron microscopy. At 6 weeks of age, KO mice LXRβ KO (Figures 4c and d) compared with WT mice (Figures 4a The level of the major CNS myelin component, myelin basic and b) had thinner myelin sheaths with higher g ratios (numerical protein (MBP), was examined in LXRβ KO mice. In 10-day-old

Figure 8. Effects of liver X receptor (LXR)β ablation on PDGFRα+/Olig2+ cell densities in the corpus callosum at P2. (a–d) PDGFRα (green) and Olig2 (red) expression in the corpus callosum (a–d) and ventricular zone/subventricular zone (VZ/SVZ) (e and f) of mice at P2 was examined by immunohistochemistry. Arrowheads indicate PDGFRα+/Olig2+ double-staining cells. The mean cell densities co-expressing PDGFRα and Olig2 per volume (mm3) of corpus callosum and SVZ were determined for wild-type (WT) (black bars) and LXRβ KO (white bars) animals at P2 (g). LXRβ KO mice had signiﬁcantly fewer PDGFRα+/Olig2+ double-staining cells in the corpus callosum compared with their WT littermates (n = 3, **Po0.01). Graphs depict mean ( ± s.e.m.) cell densities measured from four different areas of the corpus callosum and VZ/SVZ. Scale bar: a–f, 50 μm. (h) Western blot analysis showed that loss of LXRβ induced a decrease in PDGFRα and Olig2 expression, but increase in β-catenin expression in the cerebral cortical lysates compared with WT at P2.

(Figures 5a–d) and 14–day-old (Figures 5e–h) animals there Olig2 protein levels were consistently lower in cerebral cortical was a reduction in MBP level of LXRβ KO mice. Immunostaining lysates from LXRβ KO than in WT mice (Figures 7p and r). These of the brain from 6-week-old (Figures 5i–l) mice with anti-MBP results indicate that LXRβ is involved in the differentiation of antibody revealed widespread loss of fine MBP reactive fibers OPCs to mature OLs. in all layers at this late stage of myelination. Western blots Co-expression of Olig2 and PDGFRα is an indication of OPCs confirmed lower levels of MBP in LXRβ KO than in WT mice specification from Pre-OPCs or neural stem cells. We examined the (Figures 5m and n). densities of cells positive for both Olig2 and PDGFRα at P2, early postnatal development (Figure 8). Compared with WT mice Loss of LXRβ affects postnatal myelination in the corpus callosum (Figures 8a, c, e and g), there were fewer double-labeled cells α Myelin is generated by mature OLs that express adenomatous immunopositive for both Olig2 and PDGFR in the VZ/SVZ polyposis coli known as CC1 marker of postmitotic OLs. Quanti- (Figures 8e and f), medial (Figures 8a and c) and lateral (Figures β fication of CC1-positive cells at P6W showed that, compared with 8b and d) corpus callosum in the LXR KO mice (Figures 8b, d, f and g). β-catenin, which is involved in the specification of OPCs WT mice, there was a 41% (Figure 6e) decrease in CC1-labeled OLs 25,26 in corpus callosum from LXRβ-KO mice (Figures 6a–e). The number from neural stem cells, was upregulated in the cerebral β of CC1+/Olig2+ cells was reduced by 31% (Figure 6l) in multiple cortical lysates of LXR KO mice (Figure 8h). These results β fi regions, including corpus callosum of LXRβ KO mice at P14 constitute strong evidence that LXR determines OPCs speci ca- (Figures 6f–l), suggesting that lack of mature OLs could account tion through inhibition of β-catenin. for the hypomyelination in the LXRβ KO brain. LXRβ contributes to differentiation of RGCs into OPCs LXRβ regulates the number of OL progenitors in developing white RGCs, which have long been thought of as glia or glial progenitors, matter can also give rise to neurons. Brain lipid-binding protein (BLBP), Significantly fewer PDGFRα-labeled OPCs were found in the which is a reliable marker for RGC during development, combined developing corpus callosum of LXRβ − / − mice (Figures 7b, d and f) with OPC markers such as PDGFRα or Olig2, was used to study the compared with WT mice (Figures 7a, c and e) at P7 (Figures 7a effect of LXRβ on the differentiation of RGCs into OPCs. At P2 and b), P10 (Figures 7c and d) and P14 (Figures 7e and f). Western (Figures 9a–d) and P7 (Figures 9e–h), there was a large number of blots revealed that PDGFRα protein levels were consistently BLBP-labeled RGCs robustly co-expressing OPC surface marker lower in cerebral cortical lysates from LXRβ− / − than in WT mice PDGFRα in the VZ/SVZ (Figures 9c and d) and corpus callosum (Figures 7o and q). These findings indicate that LXRβ is essential (Figures 9a, b, e–h) of WT mice (Figures 9a, c, e and g), whereas the for maintaining the correct number of OPCs to populate the double-labeled BLBP+/PDGFRα+ cells were significantly reduced in developing corpus callosum. LXRβ KO mice (Figures 9b, d, f and h). The Olig2 transcription factor is a primary determinant of Early expression of Olig2 has a role in the initial specification of OL fate specification. Analysis of Olig2-positive cells reflected the the OL lineage, both in human27 and in animal models.28 We summation of changes in CC1-positive mature OLs and PDGFRα- confirmed that there was a large number of BLBP-labeled RGCs positive OPCs. Loss of LXRβ was accompanied by a significant co-expressed with Olig2 in the VZ/SVZ (Figure 10e) and corpus reduction of Olig2-positive cells compared with WT mice at P7, callosum (Figures 10a, c and k) of WT mice at P2, whereas the P10 and P14 (Figures 7g–l). Western blots revealed that double-labeled cells were significantly reduced in LXRβ KO mice

Figure 9. Effects of liver X receptor (LXR)β ablation on PDGFRα+/BLBP+ cell densities in the ventricular zone/subventricular zone (VZ/SVZ) and corpus callosum. PDGFRα (red) and BLBP (green) expressions in the corpus callosum (a,b and e–h) and VZ/SVZ (c and d) of mice at P2 (a–d) and P7 (e–h) were examined by immunohistochemistry. LXRβ knockout (KO) mice had signiﬁcantly fewer PDGFRα+/BLBP+ double-staining cells in the corpus callosum and VZ/SVZ compared with their wild-type (WT) littermates. Arrowheads indicate PDGFRα+/BLBP+ double- staining cells. Scale bar: 50 μm(a, b, c–f). P, postnatal day.

(Figures 10b, d, f and k). At P7, there were fewer double-stained significant upregulation in the proportion of NG2+ cells (51.6%) or BLBP +/Olig2+ cells in the corpus callosum of WT (Figures 10g, NG2+-positive cells from BLBP+cells (67.6%) was detected. i and k) and LXRβ KO mice (Figures 10h, j and k). DISCUSSION β Loss of LXR did not alter cholesterol contents in the cerebral Our previous studies have shown that male mice with deleted cortex LXRβ exhibit impaired performance on the rota-rod and that this There was no difference in cholesterol levels between WT and phenotype is associated with lipid accumulation and loss of motor LXRβ mice in the cerebral cortex from P10 to P6W, at the peak of neurons in the spinal cord, together with axonal atrophy and OL myelination/maturation (Figure 11). astrogliosis.18,19 Here, we further confirm that LXRβ− / − male mice display significant abnormal locomotor activity in the open-field LXR agonist promotes the differentiation of RGCs to OPCs test. Demyelination and hypomyelination in the corpus callosum of LXRβ KO mice were confirmed by electron microscopy and MBP When radial glial clone L2.3 cells were maintained in differentia- staining. Meanwhile, there was hypomyelination in the corpus μ tion medium with LXR agonist T0901317 (1 M) for 8 days, the callosum of LXRβ KO mice accompanied by an obvious defect in + percentage of NG2 (OPC surface marker) cells increased by 67% production and maturation of OLs. from 44.3 to 74% (Figures 12a, c and j; Po0.01), and percentage LXRs are known to have a key role in the homeostasis of NG2+ cells arising from BLBP+cells increased by 36% from of cholesterol,14 a major lipid constituent of the myelin sheaths 61.6 yo 83.6% (Figures 12d, f, g, i and k; Po0.05) in comparison and required for OL maturation and myelination.29,30 LXRβ is with non-treated cultures. With 0.1 μM T0901317 treatment, no expressed in a mouse OL cell line (158N) and in primary OLs

Figure 10. Effects of liver X receptor (LXR)β ablation on the BLBP+/Olig2+ cell densities in the corpus callosum of mice at P2 and P7. BLBP (green) and Olig2 (red) co-expression in the corpus callosum (a–d) and ventricular zone/subventricular zone (VZ/SVZ) (e and f) of mice at P2 and in the corpus callosum (g–j) at P7 was examined by immunohistochemistry. Arrowheads indicate BLBP+/Olig2+ double-staining cells. (k) The mean cell densities co-expressing BLBP and Olig2 per volume (mm3) of corpus callosum were determined for wild-type (WT) (white bars) and LXRβ knockout (KO) (black bars) animals at P2 and P7. LXRβ KO mice had signiﬁcantly fewer BLBP+/Olig2+ double-staining cells in the white matter compared with their WT littermates (n = 3, **Po0.01). Graphs depict mean ( ± s.e.m.) cell densities measured from four different areas of the corpus callosum. Scale bar: a–h,50μm. P, postnatal day.

isolated from neonatal and adult rats. It has a key role in the regulation of cholesterol metabolism in OLs.31,32 Deﬁciency in cholesterol synthesis by OLs leads to impaired myelination.32 Surprisingly, LXRβ KO mice exhibited defects in OL myelination/ maturation by P14 without alteration of cholesterol content in the cerebral cortex. This may infer that LXRβ affects early postnatal OL differentiation independent of cholesterol homeostasis. Indeed, generating or maintaining the correct number of OPCs is critical to populate OLs in the developing corpus callosum. PDGFRα-labeled OPCs were decreased in the corpus callosum of LXRβ KO mice from P7 to P14, strongly indicating that LXRβ is involved in the production of OPCs. It is important to note that postnatal regulation is critical for establishment of the number of OLs in the mature CNS. Spatially restricted Olig2 ablation leads to a nearly complete absence of myelination in the cortex at early postnatal stages and Figure 11. Measurement of cholesterol in the cerebral cortex of mice 33,34 from P10 to P6W. The cholesterol content of the brain was severe dysmyelination even at adulthood, suggesting that determined with a cholesterol quantiﬁcation kit (BioVision). (a) Data dorsal progenitor cells are a critical source for OL myelination in are expressed as micrograms of cholesterol per milligram of brain the developing cortex. It has also been demonstrated by the tissue. n = 4 per group; Student's t-test. P, postnatal day; W, week. anatomical fate-mapping strategy in postnatal mice that the

Figure 12. T0901317 promotes radial glial cell differentiation into oligodendrocyte progenitor cells (OPCs). (a–c) OPCs are labeled with anti- NG2 antibody 8 days after being cultured in differentiation medium. (d–i) NG2-positive cells are co-labeled with BLBP antibody 8 days after being cultured in differentiation medium; (d and g) without T0901317; (e and h) with T0901317 (0.1 μM); (f and i) with T0901317 (1 μM). (j) Quantiﬁcation of percentage of NG2+ cells (**Po0.01). (k) Quantiﬁcation of percentage of NG2+ cells in the BLBP+ cells (*Po0.05). Scale bar: a–i,50μm.

Figure 13. Schematic model of liver X receptor (LXR)β modulation of the differentiation of radial glial cells (RGCs) to oligodendrocyte progenitor cells (OPCs) during dorsal cortex oligodendrogenesis through interplay with β-catenin and Olig2. A subpopulation of RGCs in the ventricular zone/subventricular zone (VZ/SVZ), which are differentiated from the neuroepithelial cells, retain the capacity of neural stem cells in the neonate and divide asymmetrically to generate OPC and astrocyte precursor cells. (LXR)β has a key role in the transformation of RGCs to OPCs by repression of the expression of β-catenin and enhancing the expression of Olig2 transcription factors simultaneously. BLBP, brain lipid-binding protein; IOs, immature oligodendrocyte cells; MA, mantle; MBP, myelin basic protein; MOs, mature oligodendrocyte cells; MZ, marginal zone; NCs, neuroepithelial cells.

progeny of dorsal RGCs include OLs.35 In rodents, dorsal RGCs early postnatal development. Furthermore, in support of the have been implicated in generation of OPCs and OLs in the finding that an LXR agonist decreased β-catenin level in the postnatal cortex and subcortical white matter.36–38 In the fetal developing cerebellum,40 we confirmed upregulation of β-catenin brain, RGCs localized in the cortical VZ/SVZ are also involved in in the cortex at P2 by loss of LXRβ. These data suggest that the generation of OLs at a specific developmental stage, further inhibited oligodendrogenesis observed in LXRβ KO mice may confirmed by generation of OPCs from cultured isolated RGCs result from a dysregulation of Wnt/β-catenin signaling, which has under Shh influence.39 We have previously shown that LXRβ previously been shown to affect early postnatal OL differen- contributes to cerebral cortex lamination by maintenance of tiation.41,42 Thus LXRβ may interplay with β-catenin and Olig2 to RGCs.16 We noted that LXRβ was expressed in the cortical VZ/SVZ control differentiation of OPCs from RGCs. and corpus callosum with high level from P2 to P10, implying its Together, our findings indicate that LXRβ regulates oligo- possible role in oligodendrogenesis of the emerging white matter. dendrogenesis in the dorsal cerebral cortex, and may therefore This is supported by our finding of decreased densities of cells regulate the differentiation potential of newly born OPCs derived positive for BLBP+/PDGFRα+ and BLBP+/Olig2+ along the VZ and from RGCs, in addition to OL maturation/myelination (Figure 13). emerging corpus callosum at P2 and P7. Together with the Therefore, activation of LXRβ provides a new strategy to promote promotion of RGC differentiation into OPCs by LXR agonist, these OPC production, maturation and myelination. Moreover, similar to data suggest that LXRβ specifies development of RGCs to OPCs at our previous finding of an anxiogenic phenotype in female LXRβ

Molecular Psychiatry (2014), 947 – 957 © 2014 Macmillan Publishers Limited Differentiation of radial glial cells PXuet al 957 43 KO mice, signs of anxiety were also observed in the male LXRβ liver X receptor beta-/-mice, a model of adult neuron disease. J Neuropathol Exp KO mice at 3 months of age. From this we may infer that Neurol 2010; 69:593–605. impairment of OPC development in the emerging white matter 20 Wang L, Schuster GU, Hultenby K, Zhang Q, Andersson S, Gustafsson JÅ. Liver X postnatally could be involved in the pathology of psychiatric receptors in the central nervous system: from lipid homeostasis to neuronal diseases. degeneration. Proc Natl Acad Sci USA 2002; 99: 13878–13883. 21 Cui G, Qin X, Wu L, Zhang Y, Sheng X, Yu Q et al. Liver X receptor (LXR) mediates negative regulation of mouse and human Th17 differentiation. J Clin Invest. 2011; CONFLICT OF INTEREST 121:658–670. The authors declare no conflict of interest. 22 Alberti S, Schuster G, Parini P, Feltkamp D, Diczfalusy U, Rudling M et al. Hepatic cholesterol metabolism and resistance to dietary cholesterol in LXRβ- deficient mice. J Clin Invest 2001; 107: 565–573. ACKNOWLEDGMENTS 23 Li H, Babiarz J, Woodbury J, Kane-Goldsmith N, Grumet M. Spatiotemporal heterogeneity of CNS radial glial cells and their transition to restricted precursors. This study was supported by the National Nature Science Foundation of China Dev Biol 2004; 271:225–238. (No. 31271051 and 81371197), Natural Science Foundation Project of CQ CSTC 24 Markó K, Kohidi T, Hádinger N, Jelitai M, Mezo G, Madarász E. Isolation of radial 2013jjB10028, the Swedish Research Council and a grant from the Robert A Welch glia-like neural stem cells from fetal and adult mouse forebrain via selective Foundation (E-0004, J-ÅG). adhesion to a novel adhesive peptide-conjugate. PLoS One 2011; 6: e28538. 25 Chen BY, Wang X, Wang ZY, Wang YZ, Chen LW, Luo ZJ. Brain-derived neuro- REFERENCES trophic factor stimulates proliferation and differentiation of neural stem cells, possibly by triggering the Wnt/β-catenin signaling pathway. J Neurosci Res 2013; 1 Tkachev D, Mimmack ML, Ryan MM, Wayland M, Freeman T, Jones PB et al. 91:30–41. Oligodendrocyte dysfunction in schizophrenia and bipolar disorder. Lancet 2003; 26 Li Y, Lau WM, So KF, Tong Y, Shen J. Caveolin-1 inhibits oligodendroglial 362 – : 798 805. differentiation of neural stem/progenitor cells through modulating β-catenin 2 Walterfang M, Velakoulis D, Whitford TJ, Pantelis C. Understanding aberrant expression. Neurochem Int 2011; 59:114–121. white matter development in schizophrenia: an avenue for therapy? Expert Rev 27 Jakovcevski I, Zecevic N. Olig transcription factors are expressed in oligoden- 11 – Neurother 2011; :971 987. drocyte and neuronal cells in human fetal CNS. J Neurosci. 2005; 25: 10064–10073. 3 Emery B. Regulation of oligodendrocyte differentiation and myelination. Science 28 Zhou Q, Wang S, Anderson DJ. Identification of a novel family of oligodendrocyte 2010; 330: 779–782. lineage-specific basic helix-loop-helix transcription factors. Neuron 2000; 25: 4 Nava Ka. Myelination and support of axonal integrity by glia. Nature 2010; 468: 331–343. 244–252. 29 Saher G, Brügger B, Lappe-Siefke C, Möbius W, Tozawa R, Wehr MC et al. High 5 Rakic S, Zecevic N. Early oligodendrocyte progenitor cells in the human fetal cholesterol level is essential for myelin membrane growth. Nat Neurosci. 2005; 8: telencephalon. Glia 2003; 41: 117–127. 468–475. 6 Parras CM, Hunt C, Sugimori M, Nakafuku M, Rowitch D, Guillemot F. The pro- 30 Björkhem I, Meaney S. Brain cholesterol: long secret life behind a barrier. neural gene Mash1 specifies an early population of telencephalic oligoden- Arterioscler Thromb Vasc Biol 2004; 24: 806–815. drocytes. J Neurosci 2007; 27: 4233–4242. 31 Trousson A, Bernard S, Petit PX, Liere P, Pianos A, El Hadri K et al. 7 Kessaris N, Fogarty M, Iannarelli P, Grist M, Wegner M, Richardson WD. Competing 25-hydroxycholesterol provokes oligodendrocyte cell line apoptosis and stimu- waves of oligodendrocytes in the forebrain and postnatal elimination of an lates the secreted phospholipase A2 type IIA via LXR beta and PXR. J Neurochem embryonic lineage. Nat Neurosci 2006; 9:173–179. 2009; 109: 945–958. 8 Azim K, Raineteau O, Butt AM. Intraventricular injection of FGF-2 promotes 32 Nelissen K, Mulder M, Smets I, Timmermans S, Smeets K, Ameloot M et al. Liver X generation of oligodendrocyte-lineage cells in the postnatal and adult forebrain. receptors regulate cholesterol homeostasis in oligodendrocytes. J Neurosci Res Glia 2012; 60: 1977–1990. 2012; 90:60–71. 9 Franco SJ, Gil-Sanz C, Martinez-Garay I, Espinosa A, Harkins-Perry SR, Ramos C 33 Lu QR, Sun T, Zhu Z, Ma N, Garcia M, Stiles CD et al. Common developmental et al. Fate-restricted neural progenitors in the mammalian cerebral cortex. Science requirement for olig function indicates a motor neuron/oligodendrocyte 2012; 337: 746–749. 109 – 10 Malatesta P, Götz M. Radial glia—from boring cables to stem cell stars. connection. Cell 2002; :75 86. Development 2013; 140:483–486. 34 Yue T, Xian K, Hurlock E, Xin M, Kernie SG, Parada LF et al. A critical role for dorsal 26 – 11 Noctor SC, Flint AC, Weissman TA, Wong WS, Clinton BK, Kriegstein AR. Dividing progenitors in cortical myelination. J Neurosci 2006; : 1275 1280. precursor cells of the embryonic cortical ventricular zone have morphological 35 Ventura RE, Goldman JE. Dorsal radial glia generate olfactory bulb interneurons in 27 – and molecular characteristics of radial glia. J Neurosci 2002; 22: 3161–3173. the postnatal murine brain. J Neurosci 2007; : 4297 4302. 12 Kriegstein A, Alvarez-Buylla A. The glial nature of embryonic and adult neural 36 Guo C, Eckler MJ, McKenna WL, McKinsey GL, Rubenstein JL, Chen B. Fezf2 fi stem cells. Annu Rev Neurosci 2009; 32:149–184. expression identi es a multipotent progenitor for neocortical projection neurons, 80 – 13 Apfel R, Benbrook D, Lernhardt E, Ortiz MA, Salbert G, Pfahl M. A novel orphan astrocytes, and oligodendrocytes. Neuron 2013; :1167 1174. receptor specific for a subset of thyroid hormone-responsive elements and its 37 Relucio J, Menezes MJ, Miyagoe-Suzuki Y, Takeda S, Colognato H. Laminin interaction with the retinoid/thyroid hormone receptor subfamily. Mol Cell Biol regulates postnatal oligodendrocyte production by promoting oligodendrocyte 60 – 1994; 14: 7025–7035. progenitor survival in the subventricular zone. Glia 2012; : 1451 1467. fi 14 Jakobsson T, Treuter E, Gustafsson JÅ, Steffensen KR. Liver X receptor biology and 38 Tekki-Kessaris N, Woodruff R, Hall AC, Gaf eld W, Kimura S, Stiles CD et al. pharmacology: new pathways, challenges and opportunities. Trends Pharmacol Sci Hedgehog-dependent oligodendrocyte lineage specification in the tele- 2012; 33: 394–404. ncephalon. Development 2001; 128: 2545–2554. 15 Annicotte JS, Schoonjans K, Auwerx J. Expression of the liver X receptor α and β in 39 Mo Z, Zecevic N. Human fetal radial glia cells generate oligodendrocytes in vitro. embryonic and adult mice. Anat Rec A Discov Mole Cell Evol Biol 2004; 277: Glia 2009; 57:490–498. 312–316. 40 Yang Y, Tang Y, Xing Y, Zhao M, Bao X, Sun D et al. Activation of liver X receptor is 16 Fan X, Kim HJ, Bouton D, Warner M, Gustafsson JÅ. Expression of liver X receptor β protective against ethanol-induced developmental impairment of Bergmann glia is essential for formation of superficial cortical layers and migration of later-born and Purkinje neurons in the mouse cerebellum. Mol Neurobiol 2014; 49:176–186. neurons. Proc Natl Acad Sci USA 2008; 105: 13445–13450. 41 Fancy SP, Harrington EP, Yuen TJ, Silbereis JC, Zhao C, Baranzini SE et al. Axin2 17 Xing Y, Fan X, Ying D. Liver X receptor agonist treatment promotes the migration as regulatory and therapeutic target in newborn brain injury and remyelination. of granule neurons during cerebellar development. J Neurochem 2010; 115: Nat Neurosci 2011; 14: 1009–1016. 1486–1494. 42 Ye F, Chen Y, Hoang T, Montgomery RL, Zhao XH, Bu H et al. HDAC1 and HDAC2 18 Andersson S, Gustafsson N, Warner M, Gustafsson JÅ. Inactivation of liver regulate oligodendrocyte differentiation by disrupting the beta-catenin-TCF X receptor β leads to adult-onset motor neuron degeneration in male mice. interaction. Nat Neurosci 2009; 12:829–838. Proc Natl Acad Sci USA 2005; 102: 3857–3862. 43 Tan XJ, Dai YB, Wu WF, Warner M, Gustafsson JÅ. Anxiety in liver X receptor β 19 Bigini P, Steffensen KR, Ferrario A, Diomede L, Ferrara G, Barbera S et al. knockout female mice with loss of glutamic acid decarboxylase in ventromedial Neuropathologic and biochemical changes during disease progression in prefrontal cortex. Proc Natl Acad Sci USA 2012; 109:7493–7498.

Supplementary Information accompanies the paper on the Molecular Psychiatry website (http://www.nature.com/mp)