Evidence for oligodendrocyte progenitor cell heterogeneity in the adult mouse brain

Oligodendrocyte progenitor cells (OPCs) account for approximately 5% of the adult brain and have been historically studied for their role in myelination. In the adult brain, OPCs maintain their proliferative capacity and ability to differentiate into oligodendrocytes throughout adulthood, even though relatively few mature oligodendrocytes are produced post-developmental myelination. Recent work has begun to demonstrate that OPCs likely perform multiple functions in both homeostasis and disease and can significantly impact behavioral phenotypes such as food intake and depressive symptoms. However, the exact mechanisms through which OPCs might influence brain function remain unclear. The first step in further exploration of OPC function is to profile the transcriptional repertoire and assess the heterogeneity of adult OPCs. In this work, we demonstrate that adult OPCs are transcriptionally diverse and separate into two distinct populations in the homeostatic brain. These two groups show distinct transcriptional signatures and enrichment of biological processes unique to individual OPC populations. We have validated these OPC populations using multiple methods, including multiplex RNA in situ hybridization and RNA flow cytometry. This study provides an important resource that profiles the transcriptome of adult OPCs and will provide a toolbox for further investigation into novel OPC functions.


INTRODUCTION
First described in the early 1980's, oligodendrocyte progenitor cells (OPCs) are the fourth major glial subtype present in the brain. Depending on the region examined, OPCs makeup anywhere from 2-8% of the adult central nervous system (CNS) cells. 1,2 . Adult OPCs belong to the same population of progenitors that give rise to oligodendrocytes during CNS development. However, a large fraction of OPCs do not differentiate, but instead remain in a progenitor state throughout adulthood, a property not consistent with the relatively small need to generate new oligodendrocytes 2-6 . While it has been demonstrated that the differentiation of OPCs into myelinating oligodendrocytes is critical for processes such as motor learning during adulthood, recent evidence indicates that mature oligodendrocytes are a relatively stable population in the adult brain 7-9 . The slow rate at which oligodendrocytes are replaced throughout life does not correlate with the maintenance of a highly dynamic and energetically costly population of OPCs 7, 8,10 . With this discordance between the dynamics of the OPC population and the relatively small need for newly differentiated oligodendrocytes in adulthood, the field has begun to explore alternate functions of adult OPCs 11,12 .
Under homeostatic conditions, OPCs express distinct ion channel profiles that vary with both the brain region and developmental stage of the organism, indicating that subpopulations of OPCs maintain unique electrical properties and therefore may be performing multiple functions within the brain 13 . Furthermore, loss of OPCs, either globally or regionally, has been shown to result in significant depressive-like behavior, persistent weight-gain and leptin-insensitivity, as well as microglial activation and subsequent neuronal death [14][15][16] . In pathological conditions, OPCs can upregulate cytokine production in response to IL-17 signaling and greatly contribute to CNS pathogenesis 17 . Surprisingly, OPCs also upregulate antigen presentation machinery in the demyelinating CNS, and can regulate T cell function [18][19][20] . Taken together, these studies illustrate the dynamic role OPCs can play in the adult CNS and build a strong case in support of exploring adult OPCs diversity at the transcriptional level. Such an overview will provide an important resource for further functional investigation of OPCs in the CNS. author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.06.981373 doi: bioRxiv preprint Here, we have developed an inducible OPC reporter mouse strain, which expresses YFP in PDGFRα-expressing cells after tamoxifen administration. After extensive validation, we used this tool to isolate OPCs from the adult brain by fluorescent activated cell sorting (FACS) and perform single-cell sequencing. We demonstrate the presence of three novel populations of transcriptionally distinct OPCs in the adult brain. Gene Ontology (GO) term analysis and gene expression analysis of identified OPC subtypes suggest specialization of OPCs, encompassing potential functions such as immune system modulation and neuronal regulation. Sequencing results were further validated by measuring coexpression of canonical OPC markers along with cluster-specific genes identified from our sequencing dataset using RNAscope and qPCR. Taken together our results present a unique toolbox to support functional exploration of OPCs under homeostatic and pathological conditions. author/funder. All rights reserved. No reuse allowed without permission.
C57B/6J were purchased from Jackson. Mice were maintained on a 12 hours light/dark cycle with lights on at 7am. Behavior was performed on mice used in single-cell sequencing run 1. Testing consisted of sucrose preference, elevated plus maze, open field, and forced swim test. All animal experiments were approved and complied with regulations of the Institutional Animal Care and Use Committee at the University of Virginia (#3918).

Tamoxifen injections
Tamoxifen (C8267, Sigma-Aldrich) was dissolved in corn oil at 37°C overnight at 20 mg/ml. Tamoxifen was administered i.p. at 200 mg/kg with a maximum dose of 4 mg per injection. For single-cell sequencing experiments, 6 week old mice were given 2 injections of tamoxifen 3 days apart. For validation of Cre recombination in PDGFRα-Cre ER; R26-EYFP brains, 5-6 week old mice were injected with 0, 1, 2, or 3 doses of tamoxifen, each given three days apart. For those mice receiving three doses of tamoxifen, the final dose was given at 150 mg/kg.

Immunofluorescence
Mice were deeply anesthetized with pentobarbitol and subsequently perfused with 5 units/Ml heparin in saline followed by 10% buffered formalin, each for approximately 1 minute. For brain tissue, brains were rapidly dissected and post-fixed in 10% buffered formalin overnight at 4°C. Tissue was then transferred into 30% sucrose in PBS and allowed to sink for at least 24 hours. Brains were frozen in OCT, sectioned, and stored in PBS+0.02% NaAz until further staining.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.06.981373 doi: bioRxiv preprint on Mouse Blocking Reagent (MKB-2213, Vector Laboratories) according to manufacturer's instructions for at least 1 hour at room temperature. Tissue was incubated in primary antibodies overnight at 4°C with gentle agitation. Tissue was washed three times in TBS and 0.3% Triton-X 100 and incubated in secondary antibodies overnight at 4°C with gentle agitation. Following secondary incubation, tissue was stained with Hoechst (1:700, ThermoFisher Scientific, H3570) for 10 minutes at room temperature, washed 3 times TBS and Triton-X 100, and mounted on slides using Aqua Mount Slide Mounting Media (Lerner Laboratories). Images were collected on a Leica TCS SP8 confocal microscope and processed using Fiji.

Library Preparation and Sequencing
Samples were processed for single-cell sequencing according to manufacturer's instructions using the Chromium Next GEM Single-cell 3' Reagent Kit (10xGenomics) and Chromium Controller (10xGenomics). Single-cell libraries were sequenced using the NextSeq 500 Sequencing System (Illumina). Library preparation and sequencing was completed by the Genome Analysis and Technology Core at the University of Virginia.

Quantification
All steps of the quantification process were performed with Cellranger. The fastq files for the samples were quantified using the mkfastq utility, and were quantified against the mm10 mouse genome with the count utility.

Pre-processing
We used Seurat for the single-cell analysis 22,23 , and for each of the healthy brain datasets, we followed the same procedure. First, we performed a QC step to identify and remove cells that were potential outliers. This included removing potential multiplets (i.e., cells that had clear outlier gene expression) and cells that had approximately ten percent or more of mitochondrial gene expression (i.e., cells that were likely to have high technical variation). After filtering out these suspect cells, we then normalized and log-transformed the data (using the 'LogNormalize' method), regressed out unwanted sources of technical variation (i.e., the number of detected molecules and mitochondrial contribution to gene expression) 24 , and scaled the counts.

Integration
To make comparative analyses possible between the healthy brain datasets, we integrated the datasets with Seurat using the alignment strategy described previously 22 .
The first step was to select the genes to be used as the basis for the alignment. Here we author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.06.981373 doi: bioRxiv preprint took the union of the 1000 genes with highest variability in each of the datasets, and then filtered this down to only those genes found in each of the datasets, giving us 2,285 genes for the alignment. Next, we identified common sources of variation between the six datasets (3 sequencing runs with 2 samples each) by running a canonical correlation analysis (CCA) with the highly variable genes as features. By examining the strength of the thirty calculated canonical correlations (CCs), we found that the first twelve CCs were driving the variation between the datasets. We then aligned the subspaces 22 (i.e., the first twelve CCs), giving us an integrated dataset with features on a common scale.

Analysis
We then used Seurat on the aligned dataset to identify eight clusters of cells, and then used t-SNE to visualize the similarity between cells. Next, we assigned cell types to these clusters based upon the expression of pre-defined marker genes, and then identified cluster markers by finding the differentially expressed genes in one cluster compared to all other clusters (one-vs-all). Using the gene markers for each cluster, we then used gene set analysis (Fisher's exact test, as implemented in the clusterProfiler Bioconductor package 25 ) to identify GO gene sets that were enriched. To better identify markers that differentiated the three OPC clusters from each other, we excluded the other five non-OPC clusters, and then compared each OPC cluster to the other two individually (onevs-all) and identified the differentially expressed genes. As before, we used these marker genes to identify GO gene sets that were functionally enriched. All analyzed single-cell sequencing data has been uploaded in a searchable database located at http://165.22.7.10:3838/seurat_viewer/seurat_viewer_4.Rmd.

RNAscope
C57B/6J mice (8-10 weeks) from Jackson were anesthetized with pentobarbitol and subsequently perfused with ice-cold 5 units/ml heparin in saline for approximately 1 minute. Brains were rapidly dissected, flash frozen in OCT (Fisher Healthcare, 4585), and stored at -80°C until further processing. Frozen tissue was cut sagittally (15μm), immediately slide-mounted, allowed to dry for approximately 1 hour at -20°C and then stored at -80°C. All tissue was used within three months of dissection. author/funder. All rights reserved. No reuse allowed without permission.

RNAscope Quantification
Following imaging, max projected confocal images were analyzed using CellProfiler Software. RNA expression per cell was quantified using a modified version of a previously published pipeline 26 . Briefly, automated steps were used to draw nuclear masks and subsequently quantify the number of RNA puncta from each channel that colocalized with each nuclear mask. Threshold values for each channel were set based negative control images. Automatic nuclear identification was reviewed and any nuclear mask that clearly contained a large group of nuclei or was located on the edge of an image such that part of the nuclei was not visible was excluded from further analysis. Cells were considered positive for an OPC marker (Pdgfra, Cspg4, Olig1, or Olig2) if 4 or more puncta colocalized with a particular nucleus to account for background in the assay 27 . OPCs were defined by the co-expression of two canonical OPC transcripts encoding for cell surface author/funder. All rights reserved. No reuse allowed without permission.

RT-qPCR
RNA was extracted from sorted cell populations using the ISOLATE II RNA Micro Kit

RNA Flow Cytometry
Cells were isolated from the brains of C57B/6J mice (9-16 weeks, males) according to Pdgfra (Affymetrix, VB6-3197712-210). A probe targeting Actb (Affymetrix, VB1-10350-210) was used as a positive control to ensure good RNA quality. Samples were run using a 16-color Life Technologies Attune Nxt flow cytometer and data was analyzed using FlowJo software. author/funder. All rights reserved. No reuse allowed without permission.

Validation of Inducible OPC Reporter Mouse Line
In order to selectively label oligodendrocyte progenitor cells (OPCs) in the adult mouse brain with as little off-target labeling as possible, we utilized a PDGFRα-Cre ER ; R26-EYFP mouse line 6 . Animals were injected with tamoxifen at 6 weeks of age to avoid labeling the pool of OPCs that differentiate into oligodendrocytes during developmental myelination 28 .

Isolation and Sequencing of YFP+ Cells
Whole brains, including the cerebellum but excluding the spinal cord, were collected from adult PDGFRα-Cre ER ; R26-EYFP mice and processed into a single-cell suspension for FACS. Four to five brains were pooled for each sample, and male and female brains were processed separately to allow for analysis of potential sex differences in YFP+ cells (Supplemental Fig. 2A). YFP+ cells were selected by gating on live cells while excluding immune (CD45+) and red blood cells (TER119+), thus ensuring that the population of YFP+ cells collected were viable and highly enriched (Supplemental Fig. 2B). YFP+ cells were barcoded and prepared for single-cell sequencing using Next GEM reagents and Chromium microfluidics supplied by 10x Genomics. Cell sorting and sequencing was performed three independent times for a total of 6 independently sequenced samples.
Unbiased clustering of each independent run revealed overlap between distinct sequencing runs and no clustering of cells driven by sequencing run alone (Supplemental Fig. 3A). For all future analysis, all sequencing runs were combined to form one large dataset.
author/funder. All rights reserved. No reuse allowed without permission.

Profiling the Molecular Signature of OPCs in the Adult Brain
Unbiased clustering of sequenced cells using the Seurat package 22,23 revealed that cells sorted from PDGFRα-Cre ER ; R26-EYFP brains clustered into 8 distinct populations (Fig.   1A). Mature oligodendrocytes comprise one cluster, having potentially differentiated following initial tamoxifen labeling of PDGFRα expressing progenitor cells. Also captured in the sequencing were three cell-types outside the oligolineage that are known to either express PDGFRα or come from PDGFRα expressing precursors, including fibroblasts, endothelial cells, and 2 populations of pericytes [29][30][31][32] . These clusters were identified by expression of known cell type markers such as Igfbp6 and Fn1 (fibroblasts), Tek, Pecam1, and Kdr (endothelial cells), as well as Rgs5, Pdgfrb, and Des (pericytes) [33][34][35][36] .
The remaining 3 clusters of cells (OPC1, OPC2, and OPC3) expressed at least 2 of the 5 canonical OPC markers Ptprz, PDGFRα, Olig1, Olig2, and Cspg4 (Fig. 1B). Importantly, each OPC cluster of cells expressed a unique transcriptional signature distinct from the gene expression in every other cluster (Fig. 1C).
In order to further investigate how these clusters of OPCs are distinct from one another, we identified a significantly upregulated gene from each cluster that offered potential indications of distinct functions of these subpopulations (Fig. 1D). OPC1 expressed high levels of Clusterin, a secreted glycoprotein that can play either pro-apoptotic or antiapoptotic roles, depending on the splice variant that a cell expresses [37][38][39] . OPC2 expressed high levels of Lumican, an extracellular matrix protein known to bind collagen and play an important role in tissue healing [40][41][42] . OPC3 shows significant upregulation of the G-protein coupled receptor Gpr17, the only currently known marker of molecular diversity in OPCs, when compared to all other cell types sequenced (Fig. 1D) 43,44 .
Lastly, in order to gain insight into how these subpopulations of OPCs potentially differ at the functional level, we analyzed GO terms that were significantly upregulated in each OPC cluster and were not shared with either of the other OPC clusters (Fig. 1E). OPC1 showed unique upregulation in genes involved in ATP metabolic processes, cellular respiration, and oxidative phosphorylation, including Chchd10, Mdh1, and Uqcrq. Genes related to extracellular matrix organization, cytokine-mediated signaling pathways, and author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.06.981373 doi: bioRxiv preprint regulation of cell-cell adhesion were significantly upregulated in OPC2, and included genes such as Mfap4, Ifitm2, and Lgals1. Lastly, OPC3 show unique upregulation in genes involved in the positive regulation of neuron differentiation, synapse organization, and cerebral cortex cell migration, including Stmn2, Pfn2, and Dcx (Supplemental Table   1).
In sum, our single sequencing data reveal three unique subpopulations of OPCs that reside in the adult brain under homeostatic conditions and are suggestive of functional diversity based on unique gene expression profile.

Sex Differences in OPCs
Since sex differences have been implicated in multiple types of glia including microglia and astrocytes, we investigated whether OPCs isolated from male or female mice exhibited significant transcriptional differences 45,46 . Interestingly, both males and female cells were found in all 8 clusters, and the number of significantly different genes between male and female cells in each cluster fell within the range of statistical noise (data not shown). However, each cluster of OPCs did not have an equal distribution of cells from males and females, with males exhibiting a higher frequency of cells in OPC1 and females exhibiting a higher frequency of cells in OPC3. A relatively equivalent frequency was found in OPC2 (Supplemental Fig. 3C). While this data indicates that no sex-specific OPC signature found in the adult brain, the proportions of OPC subpopulations may be sexually dimorphic.

In Vivo Validation of OPC Subpopulations
Since gene expression can be altered by tissue processing before sequencing 47 , we validated the expression of each OPC cluster marker using RNAscope in adult mouse brain. OPCs were defined by the co-expression of two canonical OPC transcripts encoding for cell surface markers (Pdgfra or Cspg4) and oligolineage transcription factors (Olig1 or Olig2). Using a Cell Profiler pipeline to unbiasedly quantify RNA puncta expression per cell 26 , we subsequently quantified expression of each OPC cluster gene.
OPCs in both gray and white matter express a range of both Clusterin and Gpr17 transcripts, with a population of cells expressing little to no RNA for these markers, a author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.06.981373 doi: bioRxiv preprint population of cells showing medium expression, and a smaller population of cells expressing very high levels of these transcripts (Fig 2A, C). Lumican, the marker for OPC2, showed much lower expression in the brain, although we still observed some OPCs expressing low levels of Lumican, and many OPCs expressing no observable levels of Lumican (Fig 2B). We were therefore able to detect and validate, using the novel selected cluster markers and canonical OPC genes, each cluster of OPCs in vivo.

Expression of Canonical Markers in OPC Clusters
While single-cell sequencing provides gene expression data on an individual cell basis, its relatively shallow depth of sequencing can result in little to no detected expression of genes with known low expression 48,49 . In order to confirm that the 3 identified clusters of OPCs express canonical OPC markers using a more sensitive method of gene detection, we individually sorted each cluster of OPCs based on expression of cell surface proteins.

Clusterin and Gpr17 are exclusively expressed in OPC1 and OPC3 subsets
While both RNAscope and RT-qPCR of canonical OPC markers within each cluster have demonstrated that these clusters of OPCs are expressed within the brain and belong to the oligolineage, neither of these techniques have demonstrated that these markers characterize clusters that are unique from one another. Using PrimeFlow, a technique author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.06.981373 doi: bioRxiv preprint that allows for the combination of cellular-resolution RNA detection with the multiplexing capabilities of flow cytometry, we demonstrate that a subset of Olig2 expressing cells express Clusterin (OPC1), and a mutually exclusive population expresses Gpr17 (OPC3), with very few OPCs expressing detectable levels of both cluster markers (Fig 3. B,C).
While this does not rule out the possibility that an individual OPC might express genes enriched in different clusters at different times, it does demonstrate that, at any given point, genetic markers of these 2 clusters of OPCs largely do not overlap.
author/funder. All rights reserved. No reuse allowed without permission.

Discussion
With the development of novel tools that allow for the analysis of tissue at single-cell resolution, interest has surged in outlining how cell-types that express the same canonical cell type markers may represent more diverse subpopulations than previously thought [50][51][52][53][54] . Here, we demonstrate that OPCs from the adult brain cluster into three distinct subpopulations characterized by a transcriptional signature and Gene Ontology profile. OPCs was relatively small (approximately 300 cells) and the majority of their OPCs came from juvenile animals 54 . A more recent study from the same group characterized the transcriptional profiles of OPCs from E13.5, E17.5, and P7 mice and found three clusters of OPCs that shared similar transcriptional signatures, but clustered by the age of the cells, and one cluster of cycling OPCs 32 . From this data, they concluded that, during development, the three known waves of developmental OPCs converge into a transcriptionally homogenous group of OPCs by P7. Importantly, this sequencing dataset only profiles prenatal and early postnatal OPCs, a time window in which OPCs are preparing to generate a large population of mature oligodendrocytes to support the developmental myelination that occurs during early postnatal timepoints 28 . Therefore, it is likely that OPCs during this early stage of development may represent a relatively homogenous population of progenitors destined to give rise to myelinating glia 28 .
However, following developmental myelination, oligodendrocytes represent a relatively stable population that require minimal replacement, yet OPCs continue to represent approximately 5% of cells in the adult brain and tile every brain region 2,55 . It is therefore reasonable to hypothesize that as the CNS matures, and no longer requires the production of large numbers of mature oligodendrocytes, OPCs may develop diverse transcriptional repertoires, as demonstrated here, to perform alternative functions throughout adulthood. Indeed, recent data from Spitzer and colleagues demonstrated that author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.06.981373 doi: bioRxiv preprint OPCs throughout the brain express a diverse array of electrophysiological properties and ion channels and that these characteristics become more diverse with age 13 . Additionally, data obtained from zebrafish has demonstrated that OPCs can be categorized into two functionally distinct subpopulations that demonstrate different calcium dynamics 56 .
Interestingly, one population of OPCs was found to rarely differentiate in vivo, although these cells maintained their differentiation capacity, indicating that the main functions of this population of OPCs is likely something other than serving as a progenitor pool for mature oligodendrocytes 56 .
Our in-depth analysis of canonical OPC marker expression in each OPC cluster surprisingly indicated that both OPC1 and OPC3 expressed a unique subset of common OPC markers. While PDGFRα, CSPG4, Olig1, and Olig2 have historically all been thought to be expressed in the vast majority of OPCs 57 , recent data from zebrafish indicates that approximately 80% of gray-matter OPCs and 30% of white matter OPCs express Cspg4 transcripts 56 . This differential expression of canonical OPC markers may indicate that OPCs can express a unique array of OPC genes based on their function, and canonical OPC markers may not be as critical to OPC functioning as previously thought. This idea is supported by the recent description of the development and differentiation of a PDGFRα-independent subpopulation of OPCs 58 .
It is also important to note that both sequencing studies previously mentioned utilized different mouse lines to identify OPCs than the PDGFRα-Cre ER ; R26-EYFP used in this study, which may have also contributed to the difference between our dataset and the previous observations of a homogenous population of OPCs. Excitingly, recent sequencing data from human Alzheimer's disease patients and healthy controls demonstrated that healthy controls have three subpopulations of OPCs, and that one of these populations expressed high levels of Clusterin, one of the genes we identified as significantly upregulated in OPC1 59 . Additionally, single-cell sequencing data from human patients at fetal, adolescent, and adult timepoints reveal multiple transcriptionally distinct populations of oligo-lineage cells that largely clustered based on the age of the patient 60 .
The addition of our study to the previously published datasets, detailing the transcriptional author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.06.981373 doi: bioRxiv preprint profile of developmental OPCs will provide the field with a better understanding of how OPCs might change as the brain matures. Indeed, our works highlights how the function of OPCs might shift as development ends and the brain enters adulthood.
Many of the differentially expressed genes and related biological processes found in each OPC cluster complement emerging literature that indicates non-canonical roles for OPCs during homeostasis, and a more active role of this cell type in multiple diseases. OPC3 shows significant upregulation of the G-protein coupled receptor Gpr17 (Fig. 1D).
Importantly, Gpr17 is the only documented marker of molecular diversity in OPCs described to date and is only found in one cluster of OPCs in our dataset 43,44 . This cluster author/funder. All rights reserved. No reuse allowed without permission.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.06.981373 doi: bioRxiv preprint shows unique upregulation of genes related to neuronal differentiation and synapse organization (Fig. 1E). These processes are particularly intriguing given that OPCs are the only known glial cell to form canonical synapses with neurons, and have recently been shown to be critical in regulating circuit formation during development 12,75 . Birey and colleagues demonstrated that ablating OPCs significantly altered neuronal function and resulted in depressive and anxiety-like behavior 14 . It still remains to be seen if this effect of OPC loss on neuronal function is mediated through another cell type. Yet those studies, coupled with the sequencing data described here, makes investigation of this subpopulation of OPCs crucial in understanding how OPCs are directly influencing neuronal health, circuit functioning and formation, and overall behavioral outcomes.
While we describe the transcriptional profile of OPCs during homeostasis, it is important to note that understanding the role of OPCs in the healthy brain will provide a necessary foundation for examining any protective or detrimental novel functions in desease pathology. Previous work has demonstrated that OPCs exhibit significant transcriptional heterogeneity and disease-associated signatures in models of multiple sclerosis and cerebral ischemia, and showed significant transcriptional changes in human Alzheimer's disease patients 18,59,76 .
We believe that the work presented here provides a critical foundation and basis for the investigation of non-canonical roles of OPCs This dataset will not only assist the field in discovering novel roles for OPCs in both health and disease, but can also offer potential mechanistic explanations for intriguing phenotypes observed in OPC deletion paradigms 14-17 .
author/funder. All rights reserved. No reuse allowed without permission.