Identification of Tmem10/Opalin as a novel marker for oligodendrocytes using gene expression profiling

Background During the development of the central nervous system, oligodendrocytes generate large amounts of myelin, a multilayered insulating membrane that ensheathes axons, thereby allowing the fast conduction of the action potential and maintaining axonal integrity. Differentiation of oligodendrocytes to myelin-forming cells requires the downregulation of RhoA GTPase activity. Results To gain insights into the molecular mechanisms of oligodendrocyte differentiation, we performed microarray expression profiling of the oligodendroglial cell line, Oli-neu, treated with the Rho kinase (ROCK) inhibitor, Y-27632 or with conditioned neuronal medium. This resulted in the identification of the transmembrane protein 10 (Tmem10/Opalin), a novel type I transmembrane protein enriched in differentiating oligodendrocytes. In primary cultures, Tmem10 was abundantly expressed in O4-positive oligodendrocytes, but not in oligodendroglial precursor cells, astrocytes, microglia or neurons. In mature oligodendrocytes Tmem10 was enriched in the rims and processes of the cells and was only found to a lesser extent in the membrane sheets. Conclusion Together, our results demonstrate that Tmem10 is a novel marker for in vitro generated oligodendrocytes.


Background
Oligodendrocytes are specialized cells of the central nervous system that produce myelin, a multilayered membrane spirally ensheathing axons and facilitating rapid nerve conduction [1,2]. The development of oligodendrocytes is a gradual process, in which each step of the differentiation process is characterized by stage specific markers [3][4][5][6]. Oligodendrocytes originate from oligodendrocyte precursor cells that arise from multiple foci along the neuronal tube and migrate into the future white matter of the brain. After reaching their final position, they develop into mature post-mitotic cells that produce the myelin sheaths [7]. During the progression through the oligodendroglial lineage, the cells loose their migratory and proliferative capacities and undergo dramatic changes in their morphology by the formation of a highly branched network of processes. This transformation is accompanied by the expression of a number of gene products that are highly enriched or even specific to oligodendrocytes such as the myelin basic protein (MBP), proteolipid proteins (PLP/DM20), myelin-associated glycoprotein (MAG), cyclic nucleotide phosphodiesterase (CNP) and the glycolipids, galactosylceramide and sulfatide. The capacity of oligodendrocyte precursor cells to differentiate into oligodendrocytes that express these different gene products is intrinsic to the lineage and occurs even in the absence of neurons [4,8]. Oligodendrocytes need to provide specific sorting and transport mechanisms to enable the synthesis of an extensive amount of myelin membrane in a very short time [3,9]. Since oligodendrocytes must produce myelin at the appropriate time of neuronal development, a number of reciprocal signalling systems are likely to operate to coordinate the organisation of axonal domains and the biogenesis of myelin [10][11][12][13][14][15]. A number of recent studies have shown that neuronal-derived signalling molecules control the development of myelinforming glial cells [16][17][18][19][20][21]. We have recently shown that neurons regulate membrane trafficking in oligodendrocytes [19]. In the absence of neurons, the major myelin protein, PLP, is internalized and stored in late endosomes. After receiving an unknown soluble signal from neurons, oligodendrocytes reduce the rate of endocytosis and increase the retrograde transport of PLP from late endosomes to the plasma membrane. A fraction of PLP is released in association with exosomes [22,23]. Our previous work shows that changes in Rho GTPase activity were responsible for switching between these two modes of transport [24]. Inactivation of Rho GTPase activity reduced the transport of cargo to late endosomes and at the same time increased the mobilization of membrane from late endosomes. We found that a neuronal soluble factor was responsible for the downregulation of RhoA GTPase activity in the oligodendroglial cell line, Oli-neu [24]. The downregulation of RhoA function during morphological differentiation of oligodendrocytes is supported by a number of other studies [18,25]. In this study, we were interested in the transcriptional changes that occur after differentiation of Oli-neu cells by conditioned neuronal medium or by inactivation of Rho GTPase function. This effort led to the identification of the transmembrane protein 10 (Tmem10/Opalin) as a novel marker for oligodendrocytes. The transmembrane protein 10 is known as Tmem10/TMEM10 in mice, rats and humans, with the synonyms TMP10 or HTMP10. Recently four Tmem10 homologs of prosimian species (Eulemur macaco, Lemur catta, Microcebus murinus and Otolemur garnetti) have been named Opalin [26]. In this work the human, rat and mouse transmembrane protein 10 will be referred to as Tmem10.

Results and Discussion
As a cellular model for oligodendrocyte differentiation we use the oligodendroglial cell line, Oli-neu. The advantage of this system is that morphological differentiation of a pure oligodendroglial culture can be triggered synchro-nously by adding conditioned medium from primary neuronal cultures to the cells. To characterize the gene changes that occur after incubation of Oli-neu cells with conditioned neuronal medium, we used Affymetrix microarrays. Oli-neu cells were incubated for 16 hours with conditioned neuronal medium and compared to untreated parallel cultures. Cell fractions were used to purify mRNA for microarray analysis (the data is available at NCBI Geo accession number GSE10291). Using a 1.7 fold cut-off, we found that 716 genes were up-regulated, whereas 386 genes were down-regulated by conditioned neuronal medium as compared to the control. The eighty most highly induced genes are shown in Table 1. We have previously shown that incubation of Oli-neu cells with conditioned neuronal medium leads to RhoA GTPase inactivation. We, therefore, performed gene expression profiling analysis after treating cells with the Rho kinase (ROCK) inhibitor, Y27632 for 16 hours and compared the transcriptional changes to the ones obtained after incubation with conditioned neuronal medium. Interestingly, 70% of the genes that were controlled by both conditioned neuronal medium and by treatment with Y27632 were regulated into the same direction, indicating that these treatments affected the fate of the cells into a common path ( Table 2). Some of these genes such as the UDP galactosyltransferase 8a and CNP are known to be upregulated during the progression through the oligodendroglial lineage [4], whereas known housekeeping genes such as the ATP synthase served as internal controls and were not differentially expressed after incubation with conditioned neuronal medium or Y27632 (data not shown). One so far not characterized, but highly upregulated gene is the brain-specifically expressed, Tmem10 [27]. For our further study, we decided to focus on Tmem10 for a number of reasons. Tmem10 was the strongest up-regulated gene in our analysis of transcriptional changes induced by conditioned neuronal medium and as well highly up-regulated after treatment with Y27632. The analysis of the average fold up-regulation in both conditions shows that Tmem10 is the most up-regulated gene together with Cyp2c39 (cytochrome P450, family2, polypeptide 39) ( Table 3). In addition, a previous microarray analysis identified a 23.75 fold up-regulation of Tmem10 during oligodendrocyte differentiation [28] and in situ expression data of Tmem10 in the Allen Brain Atlas http://www.brain-map.org suggested enrichment in the white matter of the brain.
To begin our characterization of Tmem10 we raised an antibody against the C-terminal part of the protein and performed immunofluorescence analysis on primary cultures of oligodendrocytes. We did not detect Tmem10 on NG2-positive oligodendrocyte precursor cells, whereas pre-oligodendrocytes that were still NG2-positive, but also contained O4 started to express Tmem10 (Fig 1A, B). Higher expression of Tmem10 was identified on NG2negative and O4-positive or O1-positive oligodendrocytes (Fig 1A, C). Low levels of Tmem10 expression can be detected in A2B5-positive oligodendrocyte progenitors ( Fig 1D). Colocalization studies of Tmem10 with MBP indicated that Tmem10 was present in mature oligodendrocytes where it was enriched in the rims and processes of the cells and was found only to a lesser extent in the membrane sheets (Fig. 1E). Comparisons of Tmem10 expression in O4-positive/MBP-negative and MBP-positive cells shows that Tmem10 is redistributed to the rims of the membrane sheets, but the expression level does not change significantly compared to O4-positive oligodendrocytes ( Fig 1B).
Interestingly, double labelling of Tmem10 with GFAP or neuron specific βIII Tubulin showed that Tmem10 could not be detected on GFAP-positive astrocytes or neurons, which are present as a minor cell population in the same cultures, indicating a specific expression of Tmem10 in differentiating oligodendrocytes ( Fig. 2A, B). Additionally, we analysed primary cultures of astrocytes and microglia cultures, but could not detect Tmem10 on either of these cell types (Fig. 2C, D).
The analysis of the primary amino acid sequence of Tmem10 for protein domains and functional sites with InterProScan [29] revealed a predicted signal peptide (amino acid 1-15 in the mouse sequence) and a putative transmembrane domain (amino acid 31-51 in the mouse sequence) (Fig. 3A). To further characterize the protein structure and the membrane orientation of Tmem10, we used N-terminal ECFP-and C-terminal EYFP-fusion proteins of Tmem10. Oli-neu cells were transfected with either of the fusion proteins and live staining was per-formed at 4°C with anti-GFP antibody to specifically label the proteins at the cell surface. We found that only the Nterminal ECFP-fusion protein was detectable by surface staining, whereas premeabilization of the cells uncovered both fusion proteins (Fig. 3B). These results show that Tmem10, as predicted from its primary amino acid sequence, is a type I membrane protein.
After ectopic expression of both Tmem10 fusion proteins and an untagged expression construct, we found that it was mainly localized at the plasma membrane with only very little intracellular staining. Tmem10 appeared to be enriched in actin-rich membrane ruffles at the cell surface as shown by its colocalization with rhodamine-phalloidin ( Fig 4A). Additionally, comparison of Tmem10 transfected Oli-neu with untransfected control cells shows that our anti-Tmem10 antibody specifically recognizes Tmem10 ( Fig 4A).
To investigate the role of the actin cytoskeleton on the lateral mobility of Tmem10 we performed fluorescence recovery after photobleaching (FRAP) experiments in the presence or absence of the F-actin disrupting drug, latrunculin A. The mobile fraction and the half time of fluorescence recovery did not change after treatment with latrunculin A (Fig 4B), indicating that the lateral mobility of Tmem10 within the plasma membrane is not dependent on a functional cortical actin cytoskeleton. Interestingly, we did observe a redistribution of a fraction of plasma membrane Tmem10 into intracellular sites after treatment with latrunculin A (Fig 4C), suggesting a role of the actin cytoskeleton in keeping Tmem10 at the cell surface.    In summary, we have applied an expression profiling approach to identify genes upregulated during Oli-neu cell differentiation. We used two different experimental approaches -incubation with conditioned neuronal medium or treatment with the Rho kinase (ROCK) inhibitor, Y-27632. This procedure led to the identification of Tmem10, a protein that appears during oligodendrocyte differentiation. In cultured oligodendrocytes, Tmem10 was absent from bipolar precursor cells and started to be expressed after the cells had acquired the O4 epitope.
A previous comparative genome analysis suggested that Tmem10 is a mammalian-specific gene [26]. Interestingly, the comparison of the genome structure of the Tmem10 gene and its flanking region identified an evolutionary conserved region within the first intron that functions as an oligodendrocyte-specific enhancer. This domain contains binding sites for Myt1 and cAMP-response element binding protein (CREB) and the treatment of Oli-neu cells with cAMP enhanced the expression of Tmem10 [26]. Previous studies have already shown that cAMP regulates the expression of several other oligodendroglial-specific genes [30], suggesting a general role for cAMP dependent signalling in the differentiation of oligodendrocytes into myelin-forming cells. Another factor that appears to regulate Tmem10 gene expression was the leukaemia inhibitory factor (LIF) [26], which seems to be released by astrocytes in response to ATP secreted by neurons and to promote myelination by mature oligodendrocytes [31]. Together, these data suggest a function for Tmem10 in an oligodendroglial specific process. Our finding that Tmem10 colocalizes with F-actin in plasma membrane ruffles and in Factin-rich processes, points to a role in the regulation of the oligodendroglial actin cytoskeleton. This is reminiscent to another oligodendroglial-specific protein, Ermin, which has been implicated in the regulation of cell morphology by modulating the actin cytoskeleton [32]. The localization of Tmem10 to the leading edge of myelin sheets in mature oligodendrocytes suggests a role for Tmem10 in myelin membrane sheet extension. As Tmem10 localizes to the growing tip of the myelin sheet it could also be involved in the process of recognition or adhesion to potential axonal targets. Further analysis will be required to elucidate these issues.

Cell culture, transfections and immunofluorescence
Primary cultures of mouse oligodendrocytes were prepared as described previously [36]. In brief, cells were plated in MEM containing B27 supplement, 1% horse serum, L-thyroxine, tri-iodo-thyronine, glucose, Tmem10 is not detected on neurons, astrocytes and microglia glutamine, gentamycine, pyruvate, and bicarbonate on poly-L-lysine coated glass-coverslips after shaking. The minor population of neurons and astrocytes which arise together with the oligodendrocytes in the mixed brain cultures were used to assess Tmem10 expression in other cell types. Primary cultures of microglia and astrocytes were prepared as described previously [37]. In brief, microglial cells were shaken off, centrifuged and plated on poly-Llysine coated coverslips in D-MEM containing 10% FCS, glutamine, penicillin and streptomycin. The remaining astrocytes were trypsinized, centrifuged and plated on poly-L-lysine coated coverslips in D-MEM containing 10% FCS, glutamine, penicillin and streptomycin. The oligodendroglial cell line Oli-neu was cultured as described previously [38]. For microarray experiments we used Olineu cells stably expressing PLP-EGFP [19]. Cells were treated for 16 h with conditioned neuronal medium or Y27632 (Calbiochem). Conditioned neuronal medium was obtained from primary cultures enriched in neurons after culturing for 2 weeks and used directly as described previously [24]. Transient transfections were performed using FuGENE transfection reagent (Roche Diagnostics, Basel, Switzerland) according to the manufacturer's protocol. Immunofluorescence was performed as described previously [19]. For surface labelling of ECFP-Tmem10 or Tmem10-EYFP transfected, living cells were incubated with anti-GFP antibody in medium for 10 min at 4°C, washed, fixed and labelled with secondary antibody. Disruption of the actin cytoskeleton was done 16 h after transfection with 2 µM latrunculin A for 30 min at 37°C.

RNA isolation
RNA isolation was performed using the Trizol (Invitrogen, Carlsbad, CA) method according to the manufacturer's recommendations and stored at -80°C. Afterwards, the samples were DNAse I treated in order to remove genomic DNA contaminations. RNA quality was determined using the Agilent 2100 Bioanalyzer (Agilent Technologies) microfluidic electrophoresis. Only sample pairs with comparable RNA integrity numbers were selected for microarray analysis.

Experimental design and sample preparation for 2 colourmicroarrays
For gene expression profiling, a two-colour 1 × 2 design including a dye swap using 6 arrays was applied, comparing Oli-neu cells stably expressing PLP-EGFP treated for 16 h with conditioned neuronal medium or Y27632, respectively, to untreated parallel cultures.
The samples for hybridization were prepared from total RNA according to the Atlas SMART Fluorescent Probe Amplification Kit (Clonetech-Takara Bio Europe) protocol, except, that the RNA template was hydrolyzed under alkaline conditions before cDNA purification, and the PCR amplification process was monitored and stopped in the exponential phase. Quantity and Cy-dye incorporation rates of the generated target material were assessed using a NanoDrop ND-100. Cy3-and Cy5-labelled cDNA fragments, respectively, were hybridized to Agilent Technologies 44 K Mouse Whole Genome Microarrays (G4112A) for 17 h at 65°C. Post-processing washes were done according to the Agilent Technologies SSPE protocol (v2.1), replacing wash solution 3 by acetonitril, followed by immediate scanning using an Agilent G2505B scanner. Intensity data were extracted using the software 'Automatic Image processing for Microarrays'.

Statistical analysis
Normalization of the raw intensity data was done with a non-linear loess regression [42].

Uni-and multivariate designs
Differentially expressed genes were identified by an ANOVA fixed effects model [43]. The resulting P-values were adjusted with the Benjamini-Hochberg method to control the False-Discovery-Rate [44]. Normalization and statistical computation was done for two independent datasets derived from a high gain and a low gain scan, allowing replacement of saturated features in the high gain scan with data from the low gain measurement.

Sequence analysis
The prediction of the transmembrane domain of Tmem10 was done using the InterProScan algorithm provided at EMBL-EBI [29].

Microscopy and analysis
Fluorescence images were acquired on a confocal laser scanning microscope (TCS SP equipped with AOBS, Leica) with a 40× or 63× oil plan-apochromat objective (Leica). Image processing and analysis was performed using Meta Imaging Series 6.1 software (Universal Imaging Corporation). Quantification of fluorescence intensities was performed as described previously [19]. FRAP experiments were done as described in [20].