PAD2-Mediated Citrullination Contributes to Efficient Oligodendrocyte Differentiation and Myelination

Abstract

Citrullination, the deimination of peptidylarginine residues into peptidylcitrulline, has been implicated in the etiology of several diseases. In multiple sclerosis, citrullination is thought to be a major driver of pathology through hypercitrullination and destabilization of myelin. As such, inhibition of citrullination has been suggested as a therapeutic strategy for MS. Here, in contrast, we show that citrullination by peptidylarginine deiminase 2 (PAD2) contributes to normal oligodendrocyte differentiation, myelination, and motor function. We identify several targets for PAD2, including myelin and chromatin-related proteins, implicating PAD2 in epigenomic regulation. Accordingly, we observe that PAD2 inhibition and its knockdown affect chromatin accessibility and prevent the upregulation of oligodendrocyte differentiation genes. Moreover, mice lacking PAD2 display motor dysfunction and a decreased number of myelinated axons in the corpus callosum. We conclude that citrullination contributes to proper oligodendrocyte lineage progression and myelination.

Keywords: PAD2; Padi2; SILAC; chromatin accessibility; citrullination; deamination; multiple sclerosis; myelin; oligodendrocytes; proteomics.

Graphical abstract

Figure 1

Padi2 Expression Is Substantially Increased upon OL Differentiation (A) Schematic representation of the methodology used for OPC in vitro cultures. P1–P4 GFP⁺ OPCs are dissociated from brains of the transgenic mice line Pdgfra-H2B-GFP and FACS-sorted to plates to expand in the presence of growth factors (GFs). GFs are removed to induce differentiation for 2 days. (B) Comparative gene expression analysis of OPCs and 2 day differentiated OLs. Means ± SEM are shown, n = 3; ^∗p < 0.05, two-tailed t test. (C) Schematic representation of the methodology used to specifically isolate OPCs and juvenile and adult OLs from the postnatal (P1–P4), juvenile (P21), and adult (P60) brains of the transgenic mice PdgfraCre;RCE:loxP (R26R CAG-boosted EGFP); GFP⁺ cells were depleted of the OPC marker CD140a to specifically isolate OLs. (D) Comparative gene expression analysis of OPCs and juvenile and adult OLs. Means ± SEM are shown, n = 4; ^∗p < 0.05, one-way non-parametric ANOVA. (E) Western blot for PADI2 and MBP on the spinal cords of P1, P7, P14, P21, and adult wild-type mice. ACTIN signal is an internal loading control. See also Figure S1.

Figure 2

PAD Inhibition and Padi2 Knockdown Hinders OPC Differentiation (A) Immunocytochemistry for CNPase and cleaved Casp3 on 2 day differentiated Oli-neu upon 2 days treatment with DMSO or Cl-amidine 200 μM and quantification of the percentage of cleaved Casp3⁺ cells out of total DAPI in all conditions. Scale bar, 20 μm. (B) Comparative gene expression analysis of proliferating and undifferentiated (Und) Oli-neu cells and 2 day differentiated Oli-neu cells (Diff) treated with DMSO or Cl-amidine 200 μM for 2 days. Means ± SEM are shown, n = 3; ^∗p < 0.05, two-tailed t test. (C) Gene expression analysis on proliferating rat primary OPCs treated with Cl-amidine 100 μM; dashed line represents control levels, normalized to 1. Means ± SEM are shown, n = 3; ^∗p < 0.05, one-sample t test. (D) Comparative gene expression analysis on mouse primary undifferentiated (Und) and 2 day differentiated OLs (Diff) transfected with either scrambled siRNA or Padi2 siRNA targeting exon 2. Means ± SEM are shown, n = 4; ^∗p < 0.05, two-tailed t test; error bars represent SEM. See also Figure S1.

Figure 3

Padi2 Knockout Mice Display a Transient Decrease in OPC Differentiation in the Spinal Cord (A) CC1⁺ cells are presented as number of positive cells per area in the corpus callosum, anterior commissure, and spinal cord of PdgfraCre;RCE:loxP;Padi2^+/+ and PdgfraCre;RCE:loxP;Padi2^−/− juvenile (P21) and adult mice. Means ± SEM are shown, n = 4 (P21) and n = 5 (adult); ^∗p < 0.05, two-tailed t test. (B) Representative images of the spinal cord dorsal funiculus of PdgfraCre;RCE:loxP;Padi2^+/+ and PdgfraCre;RCE:loxP;Padi2^−/− mice stained with CC1 antibody and DAPI. Scale bar, 20 μm. (C) Schematic representation of the strategy used to deplete Padi2 immediately before the peak of OPC differentiation: PdgfraCreERT;RCE:loxP;Padi2^+/+ and PdgfraCreERT;RCE:loxP;Padi2^−/− mice were injected with tamoxifen at postnatal day 6 (P6) and sacrificed at P11, and brains and spinal cords were collected. (D) The number of CC1 and GFP double-positive cells was estimated out of total GFP⁺ cells in both corpus callosum and spinal cord of PdgfraCreERT;RCE:loxP;Padi2^+/+ and PdgfraCreERT;RCE:loxP;Padi2^−/− mice. Means ± SEM are shown, n = 5 (control [Ct]) and n = 4 (indKO); ^∗p < 0.05, two-tailed t test. See also Figure S2.

Figure 4

Padi2 Conditional and Full Knockout Mice Exhibit Impaired Motor and Cognitive Functions and Decreased Myelinated Axons (A–C) Behavior tests for motor (rotarod [A] and beam test [B]) and cognitive (novel object recognition [C]) performance in controls (PdgfraCre;RCE:loxP;Padi2^+/+ and wild-type [WT]), Padi2 heterozygous (Padi2 het), and Padi2 full knockout (fKO) and conditional KO (Padi2 cKO; PdgfraCre;RCE:loxP;Padi2^−/−). For the rotarod test, the time spent on rotating rod, latency to fall, was measured in seconds, and for the novel object recognition, test the discrimination index corresponds to percentage ratio between the time exploring the novel object and the total time spent exploring both objects. Means ± SEM are shown, WT n = 8, Padi2 het n = 11, Padi2 fKO n = 11 (^∗p < 0.05, one-way ANOVA), control (Ct) n = 10, and Padi2 cKO n = 10 (^∗p < 0.05, Mann-Whitney test). (D) Representative electron microcopy (EM) images for WT and Padi2 fKO corpus callosum. Scale bar, 1 μm. (E and F) Scatterplot of g-ratio as a function of axon diameter (μm) (F) and graph plot for the g-ratio distribution across different axon diameters (E) in WT (n = 4) and Padi2fKO (n = 4). (G) Percentage of myelinated axons present in the corpus callosum of WT and Padi2 fKO. Means ± SEM are shown, n = 4; ^∗p < 0.05, Mann-Whitney test.

Figure 5

Padi2 Overexpression Unveils Citrullination Targets (A) Immunocytochemistry in 2 day differentiated Oli-neu for PAD2 (yellow) and CNPase (magenta) depicting PAD2 in the nucleus and cytoplasm of cells. Scale bar, 20 μm. (B) Transfection of Oli-neu cells with Padi2 fused with ZsGreen (depicted in yellow) shows the PAD2-ZsGreen both in the cytoplasm and in the nucleus. Scale bar, 20 μm. (C) Schematic representation of the strategy used to determine the citrullination targets by means of SILAC followed by mass spectrometry. Control and overexpressing Padi2 cell lines are fed with ¹²C (light) and ¹³C (heavy) media, respectively. Proteins from proliferating and 2 days differentiating cells are collected from both light and heavy media and mixed 1:1 for further mass spectrometry analysis. (D) Boxplot representing the fold change of the identified citrullination targets in Oli-neu cells in both proliferation and differentiation conditions. Ratios are represented as log₂(H/L ratio), and PAD2-mediated citrullination targets were considered above a threshold of 1 (dashed line). (E) Table representing top citrullination targets of PAD2 in proliferation and differentiation conditions. ∞ represents proteins for which only peptides with heavy labeling were detected. (F) Gene Ontology (GO) analysis for biological processes on the PAD2 citrullinated proteins. The most significant categories (±log₁₀[false discovery rate]) were plotted. See also Figure S2 and Tables S1 and S4.

Figure 6

PAD2 Protein Interactors in Oli-neu Cells (A) Schematic representation of the strategy used to uncover PAD2-interacting proteins by means of SILAC followed by mass spectrometry. Control cells (expressing biotin ligase BirA and empty vector) and biotin-tagged Padi2 (bioPadi2) cell lines (expressing BirA and the biotin-tagged Padi2) are fed with ¹²C (light) and ¹³C (heavy) media, respectively. Immunoprecipitation (IP) with streptavidin beads is performed on the same amount of proteins collected from control or bioPadi2 cells. After IP, the streptavidin beads from the two conditions are mixed are analyzed using mass spectrometry. (B) Graph plot representing all detected immunoprecipitated proteins by mass spectrometry in both proliferating and upon 2 days differentiation of Oli-neu cells. Ratios are represented as log₂(H/L ratio) and PAD2-interacting proteins were considered for analysis when displaying a threshold above 1 and detected in both proliferation and differentiation (all targets in the upright quadrant). (C) Short list of the top ten PAD2-interacting proteins that were previously shown in the myelin proteome. (D) GO analysis for biological processes on the PAD2 interacting proteins. The most significant categories (+/− Log10 (false discovery rate)) were plotted. (E) Network analysis of all PAD2 interactors, proteins found on the cellular component GO analysis for the myelin sheath are highlighted in red. See also Tables S2 and S4.

Figure 7

PAD2 Citrullinates Histones and Regulates the Expression and Chromatin Accessibility of Genes Involved in OL Differentiation (A) Western blot for H3R2+8+17cit and H3R26cit upon treatment of proliferating Oli-neu cells with either DMSO or Cl-amidine 200 μM for 2 days. GAPDH signal is an internal loading control. Images represent n = 2. (B) Western blot analysis for PAD2, histone H3R2+8+17cit, and H3R26 on Padi2-overexpressing cell line versus control. GAPDH was used as a loading control. Representative images of n = 3. (C) Analysis of Oli-neu cell line overexpressing Padi2 or scramble control. Padi2-overexpressing cells display higher PAD activity, as assessed using the ABAP kit assay, and high mRNA Padi2, as assessed using qRT-PCR. Comparative gene expression gene analysis of proliferating and undifferentiated (Und) Oli-neu cells and 2 days differentiated Oli-neu cells (Diff) overexpressing Padi2, n = 3 (Mbp, Mog, Sox10, and Sox9). Means ± SEM are shown; ^∗p < 0.05, two-tailed t test. (D) IGV genome browser overlay views depicting chromatin accessibility (assessed using ATAC-seq) near transcription start sites, in Oli-neu cells transfected with control (Ctrl) siRNAs (blue) and siRNA against Padi2 (red) (n = 3, samples pooled for visualization, same scale for Ctrl siRNA and Padi2 siRNA). See also Tables S3 and S4.