Additional file 10 of Differential usage of DNA modifications in neurons, astrocytes, and microglia

Additional file 10: Figure S1. Cre and Tamoxifen specificity of NuTRAP induction. Brains were harvested from Camk2a-cre+; NuTRAP+ (Camk2a-NuTRAP) mice, treated or not with tamoxifen (Tam), for immunohistochemical analysis of NuTRAP allele recombination or for assessment of neuronal, glial, and endothelial maker expression in the context of EGFP/mCherry localization. A–B Compared to counterparts from mice treated with Tam (+Tam), which exhibit robust efficiency of cre- neuronal recombination (nearly all neurons are positive for mCherry and EGFP), Camk2a-NuTRAP brains of mice not exposed to Tam (−Tam) display NuTRAP allele recombination to a subset of neurons (mCherry and EGFP expression localized to some NeuN+ cells). These data show a small degree of cre recombination specific to neurons independent of Tam induction (corroborating previously published observations) that is exacerbated by 5 days of systemic Tam delivery. C Camk2a-NuTRAP brains show no cre recombination (EGFP or mCherry expression) in cells expressing CD11b (microglia) D CD31 (endothelial), or E GFAP (astrocytes). DAPI: nuclei counterstain. Scale bar: 50 μm at 20X A, B, 50 μm at 40X C–E. Figure S2. Conversion efficiency of Camk2a-NuTRAP BS/oxBS-seq. A Summary of Bisulfite-sequencing (BS-Seq) and Oxidative Bisulfite-Sequencing (oxBS-Seq) techniques. Bisulfite-converted libraries are used to determine total percent modified cytosines (mC+hmC), while oxidative bisulfite-converted libraries are used to determine percent methylated cytosines (mC). hmC values are derived by subtracting oxBS from BS values on a per base basis. B Summary of Enzymatic Methyl-sequencing. TET-converted libraries (TET+) are used to determine total percent modified cytosines (mC+hmC), while non-TET-converted libraries (TET−) are used to determine percent hydroxymethylated cytosines (hmC). mC values are derived by subtracting TET- from TET+ values. C–D) Exogenous control sequences (CEGX, Cambridge, UK) were spiked in to each sheared DNA sample (0.04% w/w) prior to oxidation and/or bisulfite conversion. Raw fastq files were read into CEGXQC v0.2 to generate summary documentation and QC reports based on the conversion efficiency of the spike-in control sequences. Conversion percentages for different cytosine modifications (C, mC, and hmC) are plotted for bisulfite-converted C and oxidative bisulfite-converted D libraries. Bisulfite-converted libraries had near complete conversion of unmodified cytosines and low over-conversion of methylated and hydroxymethylated cytosines. Oxidative bisulfite-converted libraries had high levels of conversion of unmodified and hydroxymethylated cytosines and low conversion of methylated cytosines. Note: one oxBS sample was missing the spike-in control so is not included in this plot. Figure S3. DNA modifications across neuronal genes compared to all genes. mCG A, hmCG B, and mCH C averaged over 200 nucleotide bins from 4 kb upstream, within the gene body, and 4 kb downstream of neuronal marker genes (Additional file 2) and all genes from the positive fraction. Figure S4. Single cell RNA-seq expression of DNA modification regulators. Counts (Tabula Muris) or Normalized Counts (Allen Brain Atlas, Aging Mouse Brain) of DNA 58 modification regulators were plotted from single cell RNA-seq studies (one-way ANOVA with Tukey’s multiple comparisons test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). Additional file 1: Table S1. Average whole genome modification levels across detection methods for neurons, astrocytes, and microglia. Average whole genome mCG and hmCG levels were determined from oxBS-Seq and Nanopore data. The oxBS conversion correction was performed using conversion efficiency estimations based on CEGX spike-in control sequences and equations provided in Kozlenkov et al [3]. Figure S5. Repeat element modifications detected with nanopore sequencing. mCG A and hmCG B in specific repeat elements (LINE, SINE, LTR) were assessed from nanopore sequencing data for neurons, astrocytes, and microglia. One biological replicate is depicted for each cell type. Figure S6. Methylation across regions assessed with targeted EM-seq. Methylation of neurons, astrocytes, and microglia measured with targeted EM-seq (n=3/group) in six regions found to be differentially hydroxymethylated with WGoxBS. Line plots and total mCG across each region were plotted regions corresponding to Chn1 A, Dlgap1 B, Ankrd33b C, Dab2ip D, Chst2 E, and Kalrn F (two-way ANOVA with Sidak’s multiple testing correction and single pooled variance for individual CpG differences between cell types, two-tailed unpaired t-test for average region differences between cell types; *p<0.05, **p<0.01). Figure S7. Correlation of differential methylation and differential hydroxymethylation. mCG and hmCG differences in regions having both differential methylation and differential hydroxymethylation show a significant negative correlation with one another for A Astrocyte vs Neuron, B Microglia vs Neuron, and C Astrocyte vs Microglia comparisons (Simple linear regression with best fit line (solid), 95% confidence bands (dotted), and R2 goodness of fit). Figure S8. Overlap of gene lists by expression level between cell types. Venns of non-expressed A, low expressed B, mid expressed C, and high expressed D genes between neurons, astrocytes and microglia