Abstract
Oligodendrocyte differentiation and myelination are tightly regulated processes orchestrated by a complex transcriptional network. Two bHLH transcription factors in this network, Olig1 and Olig2, are expressed exclusively by oligodendrocytes after late embryonic development. Although the role of Olig2 in the lineage is well established, the role of Olig1 is still unclear. The current studies analyzed the function of Olig1 in oligodendrocyte differentiation and developmental myelination in brain. Both oligodendrocyte progenitor cell commitment and oligodendrocyte differentiation were impaired in the corpus callosum of Olig1-null mice, resulting in hypomyelination throughout adulthood in the brain. As seen in previous studies with this mouse line, although there was an early myelination deficit in the spinal cord, essentially full recovery with normal spinal cord myelination was seen. Intriguingly, this regional difference may be partially attributed to compensatory upregulation of Olig2 protein expression in the spinal cord after Olig1 deletion, which is not seen in brain. The current study demonstrates a unique role for Olig1 in promoting oligodendrocyte progenitor cell commitment, differentiation, and subsequent myelination primarily in brain, but not spinal cord.
Introduction
Myelin ensheathment of axons dramatically increases axonal conduction velocity and provides valuable trophic support to axons (Bullock et al., 1984; Hildebrand et al., 1993; Nave, 2010). In demyelinating diseases, such as multiple sclerosis (MS), remyelination occurs, but the myelin is thinner than normal and many axons remain demyelinated (Franklin and Ffrench-Constant, 2008). Oligodendrocyte progenitor cells (OPCs) and premyelinating cells are present in MS lesions, but in general, remyelination is greatly impaired in MS tissue (Chang et al., 2000, 2002; Kuhlmann et al., 2008). Thus, understanding the molecular mechanisms driving oligodendrocyte maturation and myelination remains an extremely important research question.
Over the past decade, much of the transcriptional regulatory network of oligodendrocyte lineage progression has been elucidated (Emery, 2010a). Two transcription factors, Olig1 and Olig2, are closely related basic helix-loop-helix (bHLH) transcription factors that play critical roles in oligodendrocyte specification and differentiation (Lu et al., 2002; Zhou and Anderson, 2002). Although it appears to have an essential role in remyelination (Arnett et al., 2004), the function of Olig1 during oligodendrocyte development is still unclear. This results from the fact that several Olig1-null lines have been developed that have quite different phenotypes (Lu et al., 2002; Xin et al., 2005; Paes de Faria et al., 2014). Oligodendrocyte differentiation in developing spinal cord is delayed in the original Olig1-null mouse, yet there is no long-term developmental problem (Lu et al., 2002). It was hypothesized that this mild phenotype might result from compensation by Olig2 (Lu et al., 2002), and a second Olig1-null mouse was designed to reduce any Olig2 compensation. This animal has greater dysmyelination and dies by postnatal day (P)14 (Xin et al., 2005). Two additional Olig1-null mouse lines were recently described, which were generated by different strategies, and only mild developmental delay of spinal cord myelination is found in these mice (Paes de Faria et al., 2014).
In ongoing studies using the original Olig1-Cre mice for gene deletion in oligodendrocytes, we noted unexpected changes in myelination in the brains of homozygous-null Olig1-Cre mice. In general, the studies on the different Olig1-null mouse lines have focused on spinal cord development. However, differential regulation of oligodendrocyte differentiation and myelination has been seen in different brain regions (Fruttiger et al., 1999; Sperber et al., 2001; Bercury et al., 2014; Wahl et al., 2014). We therefore began investigating the role of Olig1 in oligodendrocyte development outside the spinal cord. These studies demonstrate that in the brain, Olig1 has a critical role in the commitment of pre-OPCs to OPCs and for subsequent OPC differentiation. The data from the current study demonstrate a unique role of Olig1 in promoting OPC commitment and differentiation, which appears more important in brain than spinal cord OPCs.
Materials and Methods
Animals.
The B6;129S4-Olig1tm1(cre)Rth/J strain of Olig1-null mice was obtained from The Jackson Laboratory (stock no. 011105). The original mixed background mouse strain has been backcrossed to C57BL/6 mice over numerous generations during the last 12 years. The total number of oligodendrocyte lineage cells and mature cells was quantified in Olig1+/− and wild-type mice at both P15 and 1 month of age and no differences were observed between them. Therefore, we used male and female Olig1+/− and Olig1−/− littermates for further analysis. Genotypes of all mice were determined by PCR analysis of tail genomic DNA based on the suggested Jackson Laboratory protocol with modified PCR primers. PCR primers specific for wild-type alleles were as follows: Olig1-WT-F, 5′-AAACGCTGCGCCCCACCAAG-3′, and Olig1-WT-R, 5′-TCACTTGGAGAACTGGGCCT-3′; for mutant alleles, they were as follows: Olig1-Mut-F, 5′-CGCCCCAGATGTACTATGC-3′, and Olig1-Mut-R, 5′-AATCGCGAACATCTTCAGGT-3′. All animal procedures were performed in an AAALAC-accredited facility in accordance with the Guide for the Care and Use of Laboratory Animals and approved by the University of Colorado Denver Institutional Animal Care and Use Committee.
Immunohistochemistry, immunocytochemistry, and TUNEL assay.
Mouse perfusion and immunohistochemistry was performed as described previously (Trapp et al., 1997), with some modifications. Free-floating cortex and cervical spinal cord sections (30 μm) were analyzed, with antigen retrieval in 10 mm sodium citrate (pH 6.0) at 65°C for 10 min as needed, using a Pelco Biowave Pro tissue processor (Ted Pella). For immunocytochemistry, oligodendrocytes were cultured on coverslips (see Primary cell culture and electroporation, below) and fixed with 4% paraformaldehyde for 15 min at RT. Cells were permeabilized with 0.1% Triton X-100 for 10 min, blocked with 3% BSA in PBS for 60 min at RT, and incubated with primary antibodies overnight at 4°C. For detection of O4 cell surface antigens, O4 antibody was diluted with media and incubated with live cells on coverslips for 1 h before fixation.
Cell death was analyzed by TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling) assay. Sections were permeabilized with 3% Triton X-100 for 30 min, and labeled with in situ cell death detection kit following the manufacturer's instructions (Roche Applied Science, no. 11684795910).
The following primary antibodies were used: guinea pig anti-NG2 (gift from Dr W. Stallcup, Burnham Institute, La Jolla, CA), rabbit anti-Olig2 and Olig1 (a gift from Dr Charles Stiles, Harvard University, Cambridge, MA), rat anti-PLP/DM20 (Clone AA3), O4 hybridoma (gift from Dr Rashmi Bansal, University of Connecticut Health Sciences Center, Farmington, CT), rat anti-BrdU (Accurate Chemical, no. YSRTMCA2060GA), rabbit anti-Sox2 (Millipore, no. ab5603), goat-anti-Sox2 (Santa Cruz Biotechnology, no. sc-17320), mouse anti-Olig2 (Millipore, no. MABN50A4), rabbit anti-Ki67 (Abcam, no. 16667), mouse anti-CC1 (Millipore, no. OP80), goat anti-Sox10 (Santa Cruz Biotechnology, no. sc-17342), rabbit anti-PDGFRα (Santa Cruz Biotechnology, no. sc-338), chicken anti-neurofilament (Neuromics, no. CH22105), and rabbit anti-MBP (Millipore, no. ab980).
Primary cell culture and electroporation.
Mouse neural progenitor cells were isolated from Olig1 heterozygous or null neocortex E12.5–E14.5 embryos to generate neurospheres and OPCs as previously described (Pedraza et al., 2008). Rat mixed glial cultures were generated from P0 to P3-d-old Sprague-Dawley rat pups as described previously (Dai et al., 2014). Oligodendrocyte cultures were typically >90% pure as assessed by immunocytochemistry for the oligodendrocyte lineage markers PDGFRα/NG2 and Olig2 and the astrocytic marker glial fibrillary acid protein. Rat OPCs were shaken from mixed cultures after 10 d and 5 × 106 cells were electroporated (Amaxa nucleofection apparatus, Lonza) in 100 μl Nucleofection solution (Amaxa basic glial cells nucleofector kit, Lonza, no. VPI-1006, IL) with siRNAs (10 μl of 20 μm rat Olig1 siRNAs (no. L100044-01) or siControl nontargeting siRNA pool (no. B002000-UB) from Dharmacon (Thermo Scientific). After electroporation, cells were resuspended and seeded onto poly-d-lysine/laminin-coated dishes or round 12 mm coverslips in DMEM supplemented with N2 (Life Technologies, no. 17502-048), Fibroblast growth factor (FGF, 10 ng/ml) and Platelet-derived growth factor (PDGF, 10 ng/ml) for 24 h, after which they were incubated in differentiation media for 1 d.
Cell counts and immunofluorescence quantification.
For oligodendrocyte lineage cell number quantification of Olig1 mutant mice, images at 40× magnification were obtained either at the midline of the corpus callosum or in the dorsal column of the cervical spinal cord on a Leica SP5 confocal microscope. Cells within the field of each image for the corpus callosum or spinal cord were counted with the ImageJ cell counting plugin. Three sections per animal were quantified from at least five animals per group. To quantify the fluorescence intensity of proteolipid protein (PLP) and myelin basic protein (MBP) immunoreactivity in corpus callosum, the corrected total cell fluorescence (CTCF) was calculated as the Integrated Density-(Area of selected cell × mean fluorescence of background readings; Burgess et al., 2010). For cultured primary oligodendrocytes, images were taken on a Zeiss Axio Imager M2. A total of six microscopic fields per coverslip with two coverslips per condition were sampled in each experiment. The total number of cells counted per condition averaged 700–1500. The total branch points/cell of O4+ cells was analyzed with Imaris filament tracer module (Bitplane).
Western blot.
Cultured cells, brain or spinal cord tissue were lysed in RIPA buffer (25 mm Tris-HCl, pH 7.5, 150 mm NaCl, 1 mm EDTA, 1% NP-40, 0.1% SDS, 0.1% DOC) supplemented with complete mini-protease inhibitor cocktail (Roche Applied Science) and phosphatase inhibitor cocktail set II (Calbiochem, no. 564652). Samples were analyzed by a standard Western blot protocol. Protein bands were detected using the LICOR Odyssey infrared scanner (LI-COR Bioscience) and proteins were quantified using the Odyssey Scanner Software 2.0. The following primary antibodies were used: rabbit anti-myelin associated glycoprotein (MAG; Cell Signaling Technology, no. 8043), rabbit anti-2′,3′-cyclic-nucleotide 3′-phosphodiesterase (CNPase; Cell Signaling Technology, no. 5664), rabbit anti-myelin oligodendrocyte glycoprotein (MOG; Abcam, no. 32760), mouse anti-MBP (Covance, no. SMI-94), mouse anti-β-tubulin (Sigma-Aldrich, no.T8328), rabbit anti-glyceraldhyde-3-phosphate dehydrogenase (Cell Signaling Technology, no. 2118). Samples were loaded at 50 μg protein/lane to detect myelin proteins in P15 cerebral samples and at 20 μg/lane for others. To detect Olig1 and Olig2, 100 μg protein/lane was loaded.
Electron microscopy.
Animals were perfused with cold PBS followed by modified Karnovsky's fixative (2% paraformaldehyde/2.5% glutaraldehyde). The brain and spinal cord were removed and postfixed overnight in the same fixative. Corpus callosum was isolated from 1 mm coronal slices of brain between −0.94 and −2.18 of bregma. All tissue was postfixed in 1% osmium tetroxide, dehydrated in graded acetone, and resin embedded in Embed 812 (Electron Microscopy Sciences) using a Pelco Biowave Pro tissue processor (Ted Pella). The corpus callosum samples were oriented such that sections could be cut midline in a sagittal plane. Ultrathin sections (80 nm) were mounted on copper grids, stained with uranyl acetate and lead citrate, and viewed at 80 kV on a Technai G2 transmission electron microscope (FEI). Electron micrographs of the corpus callosum were imaged at midline.
Statistical analysis.
For cell counts, the mean number of immunoreactive cells per image field was determined. Student's t test, paired or unpaired, was used to calculate statistical significance and graphed using OriginPro (Origin lab) or Graphpad Prism. Values represent the mean ± SEM.
Results
OPC production, differentiation, and early morphogenesis were reduced in Olig1-null mice
The impact of Olig1 on OPC specification and number in the corpus callosum during the early postnatal period was investigated. In Olig1-null mice at P1, there was a 40% reduction in the total number of oligodendrocyte lineage cells (Olig2+) in subcortical white matter compared with control (Fig. 1A–C). This reduction is consistent with a recent study that showed reduced oligodendrocyte lineage cells in Olig1-null mice and ectopic production of interneurons (Silbereis et al., 2014). Within the Olig2+ cell population in the Olig1-null mice, the percentage that were NG2+ OPCs was only approximately half that in control animals (Fig. 1A,B,D). Thus, compared with control P1 animals, where 61.9% of Olig2+ cells are NG2+, intriguingly, in the Olig1-null mice, only 29.5% of the Olig2+ cells were also NG2+ (Fig. 1D). To identify the Olig2+/NG2− cell population, we analyzed expression of a pre-OPC marker, Sox2. Sox2 is expressed in the subventricular zone (SVZ) by neural progenitors and by pre-OPCs (Olig2+/Sox2+ cells) migrating out of the SVZ (Ellis et al., 2004; Menn et al., 2006; Hu et al., 2009b; Wang et al., 2013). Once pre-OPCs reached the subcortical white matter, they reduced Sox2 expression and transitioned into Olig2+/NG2+ OPCs (Fig. 1E,F). At P1, this reduction in Sox2 expression was still in progress. Thus, there were cells with high Sox2 expression, presumably pre-OPCs, and cells with low Sox2, presumably cells beginning to differentiate to OPCs. The total number of Sox2+ cells was reduced in Olig1-null mice (Fig. 1G), which may reflect the reduction in total oligodendrocyte lineage cells in Olig1-null mice that occurs at early specification stages (Silbereis et al., 2014). The pre-OPC population was quantified as the number of Olig2+/Sox2+/NG2− cells in control or Olig1-null mice, including both high and low-expressing Sox2+ cells. Interestingly, there was a significant increase in the percentage of Olig2+ cells that were Olig2+/Sox2+/NG2− pre-OPCs in Olig1-null mice, relative to control mice (Fig. 1G). Furthermore, as can be seen in Figure 1E,F, whereas Sox2 levels were very low in NG2+ cells in control samples (Fig. 1E,H, merged image; E′, Sox2 image, compare arrowheads to arrows), in Olig1-null cells, numerous NG2+ cells retained strong Sox2 stain (Fig. 1F,H, merged image, arrowheads). This apparent delay/deficit in the downregulation of Sox2 and the transition from pre-OPC to OPC in Olig1-null mice suggest an important role of Olig1 in this very early stage of oligodendrocyte development, a stage that has not been investigated extensively in vivo previously.
Despite being a smaller population, the Olig2+ cells in Olig1 null mice were highly proliferative (Fig. 2A,B). P1 pups were injected with BrdU and killed 2 h later; tissue was analyzed for BrdU incorporation coupled with Ki67 immunohistochemistry. BrdU permanently labels cells during S phase, and their progeny retain the label, whereas Ki67 is a marker of proliferating cells currently in cell cycle (Kee et al., 2002). The number of Olig2+ cells that were either BrdU+ or Ki67+ was quantified in the subcortical white matter. The percentage of Olig2+ cells that were BrdU+ or Ki67+ in P1 Olig1-null mice was essentially twice that of control (Fig. 2C). Both the pre-OPCs (Olig2+/NG2−) and the OPCs (Olig2+/NG2+) had far more Ki67 expression (52.7% and 83.2%, respectively) than control cells (18.6% and 59.2% respectively; Fig. 2D–G). These data suggested that a significant segment of the Olig2+ cell population remained in the cell cycle in Olig1-null mice at P1.
Despite their increased proliferation, the number of Olig2+ cells remained lower in Olig1-null mice, and it seemed possible that the reduced number of cells resulted from increased cell death (Reid et al., 1999; Castedo et al., 2004). Indeed, cell death essentially doubled in the Olig1-null mice (Fig. 2H–J). Although many dying cells had lost their cell-specific markers, a number of these apoptotic cells in Olig1-null mice expressed NG2 proteoglycan, which was only rarely seen in control mice (Fig. 2K,L, arrows). Thus, the increased cell death in Olig1-null mice was likely in OPCs that failed to differentiate properly. These data indicate that OPCs were delayed in exiting cell cycle in Olig1-null mice, and the number of both proliferative (Ki67+) and dying cells (TUNEL+) in the Olig1-null mice was twice that in control mice. Thus, it appears that in the absence of Olig1, Olig2+ cells failed to effectively exit cell cycle and continued to proliferate, resulting in increased cell death.
Finally, it must be noted that the NG2+ cells in Olig1-null mice were morphologically far simpler than in control mice, with fewer processes (Fig. 2D′,E′), suggesting a role for Olig1 in early OPC differentiation. In summary, these findings provide evidence that Olig1 is important for pre-OPC to OPC commitment and subsequent OPC differentiation and morphogenesis.
Oligodendrocyte differentiation was reduced, but not the initiation of axon wrapping in P8 Olig1-null mice
We next investigated the impact of Olig1 loss on the initial stages of myelination in P8 mice. At this age, myelination had already begun in the striatum and lateral corpus callosum of Olig1 control mice (Fig. 3A), but it was significantly reduced in Olig1-null mice (Fig. 3B). The tissue was stained for Sox10, PDGFRα and CC1, and as at earlier ages, there were 43% fewer oligodendrocyte lineage cells (Sox 10+) in Olig1-null mice, with a smaller reduction in the number of PDGFRα+ OPCs and a greater reduction in the number of CC1+ mature cells (Fig. 3C,D,E,F). Thus, as a percentage of the total Sox10+ oligodendrocyte lineage cells, the progenitor cell population was increased by 27%, while the differentiated cell population was reduced by 32% (Fig. 3F), which suggested that, in addition to the delay in cell cycle exit, loss of Olig1 also delayed differentiation of the progenitor cells.
As OPCs differentiate, they move through a premyelinating stage to the myelinating stage (Trapp et al., 1997). It was important to assess whether the reduction of CC1+ cells resulted from an inability to begin to differentiate, or rather an inability to interact with axons and start myelination. At P8, a higher percentage of the Sox10+ cells remained as PDGFRα+ progenitor cells, and fewer were premyelinating PLP+ cells (Fig. 3G,H). Thus, PLP+ premyelinating oligodendrocytes only accounted for 7.2% of total Sox10+ cells in Olig1-null mice, far less than the 20.6% seen in control tissue (Fig. 3I). Thus, the initiation of differentiation from OPC to premyelinating oligodendrocyte was impaired in these mice.
As noted above, by P8, myelination was occurring rapidly in the subcortical white matter of control animals, but this was rarely seen in Olig1-null mice, where only limited myelination was seen in the lateral corpus callosum (Fig. 3A,B). This likely resulted only from the reduced numbers of CC1+/Sox10+ mature oligodendrocytes, because those that did differentiate had no deficit in initial axonal contact. Thus, the initial axonal contact, which appeared to involve PLP-expressing membranes (Fig. 3J′,K′, arrowhead), and further wrapping and segment elongation, which involved membranes containing both PLP and MBP (Fig. 3J″,J‴,K″,K‴, arrowheads), were relatively normal in Olig1-null (Fig. 3K), compared with control littermates (Fig. 3J). Indeed, as these cells continued to myelinate, they appeared relatively comparable to control cells (Fig. 3L,M). These data indicate that, in addition to a role of Olig1 in early OPC cell cycle exit, the loss of Olig1 impacted terminal differentiation of oligodendrocytes, but for cells that had terminally differentiated, Olig1 was not essential for myelination initiation per se.
Oligodendrocyte deficit persists in the brain of adult Olig1-null mice
An important question was whether the reduced oligodendrocyte number and impaired differentiation in Olig1-null corpus callosum eventually recovered as the animals matured. Oligodendrocyte maturation and myelination were therefore investigated in 2 month Olig1-null corpus callosum. Consistent with the early developmental stages, the number of Sox10+ cells at the midline of the corpus callosum of Olig1-null mice was reduced by ∼60% relative to control. This reduction resulted primarily from a reduction in mature oligodendrocytes, since there was a 68% reduction of the total number of CC1+ mature oligodendrocytes in Olig1-null mice, whereas the PDGFRα+ OPC numbers were not different from those of control littermates (Fig. 4A–C). Within the oligodendrocyte population in Olig1-null mice, the percentage of mature oligodendrocytes was reduced ∼18% in Olig1-null mice compared with control, whereas the population that remained OPCs doubled (Fig. 4D). This sharp reduction of mature oligodendrocytes had a profound impact on myelination in Olig1-null mice. Immunofluorescent quantification of the major myelin proteins PLP and MBP indicated that both PLP and MBP were decreased by 60% in Olig1-null mice, compared with control (Fig. 4E,G,I), and there appeared to be far more unmyelinated axons in Olig1-null mice, in contrast to the intensively stained myelin in control samples (Fig. 4F,H).
To establish whether there were in fact more unmyelinated axons and to further assess the impact of Olig1 loss, samples were analyzed by electron microscopy, and consistent with the immunofluorescence data, the number of unmyelinated axons in the Olig1-null animal was increased approximately twofold (Fig. 5A,B,F), and the number of myelinated axons was reduced by 47% in midline corpus callosum of Olig1-null mice relative to control mice (Fig. 5A,B,E). Furthermore, this deficit remained in 6- and 15-month-old Olig1-null mice (data not shown), excluding the possibility that over time there was long-term recovery. However, as with the data indicating that the few cells that terminally differentiated in Olig1-null mice had no problem beginning myelination (Fig. 3J,K), the g-ratios for those axons that were myelinated were essentially normal (Fig. 6C,D). Despite the persistent hypomyelination in the Olig1-null mouse, no axonal degeneration at the ultrastructural level was observed (Fig. 6B) and the total number of axons was not statistically different compared with control (Fig. 6G). In addition, axonal degeneration was further evaluated by SMI32 antibody staining of the nonphosphorylated form of neurofilament, and no significant accumulation of SMI32 was observed in Olig1-null mice at 2 months of age (data not shown).
Western blot studies confirmed the overall reduction in myelin in Olig1-null brain. Consistent with the immunofluorescent studies, total myelin protein was reduced in the Olig1-null cerebrum. Myelin proteins were quantified by Western blot, and at P15, there was already reduced myelin in Olig1-null cerebrum (Fig. 6A,C), which persisted in 2-month-old Olig1-null mice (Fig. 6B,D).
Developmental recovery and normal myelination in the spinal cord of adult Olig1-null mice
The original report of these Olig1-null mice indicated that loss of Olig1 did not result in significant dysmyelination in embryonic spinal cord (Lu et al., 2002). To confirm that these mice retained that phenotype, we analyzed oligodendrocyte development and myelination in spinal cord. By P8, the majority of the axons both in the gray and white matter of cervical spinal cord in control mice were myelinated (Fig. 7A). However, myelination was delayed in the gray matter of Olig1-null spinal cord, and numerous axons were not yet myelinated in both the dorsal and ventral columns (Fig. 7B). Strikingly, however, PLP and MBP immunofluorescence in Olig1-null spinal cord appeared relatively normal by 2 months (Fig. 7C,E), suggesting recovery from the early deficit in the spinal cord. At this time point, the majority of axons in the dorsal column were myelinated both in Olig1-null and control mice (Fig. 7D,F). Electron micrographs indicated that myelination appeared normal in the 2-month-old Olig1-null spinal cord (data not shown), which is consistent with the Arnett et al. (2004) study of Olig1-null spinal cord. Interestingly, despite the apparently normal myelin in Olig1-null spinal cord, there was still a 35% reduction of total Sox10+ oligodendrocytes. This was attributable primarily to a nearly 40% loss of CC1+ mature oligodendrocytes in Olig1-null spinal cord, compared with control littermates (Fig. 7G–I). Thus, in contrast to brain, where less mature oligodendrocytes resulted in reduced myelination, in spinal cord, there was some loss of oligodendrocytes, but the percentage of OPCs and mature oligodendrocytes was only modestly different between Olig1-null and control mice (Fig. 7J) and myelin appeared normal.
The relatively minor impact on myelination from the loss of Olig1 in spinal cord was quantified by Western blot analysis of total spinal cord. The myelin deficit had almost recovered by P15 (Fig. 8A,C), and had recovered even more so by 2 months (Fig. 8B,D). These data dramatically contrasted with the major difference in myelin proteins in cerebrum (Fig. 6).
Collectively, these data indicate that although the number of mature oligodendrocytes was sharply reduced in both brain and spinal cord of Olig1-null mice, OPC differentiation in spinal cord was relatively normal, and the remaining mature oligodendrocytes generated essentially normal amounts of myelin in the spinal cord. These data are consistent with the initial report of these mice (Lu et al., 2002).
Expression and subcellular location of Olig1 is highly coordinated with oligodendrocyte morphogenesis and the onset of myelination
Olig1 and Olig2 are highly conserved at the molecular and functional level, yet they have distinct subcellular localization in mature oligodendrocytes (Arnett et al., 2004). Olig1 is initially found in the nucleus of OPCs and as oligodendrocytes mature, Olig1 translocates to the cytoplasm of myelinating oligodendrocytes. The role for Olig1 in the cytoplasm is unclear, but our studies suggest that it may be a regulated event. We analyzed Olig1 expression in wild-type mice during the transition from premyelinating to myelinating oligodendrocytes. Strikingly, Olig1 was undetectable in premyelinating oligodendrocytes right before they contacted axons and in oligodendrocytes that were initiating axon contact (Fig. 9A,B,D,E, arrowheads). Olig1 expression then gradually increased in the cytoplasm of myelinating oligodendrocyte as myelinogenesis proceeded (Fig. 9C,F, arrowheads). This dynamic Olig1 expression and localization was even more apparent in single cells costained for both Sox10 and Olig2 (Fig. 9G–J). Furthermore, colabeling with PDGFRα (OPC) and PLP (premyelinating oligodendrocytes) confirmed stage-specific reduction of Olig1 in the premyelinating oligodendrocyte population (Fig. 9K,L).
Olig2 expression was not altered in Olig1-null brain
In earlier reports, it had been suggested that the mild impact of Olig1 loss might result from compensatory increases in Olig2 expression (Lu et al., 2002). We therefore analyzed Olig2 expression in P1, P8, P15, and 1-month-old Olig1-null and control mice (Fig. 10A–D). No obvious dysregulation of Olig2 was seen in Olig1-null oligodendrocytes in subcortical white matter by immunohistochemical analysis (Fig. 10A–D). To confirm this, Olig2 in cerebrum and spinal cord was quantified by Western blot at P15 and 2 months. Olig2 protein expression was reduced by 60% in the cerebrum of both P15 and 2-month-old Olig1-null mice compared with control littermates (Fig. 10E), which was consistent with the ∼60% reduction in total oligodendrocytes. Thus, on a per cell basis, these data suggest little change in Olig2 expression, and in particular no compensatory increase of Olig2 in cerebrum. In contrast, although we observed a 40% reduction of Olig2 in spinal cord of P15 Olig1-null mice by Western blot (Fig. 10F), Olig2 expression in Olig1-null spinal cord was comparable to control spinal cord by 2 months (Fig. 10F). Because the total oligodendrocyte number was decreased ∼35% in Olig1-null spinal cord, this finding suggests that by 2 months, the Olig2 level may actually be upregulated in individual oligodendrocytes in the Olig1-null spinal cord. An approximation of the amount of Olig2/cell was calculated for 2-month-old tissue, using data from the cell number and Western blot studies (Fig. 10F). Collectively, these results suggest that there was no compensatory change in Olig2 expression in cerebrum, where there was severe dysmyelination, but intriguingly there was potentially compensatory upregulation of Olig2 in spinal cord, where less dysmyelination was seen.
Olig1 is important for initial oligodendrocyte differentiation in vitro
To expand on the in vivo studies of Olig1-null OPCs, we examined the impact of Olig1 loss on proliferation and differentiation of purified OPCs in vitro. OPCs were generated from neurospheres derived from E12.5 to E14.5 cerebrum of Olig1-null or control mice. OPCs were plated and maintained in proliferation medium containing PDGF and FGF for 3 d. In contrast to the in vivo data (Fig. 1), Olig1-null OPCs in vitro had similar proliferation rates to control (Fig. 11D,D′,E), which suggested that Olig1 did not directly regulate the proliferation of OPCs. In contrast with the increased cell death in Olig1-null mice in vivo analysis, the number of apoptotic cells did not increase as OPCs differentiated in the absence of Olig1 in vitro (Fig. 11G).
After switching to differentiation media containing T3 for 3 d in vitro, 22.5% of control OPCs differentiated to O4+ immature oligodendrocytes, but only 11.4% of Olig1-null OPCs differentiated (Fig. 11F,F′,H). Thus, in culture, these cells had normal proliferation rates, but reduced ability to differentiate, relative to control cells. Olig2 was not upregulated in the cultured oligodendrocytes derived from Olig1-null cerebra (Fig. 11B,C), which was consistent with the lack of upregulation of Olig2 in Olig1-null cells in cerebrum in vivo (Fig. 10). As in the in vivo context, in the differentiating oligodendrocytes in vitro, Olig1 loss compromised oligodendrocyte morphological differentiation. Analysis of process complexity showed a 40% reduction of total branch points of Olig1-null O4+ cells compared with control (Fig. 11I).
To confirm that the deficit of OPC differentiation in vitro resulted only from loss of Olig1 and not from unknown compensation during early commitment of the oligodendrocyte lineage in Olig1-null mice, we analyzed rat oligodendrocytes after transient siRNA knockdown of Olig1. These cells are well past the pre-OPC to OPC transition. Acute knockdown of Olig1 in primary rat OPCs derived from postnatal mixed glia was validated by reduction of Olig1 protein after knockdown (Fig. 11J–L). Consistent with the finding from Olig1-null mouse OPCs, knockdown of Olig1 in rat OPCs did not impair their proliferation but had a similar impact on OPC initial differentiation and morphogenesis (Fig. 11M–Q). In summary, consistent with our in vivo findings, Olig1 impacted initial differentiation of cultured rat and mouse OPCs and their subsequent morphological development.
Discussion
Developmental phenotypes of different Olig1-null mouse lines
In this study, we observed the developmental delay and long-term recovery in the spinal cord first reported for this Olig1-null mouse (Lu et al., 2002), but as discussed below, we noted dramatic oligodendrocyte deficits in the brain. The initial report analyzed mice that were still on a mixed genetic background, and in the ensuing 12 years, they have been backcrossed to C57BL/6J, perhaps making the phenotype more robust.
The mild phenotype of the original Olig1-null mice had been speculated to result from compensatory upregulation of Olig2 by the cis-acting regulatory effect of the Pgk-Neo cassette retained in the Olig1 locus (Lu et al., 2002), and when that cassette was removed from the original line, the new Olig1-null mice were very severely affected, dying approximately the third postnatal week with severe dysmyelination (Xin et al., 2005). Unfortunately, Olig2 expression was not directly examined in the earlier reports on these lines, so it was unclear whether that explained the difference in phenotype.
One recent study attempted to resolve the discrepancy of these two early Olig1-null mouse lines by generating two new Olig1-null mouse lines (Paes de Faria et al., 2014). As in the initial study, much analysis was done in the spinal cord of these two new lines, and developmental delay and subsequent recovery of oligodendrocyte differentiation was observed in spinal cord for both new lines. In neither line was compensation by Olig2 noted (Paes de Faria et al., 2014). These differences among studies have led to significant speculation on the role of Olig1. Paes de Faria et al. (2014) discuss potential reasons for the dramatic differences of the Xin et al. (2005) study relative to the original Lu et al. (2002) and Paes de Faria et al. (2014), but clearly in several Olig1-null lines, spinal cord myelination is delayed but not permanently changed.
Our studies in spinal cord are consistent with these earlier studies on spinal cord development in the absence of Olig1. Nevertheless, despite apparently normal myelination in the spinal cord of Olig1-null mice, our studies do demonstrate that oligodendrocyte production and differentiation were also impaired in spinal cord. However, despite this loss of oligodendrocytes, there was no reduction in Olig2 expression (Fig. 10), which suggested upregulation of Olig2 on a per cell basis in spinal cord. The compensatory upregulation of Olig2 may explain some aspects of the relatively normal myelination in the spinal cord of Olig1-null mice.
Temporospatial control of OPC differentiation by Olig1 in brain
Oligodendrocyte lineage progression and myelin initiation are tightly controlled both temporally and spatially in the developing mouse brain (Trapp et al., 1997; Baumann and Pham-Dinh, 2001). Both extrinsic and intrinsic cues play critical roles in oligodendrocyte differentiation and myelination (Emery, 2010b; Wood et al., 2013). Olig1 and Olig2 play critical roles in oligodendrocyte specification (Lu et al., 2002; Zhou and Anderson, 2002; Silbereis et al., 2014) and early OPC differentiation (Mei et al., 2013). Silbereis et al. (2014) demonstrated that early commitment to the oligodendrocyte lineage relative to interneuron lineage is regulated by Olig1. Furthermore, within the oligodendrocyte lineage, the current studies show that Olig1 is critical for OPC cell cycle exit and subsequent differentiation of OPCs into premyelinating oligodendrocytes in cerebrum, but not spinal cord.
The pre-OPC to OPC transition has largely been studied in vitro in studies generating oligodendrocytes from human embryonic stem cells or iPS cells (Hu et al., 2009a; Wang et al., 2013). Pre-OPCs express Sox2 and Sox9 transcription factors, and the transition to OPCs involves Sox10, which is required for maintaining the differentiated state of neural progenitors by repressing the expression of Sox2 and Sox9 (Castelo-Branco et al., 2014). Based on the known interaction of Olig1 and Sox10 (Li et al., 2007), Olig1 may promote the pre-OPC to OPC commitment by interacting with Sox10 to repress Sox2 expression in newly formed OPCs, because the Sox2 level remains high in OPCs from Olig1-null mice, in contrast to the sharp downregulation of Sox2 in OPCs from control mice.
Considering the large set of common target genes downstream of Olig1 and Olig2, including most major myelin protein (Meijer et al., 2012; Weng et al., 2012) and key regulators of differentiation, such as GPR17 (Chen et al., 2009) and Sip1 (Weng et al., 2012), Olig1 and Olig2 may coordinate to regulate early OPC differentiation. The similar phenotype observed in this Olig1-null mice and the Olig2 conditional knock-out mice, in which Olig2 is selectively deleted in OPCs (Mei et al., 2013), implies the potential interplay of these two related transcription factors, which may function simultaneously in OPC differentiation in the developing mouse brain. However, in cerebrum, it appears that neither protein can fully compensate for the loss of the other protein. Although Olig2 has been shown to promote pre-OPC to OPC commitment in vitro (Hu et al., 2009b), this process was impaired in Olig1-null mice, even in the presence of Olig2.
Region-specific regulation of oligodendrocyte differentiation and myelination by Olig1 in the CNS
Both regional oligodendrocyte heterogeneity and diverse local environmental cues have been suggested to result in different myelination patterns in different CNS regions (Almeida et al., 2011; Vigano et al., 2013). Indeed, regional differences in dysmyelination have been seen in several knock-out mice lines, including conditional Fyn knock-out mice (Sperber et al., 2001), PDGFRα knock-out mice (Fruttiger et al., 1999), and mTOR, Rictor, or Raptor knock-out mice (Bercury et al., 2014; Wahl et al., 2014). The current studies are consistent with these reports. Thus, in contrast to the mild phenotype in the Olig1-null spinal cord, we observed a severe phenotype without long-term recovery in the brain of Olig1-null mice. Despite the increased proliferation of OPCs in Olig1-null mice, there was a permanent deficit of oligodendrocyte number, maturation, and myelination in Olig1-null cerebrum, which lasted through adulthood. It is likely that distinct local environmental cues result in the dramatic myelin deficit in the brain of Olig1-null mice but not in the spinal cord. Whether this difference results simply from the lack of compensatory upregulation of Olig2 in the brain or from other aspects of Olig1 function that lead to the severe reduction of myelinating oligodendrocytes is unclear at this point.
Importance of the subcellular localization of Olig1 in oligodendrocytes
In contrast to the constant expression and nuclear localization of Olig2 throughout oligodendrocyte development, dynamic expression and subcellular translocation of Olig1 was found to be highly coordinated with oligodendrocyte lineage progression and myelination initiation. Intriguingly, in contrast to the positive role of nuclear Olig1 in early OPC differentiation, Olig1 was downregulated in premyelinating oligodendrocytes before they started ensheathing axons. Olig1 then reappeared in the cytoplasmic of myelinating oligodendrocytes as the myelin segments elongated. The downregulation of Olig1 in premyelinating oligodendrocytes before initiating axon contact suggests that sustained nuclear Olig1 might play an inhibitory role in this process. However, the link between Olig1 subcellular location and function is not clear. As a transcription factor, nuclear Olig1 location in the OPCs fits its role as a positive regulator of early OPC differentiation. However, although fewer in number, Olig1-null oligodendrocytes can generate apparently normal myelin (Figs. 3, 4, 6). If the cytoplasmic Olig1 does have de novo function in myelinogenesis per se, one might predict that myelinating oligodendrocytes in Olig1-null mice either produce less myelin segments or bear thinner myelin. In contrast to this prediction, in the neocortex of P8 mice when the growing myelin segments can be distinguished, there was no apparent difference between the number of segments generated by individual myelinating oligodendrocytes of Olig1-null and heterozygous mice (Fig. 3L,M). In addition, the myelin thickness assessed by g-ratio was normal in the Olig1-null mice (Fig. 6C,D). Thus, these data do not support a requirement for cytoplasmic Olig1 in myelination. There may be an active function for Olig1 in the cytoplasm, but an interesting alternative explanation could be to prevent Olig1 entry into the nucleus during myelinogenesis, yet maintain it as a store potentially available for remyelination.
Unique function of Olig1 in myelination and remyelination
Although Olig1 loss apparently can be partially compensated for during developmental myelination, its function is essential for remyelination (Arnett et al., 2004; Whitman et al., 2012). The differentiation block in demyelinated lesions of adult Olig1-null mice resembles MS lesions where OPCs are present but many either fail to differentiate (Chang et al., 2000; Kuhlmann et al., 2008) or are blocked as premyelinating oligodendrocytes unable to differentiate further (Chang et al., 2002). Considering the role of Olig1 in both early OPC differentiation and myelination onset, it is plausible that dysregulation of Olig1 expression and localization may result in the differentiation block of OPCs and premyelinating oligodendrocytes in chronic inactive MS lesions. The current study demonstrates the critical role of Olig1 in oligodendrocyte differentiation in the brain, in parallel with the role of Olig2 in this process. In addition, phosphorylation has been shown to modify the function of both Olig1 and Olig2 in oligodendrocyte maturation and specification respectively (Li et al., 2011; Niu et al., 2012). Ongoing studies in this laboratory suggest that post-translational modification may also modulate Olig1 stability and localization. With further mechanistic insight into the regulation of Olig1 function during development, it may be possible to modulate Olig1 expression and activity to promote oligodendrocyte differentiation and subsequent myelination in MS patients.
Footnotes
This work was supported by NS08223 and an NRSA fellowship (J.T.A.).We thank Katherine E. Saul for technical assistance, Dorothy Dill for assistance with electron microscopy, and Marnie Preston for helpful discussion.
The authors declare no competing financial interests.
- Correspondence should be addressed to Dr Wendy B. Macklin, Department of Cell and Developmental Biology University of Colorado School of Medicine, 12800 East 17th Avenue, Building RC1S, Room L18-12403B, Aurora, CO 80045. Wendy.Macklin{at}ucdenver.edu