Thymic Mesenchymal Cells Have a Distinct Transcriptomic Profile

Home , FGF10, Latrophilin, Somatomedin

Thymic Mesenchymal Cells Have a Distinct Transcriptomic Profile Julien Patenaude and Claude Perreault This information is current as J Immunol 2016; 196:4760-4770; Prepublished online 29 of October 1, 2021. April 2016; doi: 10.4049/jimmunol.1502499 http://www.jimmunol.org/content/196/11/4760 Downloaded from Supplementary http://www.jimmunol.org/content/suppl/2016/04/29/jimmunol.150249 Material 9.DCSupplemental References This article cites 65 articles, 18 of which you can access for free at: http://www.jimmunol.org/content/196/11/4760.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on October 1, 2021

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2016 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Thymic Mesenchymal Cells Have a Distinct Transcriptomic Proﬁle

Julien Patenaude and Claude Perreault

In order to understand the role of mesenchymal cells (MCs) in the adult thymus, we performed whole transcriptome analyses of primary thymic, bone, and skin MCs. These three MC populations shared expression of 2850 core MC genes involved in generic processes including interactions with tissue-resident macrophages. Moreover, we discovered that 2036 genes were differentially expressed, by at least 5-fold, in the three MC populations. Genes preferentially expressed in thymic MCs are instrumental in clearance of apoptotic thymocytes by macrophages, maintenance of a noninﬂammatory milieu, and attraction-expansion of thymocyte progenitors. Thymic and bone MCs share other sets of differentially expressed genes implicated in resolution of inﬂammation and expansion of hematolymphoid progenitors. Consistent with the fact that thymic and skin MCs have to support epithelial cells, they express at higher levels genes mediating epithelial cell adhesion to basement membrane and mesenchymal–epithelial cross-talk. Differentially expressed genes preferentially expressed by bone MCs are connected to formation and remodeling of bone, whereas those prefer- Downloaded from entially expressed in skin MCs are involved in skin and hair follicle homeostasis. We conclude that MCs from different organs display substantial heterogeneity and that the transcriptome of thymic MCs is exquisitely suited for interactions with epithelial and hematolymphoid cells in an environment with a high apoptosis rate. The Journal of Immunology, 2016, 196: 4760–4770.

esenchymal cells (MCs) derive mainly from the me- MCs. During development, thymic MCs regulate the proliferation

soderm and, to a lesser extent, from the neural crest of thymic epithelial cells through production of fibroblast growth http://www.jimmunol.org/ M (ectoderm). In the adult organism, they are found in all factor (FGF)-7 and -10, insulin-like growth factor (IGF)-1 and -2, tissues and organs, where they play mechanical and metabolic and retinoic acid (7–11). However, little is known of the role of roles. MCs are the major producers of extracellular matrix, which thymic MCs during postnatal life, aside from the fact that CD248+ provides a structural framework, a scaffold for cell migration, and MCs play a role in revascularizing thymuses during infection- serves as a reservoir for cytokines and growth factors used by dependent regeneration and that FSP1+ MCs are essential for epithelial and hematopoietic parenchymal cells (1). Bone marrow the maintenance of the medullary thymic epithelium (12, 13). MCs have attracted extraordinary attention because they are key In order to gain insights into the function of thymic MCs, we components of the hematopoietic stem cell niche, and they contain elected to compare their transcriptome to that of bone and skin mesenchymal stem and progenitor cells that display an amazing MCs. Two reasons led us to select bone and skin MCs for this study. by guest on October 1, 2021 ability to regulate immune responses and coordinate tissue re- First, we included bone MCs as a reference because they have been generation (2–6). However, because extramedullary MCs are seen studied most extensively and represent by far the best characterized as supporting players, their role in normal adult physiology has MC population. Second, we reasoned that the function of MCs received much less attention than that of parenchymal (e.g., epi- might be influenced by the nature of their neighboring parenchymal thelial or hematopoietic) cells. This is notably the case for thymic cells: thymic and skin MCs have to support epithelial cells, whereas thymic and bone MCs have to support hematolymphoid cells. The Institute for Research in Immunology and Cancer, University of Montreal, Montreal, transcriptome is a critical component of systems-level under- Quebec H3C 3J7, Canada; and Department of Medicine, University of Montreal, standing of cell biology, and it can be reliably tackled in its entirety Montreal, Quebec H3C 3J7, Canada using relatively modest cell numbers (14). In the last decade, ORCID: 0000-0001-9453-7383 (C.P.). studies by the Immunological Genome Project (http://www.immgen. Received for publication November 30, 2015. Accepted for publication March 28, org/) have illustrated the relevance of transcriptome analyses by 2016. unraveling large gene networks in primary myeloid, lymphoid, and This work was supported by Grant MOP 42384 from the Canadian Institute of Health Research. The Institute for Research in Immunology and Cancer is supported in part stromal cells from various hematolymphoid organs (15). Never- by the Canada Foundation for Innovation and the Fonds de la Recherche en Sante´ du theless, several pioneer studies addressing the question of MCs’ Que´bec. transcriptomic heterogeneity are fraught with limitations because The RNA-sequencing data presented in this article have been submitted to the Gene they were based on microarray analyses performed on in vitro– Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) under accession numbers GSE73175 and GSE75598. cultured MCs (16, 17). This approach is indeed problematic be- Address correspondence and reprint requests to Prof. Claude Perreault, Institute for cause plastic adherence and in vitro culture induce major changes Research in Immunology and Cancer, P.O. Box 6128, Station Centre-Ville, Montreal, in the phenotype and biology of MCs (6). Furthermore, relative to QC H3C 3J7, Canada. E-mail address: [email protected] RNA-sequencing (RNA-seq), microarrays display lower sensitiv- The online version of this article contains supplemental material. ity and dynamic range coupled to higher technical variations (18, Abbreviations used in this article: 7-AAD, 7-aminoactinomycin D; DEG, differen- 19). We therefore elected to explore the question of MC hetero- tially expressed gene; ETP, early thymocyte progenitor; FC, fold change; FGF, fibroblast growth factor; GEO, Gene Expression Omnibus; HSC, hematopoietic stem cell; geneity by studying the transcriptome of primary freshly harvested IGF, insulin-like growth factor; IPA, Ingenuity Pathway Analysis; Lin, lineage; MC, MCs using RNA-seq. mesenchymal cell; RNA-seq, RNA-sequencing; RPKM, reads per kilobase of exon We identified three main gene sets in MCs: 6270 housekeeping per million mapped reads; TEC, thymic epithelial cell. genes, 2850 core MC genes, and 2036 differentially expressed Copyright Ó 2016 by The American Association of Immunologists, Inc. 0022-1767/16/$30.00 genes (DEGs) that were preferentially expressed in one or two MC www.jimmunol.org/cgi/doi/10.4049/jimmunol.1502499 The Journal of Immunology 4761 populations. Core MC genes are involved in generic MC functions, Sca-1+ cells. For intracellular staining, cell viability was assessed using the whereas DEGs are connected to organ-specific features such as Live/Dead fixable blue dead cell stain kit (Invitrogen). Cells were fixed/ clearance of apoptotic cells (thymic MCs), osteoclastogenesis permeabilized with the Cytofix/Cytoperm Plus kit with GolgiPlug (BD Biosciences) and stained with PE anti-CXCL12 Ab (R&D Systems) or PE (bone MCs), and hair follicle homeostasis (skin MCs). Our work anti-Nestin Ab (R&D Systems). MCs were sorted on a three-laser FACSAria provides a systems-level representation of MC heterogeneity and a (BD Biosciences) or analyzed on a three-laser LSR II (BD Biosciences) global framework to explain how MCs may adapt to organ-specific using FACSDiva software (BD Biosciences), as described (24). functions. RNA extraction and high-throughput RNA-seq RNA extraction and RNA-seq were performed as described (25, 26). Output Materials and Methods data were mapped to the Mus muculus (mm10) reference genome using Mice ELANDv2 alignment tool from the CASAVA 1.8.2 software, and transcript levels were expressed as reads per kilobase of exon per million mapped C57BL/6 mice purchased from The Jackson Laboratory (Bar Harbor, ME) reads (RPKM) (27). Mesenchymal cell RNA-seq data are accessible via were bred and housed under specific pathogen-free conditions at the In- Gene Expression Omnibus (GEO) archives under accession number stitute for Research in Immunology and Cancer and used at 3 to 4 wk of age. GSE73175 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE73175). A male/female ratio of 0.6 (three males for five female mice) was kept in all The GEO accession numbers for thymic epithelial cell (TEC) and thymocyte experiments. All procedures were in accordance with the Canadian Council RNA-seq data are GSE66873 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? on Animal Care guidelines and approved by the Comite´ de Deóntologie et acc=GSE66873) and GSE75598 (http://www.ncbi.nlm.nih.gov/geo/query/acc. Expe´rimentation Animale de l’Universite´ de Montreál. cgi?token=ydmlueiajdybveh&acc=GSE75598), respectively. In heat maps, MC isolation gene expression in TECs is presented as the mean of RPKM values in cortical and medullary TECs. To correlate gene expression in MCs Thymus. Thymic stromal cells were enriched as previously described (20).

with that of hematolymphoid stem-progenitor cells, we downloaded RNA- Downloaded from In brief, three to five whole thymi were mechanically disrupted and en- seq data from Lara-Astiaso et al. (GEO accession number GSE60101; http:// 3 zymatically digested at 37˚C for 3 15 min using 0.01% (w/v) Liberase www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE60101) (28), and all data- (Roche) and 0.1% (w/v) DNase I (Sigma-Aldrich). Final digests were sets were remapped to the mm10 reference genome using TopHat version 2.1.0. harvested, pooled, and maintained at 4˚C in FACS buffer (PBS, 0.1% [w/v] BSA, and 0.02% [w/v] NaN3) prior to cell staining for flow cytometric Statistical and bioinformatics analyses analysis and cell sorting. Statistical and bioinformatics analyses were performed using R version Bone. Femurs, tibiae, and pelvis of three to five mice were aseptically 3.1.3. Sample hierarchical clustering was performed using the R package dissected, and endosteal stromal cell enrichment was performed as de- http://www.jimmunol.org/ pvclust. Significant clusters were identified using the multiscaled bootstrap scribed by Houlihan et al. (21). In brief, bones were collected, cleaned, and resampling method included in the pvclust package. To assess differential washed three times in cold PBS. Using sharp surgical scissors, each bone expression between samples, we used the DESeq2 package for R (29). was first cut longitudinally and then transversely to generate tiny fragments Resulting p values were adjusted for multiple testing with the Benjamin- of ∼1mm2. Bone fragments were then washed three times in cold PBS and Hochberg procedure (p-adj values). Segregation of DEGs was performed incubated with agitation (80 rpm) for 3 3 20 min at 37˚C in medium in two steps. First, gene expression (spiked with 1 RPKM) was normalized containing Liberase and DNase I. Following each incubation, supernatants as Z-score: were added to HBSS medium (Invitrogen) supplemented with 2% (v/v) FBS, 10 mmol HEPES, and 1% (v/v) penicillin-streptomycin. To collect ðx þ 1Þ m additional endosteal stromal cells, remaining bone fragments were gently Z ¼ crushed in supplemented HBSS medium using a pestle and mortar (5 3 50 s by guest on October 1, 2021 gentle taps), and supernatants were pooled to previous ones. RBC lysis was Second, K-means clustering using the Hartigan and Wong algorithm was then performed on the resulting cell suspension prior to its filtration and performed using tabulated expression Z-scores as input. Biological function resuspension in FACS buffer. enrichment analyses were all performed with the Web-based application Skin. Skin MC isolation protocol was adapted from Hakkinen et al. (22) and Ingenuity Pathway Analysis (IPA) using all genes of the mouse genome Crigler et al. (23) to specifically retrieve dermal MCs. In brief, mouse as background. Biological functions with p value ,0.05 and activation trunk and dorsal skin (∼12 cm2/mouse) was aseptically dissected and in- Z-score .2 were considered significantly enriched and activated, respectively. cubated for 45 min in 0.01% (w/v) Liberase. The dermis was then To retrieve the core biological functions associated to DEG clusters, we mechanically isolated and incubated with agitation (80 rpm) at 37˚C for used a four-step approach. First, each DEG clusters was sorted according to a 2 3 30 min in medium containing Liberase and DNase I. Postincubation, relevance score that takes into account the expression level and expression supernatants were filtered and pooled in 25 ml PBS supplemented with 1% specificity. Then, sorted DEG clusters were divided in bins containing 60 FBS and 5 mmol EDTA. The final cell suspension was then centrifuged DEGs. Third, IPA analyses were performed, each time analyzing an incre- and resuspended in FACS buffer. mental number of bins per DEG cluster (first IPA analysis = bin 1 [top 60 DEGs], second IPA analysis = bin 1 and 2 [top 120 DEGs], etc.). Finally, Flow cytometry and cell sorting activated and enriched biological functions were retrieved from each IPA analysis output, and for each DEG cluster, only the redundant functions found All stromal cell suspensions were stained with the following lineage (Lin) in 100% of its IPA analyses were considered as core biological functions. Ab mixture: A700 anti-CD45 (BD Biosciences; clone 30-F11), A700 anti- Genes associated to the extracellular region were selected based on the TER119 (BioLegend; clone TER119), A700 anti-CD11b (BD Biosciences; functional annotation GO:0005576 of DAVID online bioinformatic tool clone M1/70), A700 anti-CD4 (BD Biosciences; clone RM4-5), A700 anti- (30). Unless stated otherwise, data are represented as mean 6 SD, and B220 (BD Biosciences; clone RA3-6B2), A700 anti–GR-1 (BD Biosci- statistical significance was assessed by a two-tailed paired Student t test. ences; clone RB6-8C5), PE-Cy7 anti-CD41 (BioLegend; clone MWReg30), Differences with p , 0.05 were considered significant. PE-Cy7 anti-CD2 (BioLegend; clone RM2-5), PE-Cy7 anti-CD31 (Bio- The independent dataset containing genes predicted to support hema- Legend; clone 390), and allophycocyanin-Cy7 anti-CD326 (BioLegend; topoietic stem cells (HSCs) and identified as core genes operative in site of clone G8.8). The following Abs were used to characterize the phenotype hematopoiesis originated from Charbord et al. (31). The original list of 481 of MCs: allophycocyanin anti–Sca-1 (eBioscience; clone D7), Biotin genes was filtered to retrieve only the 309 genes upregulated in HSC- anti-CD29 (BioLegend; clone HM-61-1), Biotin anti-CD105 (eBioscience; supportive mesenchymal cell lines compared with less-supportive mesen- clone MJ7/18), Biotin anti-CD140a (eBioscience; clone APA5), Biotin chymal cell lines (Supplemental Table I). anti-CD140b (eBioscience; clone APB5), PE anti-CD34 (BD Biosciences; clone RAM34), PE anti-CD44 (BD Biosciences; clone IM7), PE anti- CD51 (BD Biosciences; clone RMV-7), PE anti-CD73 (BioLegend; Results clone TY/11.8), PE anti-CD90.2 (BD Biosciences; clone 30-H12), PE anti- 2 + CD146 (BioLegend; clone ME-9F1), PE anti-Ly51 (BD Biosciences; clone Lin Sca-1 cells display canonical MC markers BP-1), PE-Cy7 anti-gp38 (BioLegend; clone 8.1.1), FITC anti-CD271 MCs were purified from freshly harvested skin, thymi, and bones (Abcam; clone MLR2), and PE-conjugated streptavidin (BD Biosci- ences). All cells were incubated at 4˚C in FACS buffer to avoid unspecific (Fig. 1A). Because primary MCs do not express markers of he- staining and stained with 7-aminoactinomycin D (7-AAD; BD Biosci- matological, epithelial, or endothelial lineage (32, 33), our gating ences) to assess cell viability. Living MCs were selected as Lin27-AAD2 strategy first hinged on elimination of leukocytes (CD45+,CD11b+, 4762 THE TRANSCRIPTOME OF PRIMARY MESENCHYMAL CELLS

B220+,CD2+,CD4+,andGr-1+), megakaryocytes (CD41+), ery- we pooled FACS-purified primary cells from 3–13 mice. The throid cells (Ter-119+), endothelial cells (CD31+), epithelial cells purity of individual replicates was verified by postsort flow cy- (epithelial cell adhesion molecule+), and dead cells (7-AAD+) tometry analysis and RNA-seq analysis of cell lineage-restricted (Fig. 1A, 1B). Two subsets of Lin2 cells were then sorted as a transcripts (Supplemental Fig. 2). Hierarchical clustering based on function of their cell-surface expression of Sca-1 (also known as global gene expression levels revealed that the three MC pop- Ly6a). Sca-1 is an 18-kDa mouse GPI-anchored cell-surface protein, ulations had distinct transcriptomic profiles (Fig. 3A). To identify whichisexpressedathighlevelsbyMCsformingtheHSCnichein DEGs, we used stringent criteria: 1) expression .1 RPKM in at bone (31). In the thymus, ∼91% of Lin2 cells are Sca-1+ (Fig. 1B). least one organ; 2) expression fold change (FC) .5; and 3) ad- In bone and skin, Sca-1+ cells represent, respectively, 5 and 79% of justed p value ,0.1 (based on DESeq2 analysis). Each one-to- Lin2 cells (data not shown). We next performed flow cytometry one comparison unveiled from 999 to 1214 DEGs (Fig. 3B, analyses on the Sca-1+ and Sca-12 subsets of Lin2 cells using 14 Supplemental Fig. 2C). In toto, we identified 2036 nonredundant markers for which a presence hasbeenreportedonMCs.Three DEGs that we then segregated, using k-means clustering, into six canonical MC markers were expressed on .97% of Lin2Sca-1+ cells clusters according to their specific gene expression pattern across from the thymus, bone, and skin: CD29, CD51, and CD140a (also MC populations (Fig 3C). Most DEGs were expressed at high known as PDGFRa) (Fig. 2). However, these markers were present levels in a single MC population (72% or 1470 DEGs; clusters on lower percentages of Lin2Sca-12 cells (#50% in the thymus; #1–3), whereas other DEGs were expressed at high levels in two Supplemental Fig. 1). We conclude that Lin2Sca-1+ cells from the MC populations (28% or 556 DEGs; clusters #4–6) (Fig. 3C, 3D). thymus, bone, and skin are genuine MCs, whereas Lin2Sca-12 cells IPA analyses of DEGs preferentially expressed in a single MC represent a heterogeneous cell population that contains some MCs population Downloaded from admixed with cells of uncertain lineage. We therefore selected Lin2Sca-1+ cells for further analyses and hereafter refer to these We first used IPA to identify the core biological functions asso- cells as MCs. The first evidence of interorgan MC heterogeneity ciated to each of the six DEG clusters. Our initial analysis revealed was given by the fact that several cell-surface molecules (e.g., that even though each DEG cluster was composed of a unique set of gp38, Ly51, CD73, and nestin) were expressed by uneven per- genes, five out of six DEG clusters were significantly enriched for centages of MCs in the three different organs (Fig. 2). genes associated to leukocyte migration and in particular to cell

movement of phagocytes (Supplemental Fig. 3). Because we http://www.jimmunol.org/ Differential gene expression in thymic, bone, and skin MCs postulated that the most biologically important MC DEGs should For each organ, we performed RNA-seq analysis on three individual be those involved in cell–cell interactions, we focused subsequent replicates of MCs (Lin2Sca-1+ cells). To create each replicate, analyses on genes coding for proteins that are secreted or located by guest on October 1, 2021

FIGURE 1. Isolation of Lin2Sca-1+ cells from the thymus, bone, and skin. (A) Isolation and analysis workﬂow for the three Lin2Sca-1+ cell populations. (B) Flow cytometry analysis showing the gating strategy used to isolate stromal Lin2Sca-1+ cells from total living cells (7-AADneg). Scatterplots were generated from thymic cells and are representative of three independent experiments (three to ﬁve mice per experiment). The same gating strategy and Ab mixture were used for isolation of Lin2Sca-1+ cells from the thymus, bone, and skin. EpCAM, epithelial cell adhesion molecule; FSC-A, forward light scatter-area. The Journal of Immunology 4763 Downloaded from

FIGURE 2. Phenotypic analysis of Lin2Sca-1+ cell populations. Flow cytometry analysis of 14 cell-surface and intracellular molecules on Lin2 Sca-1+ cells from the thymus, bone, and skin. Overlay histograms illustrate staining with the relevant Ab (in blue) and an isotype control (in http://www.jimmunol.org/ red). Each overlay histogram is representative of three independent experiments (three to ﬁve mice per biological replicate). Numbers represent the mean percentage of positive cells (6 SD). by guest on October 1, 2021

at the cell-surface membrane. We used the extracellular region clearance (efferocytosis; C3, Mfge8, and Apoe) (34–37). Upreg- annotation of DAVID to identify these genes. After this filtering ulation of genes involved in macrophage chemotaxis and effero- step, the functional annotation leukocyte migration was found to cytosis is consistent with the notion that the thymic macrophages be significantly enriched in all six DEG clusters (Fig. 4). have to clear enormous amounts of apoptotic thymocytes that do In order to sort out the most important extracellular region– not survive positive or negative selection (38). Engulfment of associated DEGs, we ranked DEGs in each cluster according to a apoptotic cells leads to alternative or M2 polarization of macro- relevance score based on two elements: gene expression level and phages. M2 macrophages promote inflammation resolution in part FC among the three organs. DEGs at the top of the hierarchy via their increased expression of anti-inflammatory cytokines were expressed at high levels and showed large interorgan FC. In (e.g., IL-10). In line with this, we noted that thymic MCs exhibit the top 40 genes of cluster #1, formed by DEGs preferentially higher expression of IL-34– and CD39-coding genes (Il34 and expressed in thymic MCs, we found 14 genes associated to leu- Entpd1), two molecules that promote M2 polarization of macro- kocyte migration (Fig. 5A). Out of these 14 genes, 7 are instru- phages (Fig. 5A) (39, 40). Besides, because thymic MCs prefer- mental in monocyte–macrophage chemotaxis (e.g., Cx3cl1, Il33, entially colocalize with blood vessels, including venules at the and Pla2g7) and 3 enhance macrophage-dependent apoptotic cell corticomedullary junction (41, 42), higher expression of Ccl19, 4764 THE TRANSCRIPTOME OF PRIMARY MESENCHYMAL CELLS Downloaded from http://www.jimmunol.org/

FIGURE 3. Differential gene expression in thymic (tMC), bone (bMC), and skin MCs (sMC). (A) Hierarchical clustering of gene expression levels based on correlation distances across MC populations and their respective biological replicates. Asterisks indicate significant clusters measured by a multiscale bootstrap resampling method. *p , 0.05. (B) Schematic representation (from top to bottom) of the analysis workflow used to identify DEGs. Numbers for each one-to-one comparison represent the amount of DEGs retrieved using DESeq2 software and the given thresholds. (C) Heat map of the 2036 DEGs categorized by k-means clustering according to their normalized gene expression (Z-score) across MC populations. Numbers to the left delimit the six DEG clusters, and numbers to the right depict the amount of genes in each cluster. (D) MC gene analysis workflow. White boxes represent genes that were not by guest on October 1, 2021 further analyzed. Blue boxes represent categories of genes that were included in this study.

Flt3l,andTslp was a noteworthy feature of thymic MCs (Fig. 5A). IPA analyses of DEGs preferentially expressed in two MC Indeed, these three genes code for molecules implicated in the ex- populations travasation of thymus-seeding progenitors across corticomedullary Alike for clusters #1–3, a recurrent theme with clusters #4–6 was the junction venules and the expansion of early thymocyte progenitors presence of diverse sets of DEGs involved in leukocyte (particularly (ETP) (43–45). From the analysis of cluster #1, we therefore con- macrophage) migration (Fig. 6). In addition, we detected two groups clude that thymic MCs show higher expression of genes instrumental of genes in cluster #4 (DEGs preferentially expressed in thymic and in clearance of apoptotic thymocytes by macrophages, maintenance bone MCs): 1) genes coding for proteins regulating hematopoietic of a noninflammatory milieu, and attraction-expansion of ETPs. stem cell expansion and early thymocyte development (Serpine2, Two main points emerged from analyses of DEGs preferentially Mmp9, Tnc, Ptlhlh,andInhba) (47–51); and 2) genes favoring expressed in MCs from bone (cluster #2) and skin (cluster #3) resolution of inflammation (Cfh, Cp, Clu,andTgfb1). The latter four (Fig. 5B, 5C). First, alike DEGs expressed at a higher level in genes have anti-inflammatory effects either directly (Tgfb1) (52), or thymic MCs; those from bone and skin MCs contained many genes implicated in leukocyte migration (e.g., Ccl2, Ccl7, Cxcl10, by enhancing efferocytosis (Cfh) (53), scavenging extracellular free and Csf2 in skin MCs). However, the sets of leukocyte migration– radicals (Cp) (54), or mediating the disposal of extracellular mis- related DEGs were entirely different in MCs from the three organs folded proteins (Clu) (55). Relevance of inflammation resolution in (Fig. 5A–C). Second, clusters #2 and #3 contained many genes the thymus and bone is linked to the fact that chronic inflammation engaged in bone- and skin-specific functions, respectively. Strik- accelerates ageing of the thymus and HSCs (56–58). ingly, out of 40 DEGs in cluster #2, 8 play a role in osteoclasto- In cluster #5 (DEGs preferentially expressed in thymic and skin genesis (Fig. 5B). Likewise, genes related to chondrocyte and MCs), we noticed two groups of genes: 1) genes implicated in osteoblast biogenesis were expressed almost exclusively in bone monocyte/macrophage chemotaxis (Eln, Rarres2, Postn, and Plat) MCs: Prg4 (proteglycan 4), Comp (cartilage oligomeric matrix and phagocytosis optimization (Rarres2 and C2 ); and 2) genes protein). and Spp1 (osteopontin). Consistent with the fact that mediating epithelial cell adhesion to basement membrane (Col6a6, procollagen types I and III are the major collagen types expressed Vtn, and Lama2) and mesenchymal–epithelial cross-talk (Postn, by skin fibroblasts (46), Col1a1, Col1a2, and Col3a1 were ex- Bmp7, Nid1, and Ngf) (Fig. 6B). The presence of the latter group pressed higher levels in skin MCs (Fig. 5C). Furthermore, nine of genes dovetails well with the fact that both thymic and skin genes involved in skin and hair follicle homeostasis were com- MCs have to support epithelial cells. DEGs in cluster #6 ponents of cluster #3. (expressed at a higher level in bone and skin) were essentially The Journal of Immunology 4765 Downloaded from http://www.jimmunol.org/ by guest on October 1, 2021

FIGURE 4. IPA of the extracellular region–associated components of the six DEG clusters. Histograms show the top 10 most activated and nonredundant biological processes per DEG cluster. All processes depicted were significantly enriched and activated (p , 0.05 and Z-score .2; IPA). Biological processes linked by a black line depict functional overlap. Processes linked to leukocyte migration are in boldface. Numbers in parentheses represent the number of genes per biological process. Black dots depict p values. connected to monocyte/macrophage chemotaxis (Cxcl1, Fn1, ranging from 10224 to 10239). When we sorted these 274 shared Thbs1, Spon2, Serpine1,andCcl6), macrophage-dependent phago- MC genes according to their mean expression level, we found at cytosis of fungi and bacteria (Ptx3 and Spon2), and modulation of the top of the hierarchy numerous genes directly coding for osteoclastogenesis (Fig. 6C). components of the extracellular matrix (Sparc, Dcn, Bgn, Pcolce, and many collagens, etc.) and its remodeling (Timp2, Mmp2, and Functional analysis of genes shared by all MC populations Mmp14) (Fig. 7E). Among genes implicated in proliferation of Having defined the landscape of genes expressed in only one or two cells and development of epithelial tissue, we identified four MC populations, we next sought to characterize the ensemble of growth factors for epithelial cells (Fgf7, Fgf10, Igf1, and Igf2) and genes relevant to MC function and shared by all MCs. To this end, one for macrophages (Csf1) (Fig. 7F). Genes involved in migra- we first identified 9120 genes that were expressed at similar levels tion of cells and cell movement included genes coding for (FC ,5) in our three MC populations (RPKM .1) (Fig. 7A). monocyte-macrophage chemoattractants (Cxcl14 and Mdk) and From this gene set, we excluded genes that were expressed at enhancers of efferocytosis (Gas6, Pros1, Cfb, and C4b). similar levels (FC , 5) in MCs, TECs, and thymocytes. These ubiquitously expressed genes (n = 6270) are housekeeping genes MCs and the HSC niche involved in generic cell processes (Fig. 7C). The resulting 2850 One reason for including bone MCs in our study was that they genes represent core MC genes shared by all MCs studied in this represent the best characterized MC population, largely because of study, of which 274 are secreted or located at the cell-surface their vital interactions with HSCs (59, 60). Because HSCs occupy membrane (Figs. 3D, 7B). As expected for MCs, IPA analysis a perivascular niche in bone, it was notable that out of the top 10 of these 274 genes revealed highly significant enrichment for biological processes enriched in shared MC genes, three were genes associated with processes such as morphology of connective directly related to blood vessels (Fig. 7C). One of them, devel- tissue, cell proliferation, development of epithelial tissues, an- opment of cardiovascular system, included 26 genes instrumental giogenesis, and cell movement–migration (Fig. 7D; p values in forming a functional HSC niche (e.g., Cxcl12, Kitl, Agptl1/2, 4766 THE TRANSCRIPTOME OF PRIMARY MESENCHYMAL CELLS Downloaded from http://www.jimmunol.org/ by guest on October 1, 2021

FIGURE 5. Top extracellular region–associated genes in DEG clusters #1–3 (A–C). Expression heat maps show the top 40 genes in each cluster. Genes were sorted based on relevance score (gene expression level and FC). Colored circles represent the biological functions associated to speciﬁc genes. bMC, bone MC; sMC, skin MC; tMC, thymic MC.

Slit2/3, Vegfa, and Wnt5a) (59) (Fig. 7G). To further explore the tors with self-renewal potential (i.e., the bone and thymus) (31, 61, relationship between MCs and HSCs, we used a list of 309 genes 62). To further evaluate the likelihood of direct interactions be- expressed at relatively high levels in HSC-supportive mesenchy- tween MCs and hematolymphoid progenitors, we identiﬁed in our mal cell lines and identiﬁed as core genes operative in site of primary MCs seven transcripts coding for ligands known to reg- hematopoiesis (31). When we compared the expression of these ulate hematopoiesis and/or thymopoiesis (Csf1, Cxcl12, Igf1, Kitl, 309 genes in our three MC populations and in thymocytes, we Plau, Slit2, and Wnt5a). Using publicly available RNA-seq data, made two observations: 1) these 309 genes were expressed at we then analyzed expression of their cognate receptors by long- higher levels in the three MC populations than in thymocytes; and term HSCs (Lin2c-Kit+Sca-1+ Flk22CD342), HSCs (Lin2c-Kit+ 2) they were expressed at higher levels in bone and thymic MCs Sca-1+Flk22CD34+), and common lymphoid progenitors (Lin2 than in skin MCs (p , 0.001; Fig. 7H). These data suggest that c-Kit+Flk2+IL-7R+) (28). We found that genes coding for these although MCs from all tested tissues constitutively express genes seven receptors were expressed (RPKM .2) in at least one pop- predicted to support HSCs, these genes are expressed at higher ulation of hematopoietic progenitors (Fig. 7I). This observation levels in MCs from organs that harbor hematolymphoid progeni- supports the relevance of direct interactions between MCs and The Journal of Immunology 4767 Downloaded from http://www.jimmunol.org/ by guest on October 1, 2021

FIGURE 6. Top extracellular region–associated genes in DEG clusters #4–6 (A–C). Expression heat maps show the top 25 genes in each cluster, Genes were sorted based on relevance score (gene expression level and FC). Colored circles represent the biological functions associated to specific genes. bMC, bone MC; sMC, skin MC; tMC, thymic MC. hematopoietic progenitors. Hence, integration of our RNA-seq coding for molecules implicated in monocyte/macrophage che- data with independent datasets (Fig. 7H, 7I) further strengthens motaxis (Figs. 5, 6). In other words, the three types of MCs the idea that genes regulating HSC and ETP expansion are ex- expressed DEGs involved in monocyte/macrophage chemotaxis, pressed at higher levels in bone and thymic MCs relative to skin but the identity of these genes was different from one organ to MCs. Nonetheless, several of these genes have pleiotropic effects. another. Because macrophages are among the most multifunc- Therefore, further transcriptomic and functional analyses on MCs tional and heterogeneous cell types in mammalian tissues (64), we from a larger repertoire of tissues and organs will be required to speculate that macrophage and MC heterogeneity may be cor- evaluate the general applicability of our conclusion on tissue- egulated as a function of the local environment. Because 94% dependent MC specialization. of thymocytes undergo apoptosis, the thymus has to deal with an inordinate amount of apoptotic cells. Furthermore, depletion Discussion of the Nra1-dependent subset of thymic macrophages impairs The present study on MC heterogeneity presents two important efferocytosis, increases proinflammatory cytokine production, and features: it was performed on freshly harvested primary cells accelerates thymic involution (65). Efferocytosis and inflamma- (without in vitro culture) and was based on whole transcriptome tion resolution are functionally linked processes because clearance sequencing. Hence, it provides a systems-level appraisal of com- of apoptotic bodies initiates M2 polarization of macrophages. monalities and discrepancies between MCs from different organs Accordingly, a notable feature of many DEGs expressed at a without being tainted by changes induced by in vitro culture (6). higher level in thymic MCs is that they code for molecules that Overall, we found that thymic, bone, and skin MCs express 6270 regulate chemotaxis, efferocytosis, and M2 polarization of mac- housekeeping genes, 2850 core MC genes, and 2036 DEGs. DEGs, rophages and that promote inflammation resolution (Figs. 5A, 6A, of which a high number establishes the extent of MC heterogeneity, 6B). These data suggest that thymic MCs play an important role in were the main focus of our analyses. the maintenance of thymic integrity by promoting efferocytosis Interactions between MCs and tissue-resident macrophages are and preventing local production of inflammatory cytokines. central to tissue homeostasis (63). Accordingly, a salient feature of The transcriptome of thymic MCs displayed a second key the six DEG clusters preferentially expressed in one or two MC feature: higher expression of genes implicated in interactions with populations was that they all contained different sets of genes hematolymphoid progenitors. First, three genes involved in at- 4768 THE TRANSCRIPTOME OF PRIMARY MESENCHYMAL CELLS Downloaded from http://www.jimmunol.org/ by guest on October 1, 2021

FIGURE 7. Genes with shared gene expression across the three MC populations. (A) Analysis workflow (top to bottom) used to identify genes with shared gene expression level across MC populations. Numbers below each one-to-one comparison represent the number of genes retrieved using DESeq2 software and the following criteria: RPKM .1 in all three MC populations and FC ,5. (B) Venn diagram showing the proportion of housekeeping(HK) genes versus core MC genes among the 9120 genes expressed at similar levels in MC populations. The 2850 core MC genes were filtered using DAVID bioinformatics database to retrieve genes associated to the extracellular region (n = 274). (C) Top 10 IPA-enriched biological functions (p , 0.05; red line) for the 6270 housekeeping genes with similar expression level across MCs, TECs, and thymocytes. Numbers in parentheses correspond to the number of genes per biological function. (D) Top 10 IPA-enriched biological functions (p , 0.05; red line) for the 274 extracellular region genes with shared gene expression across MC populations. Numbers in parentheses correspond to the number of genes per biological function. (E) Heat map depicting the expression of the top 40 most abundant core MC transcripts in five cell types: the three MC populations (thymic MC [tMC], bone MC [bMC], and skin MC [sMC]), TECs, and thymocytes (Thymo.). Genes in boldface are associated to extracellular matrix organization/regulation. (F) Expression heat map of selected genes associated to proliferation of cells and migration of cells biological functions. (G) Expression heat map of selected genes associated to the biological function development of cardiovascular system. (H) Box plot depicting expression distribution of 309 HSC-supportive genes (31) across the three MC populations and thymocytes. Numbers in parentheses represent the number of HSC- supportive genes considered as expressed (RPKM .1) in each cell population. ***p , 0.001 (paired, two-tailed Student t test). (I) Expression heat map showing expression of genes coding for cognate ligands (MCs) and receptors (long-term HSC [LT-HSC], HSC, and common lymphoid progenitor [CLP]). traction and expansion of thymus-seeding progenitors were MCs included five genes that facilitate the expansion of hema- expressed at higher level in thymic MCs (cluster #1; Fig. 5A). tolymphoid progenitors (cluster #4; Fig. 6A). Third, when we Second, DEGs preferentially expressed in both thymic and bone integrated an independent dataset with our own data, we found The Journal of Immunology 4769 that genes predicted to support HSCs were expressed at higher 16. Pelekanos, R. A., J. Li, M. Gongora, V. Chandrakanthan, J. Scown, N. Suhaimi, levels in thymic and bone MCs than in skin MCs or thymocytes G. Brooke, M. E. Christensen, T. Doan, A. M. Rice, et al. 2012. Comprehensive transcriptome and immunophenotype analysis of renal and cardiac MSC-like (Fig. 7H). populations supports strong congruence with bone marrow MSC despite main- Finally, out of the 25 top DEGs preferentially expressed in tenance of distinct identities. Stem Cell Res. (Amst.) 8: 58–73. 17. Roson-Burgo, B., F. Sanchez-Guijo, C. Del Can˜izo, and J. De Las Rivas. 2014. thymic and skin MCs, 7 code for molecules mediating epithelial Transcriptomic portrait of human Mesenchymal Stromal/Stem Cells isolated cell adhesion to basement membrane or mesenchymal–epithelial from bone marrow and placenta. BMC Genomics 15: 910. cross-talk (Fig. 6B). Collectively, components of the six gene 18. Trapnell, C., D. G. Hendrickson, M. Sauvageau, L. Goff, J. L. Rinn, and L. Pachter. 2013. Differential analysis of gene regulation at transcript resolution clusters reported in this study provide compelling evidence that with RNA-seq. Nat. Biotechnol. 31: 46–53. MCs from various organs are very heterogeneous and that their 19. Liu, Y., J. Zhou, and K. P. White. 2014. RNA-seq differential expression studies: transcriptome is remarkably well adapted to their organ-specific more sequence or more replication? Bioinformatics 30: 301–304. 20. Jain, R., and D. H. D. Gray. 2014. Isolation of thymic epithelial cells and roles, including interactions with tissue-resident macrophages. For analysis by flow cytometry. Curr. Protoc. Immunol. 107: 3.26.1–3.26.15. thymic MCs, the key roles are the promotion of efferocytosis and 21. Houlihan, D. D., Y. Mabuchi, S. Morikawa, K. Niibe, D. Araki, S. Suzuki, the support of both hematolymphoid progenitors and epithelial H. Okano, and Y. Matsuzaki. 2012. Isolation of mouse mesenchymal stem cells on the basis of expression of Sca-1 and PDGFR-a. Nat. Protoc. 7: 2103–2111. cells. Gain- and loss-of-function experiments in mutant mice will 22. Ha¨kkinen, L., L. Koivisto, and H. Larjava. 2001. An improved method for be required to evaluate the in vivo relevance of several receptor– culture of epidermal keratinocytes from newborn mouse skin. Methods Cell Sci. ligand interactions highlighted by our analyses. We therefore hope 23: 189–196. 23. Crigler, L., A. Kazhanie, T.-J. Yoon, J. Zakhari, J. Anders, B. Taylor, and that this work will provide a framework for the generation of mutant V. M. Virador. 2007. Isolation of a mesenchymal cell population from murine mice in which gene expression is modulated specifically in MCs. dermis that contains progenitors of multiple cell lineages. FASEB J. 21: 2050– 2063.

24. Dumont-Lagace´, M., S. Brochu, C. St-Pierre, and C. Perreault. 2014. Adult Downloaded from Acknowledgments thymic epithelium contains nonsenescent label-retaining cells. J. Immunol. 192: We thank Danie`le Gagne´, Gae¨l Dulude, Marianne Arteau, Jennifer Huber, 2219–2226. 25. de Verteuil, D. A., A. Rouette, M.-P. Hardy, S. Lavalleé, A. Trofimov, Se´bastien Lemieux, and Patrick Gendron for assistance and sound advice. E´ . Gaucher, and C. Perreault. 2014. Immunoproteasomes shape the tran- We also thank the personnel of the Institute for Research in Immunology scriptome and regulate the function of dendritic cells. J. Immunol. 193: 1121– and Cancer animal care facility. 1132. 26. St-Pierre, C., A. Trofimov, S. Brochu, S. Lemieux, and C. Perreault. 2015. Differential features of AIRE-induced and AIRE-independent promiscuous gene http://www.jimmunol.org/ Disclosures expression in thymic epithelial cells. J. Immunol. 195: 498–506. The authors have no financial conflicts of interest. 27. Mortazavi, A., B. A. Williams, K. McCue, L. Schaeffer, and B. Wold. 2008. Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nat. Meth- ods 5: 621–628. 28. Lara-Astiaso, D., A. Weiner, E. Lorenzo-Vivas, I. Zaretsky, D. A. Jaitin, References E. David, H. Keren-Shaul, A. Mildner, D. Winter, S. Jung, et al. 2014. Immu- 1. O¨ hlund, D., E. Elyada, and D. Tuveson. 2014. Fibroblast heterogeneity in the nogenetics. Chromatin state dynamics during blood formation. Science 345: cancer wound. J. Exp. Med. 211: 1503–1523. 943–949. 2. Meńdez-Ferrer, S., T. V. Michurina, F. Ferraro, A. R. Mazloom, B. D. Macarthur, 29. Love, M. I., W. Huber, and S. Anders. 2014. Moderated estimation of fold S. A. Lira, D. T. Scadden, A. Ma’ayan, G. N. Enikolopov, and P. S. Frenette. change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15: 550. 2010. Mesenchymal and haematopoietic stem cells form a unique bone marrow 30. Huang, W., B. T. Sherman, and R. A. Lempicki. 2009. Systematic and integrative

niche. Nature 466: 829–834. analysis of large gene lists using DAVID bioinformatics resources. Nat. Protoc. by guest on October 1, 2021 3. Morrison, S. J., and D. T. Scadden. 2014. The bone marrow niche for haema- 4: 44–57. topoietic stem cells. Nature 505: 327–334. 31. Charbord, P., C. Pouget, H. Binder, F. Dumont, G. Stik, P. Levy, F. Allain, 4. Keating, A. 2012. Mesenchymal stromal cells: new directions. Cell Stem Cell 10: C. Marchal, J. Richter, B. Uzan, et al. 2014. A systems biology approach for 709–716. defining the molecular framework of the hematopoietic stem cell niche. Cell 5. Stagg, J., and J. Galipeau. 2013. Mechanisms of immune modulation by mes- Stem Cell 15: 376–391. enchymal stromal cells and clinical translation. Curr. Mol. Med. 13: 856–867. 32. Morikawa, S., Y. Mabuchi, Y. Kubota, Y. Nagai, K. Niibe, E. Hiratsu, S. Suzuki, 6. Frenette, P. S., S. Pinho, D. Lucas, and C. Scheiermann. 2013. Mesenchymal C. Miyauchi-Hara, N. Nagoshi, T. Sunabori, et al. 2009. Prospective identifi- stem cell: keystone of the hematopoietic stem cell niche and a stepping-stone for cation, isolation, and systemic transplantation of multipotent mesenchymal stem regenerative medicine. Annu. Rev. Immunol. 31: 285–316. cells in murine bone marrow. J. Exp. Med. 206: 2483–2496. 7. Jenkinson, W. E., E. J. Jenkinson, and G. Anderson. 2003. Differential re- 33. Boxall, S. A., and E. Jones. 2012. Markers for characterization of bone marrow quirement for mesenchyme in the proliferation and maturation of thymic epi- multipotential stromal cells. Stem Cells Int. 2012: 975871. thelial progenitors. J. Exp. Med. 198: 325–332. 34. Gullstrand, B., U. Ma˚rtensson, G. Sturfelt, A. A. Bengtsson, and L. Truedsson. 8. Jenkinson, W. E., S. W. Rossi, S. M. Parnell, E. J. Jenkinson, and G. Anderson. 2009. Complement classical pathway components are all important in clearance 2007. PDGFRalpha-expressing mesenchyme regulates thymus growth and the of apoptotic and secondary necrotic cells. Clin. Exp. Immunol. 156: 303–311. availability of intrathymic niches. Blood 109: 954–960. 35. Hanayama, R., M. Tanaka, K. Miwa, A. Shinohara, A. Iwamatsu, and S. Nagata. 9. Rossi, S. W., L. T. Jeker, T. Ueno, S. Kuse, M. P. Keller, S. Zuklys, 2002. Identification of a factor that links apoptotic cells to phagocytes. Nature A. V. Gudkov, Y. Takahama, W. Krenger, B. R. Blazar, and G. A. Holla¨nder. 417: 182–187. 2007. Keratinocyte growth factor (KGF) enhances postnatal T-cell development 36. Hanayama, R., M. Tanaka, K. Miyasaka, K. Aozasa, M. Koike, Y. Uchiyama, via enhancements in proliferation and function of thymic epithelial cells. Blood and S. Nagata. 2004. Autoimmune disease and impaired uptake of apoptotic 109: 3803–3811. cells in MFG-E8-deficient mice. Science 304: 1147–1150. 10. Chu, Y.-W., S. Schmitz, B. Choudhury, W. Telford, V. Kapoor, S. Garfield, 37. Grainger, D. J., J. Reckless, and E. McKilligin. 2004. Apolipoprotein E modu- D. Howe, and R. E. Gress. 2008. Exogenous insulin-like growth factor 1 en- lates clearance of apoptotic bodies in vitro and in vivo, resulting in a systemic hances thymopoiesis predominantly through thymic epithelial cell expansion. proinflammatory state in apolipoprotein E-deficient mice. J. Immunol. 173: Blood 112: 2836–2846. 6366–6375. 11. Sitnik, K. M., K. Kotarsky, A. J. White, W. E. Jenkinson, G. Anderson, and 38. Surh, C. D., and J. Sprent. 1994. T-cell apoptosis detected in situ during positive W. W. Agace. 2012. Mesenchymal cells regulate retinoic acid receptor-dependent and negative selection in the thymus. Nature 372: 100–103. cortical thymic epithelial cell homeostasis. J. Immunol. 188: 4801–4809. 39. Hasko´, G., and P. Pacher. 2012. Regulation of macrophage function by adeno- 12. Lax, S., E. A. Ross, A. White, J. L. Marshall, W. E. Jenkinson, C. M. Isacke, sine. Arterioscler. Thromb. Vasc. Biol. 32: 865–869. D. L. Huso, A. F. Cunningham, G. Anderson, and C. D. Buckley. 2012. CD248 40. Foucher, E. D., S. Blanchard, L. Preisser, E. Garo, N. Ifrah, P. Guardiola, expression on mesenchymal stromal cells is required for post-natal and infection- Y. Delneste, and P. Jeannin. 2013. IL-34 induces the differentiation of human dependent thymus remodelling and regeneration. FEBS Open Bio 2: 187–190. monocytes into immunosuppressive macrophages. antagonistic effects of GM- 13. Sun,L.,C.Sun,Z.Liang,H.Li,L.Chen,H.Luo,H.Zhang,P.Ding,X.Sun,Z.Qin, CSF and IFNg. [Published erratum appears in 2014. PLoS One 9: doi:10.1371/ and Y. Zhao. 2015. FSP1(+) fibroblast subpopulation is essential for the maintenance annotation/a732b069-ac19-4ca4-8474-97cbe0a7d7ee.] PLoS One 8: e56045. and regeneration of medullary thymic epithelial cells. Sci. Rep. 5: 14871. 41. Odaka, C. 2009. Localization of mesenchymal cells in adult mouse thymus: their 14. St-Pierre, C., S. Brochu, J. R. Vanegas, M. Dumont-Lagace´, S. Lemieux, and abnormal distribution in mice with disorganization of thymic medullary epi- C. Perreault. 2013. Transcriptome sequencing of neonatal thymic epithelial cells. thelium. J. Histochem. Cytochem. 57: 373–382. Sci. Rep. 3: 1860. 42. Muller,€ S. M., C. C. Stolt, G. Terszowski, C. Blum, T. Amagai, N. Kessaris, 15. Immunological Genome Project Consortium. 2012. Transcriptional profiling of P. Iannarelli, W. D. Richardson, M. Wegner, and H.-R. Rodewald. 2008. Neural stroma from inflamed and resting lymph nodes defines immunological hall- crest origin of perivascular mesenchyme in the adult thymus. J. Immunol. 180: marks. Nat. Immunol. 13: 499–510. 5344–5351. 4770 THE TRANSCRIPTOME OF PRIMARY MESENCHYMAL CELLS

43. Zlotoff, D. A., A. Sambandam, T. D. Logan, J. J. Bell, B. A. Schwarz, and 54. Yang, F., W. E. Friedrichs, L. deGraffenried, D. C. Herbert, F. J. Weaker, A. Bhandoola. 2010. CCR7 and CCR9 together recruit hematopoietic progeni- B. H. Bowman, and J. J. Coalson. 1996. Cellular expression of ceruloplasmin in tors to the adult thymus. Blood 115: 1897–1905. baboon and mouse lung during development and inflammation. Am. J. Respir. 44. Jiang, Q., V. M. Coffield, M. Kondo, and L. Su. 2007. TSLP is involved in Cell Mol. Biol. 14: 161–169. expansion of early thymocyte progenitors. BMC Immunol. 8: 11. 55. Wyatt, A. R., J. J. Yerbury, P. Berghofer, I. Greguric, A. Katsifis, C. M. Dobson, 45. Anandasabapathy, N., G. Breton, A. Hurley, M. Caskey, C. Trumpfheller, and M. R. Wilson. 2011. Clusterin facilitates in vivo clearance of extracellular P. Sarma, J. Pring, M. Pack, N. Buckley, I. Matei, et al. 2015. Efficacy and safety of CDX-301, recombinant human Flt3L, at expanding dendritic cells and he- misfolded proteins. Cell. Mol. Life Sci. 68: 3919–3931. matopoietic stem cells in healthy human volunteers. Bone Marrow Transplant. 56. Chinn, I. K., C. C. Blackburn, N. R. Manley, and G. D. Sempowski. 2012. 50: 924–930. Changes in primary lymphoid organs with aging. Semin. Immunol. 24: 309–320. 46. Olsen, D. R., J. Peltonen, S. Jaakkola, M. L. Chu, and J. Uitto. 1989. Collagen 57. Ergen, A. V., and M. A. Goodell. 2010. Mechanisms of hematopoietic stem cell gene expression by cultured human skin fibroblasts. Abundant steady-state levels aging. Exp. Gerontol. 45: 286–290. of type VI procollagen messenger RNAs. J. Clin. Invest. 83: 791–795. 58. Dooley, J., and A. Liston. 2012. Molecular control over thymic involution: from 47. Ibrahim, A. A., T. Yahata, M. Onizuka, T. Dan, C. Van Ypersele De Strihou, cytokines and microRNA to aging and adipose tissue. Eur. J. Immunol. 42: 1073– T. Miyata, and K. Ando. 2014. Inhibition of plasminogen activator inhibitor 1079. type-1 activity enhances rapid and sustainable hematopoietic regeneration. Stem 59. Mendelson, A., and P. S. Frenette. 2014. Hematopoietic stem cell niche main- Cells 32: 946–958. tenance during homeostasis and regeneration. Nat. Med. 20: 833–846. 48. Heissig, B., L. R. Lund, H. Akiyama, M. Ohki, Y. Morita, J. Rømer, 60. Ding, L., and S. J. Morrison. 2013. Haematopoietic stem cells and early lym- H. Nakauchi, K. Okumura, H. Ogawa, Z. Werb, et al. 2007. The plasminogen phoid progenitors occupy distinct bone marrow niches. Nature 495: 231–235. fibrinolytic pathway is required for hematopoietic regeneration. Cell Stem Cell 1: 61. Peaudecerf, L., S. Lemos, A. Galgano, G. Krenn, F. Vasseur, J. P. Di Santo, 658–670. S. Ezine, and B. Rocha. 2012. Thymocytes may persist and differentiate without 49. Ellis, S. L., S. Y. Heazlewood, B. Williams, A. J. Reitsma, J. Grassinger, J. Borg, C. K. Heazlewood, A. P. Chidgey, and S. K. Nilsson. 2013. The role of Tenascin any input from bone marrow progenitors. J. Exp. Med. 209: 1401–1408. C in the lymphoid progenitor cell niche. Exp. Hematol. 41: 1050–1061. 62. Martins, V. C., E. Ruggiero, S. M. Schlenner, V. Madan, M. Schmidt, P. J. Fink, 50. Calvi, L. M. 2006. Osteoblastic activation in the hematopoietic stem cell niche. C. von Kalle, and H.-R. Rodewald. 2012. Thymus-autonomous T cell develop- Ann. N. Y. Acad. Sci. 1068: 477–488. ment in the absence of progenitor import. J. Exp. Med. 209: 1409–1417. 51. Licona-Limoń, P., G. Alemań-Muench, J. Chimal-Monroy, M. Macıás-Silva, 63. Davies, L. C., S. J. Jenkins, J. E. Allen, and P. R. Taylor. 2013. Tissue-resident Downloaded from E. A. Garcıá-Zepeda, M. M. Matzuk, T. I. Fortoul, and G. Soldevila. 2009. macrophages. Nat. Immunol. 14: 986–995. Activins and inhibins: novel regulators of thymocyte development. Biochem. 64. Okabe, Y., and R. Medzhitov. 2014. Tissue-specific signals control reversible Biophys. Res. Commun. 381: 229–235. program of localization and functional polarization of macrophages. Cell 157: 52. Huynh, M.-L. N., V. A. Fadok, and P. M. Henson. 2002. Phosphatidylserine- 832–844. dependent ingestion of apoptotic cells promotes TGF-b1 secretion and the resolution of inflammation. J. Clin. Invest. 109: 41–50. 65. Tacke, R., I. Hilgendorf, H. Garner, C. Waterborg, K. Park, H. Nowyhed, 53. Kang, Y.-H., B. C. Urban, R. B. Sim, and U. Kishore. 2012. Human complement R. N. Hanna, R. Wu, F. K. Swirski, F. Geissmann, and C. C. Hedrick. 2015. The

Factor H modulates C1q-mediated phagocytosis of apoptotic cells. Immunobi- transcription factor NR4A1 is essential for the development of a novel macro- http://www.jimmunol.org/ ology 217: 455–464. phage subset in the thymus. Sci. Rep. 5: 10055. by guest on October 1, 2021 Supplemental gure 1

Lin-Sca-1-

Marker Thymus Bone Skin

41% ±1 64% ±8 92% ±4 CD29 % of max % of max % of max

50% ±10 58% ±9 87% ±4 CD51 % of max % of max % of max

3% ±1 47% ±3 72% ±7 CD140a % of max % of max % of max

39% ±6 51% ±3 71% ±3 CD140b % of max % of max % of max

4% ±2 1% ±1 7% ±1 CD34 % of max % of max % of max

21% ±16 30% ±10 42% ±10 gp38 % of max % of max % of max

50% ±4 43% ±7 55% ±5 Ly51 % of max % of max % of max

49% ±3 8% ±1 73% ±6 CD90.2 % of max % of max % of max

36% ±10 35% ±10 77% ±5 CD105 % of max % of max % of max

2% ±2 47% ±8 78% ±6 CD73 % of max % of max % of max

9% ±8 8% ±3 38% ±2 CD44 % of max % of max % of max

48% ±3 27% ±1 52% ±4 CD146 % of max % of max % of max

17% ±13 11% ±8 28% ±7 Nestin % of max % of max % of max

6% ±5 23% ±29 37% ±25 CXCL12 % of max % of max % of max

SUPPLEMENTAL FIGURE 1. Flow cytometry analysis of Lin- Sca-1- cells from the thymus, bone and skin. Overlay histograms illustrate staining with the relevant antibody (in blue) and an isotype control (in red). Each overlay histogram is representative of three independent experiments (3-5 mice per biological replicate). Numbers represent the mean percentage of positive cells (+/- SD). Supplemental gure 2

A B 1000 250k M 100 tMC 98.2% K P bMC 200k R 10 sMC 1 A

- 150k 1 9 1 a 4 8 1 0 5 A 2 5 0 3 3 5 9 0 C S CD CD CD gp Ly CD CD1 FSC 100k CD14 CD140b 1000 tMC 50k bMC

M 100 sMC K

0 P Thymocytes R 10 -103 0 103 104 105 cTEC 1 mTEC SCA-1 t 5 e 1 m 1 i 4 3 3 n -k ca x c CD CD CD p o E F C tMC vs bMC tMC vs sMC bMC vs sMC 7 7 ) ) ) 6 695 6 6 365

5 426 sMC sMC bMC 5 5 4 4 4 3 3 3 2 2 2 (ReadCount (ReadCount 362 10 10 1 849 1 1 573 log log 0 log10(ReadCount 0 r = 0.476 0 r = 0.629 r = 0.607

0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 log (ReadCount tMC) log10(ReadCount bMC) log10(ReadCount tMC) 10

SUPPLEMENTAL FIGURE 2. Purity of cell populations used for RNA-Seq analyses. (A) Post-sort analysis of sorted Sca-1+ thymic MCs. (B) Upper panel, histogram depicting the expression levels (RPKM) of 10 typical MC markers in the Sca-1+ MCs from the thymus, bone and skin. Lower panel, histogram showing expression levels of genes expressed in thymocytes or TECs. (RPKM = 15; red line) (C) Scatterplots depicting the number of DEGs in each one-to-one comparison of MC populations. Numbers in bold represent the amount of DEGs associated to the respective MC population in each comparison. Pearson correlation value (r) is also depicted in the bottom right corner of each scatterplot. Supplemental figure 3

C1 - Thymus C2 - Bone Z-score Z-score 0 1 2 3 4 5 0 2 4 6 8 quantity of cells (113) quantity of blood cells (82) B migration of cells (145) quantity of leukocytes (77) leukocyte migration (125) cell movement (153) cell movement of lymphocytes (25) cell movement of leukocytes (107) cell movement of cell movement of myeloid cells (77) mononuclear leukocytes (28) Lymphocyte migration (23) cell movement of phagocytes (81) leukocyte migration (48) A cell movement of neutrophils (56) A cell movement of leukocytes (44) recruitment of phagocytes (42) migration of phagocytes (13) chemotaxis of leukocytes (53) migration of antigen presenting cells (10) chemotaxis of phagocytes (45) cell movement of antigen (21) cell movement of presenting cells mononuclear leukocytes (51) migration of phagocytes (37) orientation of cells (8) C infiltration of leukocytes (51) 0 5 10 15 20 infiltration of cells (56) - log P-value ( ) cell movement of antigen (38) 10 presenting cells

inflammatory response (95) D C3 - Skin proliferation of cells (183) E Z-score 0 1 2 3 4 angiogenesis (52) vascularization (25) B recruitment of myeloid cells (11) recruitment of cells (14) A internalization of cells (20) F recruitment of phagocytes (11) activation of myeloid cells (31) 2.4 2.6 2.8 3.0 3.2 3.4 C activation of neutrophils (12) - log10 P-value ( ) 0 10 20 30 40 50 ( ) - log10 P-value C4 - Thymus + Bone C6 - Bone + Skin

Z-score Z-score 0 5 10 15 0 5 10 15

size of body (27) differentiation of connective (21) B tissue cells B differentiation of cells (58) proliferation of cells (75) C transcription of DNA (29) development of (34) activation of DNA (26) C cardiovascular system endogenous promoter vasculogenesis (27) A angiogenesis (30) fertility (13) E

0 2 4 6 cell viability (21) - log10 P-value ( ) F

vasculogenesis (22) D C5 - Thymus + Skin angiogenesis (25) Z-score cell movement (45) 0 2 4 6 cell movement of myeloid cells (22) cell movement of phagocytes (18) migration of cells (20) A cell movement of cell movement of granulocytes (13) myeloid cells (8) A migration of myeloid cells (9) cell movement (8) of phagocytes migration of granulocytes (7)

size of body (13) B development of body axis (33) G

quantity of cells (21) C proliferation of epithelial cells (17) H

differentiation of cells (30) D adhesion of immune cells (10) I

2.0 2.5 3.0 3.5 4.0 0 1 2 3 4

- log10 P-value ( ) - log10 P-value ( ) SUPPLEMENTAL FIGURE 3. Core biological functions associated to the 6 DEG clusters. All functions depicted in histograms were signifcantly enriched (p < 0.05) and activated (Z-score > 2). Related functions were clustered (letters). Numbers in parentheses represent the number of genes per biological functions. Red/black dots represent p-values. Supplemental table 1

# Gene symbol Gene name # Gene symbol Gene name

1 1700048O20Rik RIKEN cDNA 1700048O20 gene 40 Ccl7 chemokine (C-C motif) ligand 7, MCP3

2 2610524H06Rik RIKEN cDNA 2610524H06 gene 41 Cd248 CD248 antigen, endosialin

3 4930452B06Rik RIKEN cDNA 4930452B06 gene 42 Cdk18 cyclin-dependent kinase 18

4 A4galt alpha 1,4-galactosyltransferase 43 Cdkn1c cyclin-dependent kinase inhibitor 1C; p57Kip2

5 Acpl2 acid phosphatase-like 2 44 Cdkn2b cyclin-dependent kinase inhibitor 2B (p15, inhibits CDK4)

6 Adam12 a disintegrin and metallopeptidase domain 12 (meltrin alpha) 45 Cdyl2 chromodomain protein, Y chromosome-like 2

7 Adam19 a disintegrin and metallopeptidase domain 19 (meltrin beta) 46 Cep170b centrosomal protein of 170 kDa protein B

8 Adam8 a disintegrin and metallopeptidase domain 8 47 Chst1 carbohydrate (keratan sulfate Gal-6) sulfotransferase 1

a disintegrin-like and metallopeptidase (reprolysin type) with 9 Adamts12 48 Chst2 carbohydrate sulfotransferase 2 thrombospondin type 1 motif, 12 a disintegrin-like and metallopeptidase (reprolysin type) with 10 Adamts2 49 Cib2 calcium and integrin binding family member 2 thrombospondin type 1 motif, 2

11 Adcy3 adenylate cyclase 3 50 Cldn1 claudin 1

aminoglycoside phosphotransferase domain-containing 12 Agphd1 51 Col1a1 collagen, type I, alpha 1 protein 1

13 Ahr aryl hydrocarbon receptor 52 Col1a2 collagen, type I, alpha 2

14 Akap2 a-kinase anchor protein 2 53 Col3a1 collagen, type III, alpha 1

15 Akap6 akap6 protein 54 Col4a6 collagen, type IV, alpha 6

alkaline phosphatase, tissue-nonspecific isozyme; alkaline 16 Alpl 55 Col6a2 collagen, type VI, alpha 2 phosphatase, liver/bone/kidney

17 Angel1 protein angel homolog 1 56 Coro2a coronin-2A

18 Angptl4 angiopoietin-like 4 57 Crabp1 cellular retinoic acid binding protein I

19 Aoc3 amine oxidase, copper containing 3 58 Creb3l1 cAMP responsive element binding protein 3-like 1

20 Arc activity-regulated cytoskeleton-associated protein 59 Crip2 cysteine rich protein 2, CRP-2

21 Arhgap28 rac/Cdc42 guanine nucleotide exchange factor (GEF) 6 60 Csgalnact1 chondroitin sulfate N-acetylgalactosaminyltransferase 1

22 Arhgef28 rho-guanine nucleotide exchange factor 61 Ctsh cathepsin H

23 Arhgef6 rac/Cdc42 guanine nucleotide exchange factor (GEF) 6 62 Ctso cathepsin O

24 Arrb1 beta-arrestin-1 63 Cxcl12 chemokine (C-X-C motif) ligand 12

25 Atp10a ATPase, class V, type 10A 64 Cyp26b1 cytochrome P450, family 26, subfamily b, polypeptide 1

UDP-GalNAc:betaGlcNAc beta 1,3-galactosaminyltransferase, 26 B3galnt1 65 Cyp2j6 cytochrome P450 2J6 polypeptide 1

27 B4galnt1 beta-1,4 N-acetylgalactosaminyltransferase 1 66 Dclk2 serine/threonine-protein kinase DCLK2

28 Bach2 transcription regulator protein BACH2 67 Depdc7 DEP domain containing 7

29 Bmper BMP-binding endothelial regulator protein 68 Dhrs7 dehydrogenase/reductase (SDR family) member 7

30 Btbd2 protein Btbd2 69 Dock3 dedicator of cyto-kinesis 3

31 Cables1 CDK5 and Abl enzyme substrate 1 70 Dpysl3 dihydropyrimidinase-related protein 3

32 Calcrl calcitonin receptor-like 71 Dtx3l deltex 3-like (Drosophila)

33 Capn6 calpain 6 72 Duoxa1 dual oxidase maturation factor 1

34 Car13 carbonic anhydrase 13 73 Dynap protein Dynap

35 Car5b carbonic anhydrase 5b, mitochondrial 74 E130311K13Rik RIKEN cDNA E130311K13 gene

36 Casp12 caspase 12 75 Ebf3 early B cell factor 3

37 Cbr2 carbonyl reductase 2 76 Eda ectodysplasin-A

38 Ccdc80 coiled-coil domain containing 80 77 Efna5 ephrin A5

39 Ccl5 chemokine (C-C motif) ligand 5, RANTES 78 Ehd3 EH-domain containing 3 # Gene symbol Gene name # Gene symbol Gene name

eukaryotic translation initiation factor 2C, 4, Protein 79 Eif2c4 120 Il18rap interleukin 18 receptor accessory protein argonaute-4

80 Eif4e3 eukaryotic translation initiation factor 4E member 3 121 Il1r1 interleukin 1 receptor, type I

81 Elfn1 leucine rich repeat and fibronectin type III, extracellular 1 122 Il1rl1 interleukin 1 receptor-like 1

82 Ell2 elongation factor RNA polymerase II 2 123 Il4ra interleukin 4 receptor, alpha

83 Emb embigin 124 Inf2 inverted formin-2

84 En1 homeobox protein engrailed-1 125 Inpp4b inositol polyphosphate-4-phosphatase, type II

85 Epn3 epsin 3 126 Itgb3bp centromere protein R

86 Eya4 eyes absent 4 homolog (Drosophila) 127 Itm2a integral membrane protein 2A

87 Fabp4 fatty acid-binding protein, adipocyte 128 Ivd isovaleryl coenzyme A dehydrogenase

88 Fabp7 fatty acid binding protein 7, brain 129 Jak3 tyrosine-protein kinase JAK3

89 Fam110b protein FAM110B 130 Jup junction plakoglobin

90 Fam129a protein Niban 131 Kbtbd7 kelch repeat and BTB (POZ) domain containing 7

91 Fam20a protein FAM20A 132 Kcna4 potassium voltage-gated channel subfamily A member 4

92 Fat4 FAT tumor suppressor homolog 4 (Drosophila) 133 Kirrel3 kin of IRRE like 3 (Drosophila)

93 Fbln1 fibulin 1 134 Kitl kit ligand

94 Fgf10 fibroblast growth factor 10 135 Klra6 killer cell lectin-like receptor, subfamily A, member 6

95 Fgfr2 fibroblast growth factor receptor 2 136 Lamtor4 ragulator complex protein LAMTOR4

96 Fut8 alpha-(1,6)-fucosyltransferase 137 Lgalsl galectin-related protein A

97 Fyb FYN binding protein 138 Lgmn legumain

98 Fzd1 frizzled-1 139 Lhx9 LIM homeobox protein 9

99 Fzd8 frizzled homolog 8 (Drosophila) 140 Lonrf3 LON peptidase N-terminal domain and RING finger protein 3

UDP-N-acetyl-alpha-D-galactosamine:polypeptide N- 100 Galnt13 141 Loxl1 lysyl oxidase homolog 1 acetylgalactosaminyltransferase 13

101 Gas2 growth arrest specific 2 142 Lpar1 lysophosphatidic acid receptor 1

102 Gata3 GATA binding protein 3 143 Lpar4 lysophosphatidic acid receptor 4

103 Gatsl3 GATS protein-like 3 144 Lphn2 latrophilin 2

104 Glrx glutaredoxin 145 Lpl lipoprotein lipase

105 Gnao1 guanine nucleotide binding protein, alpha O 146 Lrig3 leucine-rich repeats and immunoglobulin-like domains 3

106 Gper1 G protein-coupled estrogen receptor 1 147 Lrp11 low-density lipoprotein receptor-related protein 11

107 Gria3 glutamate receptor, ionotropic, AMPA3 (alpha 3) 148 Lrrc4c leucine rich repeat containing 4C

108 Gstt3 glutathione S-transferase T3 149 Ltbp2 latent transforming growth factor beta binding protein 2

109 H2-DMa class II histocompatibility antigen, M alpha chain 150 Ly6a lymphocyte antigen 6 complex, locus A

110 Hcn2 hyperpolarization-activated, cyclic nucleotide-gated K+ 2 151 Ly6f lymphocyte antigen 6 complex, locus F

avian musculoaponeurotic fibrosarcoma (v-maf) AS42 oncogene 111 Hoxb13 homeobox protein Hox-B13 152 Maf homolog c-Maf

112 Hoxc8 homeo box C8 153 Maged2 melanoma antigen, family D, 2

113 Hspb8 heat shock protein 8 154 Map9 microtubule-associated protein 9

114 Htra3 htrA serine peptidase 3 155 Marcks myristoylated alanine rich protein kinase C substrate

115 Ifi27l1 interferon, alpha-inducible protein 27 like 1 156 Matn2 matrilin 2

116 Ifit1 Interferon-induced protein with tetratricopeptide repeats 1 157 Medag mesenteric estrogen-dependent adipogenesis protein

117 Igf1 insulin-like growth factor 1 (somatomedin C) 158 Mest mesoderm specific transcript

118 Il13ra1 interleukin 13 receptor, alpha 1 159 Mid2 midline 2 , Probable E3 ubiquitin-protein ligase

119 Il15 interleukin 15 160 Mmp15 matrix metallopeptidase 15 # Gene symbol Gene name # Gene symbol Gene name

161 Mmp2 matrix metallopeptidase 2 202 Pear1 platelet endothelial aggregation receptor 1

162 Mmp23 matrix metalloproteinase-23 203 Phospho1 phosphatase, orphan 1

163 Mn1 meningioma 1 204 Plagl1 pleiomorphic adenoma gene-like 1

pleckstrin homology domain-containing, family A (phosphoinositide 164 Mogat2 monoacylglycerol O-acyltransferase 2 205 Plekha2 binding specific) member 2 pleckstrin homology domain containing, family G (with RhoGef 165 Morc4 microrchidia 4 206 Plekhg4 domain) member 4

166 Mrgprf MAS-related GPR, member F 207 Plxna2 plexin-A2

167 Mtss1l metastasis suppressor 1-like 208 Plxnd1 plexin-D1

168 Myo1d myosin ID 209 Pmp22 peripheral myelin protein 22

169 Nadkd1 NAD kinase domain-containing protein 1 210 Pmvk phosphomevalonate kinase

170 Ncam1 neural cell adhesion molecule 1 211 Podxl podocalyxin-like

171 Nck2 cytoplasmic protein NCK2 212 Popdc3 popeye domain containing 3

172 Nckap5 peripheral clock protein 2 213 Porcn probable protein-cysteine N-palmitoyltransferase porcupine

173 Ndrg1 protein NDRG1 214 Postn periostin, osteoblast specific factor

174 Ndrg2 N-myc downstream regulated gene 2 215 Ppap2a phosphatidic acid phosphatase type 2A

175 Nes nestin 216 Ppap2b phosphatidic acid phosphatase type 2B

176 Nfatc1 nuclear factor of-activated T-cells, cytoplasmic 1 217 Ppp1r13b protein phosphatase 1, regulatory (inhibitor) subunit 13B

177 Nid2 nidogen 2 218 Prkar1b protein kinase, cAMP dependent regulatory, type I beta

178 Notch4 notch 4 219 Prkd1 protein kinase D1

179 Nova1 RNA-binding protein Nova-1 220 Prrg1 proline rich Gla (G-carboxyglutamic acid) 1

180 Npr2 natriuretic peptide receptor 2 221 Prune2 prune homolog 2 (Drosophila)

181 Npr3 natriuretic peptide receptor 3 222 Ptn pleiotrophin

182 Nrbp2 nuclear receptor-binding protein 2 223 Ptprf protein tyrosine phosphatase, receptor type, F

183 Nudt14 uridine diphosphate glucose pyrophosphatase 224 Ptx3 pentraxin related gene

184 Nxf3 nuclear RNA export factor 3 225 Pxmp4 peroxisomal membrane protein 4

185 Nynrin NYN domain and retroviral integrase containing 226 Pygo1 pygopus homolog 1

186 Oasl1 2-5 oligoadenylate synthetase-like 1 227 Rab3il1 guanine nucleotide exchange factor for Rab-3A

187 Oasl2 2-5 oligoadenylate synthetase-like 2 228 Rac3 RAS-related C3 botulinum substrate 3

188 Onecut2 one cut domain, family member 2 229 Ralgapa1 ral GTPase-activating protein subunit alpha-1

189 Osmr oncostatin M receptor 230 Rassf2 ras association (RalGDS/AF-6) domain family member 2

190 Oxr1 oxidation resistance 1 231 Rb1cc1 RB1-inducible coiled-coil protein 1

191 Pappa pregnancy-associated plasma protein A 232 Rbpms2 RNA binding protein with multiple splicing 2

192 Pax9 paired box gene 9 233 Rgs4 regulator of G-protein signaling 4

193 Pbxip1 pre-B-cell leukemia transcription factor-interacting protein 1 234 Rhobtb1 rho-related BTB domain-containing protein 1

194 Pcdhb22 protocadherin beta 22 235 Rnf150 RING finger protein 150

195 Pde1a phosphodiesterase 1A, calmodulin-dependent 236 Ror2 receptor tyrosine kinase-like orphan receptor 2

196 Pde1b phosphodiesterase 1B, Ca2+-calmodulin dependent 237 Rorb RAR-related orphan receptor beta

197 Pde4d cAMP-specific 3',5'-cyclic phosphodiesterase 4D 238 Rps6ka6 ribosomal protein S6 kinase polypeptide 6

198 Pdgfra platelet derived growth factor receptor, alpha polypeptide 239 Rspo2 R-spondin 2

199 Pdgfrb platelet derived growth factor receptor, beta polypeptide 240 Rtkn2 rhotekin-2

200 Pdk4 pyruvate dehydrogenase kinase, isoenzyme 4 241 Rtn4rl1 reticulon 4 receptor-like 1

201 Pdlim4 PDZ and LIM domain 4 242 Rtp4 receptor transporter protein 4 # Gene symbol Gene name # Gene symbol Gene name

TAF9B RNA polymerase II, TATA box binding protein (TBP)-associated 243 S100a16 S100 calcium binding protein A16 278 Taf9b factor

244 S1pr1 sphingosine-1-phosphate receptor 1 279 Tbx20 T-box 20, hrt (zebrafish)

245 Scn1b sodium channel subunit beta-1 280 Tcf7l1 transcription factor 7-like 1, Tcf3

246 Sdc3 syndecan 3 281 Tex15 testis protein TEX15

247 Selp selectin, platelet 282 Tfap2a transcription factor AP-2, alpha

sema domain, immunoglobulin domain (Ig), short basic 248 Sema3f 283 Tgfbi transforming growth factor, beta induced domain, secreted, (semaphorin) 3F

249 Sepp1 selenoprotein P, plasma, 1 284 Thsd7a thrombospondin, type I, domain containing 7A

250 Serpinb1a serine (or cysteine) peptidase inhibitor, clade B, member 1a 285 Tiaf2 TGF-beta1-induced anti-apoptotic factor 2

251 Serpinb6b protein Serpinb6b 286 Tmeff2 tomoregulin-2

252 Serpinf1 pigment epithelium-derived factor 287 Tmem117 transmembrane protein 117

253 Sfrp2 secreted frizzled-related protein 2 288 Tmem121 transmembrane protein 121

254 Six1 homeobox protein SIX1 289 Tmem47 transmembrane protein 47

solute carrier family 1 (glial high affinity glutamate 255 Slc1a3 290 Tnfrsf23 tumor necrosis factor receptor superfamily, member 23 transporter), member 3

256 Slc27a3 solute carrier family 27 (fatty acid transporter), member 3 291 Tram1l1 translocation associated membrane protein 1-like 1

257 Slc37a2 sugar phosphate exchanger 2 292 Trim21 E3 ubiquitin-protein ligase TRIM21

258 Slc39a8 solute carrier family 39 (metal ion transporter), member 8 293 Tsc22d3 TSC22 domain family, member 3

259 Slc43a1 solute carrier family 43, member 1 294 Tspan6 tetraspanin-6

260 Slc4a4 solute carrier family 4 (anion exchanger), member 4 295 Tspan7 tetraspanin 7

261 Slfn9 schlafen 9 296 Tvp23a Family with sequence similarity 18, member A

262 Smoc2 SPARC related modular calcium binding 2 297 Ubash3b ubiquitin associated and SH3 domain containing, B

263 Snai2 zinc finger protein SNAI2, Slug 298 Ube2l6 ubiquitin-conjugating enzyme E2L 6

264 Sorcs2 sortilin-related VPS10 domain containing receptor 2 299 Ust uronyl-2-sulfotransferase

265 Sox5 SRY-box containing gene 5 300 Wdr60 WD repeat-containing protein 60

266 Sp8 trans-acting transcription factor 8 301 Wisp2 WNT1 inducible signaling pathway protein 2

267 Speg SPEG complex locus 302 Wnt10b wingless related MMTV integration site 10b

268 Spon2 spondin-2 303 Zc2hc1a zinc finger C2HC domain-containing protein 1A

269 Spp1 secreted phosphoprotein 1 304 Zdhhc15 zinc finger, DHHC domain containing 15

270 Srpx2 sushi-repeat-containing protein, X-linked 2 305 Zdhhc2 palmitoyltransferase ZDHHC2

271 St3gal1 ST3 beta-galactoside alpha-2,3-sialyltransferase 1 306 Zfp518a zinc finger protein 518A

272 St6gal1 ST3 beta-galactoside alpha-2,3-sialyltransferase 1 307 Zfp618 zinc finger protein 618

ST6 (alpha-N-acetyl-neuraminyl-2,3-beta-galactosyl-1,3)-N- 273 St6galnac2 308 Zfp639 zinc finger protein 639 acetylgalactosaminide alpha-2,6-sialyltransferase 2

274 Stat1 signal transducer and activator of transcription 1 309 Zfp709 zinc finger protein 709

275 Sulf2 sulfatase 2

sushi, von Willebrand factor type A, EGF and pentraxin domain 276 Svep1 containing 1

277 Syn1 synapsin I

SUPPLEMENTAL TABLE 1 : List of the 309 genes predicted to be operative in sites of hematopoiesis and upregulated in HSC-supportive mesenchymal cell lines (Charbord et al. 2014). Shaded boxes represent genes associated to "secreted", "extracellular" or "glycoprotein" functional annotation of DAVID online bioinformatic tool.