The myelination‐associated G protein‐coupled receptor 37 is regulated by Zfp488, Nkx2.2, and Sox10 during oligodendrocyte differentiation

Oligodendrocyte differentiation and myelination in the central nervous system are controlled and coordinated by a complex gene regulatory network that contains several transcription factors, including Zfp488 and Nkx2.2. Despite the proven role in oligodendrocyte differentiation little is known about the exact mode of Zfp488 and Nkx2.2 action, including their target genes. Here, we used overexpression of Zfp488 and Nkx2.2 in differentiating CG4 cells to identify aspects of the oligodendroglial expression profile that depend on these transcription factors. Although both transcription factors are primarily described as repressors, the detected changes argue for an additional function as activators. Among the genes activated by both Zfp488 and Nkx2.2 was the G protein‐coupled receptor Gpr37 that is important during myelination. In agreement with a positive effect on Gpr37 expression, downregulation of the G protein‐coupled receptor was observed in Zfp488‐ and in Nkx2.2‐deficient oligodendrocytes in the mouse. We also identified several potential regulatory regions of the Gpr37 gene. Although Zfp488 and Nkx2.2 both bind to one of the regulatory regions downstream of the Gpr37 gene in vivo, none of the regulatory regions was activated by either transcription factor alone. Increased activation by Zfp488 or Nkx2.2 was only observed in the presence of Sox10, a transcription factor continuously present in oligodendroglial cells. Our results argue that both Zfp488 and Nkx2.2 also act as transcriptional activators during oligodendrocyte differentiation and cooperate with Sox10 to allow the expression of Gpr37 as a modulator of the myelination process.

expression profile that depend on these transcription factors.Although both transcription factors are primarily described as repressors, the detected changes argue for an additional function as activators.Among the genes activated by both Zfp488 and Nkx2.2 was the G protein-coupled receptor Gpr37 that is important during myelination.In agreement with a positive effect on Gpr37 expression, downregulation of the G protein-coupled receptor was observed in Zfp488-and in Nkx2.2-deficient oligodendrocytes in the mouse.We also identified several potential regulatory regions of the Gpr37 gene.Although Zfp488 and Nkx2.2 both bind to one of the regulatory regions downstream of the Gpr37 gene in vivo, none of the regulatory regions was activated by either transcription factor alone.
Increased activation by Zfp488 or Nkx2.2 was only observed in the presence of Sox10, a transcription factor continuously present in oligodendroglial cells.Our results argue that both Zfp488 and Nkx2.2 also act as transcriptional activators during oligodendrocyte differentiation and cooperate with Sox10 to allow the expression of Gpr37 as a modulator of the myelination process.Sock & Wegner, 2021).Whereas some of these transcription factors (e.g., Olig2, Sox10, Myrf) have been studied mechanistically in a fairly comprehensive manner, others are less well understood regarding their mode of action and place within the gene regulatory network.
Zfp488 belongs to the less well studied transcription factors.It consists of 337 amino acids and contains two C2H2-type zinc fingers in its carboxyterminal part that flank a nuclear localization signal (Wang et al., 2006).Zfp488 has been shown to be induced by the bHLH transcription factor Olig1, to be upregulated during oligodendrocyte differentiation, interact physically and functionally with transcription factor Olig2 and to promote differentiation and myelin production by oligodendrocytes.During the process, Zfp488 is believed to mainly act as a repressor (Wang et al., 2006).
A second understudied factor is the Nk homeodomain protein Nkx2.2.Within its 273 amino acids it contains a central DNA-binding homeodomain as well as a tinman domain on its aminoterminal and an Nk-specific domain on its carboxyterminal side (Zhang et al., 2020).
Nkx2.2 is upregulated at the very onset of oligodendrocyte differentiation in premyelinating oligodendrocytes, dependent in its expression on Olig2 and the HMG-domain transcription factor Sox10 and required for an efficient and timely execution of the differentiation process (Fu et al., 2002;Liu et al., 2007;Qi et al., 2001;Weider et al., 2018).While Nkx2.2 has been characterized as a transcriptional activator during pancreas development (Mio et al., 2023), most of its functions during oligodendrocyte development have been attributed to co-repressor recruitment and thus to a repressive mode of action (Zhang et al., 2020).
While the importance of Zfp488 and Nkx2.2 for oligodendrocyte differentiation is clearly documented, much less is known about their molecular mode of action and the target genes responsible for their effects.In an attempt to shed light on these aspects, we here study the overall effects of both transcription factors on the oligodendroglial expression profile and identify the myelination-associated G protein-coupled receptor 37 (Gpr37) as a target gene of both Zfp488 and Nkx2.2.Gpr37 is upregulated during oligodendrocyte differentiation and has been described to keep differentiation and myelination in check with potential impact on the transition from pre-myelinating to myelinating oligodendrocyte and the choice of myelination substrate (Qian et al., 2021;Yang et al., 2016).As a target of Zfp488 and Nkx2.2,Gpr37 expression is directly stimulated by both transcription factors.However, stimulation is dependent on the functional interaction of Zfp488 and Nkx2.2 with the transcription factor Sox10 further emphasizing the importance of the gene regulatory network for oligodendrocyte development.
Luciferase reporter plasmids containing potential regulatory regions from the Gpr37 locus were cloned by amplifying the respective sequences from mouse genomic DNA and inserting them between KpnI and NheI sites into pTATA-luc (Peirano & Wegner, 2000).

| RNA-sequencing (RNA-seq) and bioinformatical analysis
Total RNA was isolated from rat CG4 and primary oligodendroglial cells under proliferative conditions or after 3 days of differentiation using the RNeasy Micro Kit (Qiagen) according to the manufacturer's instructions.For overexpression of Nkx2.2 and Zfp488, CG4 cells were transduced with corresponding retroviruses.
RNA samples were quantified on a Qubit 4.0 Fluorometer (Life Technologies) and RNA integrity was confirmed on an Agilent 5300 Fragment Analyzer (Agilent Technologies).RNA sequencing libraries were prepared using the NEBNext Ultra RNA Library Prep Kit for Illumina following manufacturer's instructions (NEB).Libraries were multiplexed and sequenced on an Illumina NovaSeq 6000 instrument with a 2Â 100 or 150 bp pair-end configuration.On average, 51 million reads were generated per sample library and aligned to the Rattus norvegicus rn6 genome.DESeq2 was used to identify genes differentially expressed in samples from Zfp488 and Nkx2.2 overexpressing CG4 cells compared to controls.Differentially expressed genes (DEGs) were defined by a mean base count of ≥20 gene-specific transcripts per million overall transcripts (TPM), log2-fold change ≥±0.6, and an adjusted p-value ≤0.05.EdgeR was employed for DEG identification between CG4 cells and rat OPCs under proliferative and differentiating conditions.
Gene expression datasets are deposited in GEO under GSE244592.
GO-term analysis was by GOrilla with a p-value threshold of 10 À7 (Eden et al., 2009).Gene set enrichment analysis (GSEA) (Broad Institute, http://software.broadinstitute.org/gsea/index.jsp)was used to identify changes in expression of select gene sets.
Band intensities were measured with ImageJ.

| Mice
Transgenic and control mice were on a mixed C3H Â C57Bl/6J background and housed under standard conditions with 12 h dark-light cycles and continuous access to food and water according to animal welfare laws.Both male and female mice were used for analysis.A CNSspecific knockout of Sox10 was achieved by crossing Sox10 fl/fl mice (Finzsch et al., 2010) with Olig2 Cre/+ mice (Dessaud et al., 2007).Oligodendroglial deletion of Nkx2.2 was achieved by breeding Olig1 Cre/+ (Xin et al., 2005) mice with Nkx2.2 fl/fl mice (Mastracci et al., 2013).In the

Zfp488 knockout allele, a histone 2b-GFP fusion cassette replaces the
Zfp488 open reading frame (Q.R. Lu, unpublished).Pups were killed at postnatal day (P) 7, 20, or 30 to obtain spinal cord and brain tissue.After fixation in 4% paraformaldehyde, dehydration in 30% sucrose, and freezing, tissue sections were generated on a cryotome at 10 μm thickness.

| In situ hybridization
In situ hybridizations were performed on mouse transverse spinal cord (forelimb level) and sagittal brain sections following standard procedures (Stolt et al., 2002).DIG-labeled antisense riboprobes against Gpr37 (nucleotides 1864-2702 or the full-length sequence of NM_010338.2) were used and stainings documented on a Leica MZFLIII stereomicroscope equipped with an Axiocam (Zeiss).

| Statistical analysis
Biological replicates were defined as results from independent animals, experiments, or separately and independently generated samples.
Sample size was n ≥ 3 for all mice and molecular biology experiments.
Investigators were not blinded in animal experiments.GraphPad Prism7 (GraphPad software, La Jolla, CA, USA) was used for statistical testing using two-tailed Student's t-tests and analysis of variance (ANOVA) to determine differences in cell numbers, luciferase activity, band intensities, or immunoprecipitated DNA (*p ≤ .05;**p ≤ .01;***p ≤ .001).Variance between statistically compared groups was similar.Quantifications of generated data are presented as superplots (Lord et al., 2020).

| Overexpression of Zfp488 and Nkx2.2 induces Gpr37 expression in differentiating CG4 cells
Despite the fact that both Zfp488 and Nkx2.2 are upregulated in differentiating oligodendrocytes and strongly support oligodendrocyte differentiation, little is known about their molecular mode of action and their target genes during the process.To gain mechanistic insights, we took advantage of the CG4 cell line (Louis et al., 1992).
To convince ourselves that CG4 cells are a sound model for oligodendroglial cells and their differentiation, we first performed RNA-seq on CG4 cells kept for 3 days under proliferating or differentiating conditions and compared the expression profiles to proliferating and differentiating primary oligodendroglial cells.PCA plots showed that the expression profile of CG4 cells under proliferating conditions was very similar to the one from proliferating primary oligodendroglial cells, and that differentiation for 3 days led to similar changes in the expression profile of both CG4 cells and primary oligodendroglial cells (Figure 1a).The overall number of genes upand downregulated during differentiation (mean base count >20, log2-fold change ≥±1.5, p-value ≤0.05) was comparable in CG4 and primary oligodendroglial cells (Figure 1b,c).Approximately two thirds of the genes with preferential expression in proliferating CG4 cells were also preferentially expressed in proliferating primary oligodendroglial cells and vice versa (63% and 67%, respectively) (Figure 1d).The same strong overlap was also detected between CG4 cells and primary oligodendroglial cells under differentiating conditions (68% and 72%, respectively) (Figure 1e).GO analysis of the genes upregulated during differentiation yielded nearly identical top terms for CG4 cells and primary oligodendroglial cells that were predominantly related to lipid metabolism, cell adhesion, axonal ensheathment, and myelination (Figure 1f,g).
CG4 cells also exhibited differences to primary oligodendroglial cells in their expression profile and differentiation induced changes.
According to GO analysis, expression of genes associated with regulation of metabolic processes and response to oxygen were substantially altered during differentiation in CG4 cells, but not in primary oligodendroglial cells.This may reflect an adaptation of CG4 cells to long term growth in culture.Conversely, there were significant changes in the expression of genes associated with sterol biosynthesis and immune functions during differentiation in primary oligodendroglial cultures, but not in CG4 cells.It seems reasonable to assume that alterations in the expression of immune-related genes in primary oligodendroglial cultures are caused by microglia that are present as minor contaminants in these cultures.The selective upregulation of genes associated with sterol biosynthesis during differentiation of primary oligodendroglial cells may be indicative of a substantially decreased ability of CG4 cells to produce cholesterol during differentiation.Another indicator of a reduced differentiation capacity is the lower induction of myelin gene expression in CG4 than in primary oligodendroglial cells (e.g., log2-fold change of 3.5 vs. 6.4 for Plp1 and of 4.1 vs. 7.8 for Mbp, see Figure 1h), Despite these minor differences, CG4 cells have an expression profile very similar to primary oligodendroglial cells both under proliferative and differentiating conditions and represent a good model system to study oligodendroglial cells.Therefore, we used CG4 cells to generate polyclonal lines that overexpress either Zfp488 or Nkx2.2 by retroviral transduction and consecutive fluorescence-activated cell sorting of transduced cells (Figure 2a).CG4 cells and their transduced derivatives were kept 3 days under differentiating conditions before expression profiling by RNA-seq (Figure 2a).Overexpression of Zfp488 or Nkx2.2 was confirmed in the generated lines on protein and transcript level (Figure 2b-g).Samples from Zfp488 and Nkx2.2 overexpressing CG4 lines were both clearly separated from wildtype CG4 cells in PCA plots (Figure 2h).Nkx2.2 overexpression appeared to alter CG4 gene expression more dramatically than the presence of ectopic Zfp488 (Figure 2h).Comparison of the obtained expression profiles revealed that 1018 genes were differentially expressed (mean base count >20, log2-fold change ≥±0.6, p-value ≤0.05) in Zfp488-overexpressing CG4 cells as compared to controls.Of these, 719 were upregulated and 299 downregulated (Figure 2i).For Nkx2.2, the number of DEGs amounted to 1177, with 765 upregulated and 412 downregulated DEGs (Figure 2j).As both Zfp488 and Nkx2.2 are considered to frequently act as repressors (Rowitch et al., 2002;Zhang et al., 2020), we were surprised to find more up-than downregulated genes and on average higher induction rates among the upregulated genes.The 20 most deregulated genes for the Zfp488 and Nkx2.2 overexpressing CG4 lines are listed by name in the volcano blots (Figure 2k,l).Gene ontology (GO) studies of the DEGs upon Zfp488 overexpression pointed to alterations in regulatory processes, including response to stimuli, signal transduction, and G protein-coupled receptor signaling (Figure 3a).The same terms also ranked high in the analogous GO analysis of DEGs for Nkx2.2 overexpression (Figure 3b).To our surprise, terms related to oligodendrocytes and myelin were not present among the 25 most significant GO terms for the Zfp488 or Nkx2.2 overexpressing CG4 lines.
However, transcript levels for several myelin genes, including Plp1, Mbp, Cnp1, Mog, Mobp, and Mag were significantly increased following Zfp488 overexpression (Figure 3c).The same genes showed no or much less pronounced changes in their expression level upon Nkx2.2 overexpression (Figure 3d).In line, terms like oligodendrocyte differentiation, positive regulation of myelination, and myelin sheath were significantly enriched according to GSEA of Zfp488 overexpressing CG4 cells (Figure 3e), whereas only oligodendrocyte differentiation was enriched with borderline significance in Nkx2.2 overexpressing cells (Figure 3f).These results support a direct role of Zfp488 in myelination.At the same time, they argue that Nkx2.2 mainly targets other genes to influence the differentiation process in oligodendrocytes.
To further compare the consequences of Zfp488 and Nkx2.2 overexpression in CG4 cells, we determined the number of common DEGs (Figure 4a).Among the 240 shared DEGs, 163 were similarly up-or downregulated in both overexpressing CG4 lines.Gpr37 expression increases during oligodendrocyte differentiation (Figure 4c).Its upregulation in both Zfp488 and Nkx2.2 overexpressing CG4 lines is also very robust, whereas expression of its close paralog Gpr37l1 remained unaffected in Zfp488 overexpressing cells and was even decreased in cells with ectopic Nkx2.2 expression (Figure 4d,e).Considering that Gpr37 has been shown to impact oligodendrocyte development as a modulator of terminal differentiation and myelination (Qian et al., 2021;Yang et al., 2016), we focused on Gpr37 as a potential target of Zfp488 and Nkx2.2 in differentiating oligodendrocytes.

| Gpr37 expression in oligodendrocytes depends on Zfp488 and Nkx2.2 in vivo
To study the relationship between Gpr37 and its potential regulators in vivo, we first assessed oligodendroglial Gpr37 expression in mice deficient for Zfp488.These mice exhibit a delay in developmental myelination in both brain and spinal cord (Q.R. Lu, unpublished).Using an antisense riboprobe for Gpr37, we detected Gpr37-positive cells in the spinal cord and corpus callosum of wildtype mice at P7 (Figure 5a,c).Zfp488-deficient mice, in contrast, had substantially fewer Gpr37-positive cells in both CNS regions (Figure 5b,d).Quantification confirmed the strong and statistically significant reduction of Gpr37-positive cells in Zfp488-deficient mice at P7 by 48 ± 1% in the spinal cord and by 61 ± 2% in the corpus callosum (Figure 5e,f).By P20, the number of Gpr37-positive cells had increased slightly in the spinal cord and more robustly in the corpus callosum of wildtype mice (Figure 5g,h).In Zfp488-deficient mice, Gpr37-expressing cells were still fewer.
However, the difference to the wildtype had decreased to 6 ± 2% in the spinal cord and to 8 ± 3% in the corpus callosum.Statistical significance was no longer observed at P20.We conclude from these data that Zfp488 is transiently required to ensure normal Gpr37 expression levels during oligodendrocyte differentiation.
For analyzing the relationship between Nkx2.2 and Gpr37, we used mice that carried a combination of floxed Nkx2.2 alleles and an Olig1-cre driver (C.Zhang & M. Qiu, unpublished).These mice lack Nkx2.2 in oligodendroglial cells throughout the CNS.When in situ hybridization was performed on spinal cord and corpus callosum of these mice at P7, a substantial reduction of Gpr37-positive cells was noted as compared to age-matched controls (Figure 5i-l).Quantifications revealed that the reduction amounted to approximately 58 ± 13% for spinal cord and 77 ± 6% for the corpus callosum (Figure 5m,n).Interestingly, Gpr37-positive cells were still significantly reduced in the corpus callosum of Nkx2.2 mutant mice at P30 (by 48 ± 6%, Figure 5o).Likewise, RNA-seq studies on the diencephalon of wildtype and Nkx2.2 mutant mice at P14 (M.Qiu, unpublished) revealed a 2.3-fold reduction of Gpr37 transcript levels in Nkx2.2 mutant as compared to wildtype mice (Figure 5p).These data provide strong evidence that also Nkx2.2 has a substantial impact on Gpr37 expression in differentiating oligodendrocytes in vivo.

| The Gpr37 gene locus contains several potential regulatory regions
To investigate whether Gpr37 is under direct transcriptional control of Zfp488 and Nkx2.2, we used UCSC cell and ECR browsers to screen 50 kilobases (kb) of the Gpr37 locus for regions that localize within easily accessible, open chromatin in mature oligodendrocytes (http://cells.ucsc.edu/?ds=olg-eae-ms+eae-atac+peaks#) according to scATAC-seq data (Meijer et al., 2022) (Figure 6a) and are evolutionary conserved between human, rat and mouse by at least 70% (https://ecrbrowser.dcode.org/).Three such evolutionary conserved regions (ECRs) in open chromatin were identified and localized 19 kb upstream the transcriptional start site (ECR À19 kb) as well as 3.5 and 31 kb downstream the transcriptional start site (ECR +3.5 kb and ECR +31 kb) (Figure 6b-e).As expected and in line with its localization within the transcribed region, ECR +3.5 kb showed the highest accessibility (Figure 6c,e).This was followed by ECR +31 kb (situated immediately behind the Gpr37 F I G U R E 2 Consequences of Zfp488 and Nkx2.2 overexpression on the expression profile of differentiating CG4 cells.(a) Schematic overview of the experimental approach.CG4 cells were transduced with retroviruses encoding EGFP alone or in combination with myc-tagged Zfp488 or T7-tagged Nkx2.2, sorted by FACS, expanded and differentiated for 3 days (3 d diff).Subsequently, total RNA was prepared and used for bulk RNA sequencing (RNA-seq).(b, c) Immunoblotting was performed on whole cell lysates from different batches of control (ctrl) cells or independently generated CG4 cell lines expressing ectopic Nkx2.2 or Zfp488.Overexpressed Zfp488 was detected with an anti-myc antibody, Nkx2.2 with an anti-Nkx2.gene) with still a very high accessibility (Figure 6d,e) and ECR À19 kb with a more differential one (Figure 6b).These three ECRs were inserted into a luciferase reporter under control of a minimal promoter.Upon transfection in CG4 cells, only ECR +3.5 kb and ECR +31 kb led to statistically significant increases in reporter gene expression as compared to the minimal promoter alone, with ECR +31 kb being by far the most active (Figure 6f-h).We conclude from the combination of these data that at least ECR +31 kb and possibly ECR +3.5 kb represent candidates for oligodendroglial regulatory regions of the Gpr37 gene.

| Zfp488 and Nkx2.2 cooperate with Sox10 to activate a Gpr37 downstream enhancer
To analyze the effect of Zfp488 and Nkx2.2 on the three Gpr37-related ECRs, we performed reporter gene assays in transiently transfected Neuro2a cells.Despite substantial variations in assay conditions, neither the À19 kb ECR, the +3.5 kb ECR nor the +31 kb ECR exhibited responsiveness towards the presence of Zfp488 or Nkx2.2 (Figure 6i-k and data not shown).
We had previously discovered that Myrf as a central regulator of oligodendrocyte differentiation and myelination (Emery et al., 2009) is not active on its own as a transcription factor on oligodendrocytespecific regulatory regions in Neuro2a cells.Instead it required the oligodendroglial lineage determinant Sox10 as a cooperating partner (Aprato et al., 2020;Hornig et al., 2013).To look into the possibility that the presence of Sox10 is also relevant for Zfp488 and Nkx2.2 activity, we analyzed the effect of Sox10 on Gpr37 expression.
A comparison of RNA-seq data (GEO accession number GSE136659) from oligodendroglial Oln93 cells and derivatives that had been genome-edited to inactivate the endogenously expressed Sox10 (Aprato et al., 2020;Reiprich et al., 2017) revealed that Oln93 cells contain substantially reduced Gpr37 transcript levels in the absence of Sox10 (Figure 7a).This argues that Gpr37 expression is dependent on Sox10 at least in this oligodendroglial cell line.
For verification in vivo, we performed in situ hybridization on spinal cord and corpus callosum of Sox10 fl/fl Olig2 Cre/+ mice that lacked Sox10 expression in oligodendroglial cells throughout the CNS.At P7, Gpr37-positive cells were clearly visible in the white matter of control mice but nearly absent in age-matched Sox10 mutant mice (Figure 7b-e).Quantification revealed a reduction in their number by 87 ± 5% in the spinal cord and by 98 ± 1% in the corpus callosum of Sox10 mutant mice (Figure 7f,g).Therefore, Gpr37 expression is also under control of Sox10 in vivo.
To investigate whether Sox10 effects on Gpr37 are direct and involve the identified ECRs, we next performed ChIP-PCR assays to look for enrichment of Sox10 on these ECRS in oligodendroglial chromatin (Figure 7h).In these experiments, we detected a substantial binding of Sox10 on ECR +3.5 kb and ECR +31 kb as evident from the fact that both ECRs were significantly enriched in the chromatin precipitated with anti-Sox10 antibodies relative to the pre-immune control.In contrast, no such enrichment was detected for ECR À19 kb and two unrelated genomic regions that served as negative controls.
Considering the ability of Sox10 to bind to some of the ECRs, we next employed luciferase assays in transiently transfected Neuro2a cells to study whether Sox10 is also able to activate transcription from these regions.In good agreement with the results from ChIP assays, ECR +3.5 kb and ECR +31 kb were both responsive to Sox10, whereas no Sox10-dependent effect was observed for ECR À19 kb (Figure 7i-k).Interestingly, Sox10-dependent activation through ECR +31 kb was at least an order of magnitude higher than activation through ECR +3.5 kb arguing that ECR +31 kb may be the functionally most relevant enhancer.To study whether the presence of Sox10 renders ECR +3.5 kb or ECR +31 kb susceptible to the action of Zfp488 or Nkx2.2, we performed co-transfections with luciferase reporters in Neuro2a cells and added the transcription factors in combination with Sox10.For ECR +3.5 kb, we did not detect any effect on the Sox10-dependent activity, independent of whether Zfp488 or Nkx2.2 was added (Figure 8a,c).When ECR +31 kb was analyzed instead, we reproducibly observed a 26 ± 9% increase of reporter gene activity in the presence of Sox10 and Zfp488 as compared to Sox10 alone (Figure 8b).In case of Nkx2.2, the increase amounted to 21 ± 5% (Figure 8d).No further increase was obtained in triple combinations of Sox10, Zfp488, and Nkx2.2.We conclude that Zfp488 as well as Nkx2.2 can enhance the ability of Sox10 to increase expression through the Gpr37-related ECR +31 kb.
Cut&Tag-seq data for Zfp488 (S.Cui, Y. Lin & Q.R. Lu, unpublished) furthermore confirmed the presence of Zfp488 on ECR +31 kb in oligodendrocytes, but not on ECR +3.5 kb (Figure 8e and data not shown).As comparable data were not available for Nkx2.2 in oligodendrocytes, we performed ChIP-PCR on CG4 cells transduced with a T7-tagged Nkx2.2.By using an antibody directed against the T7-tag we detected strong binding of Nkx2.2 on ECR +31 kb (Figure 8f).In contrast, neither ECR À19 kb, ECR +3.5 kb nor two unrelated genomic regions exhibited a significant enrichment in the chromatin precipitated with the anti-T7-tag antibody.In conclusion, ECR +31 kb is the only of the three identified ECRs that binds Sox10, Zfp488, and Nkx2.2 in vivo and exhibits an increased transcriptional response to Sox10 in combination with Zfp488 and Nkx2.2.

| DISCUSSION
While the zinc-finger transcription factor Zfp488 and the homeodomain transcription factor Nkx2.2 are clearly involved in oligodendrocyte differentiation (Qi et al., 2001;Wang et al., 2006), only few data exist on their molecular mode of action.By performing RNA-seq studies, we here describe the overall changes in the expression pattern of CG4 cells that are caused by overexpression of Zfp488 or Nkx2.2 under differentiating conditions.It has to be kept in mind that overexpression represents an artificial approach and that differences exist between CG4 cells as an F I G U R E 7 Gpr37 expression is dependent on Sox10.(a) Gpr37 expression in control and Sox10-deficient (Sox10 ko) Oln93 cells as FPKM values according to published RNA-seq data (Reiprich et al., 2017).(b-g) In situ hybridization for Gpr37 on spinal cord (b, c, f) and corpus callosum (d, e, g) sections of control (b, d, ctrl) and Sox10 knockout (c, e, Sox10ko) mice at P7 and respective quantifications as cells per mm 2 (f, g).Scalebar: 100 μm.(h) ChIP experiments on wildtype CG4 cells.Chromatin was precipitated with a Sox10 antiserum to detect Sox10 occupancy on ECR À19 kb, ECR +3.5 kb, ECR +31 kb or two negative control regions (neg1, neg2).The corresponding pre-immune serum served as a control (ctrl).Amounts of precipitated chromatin were determined as percent of input recovered and values for control samples set to 1. (i-k) Reporter assays were performed in Neuro2a cells with luciferase plasmids carrying ECR À19 kb (i), ECR +3.5 kb (j) or ECR +31 kb (k) in the absence or presence of Sox10 expression plasmids.Luciferase values from cells transfected with empty expression vector were set to 1. Data represent mean values ± SEM.Statistical analysis was performed with Student's two-tailed t-test (f, g, i-k) or two-way ANOVA (h) (* p ≤ .05).established cell line and cultured primary oligodendroglial cells.Nevertheless, our data also show that the expression profile of CG4 cells bears substantial resemblance to primary oligodendroglial cells both in the proliferating state and under differentiating conditions.Differentiating CG4 cells upregulate many of the same genes as differentiating primary oligodendroglial cells, including endogenous Zfp488, Nkx2.2, and the typical myelin proteins.
Interestingly, classical myelin genes are not among the genes most upregulated by Zfp488 in differentiating CG4 cells.The link between Nkx2.2 and myelin gene expression appears even weaker.Instead, genes associated with response to external stimuli, cell surface receptor signaling as well as signal transduction are much more prevalent.While this may at least in part be a consequence of the chosen approach and cell line, it may also indicate that these two transcription factors predominantly act by modulating signaling pathways or altering specific cellular properties that in turn promote oligodendrocyte differentiation and myelination.Such specific influences on select processes and pathways could also explain why differentiating oligodendrocytes have to express a whole series of zinc finger containing transcription factors including Sp7, Sip1, Zfp276, and Klf9/Klf13 that each perform at least partially unique functions (Aberle et al., 2022;Bernhardt et al., 2022;Dugas et al., 2012;Weng et al., 2012).It is also noteworthy that there is a substantial, but far from complete overlap between the Zfp488and Nkx2.2-specificDEGs as would be predicted if some functions are shared between both transcription factors in a complex gene regulatory network but not others.
Previous work had indicated that both Zfp488 and Nkx2.2 act as transcriptional repressors in differentiating oligodendrocytes (Wang et al., 2006;Weider et al., 2018;Zhang et al., 2020).Therefore it came as a surprise that we found more up-than downregulated genes upon Zfp488 or Nkx2.2 overexpression in our RNA-seq experiments.Additionally, upregulated genes were overrepresented among the genes with a particular strong deregulation.Although RNA-seq experiments cannot distinguish between direct and indirect target genes, this pattern still strongly suggests that both Zfp488 and Nkx2.2 do not exclusively act as transcriptional repressors during oligodendrocyte differentiation but may also function as transcriptional activators.
One of the genes that was significantly upregulated upon overexpression of both Zfp488 and Nkx2.2 was Gpr37.In line with an effector-target gene relationship, Gpr37 expression was also strongly reduced in Zfp488-and in Nkx2.2-deficient oligodendrocytes in vivo.
In Zfp488-deficient mice, this reduction was transient and limited to Given its function as a cell surface receptor with established roles in oligodendrocyte differentiation and myelination, Gpr37 appeared suited to learn more about the molecular mode of transcriptional activation by Zfp488 and Nkx2.2 in oligodendrocytes.By using evolutionary conservation and chromatin accessibility as criteria, we identified three potential regulatory regions of the Gpr37 gene.Unexpectedly, none of these regions was responsive to Zfp488 or Nkx2.2 alone in standard reporter gene assays after transient transfection of Neuro2a cells.We have previously observed a similar behavior of Myrf on myelin gene-related regulatory regions in reporter gene assays in heterologous cell lines (Aprato et al., 2020;Hornig et al., 2013;Muth et al., 2016) despite the fact that Myrf is required for full activation of myelin genes in vivo (Bujalka et al., 2013;Emery et al., 2009).In case of Myrf, we had furthermore found that myelin gene activation requires the simultaneous presence of Sox10.For that reason, we looked into the consequences of Sox10 co-expression.Intriguingly, we found that Gpr37 is also under transcriptional control of Sox10 and that one of the three Gpr37-associated regulatory regions is robustly activated by Sox10.In the presence of Sox10, this ECR +31 kb can be additionally activated by Zfp488 and Nkx2.2 arguing that the two transcription factors require oligodendrocyte-specific cooperation partners to implement their activity as transcriptional activators.
According to ChIP experiments ECR +31 kb is bound by Zfp488, Nkx2.2, and Sox10.This argues for a model where Sox10 and Zfp488 or Nkx2.2 have to be bound simultaneously or where Sox10 has to be present on the regulatory region to permit consecutive binding of Zfp488 or Nkx2.2.Given the fact (i) that ChIP-validated antibodies for Zfp488 and Nkx2.2 are not commercially available, (ii) that the exact binding motif for Zfp488 has not yet been established, and (iii) that the core consensus motif 5 0 -CAAT-3 0 for Nkx2.2 (Mio et al., 2023) occurs frequently and does not unambiguously identify Nkx2.2 binding sites, it is currently difficult to establish further in-depth insights into the mechanism of Zfp488-or Nkx2.2-dependent activation.
Nevertheless, our current study succeeds in better defining the role of Zfp488 and Nkx2.2 in the gene regulatory network of differentiating oligodendrocytes.

K
E Y W O R D S glia, myelin, seven transmembrane receptor, Sox, transcription factor 1 | INTRODUCTION Myelin sheaths allow saltatory conduction and are generated in the central nervous system (CNS) by oligodendrocytes as part of their terminal differentiation program.Over the years, a number of transcription factors have been identified that are part of the gene regulatory network and direct terminal differentiation in oligodendrocytes (for a recent review see using PolyJet (Signagen, SL100688) according to the manufacturer's instructions.The pCAG-IRES-EGFP carries an expression cassette under control of the artificial CAG promoter (consisting of CMV immediate early enhancer and chicken β-actin promoter) placed between the retroviral 5 0 -LTR and 3 0 -LTR sequences.Insertion of a cDNA behind CAG promoter and in front of IRES-EGFP allows co-expression of cDNA-encoded protein and EGFP in resulting viruses.In addition to pCAG-IRES-EGFP, the pCAG-myc-Zfp488-IRES-EGFP and pCAG-T7-Nkx2.2-IRES-EGFPplasmids were used.Generated retroviruses were enriched from the supernatant of transfected 293GPG cells by ultracentrifugation.CG4 cells were transduced at an MOI of 1 in basal medium with PDGF-AA and bFGF.Transduced CG4 cells were isolated by FACS for EGFP fluorescence and used to establish Nkx2.2-andZfp488-expressing polyclonal cell lines.

F
I G U R E 1 Comparison of CG4 cells and primary rat oligodendroglial cells.(a) PCA plot showing differential clustering of proliferative primary rat oligodendroglial cells (OPC, pink), proliferative CG4 cells (light green), differentiating oligodendroglial cells (OL, purple) and differentiating CG4 cells (green) after 3 days in vitro.(b, c) Pie charts depicting the number of genes with altered expression under differentiating as compared to proliferative conditions (log2-fold change >±1.5, false discovery rate (FDR) < 0.05, TPM > 20) in CG4 cells (b) and rat OPCs (c).(d, e) Venn diagram of genes preferentially expressed in primary rat oligodendroglia and CG4 cells under proliferative (d) or differentiating (e) conditions.Genes in the overlapping area are preferentially expressed in a specific state in both cell types and exhibit a log2-fold change of >1.5 in expression in at least one of the two cell types.(f, g) GO term analysis for genes upregulated in CG4 (f) and rat oligodendroglial cells (g) during differentiation.Terms are sorted by their FDR value.(h) Comparison of induction rates of select myelin genes during differentiation in CG4 and primary oligodendroglial cells, shown as log2-fold change (log2FC) of expression relative to proliferating cells (set to 1).Data represent mean values ± SEM and were statistically analyzed by Student's two-tailed t-test (**p ≤ .01;***p ≤ .001).F I G U R E 2 Legend on next page.This included 128 upregulated and 35 downregulated genes.The 30 most significantly upregulated DEGs shared between Zfp488 and Nkx2.2 are shown in Figure 4b.This includes Gpr37, a gene associated with GO terms such as response to external stimuli, signal transduction, cell surface receptor signaling, and G proteincoupled receptor signaling (see Figure 3a,b).
2 antibody.Gapdh served as a loading control.Numbers on the right indicate expected molecular weight of proteins.(d, f) Densitometric analysis of the western blots from (b, c).Analysis was performed with ImageJ and normalization was to Gapdh band intensity.Normalized band intensity of control lysates for Zfp488 and Nkx2.2 was arbitrarily set to 1. (e, g) Relative expression values of Zfp488 and Nkx2.2 from RNA-seq.Values represent 10 3 specific transcripts per million general transcripts (TPM).(h) PCA plot showing differential clustering of RNAseq data from Zfp488 (dark gray, h) and Nkx2.2 (light gray, j) overexpressing CG4 cells compared to controls (ctrl, white).(i, j) Pie charts depicting the number of up-(red) or downregulated (blue), differentially expressed genes (DEG) (log2-fold change >±0.6, p < 0.05, mean base count >20 TPM).(k, l) Volcano plots depicting these DEGs with the top 20 up-and downregulated genes annotated with their gene symbol.Data in d-g represent mean values ± SEM and were statistically analyzed by Student's two-tailed t-test (***p ≤ .001).F I G U R E 3 Zfp488 and Nkx2.2 affect signaling pathways more than myelination in CG4 cells.(a, b) GO-term analysis on differentially expressed genes (log2-fold change >±1, p < .05).Depicted are the top 25 GO terms sorted by their FDR value.(c, d) Relative transcript levels of selected myelin genes and oligodendrocyte lineage factors in Zfp488 (dark gray, c) and Nkx2.2 (light gray, d) overexpressing CG4 cells compared to controls (white).TPM values of control cells for each gene were arbitrarily set to 1. (e, f) Gene set enrichment analysis (GSEA) was performed on gene sets associated with "oligodendrocyte (OL) differentiation," "positive regulation (pos.reg.) of myelination," and "myelin sheath" on ranked gene lists of Zfp488 (e) and Nkx2.2 (f) overexpressing CG4 cells.Data in c,d represent mean values ± SEM and were statistically analyzed by Student's two-tailed t-test (*p ≤ .05;**p ≤ .01;***p ≤ .001).

F
I G U R E 5 Gpr37 expression is dependent on Zfp488 and Nkx2.2 in vivo.(a-d) In situ hybridizations for Gpr37 on spinal cord (a, b) and corpus callosum (c, d) of control (a, c) and Zfp488 knockout (Zfp488 ko) (b, d) mice at P7. (e-h) Quantifications of Gpr37-positive cells in spinal cord (e, g) and corpus callosum (f, h) sections of control and Zfp488 ko mice at P7 (e, f) and P20 (g, h) as cells per mm 2 .(i-l) In situ hybridizations for Gpr37 on spinal cord (i, j) and corpus callosum (k, l) sections of control (i, k) and Nkx2.2 knockout (Nkx2.2ko) (j, l) mice at P7. (m-o) Quantifications of Gpr37-positive cells in spinal cord (m) and corpus callosum (n,o) of control and Nkx2.2 ko mice at P7 (m, n) and P30 (o).(p) Expression data for Nkx2.2 from RNA-seq of diencephalon samples of Nkx2.2 ko and control mice at P14. Shown is the normalized count value for Gpr37 in FPKM.Data represent mean values ± SEM.Statistical analysis was performed with Student's two-tailed t-test (*p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001).Scalebars: 100 μm.

F
I G U R E 6 The Gpr37 locus contains several potential gene regulatory regions.(a) Section of UMAP plot highlighting different cell populations including mature oligodendrocytes (boxed) taken from the scATAC-seq study of Meijer et al. (2022).MiGL, microglia and related cells; OPC, oligodendrocyte progenitor cell; VLMC, vascular leptomeningeal cell.(b-d) Analysis of accessibility of three different regions from the Gpr37 locus (b, ECR À19 kb; c, ECR +3.5 kb; d, ECR +31 kb) in mature oligodendrocytes according to scATAC-seq with peak heights in single cells.(e) Schematic overview of the Gpr37 locus.Shown is the Gpr37 gene, its 2 exons (light blue), the transcriptional start site (TSS) and three evolutionary conserved regions (ECRs) in open chromatin (orange).(f-h) Reporter assays were performed in CG4 cells with luciferase plasmids carrying ECR À19 kb (f), ECR +3.5 kb (g) or ECR +31 kb (h).The relative transactivation obtained with an empty reporter plasmid was set to 1. (i-k) Reporter assays were performed in Neuro2a cells with luciferase plasmids carrying ECR À19 kb (i), ECR +3.5 kb (j) or ECR +31 kb (k) in the presence or absence of Nkx2.2 or Zfp488 expression plasmids.The relative transactivation obtained with empty expression plasmid cells was set to 1. Data represent mean values ± SEM.Statistical analysis was performed with Student's two-tailed t-test (**p ≤ .01).

F
I G U R E 8 Nkx2.2 and Zfp488 cooperate with Sox10 to activate Gpr37 ECR +31 kb.(a-d) Reporter assays were performed in Neuro2a cells with luciferase plasmids carrying ECR +3.5 kb (a, c) or ECR +31 kb (b, d) and various combinations of Sox10, Zfp488 (a, b), and Nkx2.2 (c, d) expression plasmids.Luciferase values from cells transfected with empty expression vector were set to 1. (e) Trace of Zfp488 occupancy (red) at the end of the Gpr37 gene according to Cut&Tag-seq studies (Q.R. Lu, unpublished).ECR +31 kb downstream of Gpr37 exon 2 (dark blue) is highlighted in light blue.(f) ChIP experiments on CG4 cells expressing T7-Nkx2.2.Chromatin was precipitated with an anti-T7-tag antibody to detect Nkx2.2 occupancy on Gpr37 ECRs (ECR À19 kb, ECR +3.5 kb, and ECR +31 kb) or two negative control regions (neg1, neg2).Total mouse IgG served as a control (ctrl).Amounts of precipitated chromatin were determined as percent of input recovered and values for control samples set to 1. Data represent mean values ± SEM.Statistical analysis was performed with Student's two-tailed t-test (a-d) or two-way ANOVA (f) (*p ≤ .05;**p ≤ .01;***p ≤ .001). the first three postnatal weeks arguing that other transcription factors likely compensate for the absence of Zfp488 at later times.