article open access single-cell rna-seq analysis of human ... · subject age, y sex race diagnosis...

ARTICLE OPEN ACCESS

Single-cell RNA-seq analysis of human CSFmicroglia and myeloid cells inneuroinflammationEkaterina Esaulova MS Claudia Cantoni PhD Irina Shchukina MS Konstantin Zaitsev MS

Robert C Bucelli MD PhD Gregory F Wu MD PhD Maxim N Artyomov PhD Anne H Cross MD and

Brian T Edelson MD PhD

Neurol Neuroimmunol Neuroinflamm 20207e732 doi101212NXI0000000000000732

Correspondence

Dr Edelson

bedelsonpathwustledu

or Dr Wu

wugneurowustledu

AbstractObjectiveTo identify and characterize myeloid cell populations within the CSF of patients with MS andanti-myelin oligodendrocyte glycoprotein (MOG) disorder by high-resolution single-cell geneexpression analysis

MethodsSingle-cell RNA sequencing (scRNA-seq) was used to profile individual cells of CSF and bloodfrom 2 subjects with relapsing-remitting MS (RRMS) and one with anti-MOG disorderPublicly available scRNA-seq data from the blood and CSF of 2 subjects with HIV were alsoanalyzed An informatics pipeline was used to cluster cell populations by transcriptomic pro-filing Based on gene expression by CSF myeloid cells a flow cytometry panel was devised toexamine myeloid cell populations from the CSF of 11 additional subjects including individualswith RRMS anti-MOG disorder and control subjects without inflammatory demyelination

ResultsCommon myeloid populations were identified within the CSF of subjects with RRMS anti-MOG disorder and HIV These included monocytes conventional and plasmacytoid dendriticcells and cells with a transcriptomic signature matching microglia Microglia could be dis-criminated from other myeloid cell populations in the CSF by flow cytometry

ConclusionsHigh-resolution single-cell gene expression analysis clearly distinguishes distinct myeloid celltypes present within the CSF of subjects with neuroinflammation A population of microgliaexists within the human CSF which is detectable by surface protein expression The function ofthese cells during immunity and disease requires further investigation

These authors contributed equally to this work

From the Department of Pathology and Immunology (EE IS KZ GFW MNA BTE) and Department of Neurology (CC RCB GFW AHC) Washington University School ofMedicine St Louis MO and Computer Technologies Department (KZ) ITMO University St Petersburg Russia

Go to NeurologyorgNN for full disclosures Funding information is provided at the end of the article

The Article Processing Charge was funded by the NIH NMSS and ICTS

This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 40 (CC BY-NC-ND) which permits downloadingand sharing the work provided it is properly cited The work cannot be changed in any way or used commercially without permission from the journal

Copyright copy 2020 The Author(s) Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology 1

CSF evaluation is used to aid in the diagnosis and differentia-tion of CNS disorders In inflammatory CNS diseases the CSFis typically used to assess the immunopathophysiologic pro-cesses because biopsy of CNS tissue carries significant potentialfor harm1 However relatively few cells are obtained from CSFusually on the order of 1ndash5 cellsμL Recent refinements innext-generation sequencing have enabled the efficient de-termination of individual cell gene expression within bio-specimens with relatively sparse cell populations such as theCSF Patterns identified using single-cell RNA sequencing(scRNA-seq) can uncover distinct cell types present at lowlevels within cellular communities and tissues2 scRNA-seq wasused to assess inflammatory changes within the CSF of subjectswith HIV infection identifying the presence of a ldquomicroglial-likerdquo cell3 and more recently to explore the clonal expansion ofCSF lymphocytes in MS-discordant monozygotic twin pairs4

scRNA-seq has also been used to address the issue of microglialheterogeneity within the human brain5ndash7 In addition using theprimary animal model of MS experimental autoimmune en-cephalomyelitis scRNA-seq has been used to identify severalpopulations of myeloid cells both endogenous to the CNS andfrom peripheral blood8 New methods for characterization ofmyeloid populations within the CNS during disease offer theopportunity to dissect the origin function and pathogenicity ofeach cell type with much greater resolution than previousmethods

MS is the most common inflammatory demyelinating diseaseof the CNS affecting over 600000 people in the UnitedStates9 Anti-myelin oligodendrocyte glycoprotein (MOG)disorder is a newly described CNS demyelinating disease thatshares clinical and pathologic characteristics with MS1011 MSand anti-MOG disorder appear to be distinct from one an-other and from aquaporin 4 antibody-positive neuromyelitisoptica (NMO)1012 We have applied scRNA-seq to examinethe CSF and mononuclear cells of the peripheral blood ofsubjects with relapsing-remitting MS (RRMS) and anti-MOGdisorder Individual spinal fluid samples from 2 subjects withRRMS and 1 subject with anti-MOG disorder were analyzedby using scRNA-seq In all 3 subjects we uncovered CSFpopulations of immune cells including microglial cells mon-ocytes and dendritic cells (DCs) based on gene expressionUsing CSF and blood from 7 additional subjects with RRMSanother subject with anti-MOG disorder and 3 control sub-jects we designed and tested a flow cytometry strategy thatconfirmed the presence in CSF of these cell types

MethodsSubjectsEleven subjects with inflammatory demyelinating disease (9with RRMS and 2 with anti-MOG disorder) and 3 controlsubjects (1 with amyotrophic lateral sclerosis [ALS] 1 withidiopathic intracranial hypertension [IIH] and 1 healthy control[HC]) were recruited for a study to assess the characteristics ofCSF and blood cells (table) The institutional review board ofWashington University in St Louis approved study protocolsand each subject provided informed consent Nine subjects hadRRMS based on the current diagnostic criteria13 Two addi-tional subjects were diagnosed with anti-MOG disorder onepresented with optic neuritis and the other with partial trans-verse myelitis Anti-MOG disorder was diagnosed based on the6-month sustained positive cell-based assays for MOG IgG1antibody absence of antibodies to aquaporin 4 (NMO-IgG)and absence of CSF-restricted oligoclonal bands11

To prevent alterations because of disease-modifying therapies weselected subjects either on no therapy or remote from therapySeven subjects with RRMS were naive to therapy including cor-ticosteroids Of the remaining 2 subjects with RRMS 1 (subject 7)had received fingolimod stopping 9 months before CSF testingand the other (subject 10) was treated with glatiramer acetate for 5months and had received IV steroids 2 months before CSF anal-ysisThe subjectwith transversemyelitis due to anti-MOGdisorder(subject 8) had receivedmycophenolate and then azathioprine buthad discontinued these 6 months before CSF analysis and hadreceived a course of corticosteroids 4 months before CSF exami-nation Subject 1 with anti-MOG disorder had received no treat-ment before the examination of CSF Control specimens wereobtained from a subject with IIH who presented with visual dis-turbances a patient diagnosed with ALS and a HC Each con-tributed blood and CSF cells for scRNA-seq or flow cytometrystudies (table and figure e-1 linkslwwcomNXIA244)

Standard protocol approvals registrationsand patient consentsInformed consent was obtained from all participants andapproved by the Human Research Protection Office ofWashington University in St Louis

CSF and peripheral blood mononuclearcell preparationCSF was collected in a 50-mL conical tube on ice andcentrifuged for 10 minutes at 400g Supernatant was removed

GlossaryAD = Alzheimer disease ALS = amyotrophic lateral sclerosis BAM = border-associated macrophage CCA = canonicalcorrelation analysis cDC = conventional DCDAM = disease-associated microgliaDC = dendritic cellHC = healthy controlHLA-DR = human leukocyte antigen DR IIH = idiopathic intracranial hypertension MOG = myelin oligodendrocyteglycoproteinNMO = neuromyelitis optica PBMC = peripheral blood mononuclear cell PCA = principal component analysispDC = plasmacytoid DC RRMS = relapsing-remitting MS scRNA-seq = single-cell RNA sequencing UMAP = UniformManifold Approximation and Projection

2 Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 NeurologyorgNN

without disrupting the cell pellet which was resuspended in400 μL of PBS containing 004 bovine serum albumin TheCSF cells were counted with a hemocytometer and resus-pended in a volume of PBS004 bovine serum albumin toachieve a cell concentration of 700ndash1000 cellsμL for scRNA-seq Peripheral blood mononuclear cells (PBMCs) were iso-lated by Ficoll density centrifugation and then processedalongside CSF cells for scRNA-seq or flow cytometry

Flow cytometryFlow cytometry analysis was performed using antihumanantibodies to BDCA-2 (clone 201A BioLegend San DiegoCA) CD33 (clone P676 BioLegend) CD14 (clone 61D3eBioscience San Diego CA) Lyve-1 (clone 537028 RampDSystems Minneapolis MN) CD3 (clone SK7 BioLegend)CD19 (clone HIB19 eBioscience) CD16 (clone B731BioLegend) CD1c (clone L161 BioLegend) Lox-1 (clone15C4 BioLegend) and human leukocyte antigen DR (HLA-DR) (clone Immu-357 Beckman Coulter Brea CA) Onemillion PBMCs and between 2 times 104 and 40 times 104 CSF cellswere incubated with human-Fc block (BD Biosciences SanJose CA) for 10 minutes at RT and then with antibodies for20 minutes at 4degC Subsequently the samples were washedwith PBS + 2 FBS (flow buffer) for 59 spun at 500g andresuspended in 200 μL of flow buffer The samples were runon a Gallios flow cytometer (Beckman Coulter Life Sciences)on the same day of the collection The cells were analyzedusing FlowJo software (Tree Star Ashland OR)

RNA sequencing and analysisOn the day of cell collection using fresh cells droplet-based 39(for anti-MOG disorder subject 1) and 59 (for subjects 2 and5 with RRMS) libraries were prepared using ChromiumSingle Cell 39 v2 or 59 Reagent Kits according to the manu-facturerrsquos protocol from 10times Genomics The generatedscRNA libraries were sequenced using an Illumina HiSeq4000or Novaseq sequencer Cell Ranger Single-Cell Software 22was used to perform sample demultiplexing barcode pro-cessing and single-cell 39 and 59 counting Afterward fastqfiles for each sample were processed with a cellranger countwhich was used to align the samples to GRCh38 genomefilter and quantify reads For each sample the recovered-cellsparameter was specified as 10000 cells that we expected torecover for each individual library

To reanalyze scRNA-seq data performed with SeqWell publiclyavailable from GSE117397 we downloaded count tables from 2subjects with HIV for whom both CSF and blood samples wereavailable from GEO DataSets (GSM3293822 HIV1_BldGSM3293823 HIV1_CSF GSM3293824 HIV2_BldGSM3293825 HIV2_CSF) To combine data from all samplestogether we used the canonical correlation analysis (CCA) al-gorithm from Seurat 314 Before applying CCA we removedcells from the 39 and 59 data sets that contain more than 10 ofmitochondrial RNA Cells from SeqWell data sets with less than20 of RNA from mitochondrial genes and more than 500 butless than 2500 expressed genes were considered viable cells

Table Demographic and clinical characteristics of subjects

Subject Age y Sex Race Diagnosis Presentation at onsetTime from symptomonset to LP CSF OCB Analysis

1 56 F Caucasian Anti-MOG disorder ON OD and paresthesias 8 mo 0 scRNA-seq

2 38 F Caucasian RRMS ON OD 6 wk 3 scRNA-seq

3 53 F Caucasian RRMS Paresthesias 3 y 13 Flow cytometry

4 23 F Caucasian RRMS ON OD 4 mo 7 Flow cytometry

5 34 F Caucasian RRMS ON OD 2 y 11 scRNA-seq

6 40 F Caucasian RRMS Paresthesias and tremor 15 mo 5 Flow cytometry

7 41 F Caucasian RRMS ON OS 12 y 9 Flow cytometry

8 22 M Caucasian Anti-MOG disorder Paresthesias and leg weakness 2 y 0 Flow cytometry

9 45 M Caucasian RRMS Lhermitte sign 6 mo 7 Flow cytometry

10 38 F Caucasian RRMS ON OS 6 mo 10 Flow cytometry

11 38 F Caucasian IIH TVO 1 y 0 Flow cytometry

12 70 M Caucasian ALS Weakness 4 y ND Flow cytometry

13 45 M Caucasian HC NA NA ND Flow cytometry

14 40 M Caucasian RRMS Right-sided weakness 3 mo 4 Flow cytometry

Abbreviations ALS = amyotrophic lateral sclerosis HC = healthy control IIH = idiopathic intracranial hypertension OD = oculus dextra (right eye) ON = opticneuritis OS = oculus sinistra (left eye) LP = lumbar puncture with CSF analysis MOG = myelin oligodendrocyte glycoprotein NA = not applicable ND = notdone OCB = oligoclonal band RRMS = relapsing-remitting MS scRNA-seq = single-cell RNA sequencing TVO = transient visual obscuration

NeurologyorgNN Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 3

After CCA data were scaled principal component analysis(PCA) was performed with RunPCA function A UniformManifold Approximation and Projection (UMAP) di-mensionality reduction was performed on the scaled matrixusing the first 20 PCA components to obtain a 2-dimensionalrepresentation of the cell states15 For clustering we usedFindNeighbours and FindClusters functions on 20 PCAcomponents with a resolution of 05 FindAllMarkers functionwas used to characterize clusters For heatmap representationthe mean expression of markers inside each cluster was usedHeatmaps were built with Phantasus software (artyomovlabwustleduphantasus)

Data availabilityAnonymized scRNA-seq is available to qualified investigatorsat synapseorg (syn21904732)

ResultsDiscrete immune cell populations can beidentified in the CSF and blood of subjects withneuroinflammatory diseases using scRNA-seqTo explore the extent diversity and phenotype of leukocyteinfiltration into the CNS compartment during CNS de-myelination we performed scRNA-seq of CSF cells andPBMCs from 1 subject with anti-MOG disorder (subject 1)and 2 subjects with RRMS (subjects 2 and 5) During thecourse of our data collection scRNA-seq of immune cellsisolated from the blood and CSF of subjects with HIV wasreported3 Given the small number of data sets on CSFa difficulty to access tissue we sought to optimally identifypopulations of immune cells reflective of neuroinflammationusing a recently described computational method CCA thatallows integration across different scRNA-seq technologies14

Applying this method data from a total of 23363 PBMCs and14179 CSF cells were combined and visualized using theUMAP technique15 and unsupervised clustering1617

Twenty clusters representing distinct cell populations wereidentified (figure 1A) We used the characteristic marker geneexpression to assign cellular identities to each cluster (figure1B) Several distinct lymphoid cell clusters including CD8+

T cells CD4+ T cells γδ T cells NK cells and B cells could bediscriminated in both PBMCs and CSF Myeloid populationswere also identified including microglia CD14+ and CD16+

monocytes and 2 distinct groups of DCsmdashplasmacytoid(pDCs) and conventional (cDCs) Examining each subjectindividually there were nomeaningful disparities in cell clustersbetween the subjects (figure e-2 linkslwwcomNXIA244)

We also compared the proportional abundance of specific cellclusters within either the CSF or blood (figure 1C) Weidentified a cluster of cells that exhibited features of cell stressincluding expression of the mitochondrial genes MTRNR2L8and MTRNRL12 which are indicative of low quality or dyingcells (labeled ldquoDeadrdquo in figure 1)7 These dead cells as well as

red blood cells and platelets were found almost solely withinthe blood Similarly we observed one population of NK cells(NK_1) that was exclusively present in the blood This pop-ulation shows a gene expression profile matching the recentlydefined blood NK1 cells which align with previously describedCD56dim NK cells (figure e-3 linkslwwcomNXIA244)18

Among the T and B cell clusters several exist in higherabundance in the blood (naive CD8 CD8_2 CD8_3 naiveCD4 γδ T cells and B_cells_1) CD14+ and CD16+ mon-ocytes were more prevalent in blood than CSF with the latterbeing entirely absent from the CSF cDCs were a smallerfraction among PBMC relative to CSF cells whereas pDCswere found in both compartments in similar proportionsMicroglia were exclusively found in the CSF Given thecomplexity of cells with myeloid characteristics in neuro-inflammatory diseases and the historic difficulty of studyingthese cells in CSF by conventional methods because of limitedcell numbers we dedicated subsequent analyses of our data toCSF myeloid cell populations

CSF microglia and other myeloid cell typesin neuroinflammationAnalyzing the genes expressed by the cluster of CSF cells thatwe term microglia revealed high expression of genes from themicroglial homeostatic gene signature including CX3CR1CSF1R SLC2A5 MARCKS and P2RY13 (figure 2A)71920

Recent scRNA-seq analyses of human brain microglia havedefined heterogeneity among microglia67 Sankowski et al7

defined 8 clusters of human brain microglia (termed C1ndashC3C5ndashC9) with differential gene expression among these clus-ters Expression of these microglia cluster-specific genesacross the myeloid cell populations from our scRNA-seqcomposite data set was next examined Many of these geneswere found to be strikingly microglia-specific includingAPOC1 APOE LYVE1 TREM2 C1QB GPR34 OLR1 andC3 (figure 2B) Sankowski et al also created a classifier tool todiscriminate microglial clusters which they successfully ap-plied to publicly available scRNA-seq data sets of brainmicroglia When we applied this classifier to the 394 CSF cellsin our data set that were grouped as microglia 88 of thesewere classified as belonging to C6 with 10 belonging to C7C6 and C7 were characterized by Sankowski et al as havinghigh expression of genes in the Gene Ontology pathway0048002 (ldquoantigen processing and presentation of peptideantigenrdquo) In sum these data strongly support our conclusionthat the myeloid cell populations we find exclusively in theCSF are microglial cells and suggest that they may be involvedin antigen presentation

Two previous studies using scRNA-seq to classify CSFimmune cells each identified a discrete cell populationthat was categorized as either ldquomicroglia-likerdquo cells3 orldquomonocytesrdquo4 Each study produced a list of genes thatdiscriminated these differently named clusters from otherCSF immune cells When we visualized these gene setsamong the clusters of CSF cells in our composite data set


which includes a reanalysis of the cells from 2 subjects in-cluded in the work of Farhadian et al we found markedlyenriched expression in the cluster that we term microglia(figure 2C) With this we mean that all 3 studies are in factreferring to the same subset of CSF cells and given theirexpression of core microglial genes we prefer to classifythese as microglia

Other myeloid populations could also be distinctly classified bytheir transcriptional profiles Figure 3 is a heatmap of the ex-pression of 10 genes per cell type which discriminated eachmyeloid population Some of these genes were known to dis-criminate myeloid cell types yet others are new and could berevealing different classifiers Notably expression of major

histocompatibility complex class II genes was higher in CSFcDCs than blood cDCs possibly indicating a compartmentalinfluence on the antigen presentation function of these cellsThese transcriptomic data establish the presence of onemonocyte subset 2 types of DCs andmicroglia within the CSFduring neuroinflammation

Flow cytometric characterization of CSF cellsvalidates scRNA-seq identification of microgliaand myeloid cell typesWe sought to create a flow cytometric-based strategy for thequantification of various CSF myeloid cell types includingmicroglia using commercially available antibodies specific forsurface markers identified by our gene expression studies

Figure 1 scRNA-seq characterization of leukocytes from MS anti-MOG disorder and HIV

(A) UMAP of immune cell clusters from the blood (top) and CSF (bottom) of all subjects merged (B) Characteristic marker gene expression assigned to eachcluster displayed by UMAP (C) Proportional abundance of 20 cell clusters within the blood (purple) and CSF (yellow) Y axis represents the percentage of allcells cDC = conventional DC PBMC = peripheral blood mononuclear cell pDC = plasmacytoid DC RBC = red blood cell Tgd = γδ T cells UMAP = UniformManifold Approximation and Projection


(figure e-4 linkslwwcomNXIA244) and the literature21

PBMCs and CSF cells were stained for CD3 and CD19 togate out T and B cells Within the remaining cells in both CSFand blood pDCs were identified as being HLA-DR+ andBDCA-2+ (encoded by CLEC4C) whereas cDCs wereidentified as HLA-DR+BDCA-2minusCD1c+ cells (figure 4A) Inthe CSF non-DCs could be seen to resolve as 2 distinct celltypes based on the HLA-DR and CD33 levels PBMCs onlycontained HLA-DRmidCD33high cells which were defined asmonocytes The CSF additionally contained a population ofHLA-DRhighCD33mid cells which we interpreted as microgliafurther confirmed by their expression of Lyve-1 and Lox-1(encoded byOLR1) CSF from subjects with other neurologicconditions (anti-MOG disorder ALS and IIH) and a HCsubject also contained similarly staining populations of pDCscDCs monocytes and microglia (figure 4B) The stainingintensity of HLA-DR CD33 Lyve-1 and Lox-1 discriminatedCD14+ monocytes from microglia in the CSF of all subjects(figure 4C) Overall flow cytometric characterization of my-eloid cells demonstrated the protein expression of variousmicroglial genes identified by transcriptomic profiling con-firmed the presence of microglia within the CSF and offereda ready means for identifying this cell type

DiscussionWe performed scRNA-seq to identify the cell types presentwithin the CSF of 2 subjects with RRMS and 1 person withanti-MOG disorder with comparative analysis to pairedPBMCs These and subsequent subjects were not takingdisease-modifying therapies or corticosteroids Subsequentlywe integrated publicly available data on 2 subjects with HIVon whom scRNA-seq was performed on paired PBMCs andCSF immune cells We then focused on CSF myeloid celltypes because less work has been carried out to characterizethese populations in part because of their lower abundanceWe found that (1) scRNA-seq identifies a diverse set of my-eloid cell types within the CSF including microglia (2) flowcytometry is a feasible technique for the discrimination ofthese populations of CSF myeloid cells and (3) scRNA-seqdata can be integrated across technologies to allow for optimaldiscrimination of cell populations

Because the data sets we analyzed were derived from differenttechnologies we used a recently described technique CCA tointegrate scRNA-seq data sets This allowed for an increasednumber of cells used in the analysis facilitating the optimaldiscrimination of cell clusters All CSF samples in the combinedanalysis contained cells within each cluster suggesting thatamong the diseases examined herein a common set of leuko-cytes inhabit this critical immune space within the CNS Onelimitation of CCA involves the exclusion of some genes fromdownstream analysis Nevertheless in our case 2000 geneswere still available for use in our assessment of PBMCs andCSF cells from 5 individuals Our results demonstrate thefeasibility of using CCA to integrate the growing number of

scRNA-seq data sets that will continue to be made publiclyavailable using different platforms

Our scRNA-seq analysis identified microglia in human CSFwith a gene expression pattern overwhelmingly similar tothat described for parenchymal microglia7192223 We usedgenes from a recent study describing the transcriptionalprofile of human tissue-resident microglia as a comparisonfor the CSF microglia identified in our subjects7 The topgenes identified in tissue-resident microglia were also highlyabundant in our CSF microglia populations (figure 2)Owing to their low abundance in the CSF this subset of cellshas apparently been overlooked in most previous CSFstudies using less powerful techniques One previous reportusing a different scRNA-seq platform SeqWell also identi-fied this population of cells in human CSF and termed themldquomicroglia-likerdquo3 Our reanalysis of the published data fromthe latter study confirmed the presence of cells expressinghallmark microglial genes and lacking expression of genesthat identify other myeloid cell types A more recentlypublished study involving patients with MS also identifiedseveral distinct myeloid cell types within the CSF24 Basedon the analysis of our data we believe this group used theterm ldquoMono2rdquo in reference to a population encompassingCSF microglia Indeed they describe a subset of monocytesthat expressed microglial-associated genes includingTREM2 TMEM119 and GPR34 and concluded that thesecells share features with both border-associated macro-phages (BAMs) which have been ascribed immunoregula-tory functions within the CNS22 and homeostatic microgliaIn the future it will be important to devise nomenclaturewhich prudently encompasses the appropriate degree ofsegregation among various myeloid cell populations Inparticular BAMs share substantial transcriptional overlapwith microglia25 prompting the consideration of how CSFmicroglia differ in ontogeny and function from BAMs thatwill need to be resolved with further investigation into theontogeny and function of each cell type

Microglia are long-lived self-replicating yolk sac-derivedimmune cells of the CNS parenchyma involved in CNSdevelopment and neuropathology2627 Disruption of theblood-brain barrier can allow the replacement of microglia byblood-derived progenitors with the latter eventually taking onthe near-complete gene expression profile of fetal-derivedmicroglia28 Our studies of CSF microglia cannot determinetheir ontogeny or how they gained access to the CSFMicroglia from the CNS parenchyma might traverse the piamater or choroid plexus or traffic into CSF through peri-vascular spaces or via lymphatic drainage2930 Migration ofparenchymal microglia or blood-derived progenitor cells thatbecome microglia may occur in response to chemokineswhich is supported by our finding of chemokine receptors onthese cells including CCR1 CCR5 CXCR4 and CX3CR1 Itis also possible that a unique microglial subset exists thatdevelops and permanently resides in the CSF CSF microgliaalso may not be singular in origin31


Figure 2 Relative expression of myeloid population-defining genes in neuroinflammation

(A) Expression of canonical microglial genes within the UMAP of CSF immune cell clusters (B) Heatmap representation of gene expression in blood (purple)and CSF (yellow) myeloid populations Gene list derived from Sankowski et al7 (C) Top genes reported by Farhadian et al3 as specific for ldquomicroglia-like cellsrdquo(top) and Beltran et al4 as specific for ldquomonocytesrdquo (bottom) displayed in the UMAP of immune cell clusters in the CSF of all subjects cDC = conventional DCPBMC = peripheral blood mononuclear cell pDC = plasmacytoid DC UMAP = Uniform Manifold Approximation and Projection


The source of CSF microglia may have functionally relevantimplications Microglia are known to be critical modulators ofCNS developmental processes such as synaptic pruning3233

Recent data reveal the subsets of microglia with distinctfunctions In the scRNA-seq analysis of rodent CNS BAMsand other myeloid populations have been identified within

Figure 3 Discrimination of myeloid cell populations in neuroinflammation by gene expression

Heatmap of 10 genes that discriminate myeloid cell types within the blood (purple) and CSF (yellow) Columns represent individual subjects cDC = con-ventional DC PBMC = peripheral blood mononuclear cell pDC = plasmacytoid DC


Figure 4 Flow cytometric identification of microglia and myeloid cells within the blood and CSF

(A andB) Plots were obtained by gating on CD3minusCD19minus singlets HLA-DR+BDCA-2hi cells represent pDCs cDCs are represented asHLA-DR+BDCA-2minusCD1chi Redgates in the third column represent microglia whereas blue gates represent monocytes The fourth column shows HLA-DR and Lyve-1 staining of microglia(red) andmonocytes (blue) The fifth column shows HLA-DR and Lox-1 staining of microglia (red) andmonocytes (blue) (A) Flow cytometry on CSF and bloodsamples from a subject with RRMS (B) Flow cytometry on CSF from subjects with anti-MOG disorder (anti-MOG) ALS IIH and a healthy control (C) Meanfluorescence intensity (MFI) of HLA-DR CD33 Lyve-1 and Lox-1 of CSFmonocytes andmicroglia from subjects Each pair of connected circles represents oneCSF sample Statistical significance was determined using a paired Student t test ALS = amyotrophic lateral sclerosis cDC = conventional DC IIH = idiopathicintracranial hypertension MOG = myelin oligodendrocyte glycoprotein PBMC = peripheral blood mononuclear cell pDC = plasmacytoid DC


brain choroid plexus and leptomeningeal tissue at steady-state and in the experimental autoimmune encephalomyelitismodel of MS8 Studies using animal models for Alzheimerdisease (AD) have identified a phagocytic subset of microgliatermed ldquodisease-associated microgliardquo (DAM)23 DAM ap-pear to reduce disease severity in the model and express genesassociated with AD such as Apoe and Trem2 In additionhomeostatic microglia and a subpopulation transitioning be-tween homeostatic microglia and DAM were identified inmouse AD models Experiments such as these in mice high-light not only the high resolution of CNS myeloid cell sub-types that can be obtained using scRNA-seq but also theapparent dynamic nature of myeloid populations

Using a flow cytometric panel with commercially availablereagents our data demonstrated the protein expression of someof the microglial genes identified by RNA and indicate thatmicroglia range from 02 to 55 of CSF mononuclear cellsOverall the presence of microglia within the CSF was notdisease-dependent or disease-activity dependent because theywere detected in theCSF of a healthy subject as well as in subjectswith noninflammatory neurologic diseases (IIH ALS) and in 2distinct neuroinflammatory diseases (MS and anti-MOG disor-der) Comparison of CSF microglial frequencies between dis-eases could not be performed because of low cell numbers It ispossible that subtle differences in the numbers or subtypes ofCSF microglia could reflect the state of disease in MS or otherdiseases Neither of the subjects with RRMS nor the subject withanti-MOG disorder analyzed for our study was undergoing acuterelapse which may have influenced the presence or number ofCSF microglia Our data demonstrate the feasibility of usingstandardflow cytometry in the future to address the association ofCSF myeloid populations including CSF microglia with theclinical state of inflammatory demyelination within the CNS

cDCs and pDCs were each identified in CSF by both scRNA-seq and flow cytometry Previous studies have identified thesecell types in CSF using lower resolution flow cytometry withfewer markers for discrimination of these myeloid populationsand have reported an increased number of pDCs in the CSFduring MS relapses2134 Recent scRNA-seq analysis examinedDCs and monocytes in human blood identifying 6 discretesubsets of blood DCs including 4 subsets of cDCs a subset ofclassical pDCs and a new subset termedAXL+SIGLEC6+ (AS)DCs35 ScRNA-seq analysis of the samples in our study did notresult in discrete clusters of DC subsets in the blood or CSFperhaps owing to their low abundance A dedicated scRNA-seqstudy focused on sorted CSFDCs would likely reveal additionalheterogeneity in these populations More importantly ouranalyses of differentially expressed genes between cDCs in theCSF compared with cDCs in the blood reveal that these pop-ulations although still fundamentally cDCs may alter some oftheir gene expression based on their microenvironment In-terestingly those genes thatmost discriminatedCSF cDCsweremore highly expressed by CSF microglia whereas those thatdiscriminated blood cDCsweremore highly expressed by bloodmonocytes These results suggest that environmental signals

may result in shared tissue-specific gene expression patternsacross myeloid cell populations

Using gene expression-based methods to categorize cell typeswe identified microglia monocytes cDCs and pDCs withinthe CSF The recognition that microglia are present in CSFmay have important clinical implications Abnormally func-tioningmicroglia are believed to contribute to the pathogenesisandor regulation of several diseases including AD and MSOur results indicate that it may be possible to analyzemicroglialabnormalities directly from the CSFwithout requiring biopsies

AcknowledgmentThe authors thank Drs Salim Chahin Cynthia Montana andTimothy Miller for referring subjects to these studies and forhelpful discussions

Study fundingGFW was supported by the NINDS (R01NS106289) and theNational Multiple Sclerosis Society (RG-1802-30253) AHCwas supported in part by The Manny and Rosalyn Rosenthaland Dr John L Trotter Chair in Neuroimmunology of theFoundation for Barnes-Jewish Hospital BTE was supportedby the Washington University Institute of Clinical and Trans-lational Science Clinical and Translational Research FundingProgram supported by the Foundation for Barnes-JewishHospital EE was supported by Shawn Hu and Angela Zenggraduate fellowship CC was supported by the National MSSociety Career Transition Fellowship (TA-1805-31003) Re-search reported in this publication was supported by generousdonations from the Lisa Novelly Fund of the John Trotter MSCenter and the Frala Osherow Fund for MS Research Re-search was also supported by the Washington University In-stitute of Clinical and Translational Sciences grantUL1TR002345 from the National Center for AdvancingTranslational Sciences (NCATS) of the NIH The content issolely the responsibility of the authors and does not necessarilyrepresent the official view of the NIH

DisclosureE Esaulova C Cantoni I Shchukina and K Zaitsev reportno disclosures RC Bucelli received compensation from MTPharma for advisory board and private funding for Parsonage-Turner syndrome for research equity in NeuroQuestionsLLC GF Wu received honoraria for consulting for Novartisand Genentech received research funding from Biogen EMDSerono and Roche MN Artyomov reports no disclosuresAH Cross received honoraria for consulting for BiogenCelgene EMD Serono Genentech Novartis Roche and TGTherapeutics received compensation for speaking for Gen-entech BT Edelson reports no disclosures Go toNeurologyorgNN for full disclosures

Publication historyReceived by Neurology Neuroimmunology amp NeuroinflammationSeptember 3 2019 Accepted in final form March 10 2020


References1 Alvermann S Hennig C Stuve O Wiendl H Stangel M Immunophenotyping of

cerebrospinal fluid cells in multiple sclerosis in search of biomarkers JAMA Neurol201471905ndash912 doi 101001jamaneurol2014395

2 Papalexi E Satija R Single-cell RNA sequencing to explore immune cell heteroge-neity Nat Rev Immunol 20181835ndash45 doi 101038nri201776

3 Farhadian SF Mehta SS Zografou C et al Single-cell RNA sequencing revealsmicroglia-like cells in cerebrospinal fluid during virologically suppressed HIV JCIInsight 20183121718 doi 101172jciinsight121718

4 Beltran E Gerdes LA Hansen J et al Early adaptive immune activation detected inmonozygotic twins with prodromal multiple sclerosis J Clin Invest 20191294758ndash4768 doi 101172JCI128475

5 Masuda T Sankowski R Staszewski O et al Spatial and temporal heterogeneity ofmouse and human microglia at single-cell resolution Nature 2019566388ndash392 doi101038s41586-019-0924-x

6 Geirsdottir L David E Keren-Shaul H et al Cross-species single-cell analysis revealsdivergence of the primate microglia program Cell 20191791609ndash1622e16 doi 101016jcell201911010

7 Sankowski R Bottcher C Masuda T et al Mapping microglia states in the humanbrain through the integration of high-dimensional techniques Nat Neurosci 2019222098ndash2110 doi 101038s41593-019-0532-y

8 Jordatildeo MJC Sankowski R Brendecke SM et al Neuroimmunology single-cellprofiling identifies myeloid cell subsets with distinct fates during neuroinflammationScience 2019363eaat7554 doi 101126scienceaat7554

9 Wallin MT Culpepper WJ Campbell JD et al The prevalence of MS in the UnitedStates a population-based estimate using health claims data Neurology 201992e1029ndashe1040 doi 101212WNL0000000000007035

10 Kitley J Woodhall M Waters P et al Myelin-oligodendrocyte glycoprotein anti-bodies in adults with a neuromyelitis optica phenotype Neurology 2012791273ndash1277 doi 101212WNL0b013e31826aac4e

11 Kim SM Woodhall MR Kim JS et al Antibodies to MOG in adults with in-flammatory demyelinating disease of the CNS Neurol Neuroimmunol Neuroinflamm20152e163 doi 101212NXI0000000000000163

12 Hyun JW Huh SY Shin HJ et al Evaluation of brain lesion distribution crite-ria at disease onset in differentiating MS from NMOSD and MOG-IgG-associated encephalomyelitis Mult Scler J 201925585ndash590 doi 1011771352458518761186

13 Thompson AJ Banwell BL Barkhof F et al Diagnosis of multiple sclerosis 2017revisions of the McDonald criteria Lancet Neurol 201817162ndash173 doi 101016S1474-4422(17)30470-2

14 Butler A Hoffman P Smibert P Papalexi E Satija R Integrating single-cell tran-scriptomic data across different conditions technologies and species Nat Biotechnol201836411ndash420 doi 101038nbt4096

15 Becht E McInnes L Healy J et al Dimensionality reduction for visualizingsingle-cell data using UMAP Nat Biotechnol 20193738ndash47 doi 101038nbt4314

16 Amir EAD Davis KL Tadmor MD et al ViSNE enables visualization of high di-mensional single-cell data and reveals phenotypic heterogeneity of leukemia NatBiotechnol 201331545ndash552 doi 101038nbt2594

17 Van DerMaaten L Hinton G Visualizing data using t-SNE JMach Learn Res 200892579ndash2625

18 Crinier A Milpied P Escaliere B et al High-dimensional single-cell analy-sis identifies organ-specific signatures and conserved NK cell subsets inhumans and mice Immunity 201849971ndash986e5 doi 101016jimmuni201809009

19 Butovsky O Jedrychowski MP Moore CS et al Identification of a unique TGF-β-dependent molecular and functional signature in microglia Nat Neurosci 201417131ndash143 doi 101038nn3599

20 Gautier EL Shay T Miller J et al Gene-expression profiles and transcriptional reg-ulatory pathways that underlie the identity and diversity of mouse tissue macrophagesNat Immunol 2012131118 doi 101038ni2419

21 Pashenkov M Huang YM Kostulas V Haglund M Soderstrom M Link H Twosubsets of dendritic cells are present in human cerebrospinal fluid Brain 2001124480ndash492 doi 101093brain1243480

22 Mrdjen D Pavlovic A Hartmann FJ et al High-dimensional single-cell mapping ofcentral nervous system immune cells reveals distinct myeloid subsets in healthaging and disease Immunity 201848380ndash395e6 doi 101016jimmuni201801011

23 Keren-Shaul H Spinrad A Weiner A et al A unique microglia type associated withrestricting development of Alzheimerrsquos disease Cell 20171691276ndash1290e17 doi101016jcell201705018

24 Schafflick D Xu CA Hartlehnert M et al Integrated single cell analysis of blood andcerebrospinal fluid leukocytes in multiple sclerosis Nat Commun 202011247 doi101038s41467-019-14118-w

25 Brioschi S Zhou Y Colnna M Brain parenchymal and extraparenchymal macro-phages in development homeostasis and disease Review J Immunol 2020204294ndash305 doi 104049jimmunol1900821

26 Ginhoux F Greter M Leboeuf M et al Fate mapping analysis reveals that adultmicroglia derive from primitive macrophages Science 2010330841ndash845 doi 101126science1194637

27 Colonna M Butovsky O Microglia function in the central nervous system duringhealth and neurodegeneration Annu Rev Immunol 201735441ndash468 doi 101146annurev-immunol-051116-052358

28 Shemer A Grozovski J Tay TL et al Engrafted parenchymal brainmacrophages differfrom microglia in transcriptome chromatin landscape and response to challenge NatCommun 201895206 doi 101038s41467-018-07548-5

29 Louveau A Smirnov I Keyes TJ et al Structural and functional features of centralnervous system lymphatic vessels Nature 2015523337ndash341 doi 101038nature14432

30 Ransohoff RM Kivisakk P Kidd G Three or more routes for leukocyte migration intothe central nervous system Nat Rev Immunol 20033569ndash581 doi 101038nri1130

31 Li Q Cheng Z Zhou L et al Developmental heterogeneity of microglia and brainmyeloid cells revealed by deep single-cell RNA sequencing Neuron 2019101207ndash223e10 doi 101016jneuron201812006

32 Schafer DP Lehrman EK Kautzman AG et al Microglia sculpt postnatal neuralcircuits in an activity and complement-dependent manner Neuron 201274691ndash705doi 101016jneuron201203026

33 Paolicelli RC Bolasco G Pagani F et al Synaptic pruning by microglia is necessary fornormal brain development Science 20113331456ndash1458 doi 101126science1202529

34 Longhini ALF von Glehn F Brandatildeo CO et al Plasmacytoid dendritic cells areincreased in cerebrospinal fluid of untreated patients during multiple sclerosis relapseJ Neuroinflammation 201182 doi 1011861742-2094-8-2

35 Villani A-C Satija R Reynolds G et al Single-cell RNA-seq reveals new types ofhuman blood dendritic cells monocytes and progenitors Science 2017356eaah4573 doi 101126scienceaah4573

Appendix Authors

Name Location Contribution

EkaterinaEsaulova MS

WashingtonUniversity in StLouis

Data acquisition and analysisdrafting of text and figures revisedthe manuscript for intellectualcontent

ClaudiaCantoniPhD


Study concept and design dataacquisition and analysis drafting ofthe text and figures revised themanuscript for intellectual content

IrinaShchukinaMS


Data acquisition and analysis

KonstantinZaitsev MS



Robert CBucelli MDPhD



Gregory FWu MD PhD



Maxim NArtyomovPhD


Study concept and design dataacquisition and analysis revised themanuscript for intellectual content

Anne HCross MD



Brian TEdelson MDPhD




DOI 101212NXI000000000000073220207 Neurol Neuroimmunol Neuroinflamm

Ekaterina Esaulova Claudia Cantoni Irina Shchukina et al neuroinflammation

Single-cell RNA-seq analysis of human CSF microglia and myeloid cells in

This information is current as of May 5 2020

Academy of Neurology All rights reserved Online ISSN 2332-7812Copyright copy 2020 The Author(s) Published by Wolters Kluwer Health Inc on behalf of the AmericanPublished since April 2014 it is an open-access online-only continuous publication journal Copyright

is an official journal of the American Academy of NeurologyNeurol Neuroimmunol Neuroinflamm

ServicesUpdated Information amp

httpnnneurologyorgcontent74e732fullhtmlincluding high resolution figures can be found at

References httpnnneurologyorgcontent74e732fullhtmlref-list-1

This article cites 35 articles 6 of which you can access for free at

Citations httpnnneurologyorgcontent74e732fullhtmlotherarticles

This article has been cited by 1 HighWire-hosted articles

Subspecialty Collections

httpnnneurologyorgcgicollectionmultiple_sclerosisMultiple sclerosis

httpnnneurologyorgcgicollectionamyotrophic_lateral_sclerosis_Amyotrophic lateral sclerosis

httpnnneurologyorgcgicollectionall_immunologyAll Immunology

httpnnneurologyorgcgicollectionall_demyelinating_disease_cnsAll Demyelinating disease (CNS)following collection(s) This article along with others on similar topics appears in the

Permissions amp Licensing

httpnnneurologyorgmiscaboutxhtmlpermissionsits entirety can be found online atInformation about reproducing this article in parts (figurestables) or in

Reprints

httpnnneurologyorgmiscaddirxhtmlreprintsusInformation about ordering reprints can be found online



CSF evaluation is used to aid in the diagnosis and differentia-tion of CNS disorders In inflammatory CNS diseases the CSFis typically used to assess the immunopathophysiologic pro-cesses because biopsy of CNS tissue carries significant potentialfor harm1 However relatively few cells are obtained from CSFusually on the order of 1ndash5 cellsμL Recent refinements innext-generation sequencing have enabled the efficient de-termination of individual cell gene expression within bio-specimens with relatively sparse cell populations such as theCSF Patterns identified using single-cell RNA sequencing(scRNA-seq) can uncover distinct cell types present at lowlevels within cellular communities and tissues2 scRNA-seq wasused to assess inflammatory changes within the CSF of subjectswith HIV infection identifying the presence of a ldquomicroglial-likerdquo cell3 and more recently to explore the clonal expansion ofCSF lymphocytes in MS-discordant monozygotic twin pairs4

scRNA-seq has also been used to address the issue of microglialheterogeneity within the human brain5ndash7 In addition using theprimary animal model of MS experimental autoimmune en-cephalomyelitis scRNA-seq has been used to identify severalpopulations of myeloid cells both endogenous to the CNS andfrom peripheral blood8 New methods for characterization ofmyeloid populations within the CNS during disease offer theopportunity to dissect the origin function and pathogenicity ofeach cell type with much greater resolution than previousmethods

MS is the most common inflammatory demyelinating diseaseof the CNS affecting over 600000 people in the UnitedStates9 Anti-myelin oligodendrocyte glycoprotein (MOG)disorder is a newly described CNS demyelinating disease thatshares clinical and pathologic characteristics with MS1011 MSand anti-MOG disorder appear to be distinct from one an-other and from aquaporin 4 antibody-positive neuromyelitisoptica (NMO)1012 We have applied scRNA-seq to examinethe CSF and mononuclear cells of the peripheral blood ofsubjects with relapsing-remitting MS (RRMS) and anti-MOGdisorder Individual spinal fluid samples from 2 subjects withRRMS and 1 subject with anti-MOG disorder were analyzedby using scRNA-seq In all 3 subjects we uncovered CSFpopulations of immune cells including microglial cells mon-ocytes and dendritic cells (DCs) based on gene expressionUsing CSF and blood from 7 additional subjects with RRMSanother subject with anti-MOG disorder and 3 control sub-jects we designed and tested a flow cytometry strategy thatconfirmed the presence in CSF of these cell types

MethodsSubjectsEleven subjects with inflammatory demyelinating disease (9with RRMS and 2 with anti-MOG disorder) and 3 controlsubjects (1 with amyotrophic lateral sclerosis [ALS] 1 withidiopathic intracranial hypertension [IIH] and 1 healthy control[HC]) were recruited for a study to assess the characteristics ofCSF and blood cells (table) The institutional review board ofWashington University in St Louis approved study protocolsand each subject provided informed consent Nine subjects hadRRMS based on the current diagnostic criteria13 Two addi-tional subjects were diagnosed with anti-MOG disorder onepresented with optic neuritis and the other with partial trans-verse myelitis Anti-MOG disorder was diagnosed based on the6-month sustained positive cell-based assays for MOG IgG1antibody absence of antibodies to aquaporin 4 (NMO-IgG)and absence of CSF-restricted oligoclonal bands11

To prevent alterations because of disease-modifying therapies weselected subjects either on no therapy or remote from therapySeven subjects with RRMS were naive to therapy including cor-ticosteroids Of the remaining 2 subjects with RRMS 1 (subject 7)had received fingolimod stopping 9 months before CSF testingand the other (subject 10) was treated with glatiramer acetate for 5months and had received IV steroids 2 months before CSF anal-ysisThe subjectwith transversemyelitis due to anti-MOGdisorder(subject 8) had receivedmycophenolate and then azathioprine buthad discontinued these 6 months before CSF analysis and hadreceived a course of corticosteroids 4 months before CSF exami-nation Subject 1 with anti-MOG disorder had received no treat-ment before the examination of CSF Control specimens wereobtained from a subject with IIH who presented with visual dis-turbances a patient diagnosed with ALS and a HC Each con-tributed blood and CSF cells for scRNA-seq or flow cytometrystudies (table and figure e-1 linkslwwcomNXIA244)

Standard protocol approvals registrationsand patient consentsInformed consent was obtained from all participants andapproved by the Human Research Protection Office ofWashington University in St Louis

CSF and peripheral blood mononuclearcell preparationCSF was collected in a 50-mL conical tube on ice andcentrifuged for 10 minutes at 400g Supernatant was removed

GlossaryAD = Alzheimer disease ALS = amyotrophic lateral sclerosis BAM = border-associated macrophage CCA = canonicalcorrelation analysis cDC = conventional DCDAM = disease-associated microgliaDC = dendritic cellHC = healthy controlHLA-DR = human leukocyte antigen DR IIH = idiopathic intracranial hypertension MOG = myelin oligodendrocyteglycoproteinNMO = neuromyelitis optica PBMC = peripheral blood mononuclear cell PCA = principal component analysispDC = plasmacytoid DC RRMS = relapsing-remitting MS scRNA-seq = single-cell RNA sequencing UMAP = UniformManifold Approximation and Projection















































































































Appendix Authors





ClaudiaCantoniPhD



IrinaShchukinaMS









Gregory FWu MD PhD



Maxim NArtyomovPhD



Anne HCross MD


























Reprints

















































































































Appendix Authors





ClaudiaCantoniPhD



IrinaShchukinaMS









Gregory FWu MD PhD



Maxim NArtyomovPhD



Anne HCross MD


























Reprints























Reprints




article open access single-cell rna-seq analysis of human ... · subject age, y sex race diagnosis...

Documents