Uncovering oligodendrocyte enhancers that control Cnp expression

Abstract

Oligodendrocytes (OLs) produce myelin sheaths around axons in the central nervous system (CNS). Myelin accelerates the propagation of action potentials along axons and supports the integrity of axons. Impaired myelination has been linked to neurological and neuropsychiatric disorders. As a major component of CNS myelin, 2',3'-cyclic nucleotide 3'-phosphodiesterase (CNP) plays an indispensable role in the axon-supportive function of myelin. Notably, this function requires a high-level expression of CNP in OLs, as evidenced by downregulated expression of CNP in mental disorders and animal models. Little is known about how CNP expression is regulated in OLs. Especially, OL enhancers that govern CNP remain elusive. We have recently developed a powerful method that links OL enhancers to target genes in a principled manner. Here, we applied it to Cnp, uncovering two OL enhancers for it (termed Cnp-E1 and Cnp-E2). Epigenome editing analysis revealed that Cnp-E1 and Cnp-E2 are dedicated to Cnp. ATAC-seq and ChIP-seq data show that Cnp-E1 and Cnp-E2 are conserved OL-specific enhancers. Single cell multi-omics data that jointly profile gene expression and chromatin accessibility suggest that Cnp-E2 plays an important role in Cnp expression in the early stage of OL differentiation while Cnp-E1 sustains it in mature OLs.

Keywords: Cnp; enhancers; myelin; oligodendrocyte; transcription.

Figure 1

A systematic method to find OL enhancers for Cnp.

Figure 2

TAD analysis for Cnp. Public Hi-C data for human OLs, human mammary epithelial cells (HMECs), and mouse neural progenitors (NPs) were analyzed. On the diagonal is the genome. Off the diagonal, the interaction strength between two loci is indicated by color tone. White means no interaction, and orange (panels A and D) and red (the other four panels) the strongest interaction. The CNP/Cnp promoter locations are marked by thin crossing lines. The CNP/Cnp TAD is marked by a blue box in panels D–F. A larger TAD containing it is marked by a green box. Please see Materials and Methods for the sources of these data. These figures were generated by Juicebox [78,79] and HiGlass [80]. Magnified versions of these figures are available in Supplementary Material, Fig. S2.

Figure 3

Three Cnp ECs. Rat OL ChIP-seq data were compiled for them. Also shown are the locations of three negative control (NC) regions for epigenome editing analysis. Of note, the region shown here is fully contained in the Cnp TAD demarcated in Fig. 2. iOL: immature OL. mOL: mature OL. SC: spinal cord. OL^# and OL^*: OLs treated with vehicle and lysophosphatidylcholine, respectively. For the Myrf ChIP-seq data, only peak locations are shown because the raw data is not available. The mouse Zfp24 and Klf6 ChIP-seq data were mapped to the rat genome by LiftOver. Please see Materials and Methods for the sources of these data. This figure was generated by the IGV browser [81].

Figure 4

CRISPRi interrogation of Cnp ECs. (A) RT-qPCR analysis of Cnp expression in Oli-neu cells after CRISPRi knockdown of Cnp ECs. Shown are data points (biological replicates) and their mean and standard error. ^*p < 4.17 × 10⁻² by Student’s t test with Bonferroni correction. (B) Same as panel A, except that dCas9-KRAB was brought to each target by individual gRNAs. Shown are data points (biological replicates) and their mean and standard error. ^*p < 3.01 × 10⁻² by Student’s t test with Bonferroni correction. (C) Effect of simultaneously silencing EC1 and EC2 on the expression of Cnp in Oli-neu cells. Shown are data points (biological replicates) and their mean and standard error. ^*p < 4.33 × 10⁻³ by Student’s t test with Bonferroni correction. (D) Quantitative immunofluorescence of Cnp expression in mouse OLs after CRISPRi knockdown of EC1 and EC2. Shown are representative images of the 10 samples. Scale bar, 50 μm. Zoomed in images are available in Supplementary Material, Fig. S3. (E) EGFP and Cnp signals were quantified for individual cells by CellProfiler and compared among the samples. The number of EGFP-positive cells analyzed is as follows: Scr1 (84), Scr2 (106), Pro1 (86), Pro2 (86), EC1-G1 (62), EC1-G3 (61), EC2-G1 (52), EC2-G3 (81), EC1&2-G1 (84), and EC1&2-G3 (85). ^*p < 1.86 × 10⁻² by Student’s t test with Bonferroni correction for comparison with Scr1. AU: arbitrary unit.

Figure 5

Two genes in the vicinity of the CNP/Cnp TAD. (A) Expression profile of Dnajc7 in brain cell types [23]. AS: astrocytes. iOL: immature OLs. mOL: mature OLs. MG: microglia. Endo: endothelial cells. (B) RT-qPCR analysis of Dnajc7 expression in Oli-neu cells after CRISPRi knockdown of EC1 and EC2. Shown are data points (biological replicates) and their mean and standard error. (C) Expression profile of Nkiras2 in brain cell types [23]. (D) RT-qPCR analysis of Nkiras2 expression in Oli-neu cells after CRISPRi knockdown of EC1 and EC2. Shown are data points (biological replicates) and their mean and standard error.

Figure 6

Cell type specificity of EC1 and EC2. (A) Luciferase assay results for EC1 and EC2 that were transfected into mouse OPCs cultured in the differentiation conditions for 2 days. Shown are data points (biological replicates) and their mean and standard error. ^*p < 1.45 × 10⁻³ by Student’s t test with Bonferroni correction. EC1 and EC2 are henceforth referred to as Cnp-E1 and Cnp-E2, respectively. (B) Human brain cell type-specific ChIP-seq and ATAC-seq data for Cnp-E1 and Cnp-E2. (C) Mouse single-cell ATAC-seq data for Cnp-E1 and Cnp-E2. Please see Materials and Methods for the sources of these data.

Figure 7

Temporal dynamics of Cnp-E1 and Cnp-E2. (A) Expression dynamics of OPC and OL marker genes during OL lineage progression, as defined by the monocle pseudotime analysis of the ISSAAC-seq mouse brain data. (B) Corresponding accessibility dynamics of Cnp-E1. Left: the ISSAAC-seq data visualized by the IGV browser. The y-axis represents the magnitude of the ATAC-seq signal, which was normalized by the number of cells in each group. Right: Cnp-E1 activity quantification. (C) Corresponding accessibility dynamics of Cnp-E2. Left: the ISSAAC-seq data visualized by the IGV browser. The y-axis represents the magnitude of the ATAC-seq signal, which was normalized by the number of cells in each group. Right: Cnp-E2 activity quantification. Please see Materials and Methods for the sources of these data.