redefining the heterogeneity of peripheral nerve cells in ......supplementary information for...

28
Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien Wolbert 1,* , Xiaolin Li 1,* , Michael Heming 1,* , Anne K. Mausberg 2 , Dagmar Akkermann 3 , Clara Frydrychowicz 3 , Robert Fledrich 4 , Linda Groeneweg 5 , Christian Schulz 6 , Mark Stettner 2 , Noelia Alonso Gonzalez 5 , Heinz Wiendl 1 , Ruth Stassart 3,$ , and Gerd Meyer zu Hörste 1,$ 1) Department of Neurology with Institute of Translational Neurology, University Hospital Münster, Münster, Germany. 2) Department of Neurology, University Hospital Essen, University Duisburg Essen, Essen, Germany. 3) Department of Neuropathology, University Hospital Leipzig, Leipzig, Germany. 4) Institute of Anatomy, Leipzig University, Leipzig, Germany 5) Institute of Immunology, Westfälische Wilhelms University, Münster, Germany 6) Medizinische Klinik und Poliklinik I, Klinikum der Universität München, Ludwig-Maximilians Universität, München, Germany * These authors contributed equally $ These authors co-supervised the study Correspondence to: Gerd Meyer zu Hörste, MD Email: [email protected] This PDF file includes: Supplementary Methods Figures & legends S1 to S15 Table legends S1 to S14 Supplementary references Other supplementary materials for this manuscript include the following: Tables S1 to S14 www.pnas.org/cgi/doi/10.1073/pnas.1912139117

Upload: others

Post on 27-Jul-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary Information for

Redefining the heterogeneity of peripheral nerve cells in health and

autoimmunity

Jolien Wolbert1,*

, Xiaolin Li1,*

, Michael Heming1,*

, Anne K. Mausberg2, Dagmar Akkermann

3, Clara

Frydrychowicz3, Robert Fledrich

4, Linda Groeneweg

5, Christian Schulz

6, Mark Stettner

2, Noelia

Alonso Gonzalez5, Heinz Wiendl

1, Ruth Stassart

3,$, and Gerd Meyer zu Hörste

1,$

1) Department of Neurology with Institute of Translational Neurology, University Hospital Münster, Münster, Germany.

2) Department of Neurology, University Hospital Essen, University Duisburg Essen, Essen, Germany.

3) Department of Neuropathology, University Hospital Leipzig, Leipzig, Germany.

4) Institute of Anatomy, Leipzig University, Leipzig, Germany

5) Institute of Immunology, Westfälische Wilhelms University, Münster, Germany

6) Medizinische Klinik und Poliklinik I, Klinikum der Universität München, Ludwig-Maximilians Universität, München,

Germany

* These authors contributed equally

$ These authors co-supervised the study

Correspondence to: Gerd Meyer zu Hörste, MD

Email: [email protected]

This PDF file includes:

Supplementary Methods

Figures & legends S1 to S15

Table legends S1 to S14

Supplementary references

Other supplementary materials for this manuscript include the following:

Tables S1 to S14

www.pnas.org/cgi/doi/10.1073/pnas.1912139117

Page 2: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary Methods

Supplementary Methods

Animals

C57BL/6J mice were originally purchased from the Jackson Laboratory and subsequently maintained at

the animal facility of the Medical Faculty of the Westfälische Wilhelms University Münster. Female

C57BL/6J mice were sacrificed at the age of 8-12 weeks. Icam1tm1Jcgr1

NOD mice (named ICAM-1-/-

NOD

mice for simplicity) were previously described (4) and maintained under specific pathogen free conditions

in the animal facility of the University Hospital in Essen. Mice were analyzed for clinical signs of

neuropathy as described (5). Female ICAM-1-/-

NOD mice were sacrificed at the age of 36 weeks and did

not show clinical symptoms of neuropathy. Parts of sciatic nerves of selected donors were processed for

cyro-sectioning and then stained for Hematoxylin and Eosin. Donor mice were without obvious immune

cell infiltrates (Fig. S7A). Young female, prediabetic NOD/ShiLtJ mice at an age of 8-10 weeks were used

as controls (NOD control) and obtained from Charles River laboratories. Lewis rats were originally

purchased from Harlan Laboratory and subsequently maintained at the animal facility of the University

Hospital Essen. Rats were housed under specific pathogen free conditions. Female rats were sacrificed at

the age of 10 weeks. CX3CR1-GFP mice have been described (6), were initially obtained from the

Jackson laboratories and maintained at the Institute of Immunology of the Westfälische Wilhelms-

University Münster. hGFAP-GFP mice (7) (Jax strain #007669) and PDGFRɑ-EGFP mice (8) (strain

#003257) have been described and were purchased from The Jackson Laboratory. Flt3Cre mice (9) and

mice expressing membrane-targeted tandem dimer Tomato (mT) prior to and membrane-targeted green

fluorescent protein (mG) after Cre-mediated excision from the Rosa26 locus (mT/mG mice) (10) and their

intercross (Flt3Cre-mT/mG mice) (11–13) have been described previously. Flt3Cre-mT/mG mice were

maintained at the Animal Facility of the Ludwig Maximilians Universität, München, Germany and

processed in Münster immediately after transport. Sciatic nerves were processed into single cell

suspension for flow cytometry or preserved for immunohistochemistry.

Cell extraction and purification

Sciatic nerves and the brachial nerve plexus were dissected from intracardially PBS perfused, female

C57BL/6 mice (8-12 weeks), NOD/ShiLtJ mice (8-10 weeks), Icam1tm1Jcgr1

NOD mice (36 weeks) and male

Flt3Cre-mT/mG mice (10 weeks) and transferred into ice cold Hank’s buffered salt solution (HBSS)

supplemented with 10mM HEPES (Gibco). Nerve fractions were finely chopped without epineurium

removal and teased as described(14). Nerve tissue was then digested using three different protocols to

test the optimal enzyme combination for collection of viable PNS cells.

Protocol #1 was adapted from a previous study (5) and nerves were incubated in collagenase/dispase

(Roche, 0.5 mg/ml) for 1 h at 37˚C. Protocol #2 was modified from a previous study (15). Digestion mix

contained Trypsin (Gibco, 0.25%): Collagenase II (Worthington, 1.62 U/μl): Hyaluronidase (Worthington,

1%) in a ratio of 1:1:0.04 respectively. 1 μl Pronase (Roche, 1%) was added per 50 μl of mix. Nerves were

digested for 20 minutes at 37°C with a 10 seconds vortex after 10 minutes. Protocol #3 was previously

described (16) and involved cold protease activity. In short, nerves incubated in 10 mg/ml Native Bacillus

Lichenformis protease (Creative Enzymes NATE0633) with 125 U/ml DNAse (Sigma) for 7 minutes at 6°C

Page 3: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary Methods

while shaking. Tissue was transferred to a C-tube and the gentleMACS brain_03 program was completed

twice (gentleMACS Dissociator, Miltenyi Biotec). Incubation at 6°C was repeated for 8 min and all

subsequent steps were carried out at 4°C. Digestion was in all protocols terminated by the addition of 10

ml IMDM (Gibco) with 10% FCS and the cell suspension was filtered through a 70-μm cell strainer

(Falcon). Myelin was depleted using anti-myelin beads (Miltenyi Biotec) according to the manufacturer’s

protocol. Single cells were subsequently sorted (BD FACSAria III, BD FACSDiva v8.0.1 Software) for

intact viable cells using three viability markers: Zombi NIR APC Cy7, Calcein-AM FITC, and DAPI

(Biolegend) (Fig. S1B). Cell extraction from rat sciatic nerve was performed as previously described (17).

Briefly, sciatic nerves were homogenized in DMEM with 5% FCS using a scalpel and incubated with 1 mg

collagenase/dispase (Roche, Mannheim, Germany) and 100 µg DNase I (Roche) at 37° C for 45 minutes

each. Cells were washed twice with DMEM containing 5% FCS, resuspended in cold medium, and passed

through a 70µm cell strainer. Cells were centrifuged on a 30%/70% percoll gradient (GE healthcare,

Freiburg, Germany) at 1,000g for 30 minutes. Nerve mononuclear cells were collected from the interphase

and washed in culture media.

Generation of single cell libraries and sequencing

Single cell suspensions were loaded onto the Chromium Single Cell Controller using the Chromium Single

Cell 3' Library & Gel Bead Kit v2 (both from 10X Genomics) chemistry following the manufacturer’s

instructions. Sample processing and library preparation was performed according to manufacturer

instructions using AMPure XP beads (Beckman Coulter). Sequencing was either carried out on a local

Illumina Nextseq 500 using the High-Out 75 cycle kit with a 26-8-0-57 read setup or commercially

(Microanaly, China) on a NovaSeq 6000 using the 300 cycle kit with paired end 150 read setup. All the

samples were sequenced with a sequencing depth >50,000 reads per cell. Average sequencing depth

was 82,321 ± 12,332 SEM reads/cell (Table S1).

Preprocessing of sequencing data

Processing of sequencing data was performed with the cellranger pipeline v3.0.2 (10x Genomics) and

according to the manufacturer’s instructions. Raw bcl files were de-multiplexed using the cellranger

mkfastq pipeline. Subsequent reads alignments and transcript counting was done individually for each

sample using the cellranger count pipeline with standard parameters. The cellranger aggr pipeline was

employed, to generate a single cell-barcode matrix containing all the mice samples without normalization.

The normalization of each library was subsequently performed in Seurat (see below). The cellranger

computations were carried at the High Performance Computing Facility of the Westfälische Wilhems

University Münster. The pre-quality control (QC) total cell number was 28,550 with an average of 5,350 ±

50 SEM cells per mouse sample type (Table S1).

Clustering and differential expression analysis

Subsequent analysis steps were carried out with the R-package Seurat v3.0.0 (18) using R v3.6.0 as

recommended by the Seurat tutorials. Briefly, cells were filtered to exclude cell doublets and low-quality

cells with few genes or high mitochondrial counts. Specifically, cells with <200 genes / cell (C57BL/6J

Page 4: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary Methods

mice, rat, ICAM-1-/-

NOD and NOD control mice) or >2300 genes / cell (C57BL/6J mice, ICAM-1-/-

NOD and

NOD control mice), >3000 genes / cell (rat) and cells with >8% (C57BL/6J mice, ICAM-1-/-

NOD and NOD

control mice) or >10% (rat) mitochondrial genes were filtered out. After QC the total remaining cell number

used for further analysis was 5,400 (C57BL/6J mice), 12,500 (rat), 5,250 (ICAM-1-/-

NOD mice) and 5,400

(NOD control mice) (Table S1). In order to account for differences in the total number of molecules per

cell, the UMI data were normalized using a recently described approach with regularized negative

binomial regression(19). Dimensionality reduction was done by Principal Component analysis (PCA).

Statistically significant Principal Components (PCs) were identified by a combination of a JackStraw

significance test and an elbow plot. Dimensionality reduction was done by Uniform Manifold

Approximation and Projection (UMAP) with default parameters. Clusters were identified using the

“FindNeighbors“ and “FindClusters” function in Seurat. To annotate the clusters, genes differentially

expressed in a one vs. all cluster comparison were queried for known functions in a literature search and

plotted in feature plots.

Identifying differentially expressed genes between different conditions

In order to determine differentially expressed genes between ICAM-1-/-

NOD and NOD control mice, we

performed alignment using Harmony, a newly described alignment method that projects cells into a shared

embedding to cluster cell types across multiple experiments and conditions (20). Further downstream

analysis was conducted with the resulting harmony embeddings using UMAP, “FindNeighbors” and

“FindClusters” function in Seurat. We split each cluster into ICAM-1-/-

NOD and NOD control mice data and

used the “FindMarker” function to determine differentially expressed genes between the groups. Volcano

plots were generated with the R package EnhancedVolcano. Differentially expressed (DE) genes

identified by Seurat were used as input genes. The threshold for p values was set at 0.001 and for the

average log fold change at 0.5.

Identifying cellular interactions

Molecular interactions between the cells were identified by the recently developed CellPhoneDB (21).

Normalized and filtered scRNA-seq data with the clusters previously identified by Seurat were used for

CellPhoneDB analysis. Since the current CellPhoneDB release only accepts human ensembl IDs as input,

murine ensembl IDs were converted to human ensembl IDs using biomaRt(22). In total, 1,381 (9.82%) of

the murine ensembl IDs could not be matched by the ensembl database as suitable orthologues and were

discarded. As recommended, statistical iterations were set at 1000 and genes expressed by less than

10% of cells in the cluster were removed. The interactions are based on the CellPhoneDB repository (21).

Statistical significance of the cellular interactions were calculated as described (21). Briefly, the cell

clusters were randomly permuted and the mean of the average receptor and ligand expression of each

cluster was calculated. The p value for a given receptor-ligand complex was determined by calculating the

proportion of the means which are as or more extreme than the actual mean. Significant interactions

between clusters were visualized in a heatmap and clustered with complete linkage and Euclidean

distance measure using the R package pheatmap. Network visualization was performed with Cytoscape

v.3.7.1 using the previously identified significant interactions between the clusters. The network layout was

Page 5: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary Methods

set to compound spring embedder. To improve the comparability between different datasets, we

calculated the relative cluster size (cell count in cluster / total cell count in dataset) and the proportional

number of interactions (number of cell-cell interactions / total number of interactions in dataset). The width

and transparency of connecting arrows encoded the proportional number of interactions and the

directionality of ligand/receptor interaction. Node size encodes the relative cluster size. Network analysis

such as determination of betweenness centrality was performed with the integrated NetworkAnalzyer.

Gene set enrichment analysis

We used the Enrichr tool (23) to perform gene set enrichment analysis with the top markers of the

respective clusters identified by Seurat. The following reference datasets, which are integrated in Enrichr,

were employed: TF Perturbations Followed by Expression, Transcription Factor PPI, Enrichr Submission

TF Gene Cooccurrence Enrichment Analysis, WikiPathways 2019 Mouse, KEGG 2019 Mouse, Reactome

2016 and Panther 2016. Enrichment computation was conducted by Enrichr as described (23, 24). Briefly,

Fisher exact text was used to compute enrichment for input gene lists to determine a mean rank with

standard deviation from the expected rank. A z-score for deviation from this expected rank was calculated

by using a reference table of expected ranks with variances.

Comparison with published datasets

We compared DE genes in ICAM-1-/-

NOD vs. NOD control mice in specific clusters with a published

dataset of DE genes in EAE vs. control mice (2). To improve comparability, DE genes in our dataset were

identified using MAST (25) instead of Wilcoxon rank sum test with a lower average log fold change of

0.095. All DE genes with an adjusted p-value greater than 0.05 were removed. The top up- and

downregulated DE genes of the Falcao dataset (2) (cutoff gene expression >|4|) were compared with our

top DE genes using Venn Diagrams (R package VennDiagram (26)). The ‘Interferome’ database (3) was

then searched for the 30 intersecting genes.

Immunohistochemistry on mouse sciatic nerve

For histology-based methods, sciatic nerves were dissected from intracardially PBS perfused CD57BL/6

mice, CX3CR1-GFP mice and Flt3Cre-mT/mG mice and fixed in 4% PFA (Paraformaldehyde, Merck) for

24 hours. For fixed frozen slides (FF), tissue was dehydrated in a series of 15% and 30% sucrose and

embedded into OCT (Tissue-Tek). Snap freezing of the tissue blocks was done using dry ice. Cross

sections were cut (10 μm, Cryostar NX50, Thermo Scientific) and slides were stored at -80°C until staining

with fluorescent antibodies.

For immunocytochemistry, cytospin samples were prepared from peripheral nerve cells and autologous

bone marrow of intra-cardial PBS perfused C57BL/6 mice. PNS cells were isolated with protocol #2 as

previously described. Bone marrow was isolated from the femur by flushing the bone with PBS. Cells were

immediately filtered through a 70 μm cell strainer (Falcon), followed by ammonium chloride–based

erythrocyte lysis (BD Biosciences). Cyto-centrifugation was done with ±150’000 bone marrow cells or

±2’000 PNS cells. The aliquots were centrifuged at 100 xg for 5 minutes (Rotofix 32A, Hettich). Cytospin

slides were fixed in methanol for 5 minutes and air dried before staining.

Page 6: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary Methods

For antibody staining, tissue cross sections or cytospins were permeabilized with 0.1% TritonX-100

(Sigma) in PBS, and stained with the following primary antibodies: Cd68 (1:50, rat, Biocarta), F4/80

(1:500, rat, Serotec), Cxcl4 (1:100, rabbit, Thermofisher), Cd169 (1:100, gift from Dr. Antonio Castrillo),

Cd11b (1:200, rat, BD Biosciences) and SIGNR1 (1:50, A. hamster, Invitrogen) and incubated overnight at

4°C. Cells were washed with PBS and then incubated with Alexa Fluor (AF) conjugated antibodies (AF594

anti-rabbit, AF488 anti-rabbit, AF488 anti-rat, all 1:1000, Invitrogen) in blocking reagent (Roche), for 45

minutes at room temperature. The F4/80 signal was amplified by streptavidin-HRP (1:100, 45 min, RT),

tyramid (1:100, 15 min, RT) and streptavidin-594 (1:100, 45 min, RT). Slides were mounted in

Fluoromount G with DAPI (Invitrogen). Images were taken using a three laser fluorescent microscope

(Biorevo BZ-900 microscope with BZII Viewer software, Keyence) and processed in ImageJ.

Immunohistochemistry of human sural nerve biopsies

Human samples were selected according to the histological findings in sural nerve biopsies. Sural nerve

biopsies with no major pathological findings (with respect to inflammation, axonal and myelin pathology,

vascular pathology) were selected. Samples were anonymized and processed in a blinded manner.

Selected patients were of mixed age, between >30 years and <70 years of age (inclusion criteria) and did

not suffer from a severe neurological disorder at time-point of biopsy (exclusion criteria). In total five sural

nerve samples from five independent patients were analyzed. No other criteria besides the described

characteristics were applied. The study received ethical approval by the ethic board of the University Clinic

Leipzig, Germany. The staining was performed by the fully automated immunostainer Benchmark XT

(Roche, Basel, Switzerland). Staining protocol included deparaffinization and counterstain with

hematoxylin and blue colouring reagent according to manufacturers’ instructions. No pretreatment was

performed for MBP and SMA/ACTA2. Pretreatment with cell conditioning 1 (CC1, Roche, Basel,

Switzerland) was performed for all other antibodies. The following antibodies were used: LCA (CD45)

(leukocyte common antigen, monoclonal mouse, autostainer Dako, #IS751,), CD68 (monoclonal mouse,

1:100, Dako, #M0876), CD8 (monoclonal mouse, 1:50, Dako, #M7103), CD4 (monoclonal mouse, 1:100,

Dako, #M7310), CD34 (monoclonal mouse, 1:100, Dako, #M7165), SMA/ACTA2 (monoclonal mouse,

1:300, Dako, #M0851), SOX10 (polyclonal rabbit, 1:40, Cell Marque, #383A-76), MBP (monoclonal rabbit,

1:100, Cell Marque, #295A-16).

RNA in situ hybridization

RNA in situ hybridization (ISH) was performed on fixed frozen (FF), fresh frozen and paraffin embedded

sections of sciatic nerves from intra-cardial PBS perfused C57BL/6 mice, hGFAP-GFP mice and

PDGFRɑ-EGFP mice. Three different ISH kits were used according to manufacturer’s protocol. The

ViewRNA ISH Tissue Assay Kit (1-plex) from Thermo Fisher was used on fixed frozen tissue, to test the

probes Mm-Mbp, Mm-Apod, Mm-Smoc2, Mm-Sfrp4, and Mm-Pf4 and compared to Mm-Gapdh (positive

control) and Ba-DapB (negative control). Briefly, sections were dehydrated in a series of 50%-70%-90%

ethanol each for 10 min and baked in a dry oven at 60°C for 1 h. Protease QF (1:100) was applied to the

slides and incubated for 12,5 min at room temperature (RT), followed by two PBS washes. Probes were

diluted 1:40, hybridized for 3 hours at 40°C, 0% CO2 (CO2 cell culture incubator MCO-17A1, Sanyo) and

Page 7: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary Methods

slides were washed three times with wash buffer. Amplification steps were performed by incubating with

PreAmp1 QF (1:100, 25 min, 40°C), Amp1 QF (1:100, 15 min, 40°C) and Label Probe-AP (1:1000, 15

min, 40°C) with wash buffer washes of 3x2 min in between steps. Slides then incubated in AP-Enhancer

Solution for 5 min at RT and FastRed for 30 min at 40°C (ViewRNA Chromogenic Signal Amplification Kit,

1-plex, Thermo Fisher).

The ThermoFisher ViewRNA Cell Assay Kit (multiplex) was used, in combination with the first steps of the

previously mentioned Tissue Assay kit, to detect Mm-Apod, Mm-Smoc2, Mm-Ngfr, Mm-S100b, Mm-

Sox10, Mm-Sfrp4 and Mm-Pi16 and compared to Mm-Gapdh (positive control) and Ba-DapB (negative

control) in different single and co-stain settings. For the multiplex staining on sciatic mouse nerves, the

hybridization steps were performed with the ViewRNA Tissue Assay kit as previously described.

For the multiplex amplification steps, the ViewRNA ISH Cell Assay kit (multiplex) from Thermo Fisher was

used. Briefly, steps were performed by incubating with PreAmplifier Mix (1:100, 60 min, 40°C), Amplifier

Mix (1:100, 60 min, 40°C) and Label ProbeMix (1:100, 60 min, 40°C) with wash buffer washes of 3x2 min

in between steps. Slides then incubated in DAPI (1:100, 10 min, RT) followed by a PBS wash. Slides

processed with the Thermo Fisher ViewRNA ISH kits were mounted in Fluoromount G (Invitrogen).

The BaseScopeTM

Detection Reagent Kit – RED from ACD Biotech, was used to test the following probes:

Mm-Smoc2 and Mm-Sfrp4, co-stained with an antibody for Mbp, and compared to Mm-Ppib (positive

control) and Ba-DapB (negative control). The kit was used according to manufacturer’s instructions on

paraffin embedded samples. In short, paraffin embedded nerves were cut to 5 µm sections, dried

overnight at RT, baked one hour at 60 °C and subsequently de-paraffinized in a xylol/ethanol series.

Target retrieval was performed in RNAScope target retrieval buffer for 15 min at about 99 °C, then

washed in water and ethanol shortly. Protease Plus was applied on the samples and incubated for 30 min

at 40 °C. Target and control probes were applied to the sections, slides were hybridized for 2 h at 40°C

and washed twice with wash buffer for 2 min. Amplification steps were as follows, incubation with AMP1

(30 min, 40°C), AMP2 (15 min, 40°C), AMP3 (30 min, 40°C), AMP4 (15 min, 40°C), AMP 5 (30 min, RT)

and AMP 6 (15 min, RT) with wash buffer washes of 2x2 min in between steps. Signal was visualised by

incubation with the FastRED substrate (10 min at RT).

For myelin counterstaining the samples were washed twice in PBS after RNAScope incubation and

blocked with goat serum. Samples were incubated overnight at 4°C with Mbp antibody (1:200, rabbit,

CellMarque) and with its corresponding cyanine dye (1:1000, Dianova) for 1hr at RT and mounted in Aqua

Polymount. All images were obtained with an Axio Observer Z1 (Zeiss) and processed in AxioVision and

ImageJ.

Flow cytometry of leukocytes

Flow cytometry analysis was performed on isolated PNS cells, using protocol #2 as previously described.

After myelin removal, cells were stained for extracellular antibodies (25 min, RT). As a control, autologous

brain, spleen and bone marrow were isolated and processed into single cell suspension as previously

described(5). In short, the brain was first mechanically digested followed by enzymatic digestion with

CollagenaseD (2.5 mg/ml) and DNaseI (0.05 mg/ml). The digested brain, bone marrow and spleen were

mashed through a 70 μm cell strainer (Falcon) followed by ammonium chloride–based erythrocyte lysis

Page 8: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary Methods

(BD Biosciences). The following viability dye and murine antibodies were used: zombi NIR APC-Cy7;

CD45 BV421; CD11b BV510; CD68 PE Cy7; B220 PE; CD3 FITC; CD3 BV510; CD4 APC; CD8 Pacific

Blue; NK1.1 PE Cy7; NKG2AB6 PE; NKp46 FITC; F4/80 APC; CD14 FITC; Ly6C Percp Cy5.5; CD317

Pacific Blue; CCR9 PE Cy7; CD11c AF700; MHCII PE. Samples were measured on the Gallios (10

Colors, 3 lasers, Beckman Coulter) with Kaluza for Gallios software and analysed with FlowJo_V10.

Page 9: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

C

D

scRNA-seq

sciatic nerve & b. plexus tease nerve fibers enzymatic digestion myelin depletion viable cell sorting single cell RNA seq

B

0

viable non-viable

500

1000

1500

2000ce

ll co

unts

protocol#1 #2 #3

A

−10

0

10

−15 −10 −5 0 5UMAP1

UM

AP2

fibronmSC

mySC

EC1

BC

TC

vSMC

PC

MP

batch #1batch #2batch #3

#1

●●

●●

●●

● ●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

● ●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

MC

EC2

lymph

0 50K 100K 150K 200K 250K

FSC-A

0

-103

103

104

105

Calc

ein

AM

0 50K 100K 150K 200K 250K

FSC-A

0

50K

100K

150K

200K

250K

0-103

103

104

105

DAPI

0

103

104

105

Zom

bie

NIR

SSC-

A

Figure S1: Optimizing the peripheral nerve cell extraction protocol. (A) Peripheral nerve cells were isolated from the combined sciatic nerve and brachial plexus of naive adult C57BL/6 mice after intracardial PBS perfusion . Three different protocols were tested for enzymatic digestion: #1 collagenase/dispase (Roche, 0.5 mg/ml), 1 h at 37 ̊ C; #2 Trypsin (Gibco, 0.25%): Collagenase II (Worthington, 1.62 U/μl): Hyaluronidase (Worthington, 1%) ratio 1:1:0.04 + 1 μl Pronase (Roche, 1%)/50 μl of mix, 20 min at 37°C; #3 Native Bacillus Lichenformis protease (Creative Enzymes NATE0633, 10 mg/ml) + DNAse (Sigma, 125 U/ml), 7 min at 6°C, 2x gentleMACS brain_03 program (gentleMACS Dissociator, Miltenyi Biotec), 8 min at 6°C. The proportion of viable cells (grey part) of the plot against non-viable cells (white part) after each protocol is depicted in a stacked bar plot. Viable cells were defined as Calcein-AM+Zombie-NIR-DAPI- cells. (B) Gating strategy for flow cytometry-based viable cell sorting. (C) Experimental scheme of the five step cell extraction protocol: 1) intra-cardial PBS perfusion and isolation of sciatic nerve and brachial plexus, 2) peripheral nerve dissection and mechanical dissoci-ation, 3) four enzyme digestion, 4) magnetic bead-based myelin debris removal, 5) flow sorting for viable cells. This figure was modified from Servier Medical Art, licensed under a Creative Common Attribution 3.0 Generic License (D) After multi-step purification of peripheral nerve cells, single cell (sc) transcriptomes were generated from n = 36 naive adult female C57BL/6 mice in three biological replicates (each replicate n = 12). The biological replicates are highlighted: red = batch #1, blue = batch #2, green = batch #3

Page 10: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

A

lymph

MC

MP

TC BC EC2

mySC

fibronm

SCvSM

CEC

1PC

Dhtkd1MmeDnmt1SetxSlc25a46GarsYarsAtp7aPtrh2Cntnap1Gnb4Cox6a1Dnajb2Inf2Mfn2Dctn1Med25LmnaLrsam1Kif5aRetreg1Drp2Egr2Pmp22MpzPrxSco2Abhd12Rab7Hint1Atp1a1Coa7Fgd4HarsSpg11LitafSgpl1Aifm1Bscl2Pdk3Mpv17Trim2Sbf2Ndrg1GanArhgef10Fig4Fbln5Sigmar1Sbf1Mcm3apMorc2aDync1h1Dnm2VcpPlekhg5Hoxd10Mtmr2WarsPrps1Ighmbp2Sptlc1Hspb1KarsBag3Hspb8Prps1l3Dctn2Trpv4Kif1bNagluMarsAarsmt−Atp6Hk1Surf1

−3

−2

−1

0

1

2

3 Figure S2: Many hereditary neuropathy genes show highest expression in non-glia cells. (A) Genes known to cause hereditary neuropathies were retrieved from a public database (www.molgen.ua.ac.be/CMTMutations) and plotted in the clusters we identified in peripheral nerve cells of C57BL/6 mice. The average gene expression is color-coded. mySC: myelinating Schwann cells, nmSC: non-myelinating Schwann cells, fibro: fibroblasts, vSMC: vascular smooth muscle cells, PC: pericytes, EC1: endothelial cells cluster 1, EC2: endothe-lial cells cluster 2, lymph: lymphatic vessel endothelial cells, BC: B cells, TC: T cells and natural killer cells, MC: myeloid lineage cells, MP: macrophages.

Page 11: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

●●●●

●●●●

●●●

●●●●●

●●

●●●●●

●●●●●●●●

●●●●●●

●●●●●

●●●●

MPMCBCTC

lymphPC

vSMCEC2EC1fibro

nmSCmySC

Btg2 Fth1 Mt2 Mt1Soc

s3 Jun

Fos

Apoe

Cryab Ptn

Mbp

−1012

Average Expression

Percent Expressed●

●●

406080

100

●●●●●●●●●●●

●●●

●●

●●●

●●●●●

●●●●●

●●

●●●

●●

MPMCBCTC

lymphPC

vSMCEC2EC1fibro

nmSCmySC

Sox9Mmp2

Ccl11

Ebf1 Osr2Ceb

pdSpry

2Tcf4

Lama2

Hspg2

Myoc

ApodMatn

2

0

1

2

Average Expression

Percent Expressed●

●●●

204060

80

A

B

UMAP1

UM

AP2

Ngfr

1.0

0.0

2.0Cspg4

0.75

0.0

1.5

Pdgfrb

0.8

0.0

1.6gene score

2.0

0.0

4.0

C

Ngfr Sox10 S100b0

20

40

60

80

100

% o

f all A

pod+ a

ndAp

od+ S

moc

2+ cel

ls

D

Figure S3: Localization of novel transcripts in non-myelinating Schwann cells and fibroblasts. (A-B) Dotplots of selected mySC marker genes (A) and nmSC marker genes (B) grouped by cluster. The average gene expression level per cluster is color coded and circle size represents the percentage of cells expressing the gene. Threshold was set to a minimum of 10% of cells expressing the gene. (C) Feature plots were generated to show expression of Ngfr, Cspg4 and Pdgfrb individually and as gene score combined. Plots are corresponding to Fig. 1A. Magnifications are zoomed in on the pericyte (PC) cluster. (D) This graph shows a quantification of the RNA ISH stainings performed in Fig. 2B and Fig. S4-6. The percentage of cells that co-stained for Ngfr, Sox10 or S100b was calculated within cells that expressed Apod alone or Apod together with Smoc2. Data are depicted as mean ± SEM, n=12.

Page 12: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

DAPI ApodNgfr Smoc2

Ngfr DAPIApod Smoc2

Ngfr DAPIApod Smoc2

Ngfr DAPIApod Smoc2

Ngfr DAPIApod Smoc2

Ngfr DAPIApod Smoc2

Ngfr DAPIApod Smoc2

Ngfr DAPIApod Smoc2

Ngfr DAPIApod Smoc2

Ngfr DAPIApod Smoc2

Ngfr DAPIApod Smoc2

*

Figure S4: co-staining of the nmSC markers Apod and Smoc2 with NgfrFresh-frozen sections of sciatic nerves of naive adult C57BL/6 mice were stained for Apod, Smoc2 together with the Schwann cell marker Ngfr by RNA ISH as described in the methods. This figure corresponds to Fig. 2B. Please note that each dot repre-sents a single RNA molecule. White dotted line shows the epineurium border of the sciatic nerve. Nuclei were stained with DAPI. Scale bars 20 μm (left) and 10 μm (magnification). Arrows indicate co-staining of all markers, asterisks indicate co-stain of a new marker with a known lineage marker and arrowheads indicate individual staining.

Page 13: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

DAPI ApodS100b Smoc2

S100b DAPIApod

Smoc2

*

*

*

S100b DAPIApod

S100b DAPIApod

S100b DAPIApod

S100b DAPIApod

S100b DAPIApod

Smoc2S100b DAPIApod

Smoc2S100b DAPIApod

Smoc2S100b DAPIApod

Smoc2S100b DAPIApod

*

*

*

*

Figure S5: co-staining of the nmSC markers Apod and Smoc2 with S100bFresh-frozen sections of sciatic nerves of naive adult C57BL/6 mice were stained for Apod, Smoc2 together with the Schwann cell marker S100b by RNA ISH as described in the methods. This figure corresponds to Fig. 2B. Please note that each dot represents a single RNA molecule. White dotted line shows the epineurium border of the sciatic nerve. Nuclei were stained with DAPI. Scale bars 20 μm (left) and 10 μm (magnification). Arrows indicate co-staining of all markers, asterisks indicate co-stain of a new marker with a known lineage marker and arrowheads indicate individual staining.

Page 14: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

DAPI ApodSox10 Smoc2

Sox10 DAPIApod

Smoc2

*

*

Sox10 DAPIApod

Sox10 DAPIApod

Sox10 DAPIApod

Sox10 DAPIApod

Sox10 DAPIApod

Sox10 DAPIApod

Sox10 DAPIApod

Sox10 DAPIApod

Sox10 DAPIApod

Smoc2

Smoc2

*

**

*

*

*

*

*

*

*

Figure S6: co-staining of the nmSC markers Apod and Smoc2 with Sox10Fresh-frozen sections of sciatic nerves of naive adult C57BL/6 mice were stained for Apod, Smoc2 together with the Schwann cell marker Sox10 by RNA ISH as described in the methods. This figure corresponds to Fig. 2B. Please note that each dot repre-sents a single RNA molecule. White dotted line shows the epineurium border of the sciatic nerve. Nuclei were stained with DAPI. Scale bars 20 μm (left) and 10 μm (magnification). Arrows indicate co-staining of all markers, asterisks indicate co-stain of a new marker with a known lineage marker and arrowheads indicate individual staining.

Page 15: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Pi16 DAPISfrp4DAPI Sfrp4 Pi16

DAPI ApodSmoc2 Vim

DAPI PdgfraGFP Smoc2 Apod

*

*

DAPI Apod Smoc2 Vim

DAPI PdgfraGFP Smoc2 Apod

DAPI PdgfraGFP DAPISmoc2

Apod

DAPISmoc2

Apod DAPIVim

A

C

B

Pi16 DAPISfrp4

Pi16 DAPISfrp4

Pi16 DAPISfrp4

Pi16 DAPISfrp4

Pi16 DAPISfrp4

Pi16 DAPISfrp4

Pi16 DAPISfrp4

Pi16 DAPISfrp4

Figure S7: RNA ISH staining of multiple fibro markers(A) Fresh frozen sections from C57BL/6 mice, as in Fig. 2, were stained for Apod and Smoc2 and the fibroblast marker Vim (encoding Vimentin) by ISH. (B) Fresh frozen sections from PDGFRa-GFP mice, as in Fig. 2, were stained for Apod and Smoc2 by ISH. Scale bars 50 μm (left), 20 μm (right), and 10 μm (magnification). Asterisk indicates co-stain of Apod and Smoc2, arrowhead indicates single stain of fibroblast markers Vim and PdgfraGFP. (C) Fresh-frozen sections of sciatic nerves of naive adult C57BL/6 mice, as in Fig. 2, were stained for Sfrp4 with the fibro marker Pi16 by RNA ISH as described in the methods. This figure corres-ponds to Fig. 2D. Please note that each dot represents a single RNA molecule. The GFP protein signal is more homogeneously distributed. White dotted line shows the epineurium border of the sciatic nerve. Nuclei were stained with DAPI. Scale bars 20 μm (left) and 10 μm (magnification). Arrows indicate co-staining of the two markers and arrowheads indicate indivi-dual staining.

Page 16: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

PdgfraGFP Pi16 DAPIDAPI PdgfraGFP Pi16

Sfrp4 DAPIDAPI PdgfraGFP Sfrp4B

PdgfraGFP Pi16 DAPI

PdgfraGFP Pi16 DAPI

PdgfraGFP Pi16 DAPI

PdgfraGFP Pi16 DAPI

PdgfraGFP Pi16 DAPI

PdgfraGFP

Sfrp4 DAPIPdgfraGFP

Sfrp4 DAPIPdgfraGFP

Sfrp4 DAPIPdgfraGFP

Sfrp4 DAPIPdgfraGFP

Sfrp4 DAPIPdgfraGFP

A

Figure S8: Sciatic nerve of a PDGFRαGFP reporter mouse stained with the fibro markers Pi16 and Sfrp4PFA Fixed-frozen sections of sciatic nerves of naive adult PDGFRαGFP mice were stained for Pi16 (A) and Sfrp4 (B) by RNA ISH as described in the methods. This figure corresponds to Fig. 2D. Please note that each dot represents a single RNA molecule. The GFP protein signal is more homogene-ously distributed. White dotted line shows the epineurium border of the sciatic nerve. Nuclei were stained with DAPI. Scale bars 20 μm (left) and 10 μm (magnification). Arrows indicate co-staining of the two markers and arrowheads indicate individual staining.

Page 17: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

A SC (SOX10)B mySC (MBP)C

fibro + EC (CD34)D vSMC + PC (ACTA2)E

MP (CD68)G TC (CD8)H

Leuko (CD45)F

TC (CD4)I

Figure S9: Confirmation of some cell populations in human sural nerves.Representative histological stainings of human sural nerve biopsies of a control patient (out of n=5 patients) without signs of peripheral nerve pathology are shown. (A) Semithin cross section of a representative sural nerve. cale bar 50µm. (B-I) Immunohistochemistry (IHC) stainings of control sural nerve for SOX10 (B), MBP (C), CD34 (D), smooth muscle actin (ACTA2) (E), pan leukocyte marker CD45 (F), CD68 (G), CD8 (H) and CD4 (I). Scale bars 50µm. SC: Schwann cells, mySC: myelinating Schwann cells, fibro: fibroblasts, EC: endothelial cells, vSMC: vascular smooth muscle cells, PC: pericytes, Leuko: leukocytes, MP: macrophages, TC: T cells

Page 18: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

A

B

100

101

102

103

104

105

106

100

101

102

103

104

105

106

1,03% 0,21%

3,30%10

010

110

210

310

410

510

6

100

101

102

103

104

105

106

GPVI

CD

41

live/

dead

CD45

Bone marrow Sciatic nerve

UMAP1

UM

AP2

Ms4a7

0.0

2.0Ccl4

0.0

5.02.5

0.0

0.60.3

Cx3cr1Ccl17

0.0

4.02.0

C1qb

0.0

4.02.0

Rt1-Bb

0.0

4.02.0

C

8,14%

DAPI

Cxcl4F4/80

DAPI

Cxcl4F4/80

DAPI

Cxcl4F4/80

Figure S10: Cxcl4-expressing cells in peripheral nerves are not megakaryocytes. (A) Cytospins were generated from bone marrow and sciatic nerve from a female C57BL/6 mouse. Slides were stained for F4/80, Cxcl4 and DAPI by immunocytochemistry. Arrow indicates co-stain, arrow-heads indicate single stain of Cxcl4. (B) PNS cells were purified from two female C57BL/6 mice, pooled, and analyzed by flow cytometry. The viable CD45+ cell population (left), was stained for CD41 and GPVI (right) to determine the percentage of megakaryocytes. One repre-sentative dotplot out of two independent experiments is shown. (C) Feature plots of selected myeloid genes correspon-ding to Fig. 3E (rat dataset).

Page 19: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Cx3cr1GFP Cd68Cxcl4 DAPIDAPI Cx3cr1GFP

Cxcl4 Cd68

Cx3cr1GFP Cd68Cxcl4 DAPI

Cx3cr1GFP Cd68Cxcl4 DAPI

Cx3cr1GFP Cd68Cxcl4 DAPI

Cx3cr1GFP Cd68Cxcl4 DAPI

Cx3cr1GFP Cd68Cxcl4 DAPI

Figure S11: Sciatic nerve of a CX3CR1GFP reporter mouse stained with Cxcl4 and Cd68 antibodies PFA Fixed-frozen sections of sciatic nerves from CX3CR1-GFP reporter mice were stained for Cxcl4, Cd68 and DAPI using immunohistochemistry. This figure corresponds to Fig. 3H. Scale bar 50 μm (left), 20 μm (magnification). Please note that the samples were stained on protein level, so the signal is homogeneously distributed. Arrows indicate co-stain of two markers and arrowheads indicate individual staining.

Page 20: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

A

B

m-tdTomatom-GFP

% w

ithin

CD

11b+ C

D68

+

brain PNS BM SPC 0

25

50

75

10075.7%liv

e/de

adC

D11

b

CD68 m-tdTomato

m-G

FP

0.86% 66.0%

34.0%

C

●●●

●●

●●

●●

●●●

●●●

●●●●●

●●●

●●●

●●●

●●

●●

MPMCBCTC

lymphPC

vSMCEC2EC1fibro

nmSCmySC

Tgfbr1 Hexb Frcls Sparc Tmem119 Pry12

Average Expression

−1012

Percent Expressed ●●●

255075

D

Yolk sac hematopoiesis (<E10)

Microglia&

Some tissue macrophages

Yolk sacstem cell

M

EMP

YSC

Erythro-myeloidprogenitors

Definitive hematopoiesis (>E10)

TC BC

HSC

MPP

LPMP

GRM MK

Myeloid progenitor

Lymphoid progenitor

Multi potentprogenitor

Hematopietic stem cellFlt3- or Flt3+

MacrophagesGranulocytes

MegakaryocytesT cells

B cells

m-GFPm-tdTomato

m-GFPm-tdTomato

Figure S12: Nerve-associated macrophages are heterogeneous and transcriptionally different from microglia. (A) Simplified schematic overview of hematopoiesis, that takes place until embryonic day E8.0, and definitive hematopoiesis, starting from embryonic day E12.5. Colors of schematic cells indicate their expression of membrane-targeted tdTomato (mT; red) and membrane-targeted GFP (mG; green) in Flt3Cre-mT/mG mice. (B) Representative gating of pooled leukocytes extracted from the PNS of three Flt3Cre-mT/mG mice and stained against Cd11b and Cd68. (C) Three male adult Flt3Cre-mTmG mice were intracardially perfused with PBS. The proportion of viable Cd11b+Cd68+ cells expressing mT and mG in brain, peripheral nerve cells (PNS), bone marrow (BM), and spleen (SPC) was quantified by flow cytometry as in B. No CD45 antibodies were intravenously injected (1). Bar graph shows the proportion of cells in the m-GFP+ and m-tdTomato+ gates. One out of two expe-riments is shown. (C) Fresh frozen sections of sciatic nerves of a male Flt3Cre-mT/mG mouse were stained for DAPI. Scale bar 50 μm (left), 20 μm (right), arrowheads indicate single stain. (D) Dotplot of the expression of microglia markers in the cell clusters derived from naive C57BL/6 mice is shown. The circle size correlates with the percentage of cells expressing the gene and the average gene expression level per cluster is color-coded. Threshold was set to a minimum of 1% of cells in the cluster expres-sing the gene.

Page 21: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

A

B

UMAP1

UM

AP2

0.0

2.01.0

H2-DMb2

0.0

2.01.0

Ms4a1

0.0

2.01.0

Ighd

0.0

6.03.0

Ighg1

0.0

2.01.0

Cx3cr1

0.0

2.01.0

Vps37b

0.0

2.01.0

Cxcr6

0.0

2.01.0

Cd8a

0.0

2.01.0

Klrc1 Xcl1

0.0

3.01.5

Adgre1

0.0

1.0Pf4

0.0

2.0Cd86

0.0

1.6

0.8

0.0

1.00.5

Flt3 Siglech

0.0

2.01.0

C Pmp22

2.0

0.0

4.0

1.50.0

3.0

2.5

0.0

5.0

2.0

0.0

4.0

Gsn Mpz

2.0

0.0

4.0

1.50.0

3.0

1.25

0.0

2.5Sostdc1

1.00.0

2.0

NO

D

ICAM

-1-/-N

OD

2.0

0.0

4.0

2.5

0.0

5.0

Ccl5Psmb8

0.8

0.0

1.6

1.0

0.0

2.0

B2m

1.5

0.0

3.0

1.50.0

3.0

H2-K1

1.0

0.0

2.0

1.25

0.0

2.5

NO

D

ICAM

-1-/-N

OD

UMAP1

UM

AP2

Figure S13: Pre-inflammatory stage of sciatic nerve and gene expression of ICAM-1-/-NOD mice. (A) Longitudinal cryo-section of the sciatic nerve of pre-clinical female ICAM-1-/-NOD mice stained with hematoxylin and eosin. One represen-tative out of 10 mice is shown. Scale bars represent 200μm and 50μm. (B) Feature plots of selected leukocyte markers corresponding to Fig. 4A. Insets show higher magnification of smaller clusters of interest. Intensity of red indicates expression level. (C) Genes differentially expressed (DE) in selected cell clusters between NOD control (top rows) and ICAM-1-/- NOD (bottom rows) derived samples were calculated. Selected DE genes are shown in feature plots corres-ponding to Fig. 4A

Page 22: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

NF1

ISRE

STAT

ICSBP

IRF

IRF7

STAT1

IRF8

IRF1

STAT3

NFKB

IRF3

NFKAPPAB65

NFKAPPAB50

NFKAPPAB

SIRT6

-1500bp -1000bp -500bp 0bp 500bp

Rtp4H2-D1

Ifi27Calm2

Iigp1Psmb10

ApodPlp1

Shisa5Bst2

TapbpTap2Tap1

Oasl2Psmb8H2-K1H2-Q4

Irf1Hsp90b1

Hspa5Gas7B2m

Gbp7Igtp

Irgm2Irgm1

Ifi35H2-T23

Stat1Gbp3

Figure S14: Transcriptional response to autoimmunity is conserved between oligos and mySC and mimicking an IFN-responseDifferentially expressed genes (up- or down-regulated) were identified in the mySC cluster in ICAM-1-/-NOD vs. NOD control mice. Published DE genes of oligos in EAE vs. control CNS were obtained (2). DE gene lists from both sources were tested for overlap (compared with Venn diagram in Fig. 4G). Shared DE genes were submitted to the Interferome database (3) and the plot generated by the database in ‘transcription factor (TF) analysis’ was downloaded. Binding sites of the indicated TF are depicted as colored boxes relative to the transcriptional start site (0 bp) of the indicated genes.

Page 23: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

CD

74_MIF

APP_CD

74SELL_SELPLGG

AS6_AXLSELL_C

D34

CXC

L12_CXC

R4

SPP1_CD

44C

CL11_C

CR

2C

XCL12_A C

KR3

GR

N_TN

FRSF1A

CC

L11_DPP4

CXC

L12_DPP4

PDG

FA_PDG

FRA

a7b1 complex_LAM

C1

a6b4 complex_IG

F1a6b1 com

plex_LAMC

1C

CL2_C

CR

2C

CL7_C

CR

2C

CL24_C

CR

2C

CL8_C

CR

2C

CL7_C

CR

5TN

F_LTBRC

CL8_C

CR

5PTPR

C_M

RC

1TSLPR

_CR

LF2PSAP_G

PR37L1

JAM2_JAM

3FG

FR1_N

CAM

1JAG

1_NO

TCH

2TSLPR

_TSLPC

D48_C

D244

CC

L5_CC

R5

LTB_L TBRD

LL4_NO

TCH

2C

CR

5_CC

L4SEM

A4D_C

D72

LILRB4_LAIR

1EN

TPD1_AD

OR

A2AIN

SL6_NO

TCH

1D

LL4_NO

TCH

1PTPR

C_C

D22

SELL_POD

XLTG

FBR3_TG

FB1PR

OS1_AXL

aVb5 complex_FN

1H

BEGF_C

D44

PTN_PTPR

SC

D8 receptor_LC

KIG

F1R_IG

F1JAG

1_NO

TCH

3PTH

LH_PTH

1RJAG

1_NO

TCH

1TN

F_TNFR

SF1AFAS_TN

FIL34_C

SF1RD

LL4_NO

TCH

3C

SF1_CSF1R

C3_C

3AR1

BC−BCBC−TC(CD8)BC−fibroBC−mySCBC−MCBC−EC1BC−vSMC/PCBC−pDCBC−EC2TC(CD8)−(TC)CD4 TC(CD8)−nmSCTC(CD8)−BCTC(CD8)−TC(CD8)TC(CD8)−fibroTC(CD8)−mySCTC(CD8)−MCTC(CD8)−EC1TC(CD8)−vSMC/PCTC(CD8)−pDCTC(CD8)−EC2fibro−TC(CD4)fibro−nmSCfibro−BCfibro−(TC)CD8fibro−fibrofibro−mySCfibro−MCfibro−EC1fibro−vSMC/PCfibro−pDCfibro−EC2mySC−TC(CD4)mySC−nmSCmySC−BCmySC−TC(CD8)mySC−fibromySC−mySCmySC−MCmySC−EC1mySC−vSMC/PCmySC−pDCmySC−EC2MC−TC(CD4)MC−nmSCMC−BCMC−TC(CD8)MC−fibroMC−mySCMC−MCMC−EC1MC−vSMC/PCMC−pDCMC−EC2EC1−TC(CD4)EC1−nmSCEC1−BCEC1−TC(CD8)EC1−fibroEC1−mySCEC1−MCEC1−EC1EC1−vSMC/PCEC1−pDCEC1−EC2vSMC/PC−TC(CD4)vSMC/PC−nmSCvSMC/PC−BCvSMC/PC−TC(CD8)vSMC/PC−fibrovSMC/PC−mySCvSMC/PC−MCvSMC/PC−EC1vSMC/PC−vSMC/PCvSMC/PC−pDCvSMC/PC−EC2pDC−TC(CD4)pDC−nmSCpDC−BCpDC−TC(CD8)pDC−fibropDC−mySCpDC−MCpDC−EC1pDC−vSMC/PCpDC−pDCpDC−EC2EC2−TC(CD4)EC2−nmSCEC2−BCEC2−TC(CD8)EC2−fibroEC2−mySCEC2−MCEC2−EC1EC2−vSMC/PCEC2−pDCEC2−EC2

0

1

2

3

4

5

6BA

TC(CD8)

MC

EC2

pDC

vSMC/PC

mySC

EC1

fibro

TC (CD4)

nmSC

BC

NOD control

C ICAM-1-/-NOD

mySC

MC

fibro

BC

vSMC/PC

EC1

pDC

TC(CD8)

TC (CD4)

nmSC

EC2

D

CD

74_MIF

APP_CD

74C

XCL12_C

XCR

3C

CL11_C

CR

2C

CL7_C

CR

5C

CL7_C

CR

2C

XCL12_C

XCR

4EG

FR_TG

FB1TG

FBR3_TG

FB1C

CL2_C

CR

2SELL_SELPLGTG

Fbeta receptor1_TGFB1

SELL_CD

34PSAP_G

PR37L1

IFNG

_Type II IFNR

a4b1 comple x_FN

1a4b7 com

plex_FN1

DLL4_N

OTC

H2

a6b4 compl ex_IG

F1a6b1 com

pl ex_LAMC

1PD

GFA_PD

GFR

Aa7b1 com

plex_LAMC

1aVb1 com

plex_VTNaVb1 com

plex_FN1

CD

86_CTLA4

PLAUR

_a4b1 comple x

CXC

L16_CXC

R6

CXC

L9_CXC

R3

CXC

L10_CXC

R3

CC

L8_CC

R5

CC

L8_CC

R2

PECAM

1_CD

38JAM

2_a4b1 complex

TNFR

SF1A_ FASLGG

RN

_TNFR

SF1BPTN

_PTPRS

JAM2_JAM

3a1b1 com

plex_CO

L18A1EFN

A1_EPHA5

a1b1 complex_C

OL27A1

a1b1 complex_C

OL16A1

a1b1 complex_C

OL5A2

a1b1 complex_C

OL5A3

a1b1 complex_C

OL1A1

a1b1 complex_C

OL4A2

a1b1 complex_C

OL4A1

a1b1 comple x_C

OL6A3

a1b1 complex_C

OL1A2

a1b1 complex_C

OL3A1

RAR

RES2_C

MKLR

1TH

Y1_aXb2 complex

ICAM

1_aXb2 complex

ALCAM

_CD

6C

D8 receptor_LC

KC

CL5_C

CR

5aLb2 com

plex_ICAM

1IC

AM1_ITG

ALC

CR

5_CC

L4PTPR

C_C

D22

CD

28_CD

86SEM

A4D_C

D72

LILRB4_LAIR

1aVb5 com

plex_FN1

a11b1 complex_C

OL6A3

a11b1 compl ex_C

OL1A2

a11b1 complex_FN

1a11b1 com

plex_CO

L8A1a11b1 com

plex_CO

L6A1a11b1 com

ple x_CO

L15A1FG

F2_FGFR

1a11b1 com

plex_CO

L6A2a11b1 com

pl ex_CO

L18A1a11b1 com

plex_CO

L5A2a11b1 com

plex_CO

L4A2a11b1 com

plex_CO

L4A1a11b1 com

pl ex_CO

L3A1a11b1 com

plex_CO

L5A3a11b1 com

pl ex_CO

L1A1C

XCL12_D

PP4C

CL11_D

PP4N

RP2_SEM

A3CN

RP2_VEG

FALR

P1_MD

KSELL_PO

DXL

TNF_TN

FRSF1A

CXC

L9_DPP4

CXC

L10_DPP4

TGFbeta receptor1_TG

FB3C

D44_H

GF

EFNA4_EPH

A2SPP1_C

D44

a4b1 comple x_SPP1

aLb2 compl ex_F11R

aLb2 complex_IC

AM2

FASLG_FAS

CLEC

2D_FAM

3CVC

AM1_a4b1 com

plexSEM

A4D_PLXN

B1IL34_C

SF1RH

BEGF_C

D44

a4b7 comple x_VC

AM1

CC

L5_CC

R1

CC

L18_CC

R1

CC

L15_CC

R1

CC

L8_CC

R1

C3_C

3AR1

PDG

FB_PDG

FRB

IGF1R

_IGF1

PDG

FB_PDG

FRA

ANG

PT2_TEKN

RP1_VEG

FAD

LL4_NO

TCH

3a1b1 com

pl ex_CO

L17A1EFN

A1_EPHA2

TNFSF12_TN

FRSF12A

FGFR

1_NC

AM1

EFNA5_EPH

A5a11b1 com

pl ex_CO

L27A1a11b1 com

pl ex_CO

L16A1a11b1 com

plex_CO

L17A1EFN

A5_EPHA2

LRP1_PD

GFB

RSPO

1_LGR

4PD

GFR

complex_PD

GFB

CC

L7_CC

R1

CSF1_C

SF1RO

SMR

_OSM

F AS_TNF

LIFR_O

SMEFN

A4_EPHA5

FCER

2_aXb2 complex

SEMA7A_a1b1 com

ple xC

D96_PVR

HBEG

F_EGFR

TNF_TN

FRSF1B

TGFBR

3_TGFB3

a11b1 complex_C

OL14A1

a11b1 complex_C

OL5A1

a1b1 compl ex_C

OL8A1

FLT1_VEGFA

a1b1 compl ex_C

OL14A1

a1b1 comple x_C

OL5A1

ENTPD

1_ADO

RA2B

a1b1 comple x_C

OL15A1

a1b1 complex_C

OL6A2

a1b1 complex_C

OL6A1

GAS6_AXL

CXC

L12_ACKR

3PR

OS1_AXL

GR

N_TN

FRSF1A

TC(CD4)−TC(CD4)TC(CD4)−nmSCTC(CD4)−BCTC(CD4)−TC(CD)8TC(CD4)−fibroTC(CD4)−mySCTC(CD4)−MCTC(CD4)−EC1TC(CD4)−vSMC/PCTC(CD4)−pDCTC(CD4)−EC2nmSC−TC(CD4)nmSC−nmSCnmSC−BCnmSC−TC(CD8)nmSC−fibronmSC−mySCnmSC−MCnmSC−EC1nmSC−vSMC/PCnmSC−pDCnmSC−EC2BC−TC(CD4)BC−nmSCBC−BCBC−TC(CD8)BC−fibroBC−mySCBC−MCBC−EC1BC−vSMC/PCBC−pDCBC−EC2TC(CD8)−TC(CD4)TC(CD8)−nmSCTC(CD8)−BCTC(CD8)−TC(CD8)TC(CD8)−fibroTC(CD8)−mySCTC(CD8)−MCTC(CD8)−EC1TC(CD8)−vSMC/PCTC(CD8)−pDCTC(CD8)−EC2fibro−TC(CD4)fibro−nmSCfibro−BCfibro−TC(CD8)fibro−fibrofibro−mySCfibro−MCfibro−EC1fibro−vSMC/PCfibro−pDCfibro−EC2mySC−TC(CD4)mySC−nmSCmySC−BCmySC−TC(CD8)mySC−fibromySC−mySCmySC−MCmySC−EC1mySC−vSMC/PCmySC−pDCmySC−EC2MC−TC(CD4)MC−nmSCMC−BCMC−TC(CD8)MC−fibroMC−mySCMC−MCMC−EC1MC−vMC/PCMC−pDCMC−EC2EC1−TC(CD4)EC1−nmSCEC1−BCEC1−TC(CD8)EC1−fibroEC1−mySCEC1−MCEC1−EC1EC1−vSMC/PCEC1−pDCEC1−EC2vSMC/PC−TC(CD4)vSMC/PC−nmSCvSMC/PC−BCvSMC/PC−TC(CD8)vSMC/PC−fibrovSMC/PC−mySCvSMC/PC−MCvSMC/PC−EC1vSMC/PC−vSMC/PCvSMC/PC−pDCvSMC/PC−EC2pDC−TC(CD4)pDC−nmSCpDC−BCpDC−TC(CD8)pDC−fibropDC−mySCpDC−MCpDC−EC1pDC−vSMC/PCpDC−pDCpDC−EC2EC2−TC(CD4)EC2−nmSCEC2−BCEC2−TC(CD8)EC2−fibroEC2−mySCEC2−MCEC2−EC1EC2−vSMC/PCEC2−pDCEC2−EC2

0

1

2

3

4

5

6

Figure S15: Comparative cell-cell interactomes in healthy and pre-neuropathic mice. (A, C) Cellular interaction networks predicted from genes expressed in the peripheral nervous system of NOD control (A) and ICAM-1-/-NOD mice. Nodes (circles) repre-sent cell clusters and node size correlates with the relative cell count. Significant cell-cell interacti-ons were predicted by CellP-honeDB and represent edges (arrows) in the network. The width and transparency of the edges correlate with the amount of interactions; arrows indicate the directionality of ligand/receptor interactions. Layout was set to compound spring embedder. (B, D) Heatmaps depicting the color-coded negative natural logarithm of the p-value (Methods) of the signifi-cant cell-cell interactions of NOD control (B) and ICAM-1-/-NOD (D) samples. Only cell-cell interactions are shown that show at least one significant interaction. mySC: myelinating Schwann cells, nmSC: non-myelinating Schwann cells, fibro: fibroblasts, vSMC/PC: vascu-lar smooth muscle cells and pericy-tes, EC1: endothelial cells cluster 1, EC2: endothelial cells cluster2, BC: B cells, CD4: CD4 T helper cells, CD8: cytotoxic CD8 T cells and natural killer cells, MC: myeloid lineage cells, pDC: plasmacytoid dendritic cells.

Page 24: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary Table Legends

Supplementary Table Legends

Table S1: A summary of technical information regarding samples and sequencing.

STDEV: Standard deviation; SEM: Standard error of the mean

Table S2: Marker genes of cluster in naive mice (corresponding to Fig. 1A)

avg_log_FC: log fold change of the average expression between the cluster vs. all remaining clusters;

pct.1: percentage of cells with the gene detected in the cluster; pct.2: percentage of cells with the gene

detected in all remaining clusters; p_val_adj: adjusted p values (based on Bonferroni correction). Average

log FC threshold was set to 0.25.

Table S3: Marker genes of mySC and nmSC in naive mice with lowered threshold

Threshold was lowered to detect panSC markers in mySC and nmSC, genes of interest are highlighted in

yellow. avg_log_FC: log fold change of the average expression between the cluster vs. all remaining

clusters; pct.1: percentage of cells with the gene detected in the cluster; pct.2: percentage of cells with

the gene detected in all remaining clusters; p_val_adj: adjusted p values (based on Bonferroni correction).

Average log FC threshold was set to 0.05.

Table S4: Curated list of transcripts not previously identified in myelinating Schwann cells.

avg_log_FC: log fold change of the average expression between the mySC cluster vs. all remaining

clusters; pct.1: percentage of cells with the gene detected in the mySC cluster; pct.2: percentage of cells

with the gene detected in all remaining clusters; p_val_adj: adjusted p values (based on Bonferroni

correction).Threshold was set on 0.25 avg logFC; PNS-SC/myelination: these genes have been previously

described to be expressed by Schwann cells of the PNS or to be associated with PNS myelination. CNS

glia cells/myelination: these genes have been previously described to be expressed by glial cells of the

CNS or to be associated with CNS myelination. SC: Schwann cell, PNS: peripheral nervous system, CNS:

central nervous system, OL: oligodendrocyte, OPC: oligodendrocyte precursor cell, DRG: dorsal root

ganglion

Table S5: GSEA results of marker genes in the mySC cluster in naive mice

p_val_adj: FDR-adjusted p values

Table S6: GSEA results of marker genes in the nmSC cluster in naive mice

p_val_adj: FDR-adjusted p values

Table S7: GSEA results of marker genes in the fibro cluster in naive mice

p_val_adj: FDR-adjusted p values

Page 25: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary Table Legends

Table S8: Marker genes of cell clusters in rat samples (corresponding to Fig. 3F)

avg_log_FC: log fold change of the average expression between the cluster vs. all remaining clusters;

pct.1: percentage of cells with the gene detected in the cluster; pct.2: percentage of cells with the gene

detected in all remaining clusters; p_val_adj: adjusted p values (based on Bonferroni correction),

Threshold for average logFC was set to 0.25.

Table S9: Marker genes of cell clusters in ICAM-1-/-

NOD and NOD control mice (corresponding to Fig. 4A)

avg_log_FC: log fold change of the average expression between the cluster vs. all remaining clusters;

pct.1: percentage of cells with the gene detected in the cluster; pct.2: percentage of cells with the gene

detected in all remaining clusters; p_val_adj: adjusted p values (based on Bonferroni correction).

Threshold for average logFC was set at 0.25.

Table S10: Differentially expressed genes in the EC cluster in ICAM-1-/-

NOD vs. NOD control mice

avg_log_FC: log fold change of the average expression between ICAM-1-/-

NOD and NOD control mice

(positive values indicate a higher gene expression in ICAM-1-/-

NOD mice); pct.1: percentage of cells with

the gene detected in ICAM-1-/-

NOD mice; pct.2: percentage of cells with the gene detected in NOD control

mice; p_val_adj: adjusted p values (based on Bonferroni correction). Threshold for average logFC was set

to 0.25.

Table S11: Differentially expressed genes in the nmSC cluster in ICAM-1-/-

NOD vs. NOD control mice

avg_log_FC: log fold change of the average expression between ICAM-1-/-

NOD and NOD control mice

(positive values indicate a higher gene expression in ICAM-1-/-

NOD mice) ; pct.1: percentage of cells with

the gene detected in ICAM-1-/-

NOD mice; pct.2: percentage of cells with the gene detected in NOD control

mice; p_val_adj: adjusted p values (based on Bonferroni correction). Threshold for average logFC was set

at 0.25.

Table S12: Differentially expressed genes in the mySC cluster in ICAM-1-/-

NOD vs. NOD control mice

avg_log_FC: log fold change of the average expression between ICAM-1-/-

NOD and NOD control mice

(positive values indicate a higher gene expression in ICAM-1-/-

NOD mice) ; pct.1: percentage of cells with

the gene detected in ICAM-1-/-

NOD mice; pct.2: percentage of cells with the gene detected in NOD control

mice; p_val_adj: adjusted p values (based on Bonferroni correction). Threshold for average logFC was set

at 0.25.

Page 26: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary Table Legends

Table S13: Differentially expressed genes in the oligo cluster in naive vs. EAE mice (2), in the mySC

cluster in ICAM-1-/-

NOD vs. NOD control mice and the overlapping genes

Top DE genes (up- and down-regulated) were identified in the mySC cluster in ICAM-1-/-NOD vs. NOD

nerves. Published genes DE in oligo in EAE vs. control CNS (2) were obtained. DE gene lists from both

sources were tested for overlap (Venn diagram, Fig. 4G). Shared DE genes were submitted to the

Interferome database and the plot generated in TF analysis was downloaded. Binding sites of the

indicated TF are depicted as colored boxes relative to the transcriptional start site (0 bp) of the indicated

genes.

Table S14: Network characteristics in NOD control and ICAM-1-/-

NOD mice (corresponding to Fig. S9A

and Fig. S9C).

Page 27: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary References

Supplementary References

1. E. M. Steinert, et al., Quantifying Memory CD8 T Cells Reveals Regionalization of

Immunosurveillance. Cell 161, 737–749 (2015).

2. A. M. Falcão, et al., Disease-specific oligodendrocyte lineage cells arise in multiple sclerosis. Nat.

Med. 24, 1837–1844 (2018).

3. I. Rusinova, et al., Interferome v2.0: an updated database of annotated interferon-regulated genes.

Nucleic Acids Res. 41, D1040–6 (2013).

4. S. Martin, et al., Dominant role of intercellular adhesion molecule-1 in the pathogenesis of

autoimmune diabetes in non-obese diabetic mice. J. Autoimmun. 17, 109–117 (2001).

5. G. Meyer zu Horste, et al., Thymic epithelium determines a spontaneous chronic neuritis in

Icam1(tm1Jcgr)NOD mice. J. Immunol. 193, 2678–2690 (2014).

6. S. Jung, et al., Analysis of fractalkine receptor CX(3)CR1 function by targeted deletion and green

fluorescent protein reporter gene insertion. Mol. Cell. Biol. 20, 4106–4114 (2000).

7. T. G. Hamilton, R. A. Klinghoffer, P. D. Corrin, P. Soriano, Evolutionary divergence of platelet-derived

growth factor alpha receptor signaling mechanisms. Mol. Cell. Biol. 23, 4013–4025 (2003).

8. L. Zhuo, et al., Live astrocytes visualized by green fluorescent protein in transgenic mice. Dev. Biol.

187, 36–42 (1997).

9. C. Benz, V. C. Martins, F. Radtke, C. C. Bleul, The stream of precursors that colonizes the thymus

proceeds selectively through the early T lineage precursor stage of T cell development. J. Exp. Med.

205, 1187–1199 (2008).

10. M. D. Muzumdar, B. Tasic, K. Miyamichi, L. Li, L. Luo, A global double-fluorescent Cre reporter

mouse. Genesis 45, 593–605 (2007).

11. S. W. Boyer, A. V. Schroeder, S. Smith-Berdan, E. C. Forsberg, All hematopoietic cells develop from

hematopoietic stem cells through Flk2/Flt3-positive progenitor cells. Cell Stem Cell 9, 64–73 (2011).

12. A. E. Beaudin, et al., A Transient Developmental Hematopoietic Stem Cell Gives Rise to Innate-like B

and T Cells. Cell Stem Cell 19, 768–783 (2016).

13. E. Gomez Perdiguero, et al., Tissue-resident macrophages originate from yolk-sac-derived erythro-

myeloid progenitors. Nature 518, 547–551 (2015).

14. N. D. Andersen, S. Srinivas, G. Piñero, P. V. Monje, A rapid and versatile method for the isolation,

purification and cryogenic storage of Schwann cells from adult rodent nerves. Sci. Rep. 6, 31781

(2016).

Page 28: Redefining the heterogeneity of peripheral nerve cells in ......Supplementary Information for Redefining the heterogeneity of peripheral nerve cells in health and autoimmunity Jolien

Supplementary References

15. M. P. Clements, et al., The Wound Microenvironment Reprograms Schwann Cells to Invasive

Mesenchymal-like Cells to Drive Peripheral Nerve Regeneration. Neuron 96, 98–114.e7 (2017).

16. M. Adam, A. S. Potter, S. S. Potter, Psychrophilic proteases dramatically reduce single-cell RNA-seq

artifacts: a molecular atlas of kidney development. Development 144, 3625–3632 (2017).

17. A. K. Mausberg, et al., Trapped in the epineurium: early entry into the endoneurium is restricted to

neuritogenic T cells in experimental autoimmune neuritis. J. Neuroinflammation 15, 217 (2018).

18. A. Butler, P. Hoffman, P. Smibert, E. Papalexi, R. Satija, Integrating single-cell transcriptomic data

across different conditions, technologies, and species. Nat. Biotechnol. 36, 411–420 (2018).

19. C. Hafemeister, R. Satija, Normalization and variance stabilization of single-cell RNA-seq data using

regularized negative binomial regression. bioRxiv, 576827 (2019).

20. I. Korsunsky, et al., Fast, sensitive and accurate integration of single-cell data with Harmony. Nat.

Methods 16, 1289–1296 (2019).

21. R. Vento-Tormo, et al., Single-cell reconstruction of the early maternal–fetal interface in humans.

Nature 563, 347–353 (2018).

22. S. Durinck, P. T. Spellman, E. Birney, W. Huber, Mapping identifiers for the integration of genomic

datasets with the R/Bioconductor package biomaRt. Nat. Protoc. 4, 1184–1191 (2009).

23. M. V. Kuleshov, et al., Enrichr: a comprehensive gene set enrichment analysis web server 2016

update. Nucleic Acids Res. 44, W90–7 (2016).

24. E. Y. Chen, et al., Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool.

BMC Bioinformatics 14, 128 (2013).

25. G. Finak, et al., MAST: a flexible statistical framework for assessing transcriptional changes and

characterizing heterogeneity in single-cell RNA sequencing data. Genome Biol. 16, 278 (2015).

26. H. Chen, P. C. Boutros, VennDiagram: a package for the generation of highly-customizable Venn and

Euler diagrams in R. BMC Bioinformatics 12, 35 (2011).