miR-23 regulation of lamin B1 is crucial for oligodendrocyte development and myelination

Lamins are major components of the nuclear envelope and are essential for maintaining nuclear integrity, gene expression and many other functions (Broers et al., 2006). Lamins can be categorized into two subfamilies (A and B types) that are encoded by different genes (LMNA and LMNB). LMNA mutations have been linked to a variety of diseases such as muscular dystrophy, cardiomyopathy, lipodystrophy and progeria (Capell and Collins, 2006). To date, autosomal dominant leukodystrophy (ADLD) is the only human disease that has been linked to an LMNB1 mutation (Padiath et al., 2006). In these patients, elevated levels of the lamin B1 transcript and protein result from a duplication of the LMNB1 gene. Post-mortem examination of the brains from ADLD patients showed severe myelin loss, although the axons in the white matter lesions were relatively spared from demyelination (Coffeen et al., 2000). In contrast to multiple sclerosis, oligodendrocytes are preserved in ADLD and no signs of inflammatory infiltrate can be detected. Nerve conduction velocity studies showed no evidence of demyelinating features in the peripheral nervous system suggesting that LMNB1 duplications preferentially lead to myelin loss in the central nervous system (CNS).

Lamins provide anchorage sites for heterochromatin and thus epigenetically regulate transcription (Cohen et al., 2001). Specific mutations in LMNA that lead to familial partial lipodystrophy or those that cause Hutchison-Gilford progeria syndrome cause aberrant localization of heterochromatin protein 1 in the perinuclear area, altered histone modification and mislocalization of nuclear pore complexes (Scaffidi and Misteli, 2006; Shumaker et al., 2006). It is therefore suggested that LMNA mutations cause these disorders by altering the epigenetic regulation of gene transcription and perturbing trafficking across the nuclear envelope. However, little is known about whether overexpression of nuclear lamin can lead to alterations in the organization of nuclear envelope components and of chromatin structure. In contrast to manifestations of LMNA mutations in diverse tissues (Muchir and Worman, 2004), CNS demyelination is the only recognized defect associated with the LMNB1 mutation (Padiath et al., 2006). Dominant-negative lamin B expression disrupts spindle assembly during mitosis (Tsai et al., 2006), suggesting that lamin B functions actively in modulating mitotic organization. Homozygous truncated Lmnb1 in mice leads to defective lung and bone development and neonatal lethality, indicating the involvement of lamin B1 in development (Vergnes et al., 2004). Although a regulatory function of lamin B1 has been demonstrated in cell mitosis and embryonic organogenesis, little is known about its role in the developing CNS. The demyelination in ADLD is accompanied by preservation of axons and oligodendrocytes and by decreased numbers of astrocytes with abnormal morphology suggesting differential susceptibility of cell types to lamin B1 overexpression. However, the connection between lamin B1 and glial development is unclear.

Recent progress in understanding small noncoding RNAs has led to the identification of their regulatory roles in many biological functions (Kloosterman and Plasterk, 2006). MicroRNAs (miRNAs) have been implicated in normal physiological processes and diseases (Stefani and Slack, 2008). In the developing CNS, many miRNAs show a distinct expression pattern (Landgraf et al., 2007) supporting the idea that they might play important roles during mammalian brain development, particularly in cell type differentiation in the CNS (Johnston and Hobert, 2003). Here, we investigated the mechanism by which elevated LMNB1 gene dosage leads to myelin loss in ADLD. We show that increased lamin B1 expression results in disturbances in nuclear envelope organization and nuclear pore transport. We demonstrate that the expression level of lamin B1 is crucial in determining the progress of oligodendrocyte maturation and myelin formation, and therefore uncover a novel function for a nuclear structural protein. We also identified miR-23 as a negative regulator of lamin B1 that can counteract the defects caused by increased lamin B1 dosage. This finding allows us to place miR-23 and lamin B1 as new components in the regulatory networks of oligodendroglia biology and of myelin sheath formation and maintenance in vitro.

Lamin B1 is ubiquitously expressed in various tissues (Broers et al., 1997), different parts of the brain (Padiath et al., 2006) and in both neurons and glia (Fig. 1A). It interacts with nuclear and integral membrane proteins including lamina-associated polypeptide (LAP2) and lamin B receptor (LBR) (Dreger et al., 2002). In order to examine the effects of LMNB1 overexpression on the integrity of nuclear membrane proteins, ectopic overexpression of LMNB1 was performed in neuronal (C17.2), astrocytic (SVG p12) and oligodendrocytic (N20.1) cell lines. Abnormal nuclear morphology such as extensive folding, blebbing and lobulation in the nuclear envelope was observed in all three cell types when LMNB1 was overexpressed. This abnormality did not lead to significant changes in other nuclear membrane components in C17.2 and SVG cells (data not shown). Western blot analysis revealed that LMNB1 (but not LMNA or LMNB2) overexpression leads to a reduction of LAP2 levels in N20.1 cells (Fig. 1B) and disrupts its co-localization with LMNB1 in the nuclear envelope (Fig. 1C). LBR co-localization and protein levels were not affected by LMNB1 overexpression. In addition, no abnormality was observed for Emerin and MAN1 under these conditions (data not shown). These results showed that LMNB1 overexpression affects the subcellular localization and protein levels of LAP2 in oligodendrocytes.

Since abnormalities in nuclear morphology were induced by LMNB1 overexpression, we examined its effect on chromatin organization. In fibroblasts overexpressing LMNB1, immunocytochemical analysis revealed that the localization of heterochromatin protein 1 β(HP1β) and methylated histone 3 (K3H9) were disrupted (Fig. 1D). The HP1βdistribution in the nuclei of these cells was more distant from the nuclear envelope than in control cells, indicating altered chromatin organization.

Nuclear lamins interact directly with heterochromatin and regulate DNA synthesis and transcription. In addition, LAP2 is involved in regulating gene expression (Nili et al., 2001). We therefore sought to determine whether LMNB1 overexpression affects transcription in a cell-type-specific manner in the CNS. Luciferase reporter analysis revealed that LMNB1 overexpression did not change the transcription from the neurogenic differentiation 1 (NeuroD) (neuron specific) promoter (Fig. 2A) but increased transcription from the GFAP (astrocyte specific) promoter (Fig. 2B). In addition, LMNB1 overexpression significantly repressed transcription from the oligodendrocyte-specific promoters of myelin basic protein (MBP), proteolipid protein (PLP) and myelin oligodendrocyte glycoprotein (MOG) (Fig. 2C–E). Therefore, LMNB1 overexpression exhibits direct effects on transcriptional regulation in an oligodendrocyte- and astrocyte-specific manner.

Lamins have been reported to associate with the nuclear pore proteins and affect their recruitment and maintenance in the nuclear pore complex (Smythe et al., 2000). To assess nuclear transport following LMNB1 overexpression, we compared relative amounts of glutathione S-transferase (GST)-GFP proteins that were fused to either a nuclear localization sequence (NLS) derived from the simian virus 40 (SV40) T antigen or a nuclear export sequence (NES) derived from HIV Rev. The GST-GFP fusion proteins were large enough in size that they could not freely diffuse through the nuclear pore. GST-GFP fusion proteins containing the NLS (NLS-GFP) or the NLS plus NES (NLS-GFP-NES) were introduced into HEK293 cells with or without a cDNA construct expressing LMNB1 or LMNB2. The cytoplasmic and nuclear levels of GST-GFP were then assessed by western blot analysis. NLS-GFP accumulated equally in the cytoplasm and nuclei of cells co-transfected with either a vector or the cDNA construct expressing LMNB1 or LMNB2 (Fig. 3A,B). In contrast, NLS-GFP-NES protein accumulated in higher levels in the nuclei of cells co-transfected with the cDNA construct expressing LMNB1 compared with in the nuclei of cells transfected with either the vector or the cDNA construct expressing LMNB2 (Fig. 3B). NLS-GFP-NES protein levels in the cytoplasmic fractions were similar for all three constructs (Fig. 3A). To further examine whether the increased accumulation of NLS-GFP-NES protein in nuclei was the result of enhanced nuclear import and/or impaired nuclear export following LMNB1 overexpression, the cells were treated with leptomycin B (LMB, an inhibitor of nuclear export in HIV Rev). In the presence of LMB, nuclear NLS-GFP-NES protein levels were similar in cells co-transfected with the vector control or with the cDNA construct expressing LMNB1 or LMNB2, suggesting that LMNB1 overexpression impairs nuclear export. Furthermore, western blot analysis of whole cell lysates showed that LMNB1 overexpression also resulted in reduction of Nup153 (Fig. 3C), a nucleoporin that interacts with lamin B (Smythe et al., 2000). Taken together, these results demonstrate that LMNB1 overexpression impairs nuclear-cytoplasmic shuttling and leads to reduced Nup153 levels.

Although lamin B1 is widely expressed in various parts of the brain (Allen Brain Atlas, www.brain-map.org), its temporal expression pattern has not been studied. In addition, ADLD is an adult-onset disease; patients have no symptoms until the forth or fifth decades of life. We set out to examine the temporal expression profile of lamin B1 during mouse brain development using quantitative RT-PCR and western blotting. mRNA and protein levels of lamin B1 peak at birth or postnatal day 1, followed by a gradual decrease from postnatal day 1 to 10 months of age (Fig. 4A,B). Therefore, lamin B1 is developmentally regulated. This is consistent with the previous finding that Lmnb1 mRNA levels are gradually downregulated during oligodendrocyte maturation in vitro (Dugas et al., 2006). Interestingly, the expression pattern of Lmnb1 shows a reverse correlation with other myelin-specific proteins [CNP, MBP and MAG (myelin-associated glycoprotein)] (Fig. 4B), suggesting a possible role for Lmnb1 in the regulation of oligodendrocyte development.

miRNAs play essential roles in the development of various organisms from nematodes to mammals. To test whether miRNAs play a role in the developmental regulation of lamin B1, we sought to identify miRNAs that target mouse Lmnb1 mRNA. The 3′ untranslated region (UTR) of Lmnb1 mRNA was scanned for potential miRNA binding sites. Five sites were chosen because of their conservation in human, chimpanzee, mouse, rat, dog and chicken genomes (supplementary material Fig. S1A). The Lmnb1 3′UTR was fused to the luciferase reporter in an expression plasmid before co-transfection into HEK293 cells with each of these miRNAs to examine their repressive effect. Transfection of miR-23a and miR-23b resulted in significantly reduced luciferase activities (Fig. 5A). Western blot was used to reveal that endogenous LMNB1 levels were reduced by miR-23a and miR-23b (Fig. 5B) in HEK293 cells. No change in LMNB1 mRNA levels was noted (data not shown). In addition, knocking down endogenous miR-23 with synthetic antisense oligonucleotides resulted in increased LMNB1 levels (Fig. 5C). Further characterization of the Lmnb1 3′UTR revealed that there are three separate miR-23 binding sites within this region (see supplementary material Fig. S1 and Table S1). The coding sequences of the GFP-Lmnb1 fusion protein were then fused to either a wild-type Lmnb1 3′UTR or a Lmnb1 3′UTR with mutated miR-23 sites to examine the accessibility of miR-23 to the 3′UTR of Lmnb1 mRNA. Introduction of these GFP-Lmnb1 plasmids into HEK293 cells led to nuclear envelope localization (data not shown). When the wild-type 3′UTR construct was transfected, addition of miR-23a or miR-23b into the system resulted in a significant reduction in the GFP-Lmnb1 levels (Fig. 5D). The GFP-Lmnb1 level was not affected in cells transfected with the mutated 3′UTR construct, indicating that the miR-23-binding sites in Lmnb1 3′UTR are authentic. We next examined the temporal expression pattern of miR-23 in the mouse brain. Northern blotting revealed that miR-23a and miR-23b are similarly expressed in the brain and that their levels gradually increase with age in a pattern that is complimentary to that of Lmnb1 (Fig. 5E,F). Taken together, these results indicate a role for miR-23 in downregulating Lmnb1 expression, especially in adulthood.

Overexpression of LMNB1 or miR-23 in glial cultures was performed to test the roles of lamin B1 and miR-23 in glial development. To achieve miR-23 or LMNB1 overexpression, we used a lentiviral vector that permits efficient expression of exogenous miR-23 or LMNB1 in mouse primary glial cultures. Overexpression of miR-23 led to significantly increased numbers of cells with positive immunoreactivity to CNP (Fig. 6A,B) and MBP (data not shown). This increase was accompanied by increased levels of CNP (Fig. 6C) and MBP (data not shown) suggesting that miR-23 causes enhanced oligodendrogenesis. In contrast, LMNB1 overexpression caused a reduction in the numbers of CNP- or MBP-positive cells and decreased CNP and MBP protein levels, implying a repressive effect of LMNB1 on oligodendrogenesis. Neither miR-23 nor LMNB1 overexpression exerted significant effects on GFAP expression in mixed glial cultures. These results indicate that oligodendrocyte development is specifically suppressed by lamin B1 and enhanced by miR-23. miR-23 and LMNB1 were also individually expressed in cultured astrocytes or oligodendrocytes that had been purified from mouse brains. Consistent with previous results, miR-23 enhanced, and LMNB1 repressed, oligodendrocyte differentiation in oligodendrocyte-enriched cultures. Neither miR-23 nor LMNB1 overexpression caused any alteration in GFAP immunoreactivity in astrocytic glial cultures (data not shown).

Regulation of oligodendrocyte development is a complex biological process that involves intricate transcriptional regulation and many signaling pathways among neurons, astrocytes and oligodendrocytes. Astrocytes are very abnormal in postmortem ADLD brains – one possibility is that oligodendrocyte differentiation could be regulated by factors secreted by astrocytes. To address this possibility, conditioned media from astrocytes overexpressing miR-23 or LMNB1 was used to culture mixed glial cells. This media did not affect oligodendrocyte differentiation in primary cultures (as assessed by immunocytochemistry and Western blot) (supplementary material Fig. S2), indicating that the observed differential effect of miR-23 and LMNB1 on cultured oligodendrocytes was not mediated by soluble factors secreted from astrocytes but by the direct effects of miR-23 and LMNB1. Indeed, immunocytochemical analysis in purified oligodendrocyte progenitor cells (OPC) recapitulates our previous findings in mixed glial cultures (Fig. 6D). We concluded that miR-23 acts as a positive regulator of oligodendrocyte differentiation whereas LMNB1 is a suppressor of oligodendrocyte maturation, raising the possibility that the finely regulated expression of LMNB1 might be important in the control of oligodendrocyte development and myelin formation in vivo.

The process of OPC differentiation into myelinating oligodendrocytes occurs in distinct temporal stages. To gain more insight into whether the lamin B1-mediated inhibitory effect on oligodendrogenesis is specific to certain developmental periods, immunocytochemical experiments were carried out with stage-specific markers. These antigenic characterizations were conducted in OPC cultures with proliferation (+PDGF) or differentiation (+T3) medium. LMNB1 overexpression did not alter the pattern of NG2 (neural-glial antigen 2) staining but did cause decreases both in the number of branched processes visualized by galactocerebroside (GalC) and in CNP immunoreactivity (Fig. 7A). In control cells, the presence of MBP and PLP was distributed in the cell bodies and distal cellular processes, whereas in LMNB1-overexpressing cells the majority of MBP was found in the cell nuclei and PLP was localized to perinuclear regions and proximal portions of processes. The repressive effect of LMNB1 overexpression was also reflected in the reduced levels of CNP and the myelin proteins MBP, PLP and MAG (Fig. 7B). MBP immunoreactivity was used to assess process branching using Sholl analysis, a quantitative method for radial distribution of dendritic morphology and complexity (Fig. 7C). There was a dramatic decrease in the number of intersections of processes in oligodendrocytes overexpressing LMNB1 compared with in control cells. To determine whether the aberrant localization of MBP and PLP was the result of a defect in process formation, immunocytochemistry was conducted using an antibody against MAG, which correlates with late stage myelinating oligodendrocytes. MAG was present in the distal branching processes of oligodendrocytes that overexpressed LMNB1 despite their reduced branching morphology versus control cells (Fig. 7A). These experiments were also carried out by overexpressing mouse Lmnb1 and similar results were obtained (data not shown). These data suggest that lamin B1 overexpression can alter the subcellular localization of MBP/PLP and reduce the branching of maturing oligodendrocytes.

To further validate our finding that lamin B1 can inhibit oligodendrocyte maturation in vivo, we obtained brain tissues from two ADLD patients. Protein extracts from these brain tissues were used in western blotting and the levels of oligodendrocyte- (CNP) and myelin- (MBP, PLP, MAG) specific proteins were determined with specific antibodies. In agreement with our in vitro finding, the levels of these proteins were all reduced in ADLD brain tissues when compared with control brain tissue (Fig. 7D). In addition, the levels of reduction were greater in tissues with higher lamin B1 overexpression (Fig. 7E), supporting a dosage effect on the oligodendrocyte/myelin inhibition by lamin B1.

Since miR-23 suppresses lamin B1 and exhibits positive regulation of oligodendrocyte maturation, we investigated whether miR-23 could rescue the oligodendrocyte maturation defects caused by lamin B1 overexpression. First, we set out to determine the dosage effect of miR-23 on Lmnb1 expression by transfecting HEK293 cells with various ratios of miR-23 and the construct expressing GFP-tagged Lmnb1. Western blot analysis revealed a dose-dependent repressive effect of miR-23 on Lmnb1 expression (Fig. 8A). Notably, at a dose ratio of 50:1, miR-23 reduced both the exogenous Lmnb1 (GFP-LB1) and endogenous LMNB1 (LB1) proteins. We then monitored the extent of oligodendrocyte differentiation under different dose combinations using anti-MBP immunocytochemistry. No alteration in MBP nuclear localization was observed when co-expressing miR-23 with the construct expressing GFP-Lmnb1 at a ratio of 1:1 compared with co-transfecting a vector control instead of miR-23 (Fig. 8B). At an expression ratio of 10:1, cellular processes were clearly visible but reduced in number. At a higher expression ratio of 50:1, a substantial (although not complete) rescue of MBP distribution in oligodendrocytes was found. Increasing amounts of miR-23 enhanced this rescue indicating that this is a dosage-dependent biological effect. In contrast, when constructs carrying Lmnb1 with mutated miR-23 binding sites were used, no rescue was found even at a dose ratio of 50:1 (Fig. 8C,D). In agreement with our previous observation, co-expression of miR-23 and the GFP-Lmnb1 construct carrying either the wild-type or the mutant Lmnb1 3′UTR did not alter the MAG immunoreactivity in distal oligodendrocyte processes. Together, these findings suggest that increased expression of lamin B1 causes premature arrest of oligodendrocyte maturation and that miR-23 can enhance oligodendrocyte development by suppressing lamin B1.

Lamin B1 is one of the structural proteins in the nuclear envelope. Low-level overexpression of structural proteins is usually tolerable and does not cause significant pathological consequences (Abe and Oshima, 1990). However, high levels of overexpression can result in detrimental cellular effects such as overloaded protein trafficking or degradation systems (Garbern, 2007). Increased gene dosage accompanied by excessive gene products has been reported in association with many neurodegenerative diseases (Singleton et al., 2003; Rovelet-Lecrux et al., 2006). Excessive LMNB1 production owing to LMNB1 duplication is associated with the CNS leukodystrophy phenotype in ADLD patients, demonstrating that myelin genesis and/or maintenance is sensitive to lamin B1 levels (Padiath et al., 2006). In this study, we showed that LMNB1 overexpression caused abnormal nuclear envelope morphology and altered the expression and distribution of LAP2. Overexpression of B-type lamins has been shown to promote nuclear membrane growth and intranuclear membrane formation in amphibian oocytes and epithelia and in mammalian kidney cells in a CaaX-motif-dependent manner (Prufert et al., 2004; Ralle et al., 2004). In these systems, expression of different lamins at moderate levels induced nuclear envelope growth whereas expression at higher levels led to the formation of intranuclear membranes. Examination by electron microscopy showed that these intranuclear membranes were not continuous with the inner nuclear membrane and were devoid of pore complexes. In agreement with these findings, our ectopic overexpression of LMNB1 in established neuronal and glial cells increased the surface area of the nuclear membrane and the number of intranuclear aggregates. One possibility is that LMNB1 overexpression, through formation of these intranuclear membranes, leads to the altered subcellular localization of LAP2 and perturbed nuclear transport. It is noteworthy that in the primary culture system, lentivirus induces moderate overexpression of LMNB1 in oligodendrocytes leading to differentiation defects and drastic effects on myelin protein expression accompanied by minimal nuclear envelope distortion and growth when compared with established culture systems.

Loss of function owing to truncated Lmnb1 leads to postnatal lethality with defective lung and bone development in mice (Vergnes et al., 2004), however it is not known whether this truncation is associated with any neurological deficits in the brain. Organization of the nuclear genome in relation to the nuclear lamina plays an important role in the regulation of gene expression (Taddei, 2007). Fibroblasts cultured from Lmnb1 knockout mice exhibit drastic changes in the transcription level of 498 genes (Malhas et al., 2007). Moreover, Drosophila lamin B has been found to interact directly with about 500 genes that are clustered in the genome and are developmentally expressed (Pickersgill et al., 2006). The alteration in chromatin organization that we observed here could also be attributed to the aberrant intranuclear membrane formation caused by LMNB1 overexpression (Prufert et al., 2004; Ralle et al., 2004). These changes might then lead to alterations in transcriptional regulation and DNA replication (Somech et al., 2007). Cells with different specifications have distinct distribution signatures of chromosome territories and pericentromeric heterochromatin (Mayer et al., 2005). Distinct gene expression patterns have been demonstrated in neurons, astrocytes and oligodendrocytes despite their identical genomic structure (Cahoy et al., 2008). Indeed, the specific effects of LMNB1 overexpression on gene transcription were confirmed by the repression of myelin-specific genes and the activation of GFAP transcription. These results suggest that overexpression of LMNB1 could disturb the unique expression patterns in individual CNS cell types and that phenotypes would only appear in certain cell types that are vulnerable to the radical transcriptional changes.

Despite the ubiquitous spatial expression in neuronal and glial cell types, the temporal expression of lamin B1 is regulated such that the pattern is complementary to that of the major oligodendrocyte and myelin proteins (including CNP, MBP and MAG). Lamin B1 overexpression exerts inhibitory effect on the genesis of oligodendrocytes. In contrast, ectopic expression of miR-23, a negative regulator of lamin B1, can enhance oligodendrogenesis. These results established miR-23 and lamin B1 as putative regulators in oligodendroglia development and myelin sheath formation. To date, accumulating reports provide evidence that miRNAs function in the regulation of many physiological processes such as neuronal development, cell-fate specification, differentiation and synaptogenesis (Schratt et al., 2006; Visvanathan et al., 2007). In the presence of excess miR-23, a greater proportion of cells express mature oligodendrocyte markers and increased levels of mature myelin proteins together with having a multipolar morphological appearance, indicating that miR-23 can enhance oligodendrogenesis. In contrast, excessive lamin B1 leads to lower numbers of cells expressing mature markers with reduced levels of mature myelin proteins, suggesting defective differentiation of oligodendrocytes. These in vitro findings will need to be validated in an in vivo system. One piece of evidence supporting the in vivo relevance is that we observed reduced levels of oligodendrocyte- and myelin-specific proteins in ADLD brain tissues. It is possible that miR-23 can also enhance oligodendrocyte development through other lamin B1-independent pathways. In this case, excessive lamin B1 production in the cells might sequester miR-23 during maturation, thereby further adding to the deteriorating myelin loss that results from lamin B1 overexpression in the CNS. Since many transcription factors and environmental cues are involved in the complex control mechanisms of oligodendrogenesis and myelin sheath formation, identification of possible, additional downstream targets of miR-23 in oligodendrogenesis pathways might provide new insight into the mechanisms of oligodendrocyte development, myelin formation and maintenance.

In oligodendrocytes, increased LMNB1 gene dosage caused premature arrest of differentiation with the principal phenotype being a lack of MBP and PLP at the cell surface. The absence of MBP and PLP induced by lamin B1 overexpression is specific since MAG, a myelin protein sharing the same biosynthetic pathway as PLP, was present on the cell surface of oligodendrocytes. The decreased appearance of MBP and PLP at branching processes could be caused by lower protein levels owing to suppressed transcription. It is also possible that defective nuclear export contributes to this phenomenon. MBP is severely affected by lamin B1 overexpression and the lack of functional MBP results in a lack of myelin in mouse brain tissue. Re-introducing MBP into this mutant mouse improved the degree of myelin assembly and major dense line formation (Popko et al., 1987; Readhead et al., 1987). Because MBP can be transported between the cytoplasm and the nucleus (Pedraza et al., 1997), and because of its distribution in the oligodendrocyte nuclei, soma and myelin sheathes, it might have a regulatory role in gene transcription and oligodendrocyte processing during myelination (Verity and Campagnoni, 1988). The regulatory role of MBP is supported by an increased level of MBP in the nuclei and cytoplasm of oligodendrocytes undergoing myelinogenesis or myelin maintenance; in addition, plasmalemmal MBP occurs in quiescent oligodendrocytes (Hardy et al., 1996). To a lesser extent, lamin B1 overexpression also disrupted the appearance of PLP in distal processes of oligodendrocytes. PLP functions in maintaining the adhesion and stabilization of the extracellular surfaces of the myelin sheath (Klugmann et al., 1997). Interestingly, PLP is overexpressed in classical Pelizaeus-Merzbacher disease (PMD) and, instead of reaching the cell surface, mutated PLP is aberrantly retained in the endoplasmic reticulum, Golgi apparatus or nuclear envelope in X-linked spastic paraplegia (Koizume et al., 2006). These mutations induce in vivo hypomyelination, arrest oligodendrocyte maturation and lead to death of oligodendrocytes (Kagawa et al., 1994; Readhead et al., 1994). Despite the similarity in both the aberrant subcellular localization of PLP and the developmental defects in oligodendroglia, lamin B1 overexpression does not result in the demise of oligodendrocytes, suggesting that the myelin defects induced by lamin B1 overexpression are fundamentally distinct from those caused by mutations in the myelin genes. The appearance of normal numbers of NG2 cells, which are morphologically normal, suggests that lamin B1 overexpression does not affect cell specification and patterning during the early development of oligodendrocytes. Our results indicate that the developmental defects induced by lamin B1 overexpression occur mainly during terminal differentiation. It is possible that a reduction in MBP and PLP assembled into the myelin membrane can make the sheath less compact and more unstable, therefore leading to accelerated myelin breakdown which is reflected by the observation of demyelination, without death of the oligodendrocytes, in the brains of ADLD patients.

We showed that the defects induced by lamin B1 overexpression can be reversed by miR-23 in a dosage-dependent manner demonstrating that lamin B1 has a threshold effect on oligodendrocyte development and myelin production. It is not surprising that a threshold effect can apply to normal myelin proteins since dysmyelination and demyelination occur with increased or decreased dosage of myelin proteins such as PLP and PMP22 (Lupski et al., 1991; Inoue et al., 1996). Unexpectedly, the ubiquitously expressed nuclear lamina protein, lamin B1, potently influences the amount and quality of myelin formation in the CNS. miR-23 is important for the process of transitioning of oligodendrocyte progenitors into mature oligodendrocytes, where it acts, at least in part, by antagonizing lamin B1 levels. Clearly, further exploration of the mechanisms that link miR-23 and lamin B1 to oligodendrocyte maturation and myelin maintenance might provide insights into new therapeutics for ADLD and other demyelinating disorders like multiple sclerosis.

The cDNA encoding mouse Lmnb1 was amplified by polymerase chain reaction (PCR) from mouse brain total cDNA and was subcloned into pCMV tag2 (Stratagene), pEGFP-C3 (Clontech) and pSicoR (Ventura et al., 2004). The 3′UTR from mouse Lmnb1 was cloned into the pRL-TK vector (Promega) and mutations were introduced into the wild-type Lmnb1 3′UTR by PCR. microRNA precursors with a 50-80 base pair flanking sequence on each site were amplified by PCR from mouse genomic DNA (Promega) and then inserted into the pSuper (OligoEngine) and the pSicoR lentiviral vectors. All constructs used in this study were verified by sequencing (Genomics Core Facility, UCSF). GFP fusion lamin B1 (Daigle et al., 2001), lamin A (Ostlund et al., 2001) and lamin B2 constructs (BC 006551) (Open BioSystems) were made in pEGFP-C3 and GFP-GST vectors containing NLS, or NLS and NES (Walther et al., 2003). Anti-miR analysis of microRNA knockdown was performed using single-stranded RNA oligonucleotides designed to inhibit miR-23a and miR-23b (Dharmacon).

HEK293, C17.2 and N20.1 cells were grown as described previously (Snyder et al., 1992; Verity et al., 1993). The mouse fibroblasts were isolated from adult LMNB1-overexpressing transgenic mice (unpublished data) following standard procedures (Xu et al., 2007). Primary glial cultures were performed using standard methods (McCarthy and de Vellis, 1980; Armstrong et al., 1992; Dugas et al., 2006). Plasmid or single-stranded RNA transfection was performed by using FuGene HD (Roche) or nucleofector electroporation with the Amaxa system (Amaxa Biosystems).

Lentiviruses were generated essentially as described previously (Ventura et al., 2004). Lentiviral vectors and each packaging vector were co-transfected into HEK293T cells. Supernatants were collected 36-48 hours after transfection and were passed through a 0.45 μm filter. The viral supernatant was centrifuged at 100,000 g for 1.5 hours. The viral pellet was resuspended in PBS and incubated overnight at 37°C with the cultured primary cells. Two to four rounds of infection were performed 24 hours apart.

Luciferase and β-galactosidase activities in the cell extracts were assayed 48 hours after transfection as described previously (Zhao et al., 2005). pGL3-promoter or pSV-βGAL expression constructs were co-transfected to normalize for transfection efficiency. NeuroD-Luc, MBP-Luc, PLP-Luc, MOG-Luc (PCR cloning from −657~+80) or GFAP-Luc were used in this study.

RNA was isolated from cells and mouse brain using Trizol (Invitrogen) and microRNA northern blotting was performed as described previously (Zhao et al., 2005). Quantitative PCR was performed using a RotorGene RG3000 real-time PCR system (Corbett Research) with the SYBR-green-containing PCR kit (BioRad). All PCR fragments were sequenced to validate the specificity.

Mice were perfused with 4% paraformaldehyde and the brains were postfixed overnight at 4°C. Cryosections (20 μm) were permeabilized with 0.3% Triton X-100 in PBS and then blocked with 20% normal goat serum. Cells grown on coverslips were fixed in methanol at −20°C or 4% paraformaldehyde in PBS at room temperature. Cells were then permeabilized with 0.1% Triton X-100 in PBS followed by blocking with 5% nonfat dry milk. These cells or brain sections were subjected to primary antibody followed by Cy2- or Cy3-conjugated antibodies. Photographs were taken on a Zeiss Pascal confocal microscope with a 63′ oil-immersion objective.

Cell lysates were loaded onto 10% SDS-PAGE gels and were then transferred onto PVDF membrane (Bio-Rad). Blots were blocked with 5% nonfat dry milk followed by incubation with adequate primary antibody in 2.5% nonfat dry milk. HRP-conjugated secondary antibody was then applied to the blots and bands were visualized using an ECL chemiluminescence kit (Amersham) and autoradiograph (Pierce X-OMAT film).

NIH ImageJ software with Sholl analysis plugin was performed in oligodendrocytes following MBP immunocytochemistry. Traced cells were analyzed by centering nested spheres on the cell body with each spheres spaced 10 μm apart (from 10 to 100 μm). The complexity of oligodendrocyte morphology was presented by counting how many times the branches intersect with the circumference of these circles.

Data were presented as mean±s.e.m. and data comparison was undertaken using the Student’s t-test or one-way ANOVA with Newman-Keuls post hoc test. The significant difference was set at P<0.05 unless otherwise stated.

The primers for quantitative real-time PCR were as follows: Lmnb1 F: 5′-CAGGAATTGGAGGACATGCT-3′ and R: 5′-GAAGGGCTTGGAGAGAGCTT-3′ (40 cycles of 95°C for 10 seconds, 59°C for 10 seconds, 72°C for 20 seconds; detection temperature at 82°C).

GAPDH F: 5′-AACTTTGGCATTGTGGAAGG-3′ and R: 5′-ACACATTGGGGGTAGGAACA-3′ (40 cycles of 95°C for 10 seconds, 60°C for 10 seconds, 72°C for 15 seconds; detection temperature at 82°C).

Oligodeoxynucleotides used as northern probes were as follows: miR-23a, 5′-GGAAATCCCTGGCAATGTGAT-3′; miR-23b, 5′-GGTAATCCCTGGCAATGTGAT-3′; U6, 5′-GCAGGGGCCAT-GCTAATCTTCTCTGTATCG-3′.

Monoclonal or polyclonal antibodies used in immunohistochemistry and immunocytochemistry were as follows: CNP at a dilution of 1:250 (Abcam), GalC at 1:200 (Chemicon), GFAP at 1:500 (Chemicon), GFP at 1:1000 (Abcam), HP1βat 1:500 (Chemicon), trimethyl-histone H3 (Lys4) at 1:250 (Upstate), tri-methyl-histone H3 (Lys9) at 1:250 (Upstate), Lamin B1 at 1:500 (Abcam), LAP2 at 1:1000 (BD bioscience), LBR at 1:1000 (Dreger et al., 2002), MAG at 1:250 (Chemicon), MBP at 1:500 (Chemicon), NeuN at 1:250 (Chemicon), NG2 at 1:400 (Chemicon) and PLP at 1:100 (Abcam).

Monoclonal or polyclonal antibodies used in western blotting were as follows: CNP at a dilution of 1:1000 (Abcam), Flag M2 at 1:5000 (Sigma), GAPDH at 1:5000 (Chemicon), GFAP at 1:3000 (Chemicon), GFP at 1:5000 (Abcam), HDAC1 at 1:5000 (ABR), LAP2 at 1:1000 (BD bioscience), Lamin B1 at 1:1000 (Abcam), MBP at 1:1000 (Santa Cruz), LBR at 1:5000 (Dr Harald Herrmann), MAG at 1:500 (Zymed), NeuN at 1:1000 (Chemicon) and Nup153 at 1:1000 (Abcam).

We thank E. Synder for the C17.2 cell line; A. Campagnoni for the N20.1 cell line; H. Worman for the lamin A cDNA construct; J. Ellenberg for the GFP-lamin B1 (human) construct; M.-J. Tsai for the NeuroD-luciferase construct; R. Miskimins for the MBP-luciferase construct; S. Goebbels and K. Nave for the PLP-luciferase construct; I. Rozovsky for the GFAP-luciferase construct; E. Atlas and R. Haché for the NLS-GFP and NLS-GFP-NES constructs; H. Herrmann for the LBR antibody; K. Luo for the MAN1 antibody; Q. Padiath for protein extracts of ADLD brains; D. Srivastava for microRNA targeting predictions; and L. Ptá3ek and S. Pleasure for careful reading of the manuscript. We thank members of the Fu and Ptá3ek labs for discussions and technical assistance. We especially thank B. Barres for helping us to implement the various primary culturing methods from his laboratory. This work was supported, in part, by the Sandler Neurogenetics fund (to Y.-H.F.).