MicroRNA miR-204 regulates proliferation and differentiation of oligodendroglia in culture

Oligodendrocytes wrap and physically shield axons of the central nervous system with myelin sheaths, resulting in rapid signal transduction and accurate neuronal function. The complex oligodendroglial development from immature oligodendrocyte precursor cells (OPCs) to myelinating oligodendrocytes (OLs) is profoundly depen-dent on the activity of transcription factors of the Sox protein family. Target genes of the crucial regulator Sox10 have recently been expanded to microRNAs. Here, we report miR-204 as a novel transcriptional target of Sox10. Regulatory regions of miR-204 show responsiveness to and binding of Sox10 in reporter gene assays and electromobility shift assays. Once expressed, miR-204 inhibits OPC proliferation and facilitates differentiation into OLs in the presence of Sox10 as evident from overexpression in primary rat and mouse oligodendroglial cultures. Phenotypes are at least in part caused by miR-204-dependent repression of the pro-proliferative Ccnd2 and the differentiation inhibiting Sox4. These findings argue that the transcriptional activator Sox10 forces oligodendroglial cells to exit the cell cycle and start differentiation by gene inhibition via miR-204 induction.


| INTRODUCTION
MicroRNAs (miRs) are increasingly recognized as important regulatory factors in oligodendroglia in development and disease. They are necessary when myelinating glia of the central nervous system (CNS) develop from oligodendrocyte precursor cells (OPCs) either during embryonic myelination or during remyelination in de-and dysmyelinating diseases such as multiple sclerosis and leukodystrophies (for review, see Galloway & Moore, 2016). For instance, measurement of miR levels in different subtypes of multiple sclerosis lesions showed a significant dysregulation in the expression of at least 50 miRs (Junker et al., 2009;Noorbakhsh et al., 2011). The importance of miRs is even more obvious from diverse mouse mutants with deletion of components of the miR processing machinery such as the endoribonuclease Dicer. In the absence of Dicer, mature miRs are not cleaved from pre-miR precursors so that mature miRs are completely missing. Early Olig1-Cre or Olig2-Cre driven deletion of Dicer results in strongly reduced differentiation into mature myelinating cells during development. Deletion by CNP-Cre at a slightly later stage, before OPCs start to differentiate, impairs myelination as well and causes tremor and motor deficits typical for dysmyelination (Dugas et al., 2010;Zhao et al., 2010). Inducible Dicer knockout in already mature oligodendrocytes (OLs) by PLP-Cre-ERT2 induces CNS demyelination showing that miRs also help to maintain myelin in the adult (Shin, Shin, McManus, Ptacek, & Fu, 2009). Since deletion of Dicer affects all miRs, these mouse mutants show that the presence of miRs is crucial for maturation and maintenance of myelinating OLs at embryonic and postnatal stages.
Nonetheless, specific functions of single miRs still have to be identified. Only few miRs are characterized in their role of fine-tuning the complex network of transcription factors that controls differentiation from the OPC to the mature OL. The small inhibitory RNAs recognize complementary binding sites in target mRNAs and lead to their degradation or inhibition of translation. For example, miR-338, miR-219, and miR-138 support differentiation of oligodendroglia by targeting mRNAs of factors that keep the cells in a precursor stage (Dugas et al., 2010;Zhao et al., 2010). On the other hand, repression of the pro-myelinating transcription factor Myrf by miR-145 keeps cells in an undifferentiated stage and inhibits precocious maturation (Hoffmann et al., 2014). The transient expression of Tcf7l2, another transcription factor involved in OL differentiation, is under control of miR-338 and miR-155. Bioinformatic modeling suggests many more miRs to be involved in the network (Cantone et al., 2019).
Transcription factors are not only targets of miRs, but also regulators of miR expression. The HMG-box protein Sox10 is indispensable for proper differentiation of OLs. In the absence of Sox10, OPCs are specified, but do not differentiate (Stolt et al., 2002). Among the target genes of Sox10 are also miRs with relevance in OL development. Sox10 binds and activates regulatory regions of miR-338, miR-335, and miR-155 (Cantone et al., 2019;Reiprich et al., 2017). Some additional miRs show changes in expression when Sox10 is missing. Their role needs to be determined.
Here, we identify miR-204 as a regulator of OL proliferation and differentiation. We report that miR-204-5p expression depends on the oligodendroglial differentiation stage and on the presence of Sox10.
Sox10 binds and activates evolutionary conserved regions (ECRs) from the miR-204 genetic locus. We show that miR-204 reduces proliferation and induces differentiation of primary OLs, partially by regulating Ccnd2 and Sox4. Therefore, miR-204 fulfills an important function in the progression from OPCs to OLs downstream of Sox10.

| Plasmids and viruses
To analyze the transactivation capability by Sox10, ECRs upstream of the miR-204 genetic locus were amplified from rat genomic DNA by The Sox10 expression plasmid was based on pCMV5 and has been previously described (Kuhlbrodt, Herbarth, Sock, Enderich, et al., 1998).   Extracts used as protein source for electromobility shift assays (EMSAs) were prepared as previously described (Kuhlbrodt, Herbarth, Sock, Enderich, et al., 1998). In brief, HEK293 cells were transfected with 16 μg pCMV-Sox10 expression plasmid per 100-mm dish by polyethylenimine treatment and harvested 48 hr posttransfection, followed by preparation of whole cell extracts. Double-stranded oligonucleotides S1-S7 of ECR2-4, which contained putative Sox10 binding sites, and an oligonucleotide containing Site B of the Mpz gene were labeled with radioisotope 32 P in order to perform EMSAs. In the presence of HEK293 extracts and poly-dGdC as unspecific competitor, labeled oligonucleotides of 28 bp length were analyzed. Site B served as control for monomeric Sox10 binding (Peirano, Goerich, Riethmacher, & Wegner, 2000).

| RNA preparation from mice and cultured cells
Mice were kept at 12/12 hr light-dark cycles with permanent access to drinking water and food in accordance with animal welfare laws. All mice used for this publication were on C3H background. Tissue was prepared at postnatal days (P) 7, 14, and 21 of both males and females. Animals were killed by decapitation and brains were isolated.
Callosal areas were excised and RNA was isolated using TRIZOL reagent (Invitrogen

| Bioinformatics and statistical analysis
TargetScan 7.1 was used for the prediction of target genes of murine miR-204-5p (Lewis, Burge, & Bartel, 2005) and the analysis of target 3 0 -UTR profiles (Nam et al., 2014). Gene ontology terms-subcategory biological processes (GO-BP) for predicted miR-204-5p targets were determined by the use of DAVID 6.8 ( ECRs were checked for potential transcription factor binding sites using P-Match (Kel et al., 2003). Quantification of Western blotting was performed using Fiji ImageJ (Schindelin et al., 2012). Experimental results from independent animals, as well as from independent transfection and transduction experiments were treated as biological replicates (n ≥ 3). Differences in cell numbers, transcript expression, protein levels, or luciferase activities were analyzed by two-tailed Student's t test for statistical significance (*, p ≤ .05; **, p ≤ .01; ***, p ≤ .001).

| miR-204-5p is differentially expressed in oligodendroglia
In a quantitative RT-PCR (qrt-PCR)-based high-throughput screening of miR expression in the oligodendroglial cell line Oln93, we identified a number of miRs that were strongly downregulated in the absence of Sox10 (Reiprich et al., 2017). We analyzed predicted target genes of the downregulated miRs for associated biological pro- To check for differential expression during differentiation, we isolated RNA from primary rat oligodendroglia in the OPC stage (d0) and after 6 days in differentiation medium (d6). In the differentiated stage, miR-204-5p expression was increased 3.5-fold compared to the precursor stage ( Figure 1c).
Next, we wanted to see, whether miR-204-5p expression also increases in vivo during developmental myelination. As miR-204-5p expression has been reported in neurons (Conte et al., 2014;Mohammed et al., 2016), we isolated tissue enriched for mouse corpus callosum as the region with the highest density of oligodendroglia at P7, P14, and P21, to specifically assess oligodendroglial miR-204-5p expression during the active phase of developmental myelination.
In corpus callosum cDNA, we determined a 3.4-fold increase in miR-204-5p expression by qrt-PCRs from P7 to P21 coincident with maximum induction of myelin gene expression ( Figure 1d). We conclude from expression analyses and GO-BP studies that miR-204-5p is involved in oligodendroglial development, and that its expression depends on the presence of Sox10 and increases during differentiation of oligodendroglia.

| Sox10 binds and activates regulatory regions of the miR-204 genetic locus
To analyze whether reduction of miR-204-5p expression in the Sox10-deficient Oln93 cell line is a direct effect, we reintroduced A more detailed sequence analysis yielded three potential Sox protein binding sites in ECR2 (S1, S2, S3), two potential binding sites in ECR3 (S4, S5) and two in ECR4 (S6, S7) ( Figure 3a). Potential binding sites were defined as completely matching the Sox consensus Sites closely spaced in the typical arrangement for dimeric binding were not present in the ECRs so that all potential sites are predicted to be monomeric binding sites . In EMSAs, we observed binding of fulllength Sox10 to oligonucleotides containing site S2 or S3 of ECR2, F I G U R E 1 Pattern of miR-204-5p expression and gene ontology term (GO) analysis suggest a role in oligodendroglia. (a) Predicted targets of miR-204-5p were analyzed for enrichment in GO-BP terms. Upper axis relates to bars, indicating fold enrichment; lower axis relates to dots, indicating p-values of corresponding enrichments (spec., specification; dev., development; diff., differentiation; pos., positive; reg., regulation; prol., proliferation;). (b-d) Analysis of miR-204-5p transcript levels by qrt-PCR. miR-204-5p expression levels were normalized to U6 snRNA levels. (b) miR-204-5p levels of wild-type (WT) and Sox10-deficient (KO) Oln93 cells were compared on transcript level. For miR quantification, miR-204-5p transcript levels in the WT cell line were arbitrarily set to 1, while levels in Sox10-deficient cells are shown relative to the WT (mean + SD; n = 3). (c) Transcript levels of miR-204-5p were measured in undifferentiated (d0) and 6 days differentiated (d6) primary rat oligodendrocytes. Transcript levels of undifferentiated cells were arbitrarily set to 1, while levels in differentiated cells are shown relative to d0 (mean + SD; n = 3). (d) miR-204-5p expression was analyzed in corpus callosum tissue of mice at stages P7, P14, and P21. miR-204-5p levels at P7 were arbitrarily set to 1, while levels at other stages are shown relative to P7 (mean + SEM; n ≥ 6). Statistically significant differences between samples and controls are indicated (Student's t test; *, p ≤ .05; **, p ≤ .01; ***, p ≤ .001) but not to S1 (Figure 3b). Site S5 in ECR3 showed the strongest binding of Sox10, while S4 was not bound. In ECR4, the oligonucleotide containing S7 showed complex formation with Sox10, whereas S6 was negative for Sox10 binding (Figure 3b,c). In further analyses, no

| miR-204 induces differentiation of primary oligodendroglia
Next, we wanted to know whether increased exit from the cell cycle coincides with higher rates of differentiation. Therefore, we trans- (b) Electromobility shift assays (EMSAs) were performed with oligonucleotides containing potential binding sites S1-S7. S1-S7 were incubated without cell extract (−) or with extract from human embryonic kidney (HEK) 293 cells transfected with empty pCMV vector (C) or full length Sox10 expression vector (S10). Site B of the Mpz promoter served as positive control for monomeric binding . (c) Sequence of the Sox10 consensus motif and putative binding sites S1-S7 analyzed in EMSAs. Sites bound by Sox10 3.5 | miR-204 regulates oligodendroglial proliferation and differentiation partially by targeting Ccnd2 and Sox4 As we have observed reduced proliferation and increased differentiation, we asked which target genes of miR-204 might contribute to these phenotypes. Based on previous studies (Wu, Pan, et al., 2015;Wu, Zeng, et al., 2015) and target gene prediction programs, we hypothesized that miR-204-5p may target Ccnd2 and Sox4 mRNAs in oligodendroglia. To study the impact of miR-204 on the putative targets, fragments of the 3 0 -UTRs of the corresponding mRNAs with predicted miR binding sites were inserted downstream of the luciferase coding sequence into reporter plasmids in wild-type or mutated versions (Figure 7a,b). A reporter with the wild-type fragment of the Ccnd2-3 0 -UTR behind the luciferase coding sequence showed responsiveness to miR-204 in Neuro2a cells (Figure 7c). In the presence of Having seen that miR-204-5p can regulate Ccdn2 and Sox4 expression, we wanted to verify that these target genes also contribute to the phenotypes that we observed during oligodendrogenesis.
Therefore, we inhibited Ccnd2 or Sox4 by RNAi and analyzed if this would mimic the effects of miR-204 overexpression in cultured oligodendroglia. Knockdown efficiencies were 58% for Ccnd2 and 57% for Sox4 as determined by Western blots of HEK293 extracts after cotransfection of expression plasmids for proteins with antisense RNA or scrambled control (Figure 7h,i). the other lineage determining transcription factor of oligodendroglia (Küspert, Hammer, Bösl, & Wegner, 2011;Weider et al., 2015;Weider et al., 2018). With Sox6 and Sox9, Sox10 modulates other Sox proteins that are involved in oligodendroglial development (Stolt et al., 2003;Stolt et al., 2006). At the initiation of terminal differentiation, Sox10 then induces myelin gene expression by activating expression of the pro-differentiation factors Nkx2.2 and Myrf and of the myelin gene Mbp itself, in part with other transcription factors such as Nfatc2 (Hornig et al., 2013;Liu et al., 2007;Stolt et al., 2002;Weider et al., 2018). Among miRs, miR-338, miR-335, and miR-155 are known to be activated by Sox10 (Cantone et al., 2019;Reiprich et al., 2017).
We The miR-204 sequence is located in intron VI of the Trpm3 gene in the mouse, which encodes a member of the melastatin-like subfamily of TRP channels (Grimm, Kraft, Sauerbruch, Schultz, & Harteneck, 2003). Trpm3 has previously been shown to be involved in the regulation of sphingosine-induced cellular Ca 2+ influx during oligodendroglial differentiation (Hoffmann et al., 2010). Whether activation of the ECRs upstream of the Trpm3/miR-204 genetic locus only affects expression of miR-204 or also expression of the host gene Trpm3 will be subject of further studies.
Sox10 is known as a transcriptional activator, but through the activation of miRs, which exert repressive functions, it can also indirectly inhibit expression of target genes. Activation and repression of gene expression are both necessary to proceed through development.
While proliferation and progenitor characteristics need to be shut off, differentiation has to be activated for the onset of myelination.
miR-204 inhibits translation of the Ccnd2 mRNA, which encodes the pro-proliferative CyclinD2 protein, and at the same time translation of the Sox4 mRNA, which has differentiation preventing effects. Both proteins are subject to rapid turnover at least in nonglial cell types, which makes them prone to dynamic posttranscriptional regulation (Beekman et al., 2012;Kida, Kakihana, Kotani, Kurosu, & Miura, 2007). Thus, miR-204 influences two decisive steps towards myelination, namely exit from the cell cycle and entry into differentiation.
Despite its positive effect on differentiation, overexpression of miR-204 was not able to revert the differentiation failure caused by Sox10 deletion in primary mouse oligodendroglia. This argues that induction of differentiation and myelin genes by Sox10 is distinct from the differentiation promoting effects of Sox10-induced miR-204 expression.
Supporting the direct and active role of Sox10 in the induction of myelin gene expression, repression of differentiation inhibitors by miR-204 is by itself not sufficient to permit OL differentiation. This is in line with a predominantly fine-tuning function of miRs.
miR-204 has not been described in oligodendroglia before. It is mildly expressed in OPCs and increases during differentiation correlating to the observed role in progression of maturation. An antiproliferative effect of miR-204 as described here in oligodendroglia has also been observed in other studies. miR dysregulation is seen in many types of cancer where miRs often are involved in the regulation of proliferation. Expression of miR-204 is downregulated in highly proliferative human glioblastoma tissue compared to healthy tissue (Song, Fajol, Tu, Ren, & Shi, 2016). Vice versa, ectopic expression of miR-204 reduces cell proliferation. The phenotype was mostly ascribed to inhibition of the transcription factor ATF2 as a miR-204 target gene. Similarly, miR-204 downregulation is seen in retinoblastoma tissue and cell lines, where restoration of miR-204 levels can inhibit tumor growth (Wu, Zeng, et al., 2015). In this cellular context, Ccnd2 and MMP-9 are identified as target genes of miR-204. In the present study, we confirm that miR-204 regulates Ccnd2 mRNA via the same recognition seed sequence in its 3 0 -UTR in the mouse as described for human retinoblastoma. Whether ATF2 and MMP-9 are also targets in oligodendroglia, could be a question of further studies, although expression databases argue for low amounts or even absence of these proteins in oligodendroglial cells.
A single-base mutation in human MIR-204 is associated with inherited retinal dystrophy (Conte et al., 2015). The mutation within the seed sequence leads to a shift in the set of target genes, some target sites are not recognized anymore, but many more new targets arise as a result of the mutated seed sequence. Phenotypically, this results in an increase in apoptosis and reduced differentiation of functional photoreceptors. In the eye, proliferation is not affected by the loss of wild-type miR-204. This may be consequence of the new targets that are generated by the mutation. This complication makes a target comparison between this and other studies impossible.
Sox4, the other miR-204 target that we identified, is expressed in OPCs, and needs to be downregulated before the onset of myelination. Transgenic mice with prolonged Sox4 expression beyond the OPC stage present a severe hypomyelination although oligodendroglia are present in normal numbers and proliferation is not significantly changed (Potzner et al., 2007). At early stages, Sox4 seems to be important to keep the cells in a precursor stage and to inhibit premature differentiation. Recently, the effect of Sox4 on differentiation has also been studied in primary OPCs by shRNA-mediated knockdown (Braccioli, Vervoort, Puma, Nijboer, & Coffer, 2018). Stainings showed an increase of mature OLs in the absence of Sox4. Here we argue, that miR-204-5p comes into play at the onset of differentiation and downregulates Sox4 by recognizing a binding site in its 3 0 -UTR. This binding site is also targeted in human renal cell carcinoma cells, where overexpression of miR-204 not only leads to reduced levels of Sox4, but also to decreased proliferation, migration and invasion (Wu, Pan, et al., 2015). Although Sox4 is the only described target of miR-204 in this study, further targets are likely to exist. Thus, reduced proliferation in renal cell carcinoma may also be a consequence of Ccnd2

DATA AVAILABILITY STATEMENT
All data generated in this study are contained within the manuscript and available.