Vivo Blood and Tonsillar Dendritic Cells in Functionally Associated

Gene Family Clustering Identifies Functionally Associated Subsets of Human In Vivo Blood and Tonsillar Dendritic Cells

This information is current as Malin Lindstedt, Kristina Lundberg and Carl A. K. of September 26, 2021. Borrebaeck J Immunol 2005; 175:4839-4846; ; doi: 10.4049/jimmunol.175.8.4839 http://www.jimmunol.org/content/175/8/4839 Downloaded from

References This article cites 28 articles, 12 of which you can access for free at: http://www.jimmunol.org/content/175/8/4839.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 September 26, 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 © 2005 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Gene Family Clustering Identiﬁes Functionally Associated Subsets of Human In Vivo Blood and Tonsillar Dendritic Cells1

Malin Lindstedt,2* Kristina Lundberg,2* and Carl A. K. Borrebaeck3*

Human dendritic cells (DCs) are a distinct but heterogeneous lineage of APCs operating as the link between innate and adaptive immune responses, with the function to either maintain tolerance or trigger immunity. The DC lineage consists of several subpopulations with unique phenotypes; however, their functional characteristics and transcriptional similarities remain largely unknown. To further characterize the phenotypes and transcriptomes of the subsets, we purified myeloid CD16؉, blood DC Ag .1؉ (BDCA1؉), and BDCA3؉ DC populations, as well as plasmacytoid CD123؉ DCs, from tonsillar tissue and peripheral blood Transcriptional profiling and hierarchical clustering visualized that BDCA1؉ DCs clustered with BDCA3؉ DCs, whereas CD16؉ Downloaded from DCs and CD123؉ DCs clustered as distinct populations in blood. Differential expression levels of chemokines, ILs, and pattern recognition receptors were demonstrated, which emphasize innate DC subset specialization. Even though highly BDCA1؉ and BDCA3؉ DC-specific gene expression was identified in blood, the BDCA1؉ DCs and BDCA3؉ DCs from tonsils displayed similar transcriptional activity, most likely due to the pathogenic or inflammatory maturational signals present in tonsillar tissues. Of note, plasmacytoid DCs displayed less plasticity in their transcriptional activity compared with myeloid DCs. The data demon-

strated a functionally distinct association of each of the seven subsets based on their signatures, involving regulatory genes in http://www.jimmunol.org/ adaptive and innate immunity. The Journal of Immunology, 2005, 175: 4839–4846.

endritic cells (DCs)4 constitute a distinct lineage of leu- ertheless, plasmacytoid CD123ϩ DCs and myeloid CD11cϩ DC pop- kocytes with exclusive features, such as initiation of T ulations have been defined in human thymus (4, 5), spleen (6), and D cell responses (1). Serving as the gatekeeper between the tonsils (7). The majority of human ex vivo DC studies have been innate and adaptive immune system, DCs are involved in the con- performed on DCs isolated from peripheral blood, which is a rela- trol of both tolerance and immunity. DCs have an outstanding tively consistent source of DCs. Based on lineage-specific marker capacity to present antigenic peptides in the context of MHC and negativity (LinϪ) and HLA-DRϩ, these DC populations have been selectively respond to environmental factors and pathogens fractionated primarily by using Abs against CD11c and CD123. Abs by guest on September 26, 2021 through the repertoire of pattern recognition receptors (PPRs). De- to the DC-specific markers blood DC Ag 1–4 (BDCA1 to -4) have spite many common features within the lineage, DCs comprise a enabled further characterization of different DC subpopulations (8). heterogeneous population of cells that originate from both lym- Thus far, five DC subsets, expressing unique phenotypes, have re- phoid and myeloid progenitors (2). Recent evidence also shows cently been identified in human blood based on their expression of that plasmacytoid DCs can originate from both lymphoid and my- CD123, CD1c/b, BDCA3, CD34, and CD16 (9). eloid pathways (3); however, the contribution of the respective pre- Even though several different DC subsets have been identified in cursors to the different DC subsets remains unclear. Multifaceted phe- blood, more information about their relationship, functional proper- notypes and specialized functions of the different DC subpopulations, ties, and tissue distribution is necessary. In this study, we further char- as well as their wide distribution, are factors that complicate the area acterized the CD1cϩ (BDCA1ϩ) DCs, BDCA3ϩ DCs, CD16ϩ DCs, of DC characterization. Their migratory behavior and membrane re- and CD123ϩ DCs isolated from peripheral blood, as well as identified organization during maturation further adds to the complexity. and phenotypically characterized BDCA1ϩ DCs, BDCA3ϩ DCs, and In contrast to the numerous studies performed in mice, only limited CD123ϩ DCs in human tonsils. In addition, we performed a global data are presently available on human DCs. Limited knowledge of transcriptional analysis of these discrete peripheral blood and tonsillar functionally associated DC subtypes, in combination with low fre- tissue DC subsets. We identified large clusters of DC subtype-specific quencies of tissue DCs, have hampered their characterization. Nev- gene expression and further analyzed their transcriptional relationship. Our data suggest that the defined subtypes are unique DC populations, exhibiting different repertoires of chemokine receptors, TLRs, and Department of Immunotechnology, Lund University, Lund, Sweden C-type lectins, indicating different functional roles. In addition to DC Received for publication April 27, 2005. Accepted for publication July 5, 2005. subtype-specific gene expression, we also identified transcription sig- ϩ ϩ The costs of publication of this article were defrayed in part by the payment of page natures common to BDCA1 DCs and BDCA3 DCs. These clusters charges. This article must therefore be hereby marked advertisement in accordance of DC-specific markers offer new insights in the ontogeny and func- with 18 U.S.C. Section 1734 solely to indicate this fact. tional role of the different DCs subsets, and for identification of novel 1 This work was supported by a grant from Vetenskapsra˚det. DC-specific markers. 2 M.L. and K.L. contributed equally to this article. 3 Address correspondence and reprint requests to Prof. Carl A. K. Borrebaeck, De- Materials and Methods partment of Immunotechnology, Lund University, P.O. Box 7031, S-220 07 Lund, Isolation of DC populations from tonsils and peripheral blood Sweden. E-mail address: [email protected] 4 Abbreviations used in this paper: DC, dendritic cell; BDCA, blood DC Ag; PPR, Human tonsils were obtained from children undergoing tonsillectomy at pattern recognition receptor; CLECSF, C-type lectin superfamily member. Lund University Hospital (Lund, Sweden). Tonsils were minced in RPMI

1640 medium (Sigma-Aldrich) supplemented with 0.2% gentamicin (In- analyzer (Agilent Technologies), and denatured at 94°C before hybridiza- vitrogen Life Technologies), and the tissue was digested with 2 mg/ml tion. The samples were hybridized to the Human Genome U133 Plus 2.0 collagenase IV and 100 U/ml DNase I for 15 min in room temperature. Array at 45°C for 16 h by rotation (60 rpm) in an oven. The arrays were Released cells were filtered through 70-␮m nylon cell strainers (Falcon; then washed, stained with streptavidin-PE (Molecular Probes), washed BD Biosciences) and washed once in gentamicin-supplemented RPMI again, and scanned with a GeneArray Scanner (Affymetrix). 1640. PBMC were isolated from leukocyte-enriched buffy coats (Lund University Hospital) by Ficoll-Paque (Amersham Biosciences) density gra- Microarray data analysis dient centrifugation. T lymphocytes were depleted by rosetting with sheep erythrocytes treated with neuraminidase (Sigma-Aldrich). Further deple- The fluorescence intensity was analyzed, using the GeneChip Operating tion of B cells, T cells, and monocytes was performed by magnetic sepa- Software (GCOS) 1.1 (Affymetrix), and scaled to a target value of 100. ration, using anti-CD19-, anti-CD3-, and anti-CD14-coated beads (Dynal Further data analysis was performed with GeneSpring 7.1 software. For Biotech). Negatively selected cells enriched for DCs were incubated with clustering, the samples were normalized both per chip, to the 50th percen- FITC-conjugated mAbs against CD3 (BD Biosciences), CD14 and CD19 tile, and per gene, which makes the median value for each gene across the (DakoCytomation), allophycocyanin-conjugated anti-HLA-DR (BD Bio- samples equal to 1. A tree clustering was performed on the individual sciences), and either PE-conjugated mAb against BDCA3 (Miltenyi Bio- samples to distinguish sample relationship and replicate similarities. The tec) or CD123 (BD Pharmingen). The DC-enriched cells from peripheral tree clustering algorithm, based on Spearman correlation, was used on blood were also labeled with PE-conjugated CD16 mAb (BD Biosciences). genes denoted P (present) in at least one DC population (three replicates) ϩ and displaying a fold change in mean expression level of Ϯ2 between two BDCA1 DCs were incubated with a lineage-specific PE-labeled Ab mix- ture (DakoCytomation), HLA-DR-allophycocyanin (BD Biosciences), and populations, giving a total of 13,931 genes. Thus, the degree of similarity between the samples was based on differentially expressed transcripts, thus BDCA1-FITC (Miltenyi Biotec). Lineage-negative and HLA-DRhigh DCs, positive for either BDCA1, BDCA3, CD123, or CD16, were then sorted on excluding genes not changing in expression level between the DC popu- a FACSDiVa or a FACSAria (BD Biosciences) and collected in tubes lations, such as housekeeping genes, or absent genes. K-means clustering containing 1 ml of 100% FBS. A total of 40,000–100,000 cells of each was performed to identify expression patterns in both blood and tonsillar Downloaded from unique DC population was isolated. The purity of each blood or tonsil DC DCs, and the identified clusters were further evaluated with filtering pro- subpopulation was directly confirmed by reanalysis using a FACScan (BD cedures. To identify DC subset-specific gene transcription, we performed a Biosciences), and the cells were routinely Ͼ98% pure. DC samples were stringent analysis with criss-cross comparisons between the triplicates of all DC subtypes, in both blood and tonsils. For instance, for transcripts to lysed in TRIzol reagent (Invitrogen Life Technologies) and stored ϩ at Ϫ20°C. be identified as CD16 DC-selective, each replicate had to display differential expression compared with all other DC replicates in blood, based on Flow cytometry Ͼ2-fold expression change, present call (P) and raw signal intensity Ͼ20 to eliminate borderline expression consisting of random noise. This strat- http://www.jimmunol.org/ Cell surface staining of blood and tonsillar DCs was performed on cells egy of pairwise comparisons was used to extract the subset-specific tran- isolated either by magnetic cell sorting (MACS), with the BDCA1, scriptomes for each DC population in blood and tonsils. All data are MI- BDCA3, or BDCA2 isolation kits (Miltenyi Biotec), or by FACSDiVa (BD AME compliant and have been submitted to ArrayExpress database Biosciences) sorting on lineage-negative cells. PBS, containing 1% BSA (͗www.ebi.ac.uk/miamexpress͘; accession no. E-TABM-34). (w/v) and 2.5 mM EDTA, was used in all cell labeling and washing steps. Gates were set to exclude debris and nonviable cells on the basis of light Results scatter properties. To reduce unspecific binding, freshly sorted blood or tonsillar DCs were first blocked with ChromPure Mouse IgG (Jackson Isolation and phenotypic characterization of DC subsets ImmunoResearch Laboratories) before incubation with the specific mouse The composition of DC populations in tonsils was initially studied mAbs for 30 min at 4°C. Analysis was performed on a FACScan or a by phenotypic analysis of the lineage-negative mononuclear cells. by guest on September 26, 2021 FACSDiVa (BD Biosciences). The following mAbs were used for flow ϩ ϩ cytometry: FITC-conjugated CD11c, CD13, CD14, CD19, CD20, CD86 We could identify the myeloid BDCA1 DCs, BDCA3 DCs, and (DakoCytomation), HLA-DR, CD4, CD11b, CD44 (BD Biosciences), CD3, CD40, CD45RA (BD Pharmingen), CD80 (Immunotech), BDCA1 (Miltenyi Biotec), and CD44v7 (Chemicon International); PE-conjugated CD23 (BD Biosciences), CD123 (BD Pharmingen), BDCA3 (Miltenyi Biotec), and CD44v7 (Chemicon International); allophycocyanin-conjugated HLA-DR (BD Biosciences), BDCA3, and BDCA4 (Miltenyi Bio- tec); PE-Cy5-conjugated CD3, CD19 (DakoCytomation), biotin-conjugated HLA-DR, CXCR4 (BD Pharmingen), and unconjugated CD45RO (DakoCytomation). FITC- or PE-conjugated streptavidin (DakoCytoma- tion), allophycocyanin-Cy7-conjugated streptavidin (BD Pharmingen), and PE- or PE-Cy5-conjugated rabbit anti-mouse Igs (DakoCytomation) were used as secondary reagents. For graphic three-dimensional visualization and color coding, the analyzed fcs-file, generated by the FACSDiVa, was imported into GeneSpring 7.1 software (Silicon Genetics). Preparation of cRNA and gene chip hybridization BDCA1ϩ DCs, BDCA3ϩ DCs, and CD123ϩ DCs from tonsils, and BDCA1ϩ DCs, BDCA3ϩ DCs, CD16ϩ DCs, and CD123ϩ DCs from peripheral blood were sorted from three different donors, giving a total of 21 samples, each containing 40,000–100,000 cells. Fragmentation, hybridization, and scanning of the Human Genome U133 Plus 2.0 Arrays were performed according to the manufacturer’s protocol (Affymetrix). The preparation of labeled cRNA was performed according to the Two-cycle Eukaryotic Target Labeling assay protocol, using the GeneChip Expression 3Ј amplification two-cycle labeling and control reagents kit (Affymetrix). Briefly, cDNA was generated from total RNA (20–150 ng), using Super- Script II (Invitrogen Life Technologies) and a T7-oligo(dT) promoter primer (Affymetrix). After a second-strand cDNA synthesis, cDNA was FIGURE 1. Flow-cytometric analysis of lineage-negative mononuclear converted to cRNA by an in vitro transcription reaction (Ambion MEGA- cells, isolated from human tonsillar tissue, identifies three unique DC sub- script T7 kit). Thereafter, the cRNA was purified using RNeasy Mini kit (Qiagen), and the yield was controlled with a spectrophotometer. A second sets. Gates were set to include only viable cells, as determined by forward- cycle of cDNA synthesis was performed, followed by the same cleanup as and side-scatter characteristics. Furthermore, cells were gated to be both above and a second in vitro transcription reaction cycle with biotin-labeled lineage negative and either BDCA1 (blue), BDCA3 (green), or CD123 ribonucleotides and T7 RNA polymerase. Labeled cRNA was purified, (red) positive. The percentages represent relative levels of the different DC using RNeasy Mini kit (Qiagen), quality controlled with Agilent 2100 Bio- populations from one representative experiment. The Journal of Immunology 4841

Table I. Phenotype of peripheral blood and tonsillar tissue DCs, as determined by ﬂow cytometry

Blood DCs Tonsillar DCs

CD123ϩ BDCA1ϩ BDCA3ϩ CD16ϩ CD123ϩ BDCA1ϩ BDCA3ϩ

CD11b Ϫ ϩ Ϫ ϩϩ Ϫ ϩϩ ϩϩ CD11c Ϫ (ϩ)a ϩϩϩ ϩϩϩ ϩϩϩ Ϫ ϩϩϩ ϩϩ/Ϫ CD13 Ϫ (ϩ) ϩϩ ϩϩϩ ϩϩ Ϫ ϩϩ ϩ (ϩϩϩ) CD4 ϩϩ ϩϩϩϩϩ CD40 ϪϪ ϪϩϪϩϩϩϩ CD44 ϩϩ ϩϩ ϩϩ ϩϩ ϩϩ ϩϩ ϩϩ CD44 v7 ϪϪ ϪϪϪϩϩϩϩ CD45RA ϩϩ Ϫ Ϫ ϩϩ ϩϩϩ Ϫ ϩϩ/Ϫ CD45RO Ϫϩ ϩϩϪϩϩϩ CD80 ϪϪ ϪϪϪϪϪ CD86 Low Low Low Low Ϫ Low Low CXCR4 ϪϪ ϪϪϪϩϩϩϩ HLA-DR ϩϩ ϩϩϩ ϩϩ ϩϩ ϩϩϩ ϩϩϩ ϩϩϩ

a Parentheses indicate the cell surface marker positivity of a minor subset of cells. Downloaded from plasmacytoid CD123ϩ DCs in both blood and tonsillar tissue, al- these seven DC populations were analyzed by ﬂow cytometry (Ta- though the BDCA3ϩ DC subset was less distinctive in tonsils ble I). A total of 40,000–100,000 cells from three independent compared with blood, whereas the CD16ϩ DC subset was present donors were sorted to a purity exceeding 98% (data not shown) only in blood. Furthermore, similar to blood DCs, the tonsillar from both tonsils and blood and immediately frozen for RNA ex- BDCA1ϩ DCs, BDCA3ϩ DCs, and CD123ϩ DCs could be iden- traction. It was evident that the populations consisting of BDCA1ϩ ϩ

tiﬁed as unique DC subsets (Fig. 1), even though a minor fraction DCs and BDCA3 DCs exhibited a fairly similar phenotype both http://www.jimmunol.org/ of CD123ϩBDCA3ϩ DCs was recognized. The phenotypes of in blood and tonsils. Also, the phenotype of tonsillar CD123ϩ DCs by guest on September 26, 2021

FIGURE 2. Scheme for sorting DC subpopulations from peripheral blood and tonsillar tissue. Mononuclear cells, enriched for DCs by depletion of CD3-, CD14-, and CD19-positive cells, were stained with Lineage, HLA-DR, and one of the CD123, CD16, BDCA1 or BDCA3 mAbs. Gates were set to include viable cells, as determined by forward- and side-scatter characteristics. Thereafter, gates were set to exclude Linϩ cells (A). Finally, LinϪ cells were positively sorted based on the phenotype of HLA-DRϩDC-markerϩ (B). A total of 40,000–100,000 cells of the CD16ϩ, CD123ϩ, BDCA1ϩ, and BDCA3ϩ DC populations were sorted, from three donors, from blood and tonsils, to a purity of Ͼ98%. 4842 TRANSCRIPTIONAL ANALYSIS OF HUMAN BLOOD AND TONSILLAR DCs was similar to that of blood CD123ϩ DCs. Among the cell surface Hierarchical gene clustering and data filtering markers that actually did vary between the blood DC subsets were To determine the relationship between DC subsets and correlation CD11b, CD11c, CD13, CD40, CD45RA, and CD45RO. In tonsils, ϩ ϩ between triplicate samples, hierarchical clustering was performed BDCA1 DCs and BDCA3 DCs displayed up-regulation of on differentially expressed genes in the entire data set (Fig. 4). The CD40, CD44v7, and CXCR4, compared with their blood counter- plasmacytoid CD123ϩ DCs from blood and tonsils clustered to- parts. The DC populations in blood and tonsils were FACS sorted gether and were separated from the myeloid CD16ϩ, BDCA1ϩ, from mononuclear cells depleted of monocytes, B cells, and T cells and BDCA3ϩ DC subsets. The gene expression of BDCA1ϩ DCs (Fig. 2). in blood was more similar to the BDCA3ϩ DC population in blood than the BDCA1ϩ DC population in tonsils (and vice versa for BDCA3ϩ DCs). However, the CD16ϩ DCs were separated from Validation of CD markers by transcriptional analysis ϩ ϩ both the BDCA1 DCs and the BDCA3 DCs isolated from blood RNA from each DC population was hybridized to Affymetrix Hu- and tonsillar tissue. man Genome U133 Plus 2.0 arrays containing Ͼ54,000 probe sets, which cover 38,500 human genes. Arrays were run in triplicate, Comparison of gene expression patterns between tonsillar and with RNA from three different donors, for each DC population. For blood DC subsets each subset, the intensity signals for selected marker genes were Patterns of transcription in blood and tonsillar DCs were identified assessed that were expected to be either expressed and thus de- by K-means clustering analysis on the differentially expressed noted present, or absent (Fig. 3). CD1c (BDCA1), CD16, the my- genes selected above. In addition to the subtype-specific gene ex- eloid cell marker CD11c, as well as the plasmacytoid DC markers pression, a set of genes expressed by both BDCA1ϩ DCs and Downloaded from CD123 and BDCA2 were selected as positive controls (Fig. 3A). A BDCA3ϩ DCs was recognized (Fig. 5A). A significant number of probe set representing the BDCA3 gene was not present on the genes were found to be differentially transcribed between the var- array and could thus not be included. CD11c was, as expected, ious subsets, in that 808 and 579 genes were preferentially ex- ϩ ϩ ϩ expressed by the CD16 , BDCA1 , and BDCA3 subsets, but not pressed by plasmacytoid DCs in blood and tonsils, respectively, ϩ by CD123 DCs. CD123, CD16, and BDCA1 were expressed pri- whereas 516 transcripts were identified as CD16ϩ DC-selective. marily by the appropriate populations, even though low expression Furthermore, BDCA1ϩ DCs and BDCA3ϩ DCs in blood also ex- http://www.jimmunol.org/ of these markers could be detected in the BDCA3ϩ DC subset. To hibited selective transcription with 93 and 192 genes, respectively, exclude the possibility of contaminating cell types in the DC pop- in addition to a common set of 137 expressed genes. A similar ulations, a selection of B cell, T cell, monocyte, and NK cell pattern of expression was detected in tonsillar DCs, with an even marker genes were analyzed (Fig. 3B). These negative controls, greater overlap in expression (1132 genes) in BDCA1ϩ DCs and consisting of CD3, CD19, CD14, CD56, CD8, and TCR genes, BDCA3ϩ DCs. were either denoted absent by the detection call or, in a few cases, expressed at low levels. Overall, the concordance between the high Selective expression of ILs, chemokines, C-type lectins, and expression levels of the indicator genes and the low or absent TLRs by DC subsets by guest on September 26, 2021 expression of lineage-marker genes, clearly show that the DC iso- We next turned to the identification of gene families, indicating lations have been performed without contamination. different roles of the isolated DC subsets. Within the data set of

FIGURE 3. Gene expression level of selected control genes. A, Mean values from DC subset triplicates are represented for positive control transcripts; CD123 (NM_002183), CD1c/BDCA1 (NM_001765), CD16 (NM_000570), BDCA2/CLECSF7 (NM_130441), and CD11c (M81695). If two or more copies of a speciﬁc gene, with different Gen- Bank numbers, were present on the array, the mRNA reference sequence was chosen. The y-axis displays raw signal intensity and the x-axis numbers represent the following DC subsets: tonsil BDCA1ϩ DCs (1), blood BDCA1ϩ DCs (2), tonsil BDCA3ϩ DCs (3), blood BDCA3ϩ DCs (4), blood CD16ϩ DCs (5), tonsil CD123ϩ DCs (6), and blood CD123ϩ DCs (7). B, Expression levels of selected NK cell-, B cell-, T cell-, and monocyte-speciﬁc genes as negative controls. Mean value of replicate samples are represented. The minus sign (Ϫ) denotes samples with an absent (A) detection call in one to three of the replicate samples. The Journal of Immunology 4843

BDCA3ϩ DCs, irrespectively of tissue source. Furthermore, in blood, IL1R1, IL1R2, and IL13RA1 were primarily expressed by BDCA1ϩ DCs, whereas IL21R was expressed by CD16ϩ DCs and IL18R1 by plasmacytoid DCs. Overall, the transcriptional signatures of CD123ϩ DCs in blood and tonsils were quite similar, and the CD123ϩ DCs expressed a common set of TLRs, chemokine receptors, IL receptors, and C-type lectins. In contrast, DC subtype-speciﬁc gene expression was evident among the myeloid cells in blood, whereas the tonsillar BDCA1ϩ DCs and BDCA3ϩ DCs expressed a majority of the functionally associated markers (sum- marized in Table II).

Discussion Identiﬁcation and characterization of distinct subgroups of DCs are important to advance our understanding of their role in regulating the immune response in health and disease. In this study, we at- tempt to phenotypically describe the myeloid and plasmacytoid DCs found in circulation and in secondary lymphoid tissue and to

illustrate their relationship and subset-speciﬁc transcriptionally Downloaded from clusters, with emphasis on immune response-associated gene families. Characterization of the complete human transcriptome with gene expression proﬁling is well suited for addressing any global disparity between DC subsets, identifying novel DC markers and describing the complexity of biological processes. Thus, the ex-

pression profiling not only describes the specialization of the re- http://www.jimmunol.org/ spective subtype, but also portrays the transcriptional changes involved in the process of migration and functional reprogramming during relocation of DCs from peripheral blood to secondary lym- FIGURE 4. Reproducibility of transcriptional data and subset relation- phoid organs. Consequently, we isolated DC subsets found in pe- ship determined by hierarchical clustering of replicate DC samples. Clus- ripheral blood (9) and compared them with the tonsillar DC tering, using Spearman correlation, was performed on genes changing in expression level with at least a 2-fold change between two DC populations counterparts. (as described in Materials and Methods). The node number (the distance, The phenotype of tonsillar DC populations has previously been displayed in log scale) represents the degree of similarity between the described based on their relative expression of HLA-DR, CD11c, samples. Thus, if two samples are grouped under one branch with a dis- CD123, and CD13 (7). The DC subsets were, for instance, described by guest on September 26, 2021 tance 0.05, the correlation is 0.95. as HLA-DRhighCD11cϩ DCs, HLA-DRmodCD11cϩCD13ϩ DCs, and HLA-DRmodCD11cϪCD123ϩ DCs. Identification of the subset- specific DC markers enabled us to use positive selection of these 3508 differentially regulated genes (Fig. 5A), we identified a vast subsets. In a phenotypical comparison between our data and the study number of functionally associated genes, involved in, e.g., Ag up- by Summers et al. (7), it appears that the isolated BDCA1ϩ DCs and take, signaling, migration, etc., suggesting that the distinct DC BDCA3ϩ DCs in our system resembles the HLA-DRhighCD11cϩ subsets are specialized for different tasks. The distribution of che- DCs and HLA-DRmodCD11cϩCD13ϩ DCs, respectively, based on mokines and ILs, their receptors, as well as C-type lectins and their expression of, e.g., CD45RO and CD45RA. TLRs, are presented in Fig. 5B. It was evident that the BDCA1ϩ Hierarchical clustering emphasizes the similarities between trip- DCs and BDCA3ϩ DCs isolated from tonsils displayed the most licate samples and visualizes the transcriptional relationship be- similar expression profiles of these functionally clustered genes, tween the DC subtypes. As expected, the plasmacytoid DCs are although a few subtype-specific transcripts could be identified as separated from the other subtypes classically referred to as the well. Furthermore, in the blood DC populations, BDCA3ϩ DCs myeloid DCs. Even though CD16ϩ DCs are distinct from the primarily expressed TLR3, CD16ϩ DCs expressed TLR8, whereas BDCA1ϩ DCs and BDCA3ϩ DCs, they still cluster with the my- plasmacytoid DCs expressed TLR7 and TLR9. Among the C-type eloid cells. The relationship between BDCA1ϩ DCs and BDCA3ϩ lectins, BDCA3ϩ DCs in blood demonstrated the highest levels of DCs in blood and tonsils, demonstrates a subset similarity and that MRC2, and this subset selectivity was also found in tonsils. Blood their transcription is determined by tissue distribution. Our results BDCA1ϩ DCs, in contrast, expressed DCIR, C-type lectin super- suggest that CD123ϩ DCs and CD16ϩ DCs arise from separate family member (CLECSF)14, MRC1, and DC-SIGN, whereas plas- precursor cells, whereas BDCA1ϩ DCs and BDCA3ϩ DCs may macytoid cells expressed CLECSF7 (BDCA2). have a common origin and rather represent two different stages of The majority of differentially expressed ILs, chemokines, and a similar subset. Even though the identified subset-selective genes their receptors were found in the tonsillar BDCA1ϩ DCs and do not necessarily represent unique markers for a given population, BDCA3ϩ DCs. In blood, CX3CR1 was CD16ϩ DC-selective, the pronounced difference in expression levels and the vast number whereas CCR9 was expressed primarily by BDCA3ϩ DCs. We of genes differentially expressed between DC populations indicate also identified expression of chemokine-like factor (CKLF) and that these cell types indeed have specialized functions that are chemokine orphan receptor (CMKOR1) in the myeloid DC popu- related to their distinct lineage. In addition, among the numerous lations, which previously has not been reported to be expressed by genes differentially expressed between the DC subtypes, some DCs. CMKOR1 was primarily expressed by the CD16ϩ DCs and transcripts might not result in protein product. However, the avail- BDCA1ϩ DC subsets in blood and by the myeloid DC populations ability of the raw data in ArrayExpress will allow investigators to in tonsils, whereas CKLF was expressed by BDCA1ϩ DCs and perform further analysis. 4844 TRANSCRIPTIONAL ANALYSIS OF HUMAN BLOOD AND TONSILLAR DCs Downloaded from http://www.jimmunol.org/ by guest on September 26, 2021

FIGURE 5. Hierachical clustering of differentially expressed genes in tonsillar and peripheral blood DC subsets. A, Subset-specific transcriptional patterns, identified by K-means analysis and filtering (described in Materials and Methods) are illustrated by gene clustering. Color changes, within a row, indicate expression levels relative to the median of the sample population. Because the samples are normalized to a median value of 1, the color bar range of 5 (red) to 0.2 (green) represents high and low expression levels, respectively. B, Subset-specific gene expression shown according to their distribution in functional gene clusters.

To pinpoint the innate diversity between DC subsets, we fo- show that the BDCA3ϩ and CD16ϩ DC subsets also display spe- cused on expression of PPRs, cytokines, and chemokines. Both cialized TLR repertoires. Blood-derived BDCA3ϩ DCs selectively myeloid and plasmacytoid DCs use a sophisticated repertoire of expressed mRNA for TLR3, whereas the tonsillar BDCA1ϩ DCs PPRs, such as the TLRs (10) and C-type lectins (11), to recognize and BDCA3ϩ DCs both expressed TLR3. ssRNA were recently pathogen-associated molecular patterns. A specialization of micro- identiﬁed as natural ligands for the structurally related TLR7 and bial molecule recognition by the peripheral blood CD1cϩ TLR8 (13, 14). However, agonists to TLR7 and -8 have shown that (BDCA1ϩ) DC and plasmacytoid DC subsets has been demon- they are functionally distinct in innate immunity, because TLR7 strated on the mRNA level by PCR (12). Whereas CD1cϩ DCs ligation induces IFN-␣ production by plasmacytoid DCs, whereas expressed TLR2, -4, -5, -6, and -8, plasmacytoid DCs expressed TLR8 ligation stimulates IL-12 production by monocyte-derived only TLR7 and -9. In this report, we conﬁrm these data and further DCs and CD11cϩ blood DCs (15). Based on our microarray data, The Journal of Immunology 4845

Table II. Summary of markers expressed in DC subsets isolated from blood and tonsilsa

DC Subsets Markers

Blood DCs CD123ϩ CCR2, CXCR3, CMKLR1, IL18R1, IL3RA, CLECSF7, TLR7, TLR9 BDCA1ϩ IL1R1, IL1R2, IL13RA1, DC-SIGN, DCIR, CLECSF14, MRC1 BDCA3ϩ CCR9, CLEC1, MRC2, TLR3 CD16ϩ CX3CR1, HM74, IL7, IL15, IL21R, TLR8 BDCA1ϩ and BDCA3ϩ CXCL9

Tonsillar DCs CD123ϩ CCR2, CCR3, CMKLR1, IL18R1, IL3RA BDCA1ϩ CMKOR1 BDCA3ϩ MRC2 BDCA1ϩ and BDCA3ϩ CCL19, CCR6, CCR7, CXCL1, CXCR1, CXCL10, CXCL16, CXCL9, CKLF, HM74, IL1R1, IL1R2, IL1B, IL12B, IL13RA1, IL15, IL18, IL21R, IL7R, IL8, Langerin, Dectin-1, CLECSF13, CLECSF14, MRC1, TLR2, TLR3, TLR4, TLR5, TLR8

a Differentially transcribed genes, deﬁned as DC subset speciﬁc, were selected as described in Materials and Methods

ϩ the CD16 DC population may exert such a specialization toward CCR9. Interestingly, the majority of differentially expressed che- Downloaded from ssRNA, because the subset preferentially expresses TLR8 in blood. mokines/chemokine receptors were induced in the myeloid DCs in In addition to differential TLR expression patterns among the DC tonsils. Myeloid DCs are known to up-regulate CCR7 during mat- subsets, we also identified selective expression of C-type lectins in uration and acquire responsiveness to the CCR7 ligand CCL19 ϩ blood-derived BDCA1 DCs (DCIR, CLECSF14, MRC-1, DC- (ELC) (23). Of note, both CCR7 and CCL19 were up-regulated in ϩ SIGN) and BDCA3 DCs (CLEC1, MRC-2), with particularly pro- the myeloid DCs in tonsils and, to a lesser extent, in tonsillar ϩ ϩ nounced common expression by BDCA1 DCs and BDCA3 plasmacytoid DCs, suggesting that CCL19 can act in an autocrine http://www.jimmunol.org/ DCs in tonsils (Langerin, Dectin-1, CLECSF13, CLECSF14, and paracrine fashion to keep the CCR7ϩ cells in the vicinity. MRC-1). DC-SIGN, mannose receptor (MRC-1), and dectin-1 Furthermore, the wide range of chemokines/chemokine receptors (CLECSF12) are receptors for fungal-derived carbohydrates, such expressed by tonsillar myeloid DCs may also be important for as mannose, fucose, or ␤-glucans (16). The expression of these selective attraction of specific cell types within the tissue, e.g., lectins by BDCA1ϩ DCs may enable them to selectively respond CXCL16 (24), in combination with responses to exogenous medi- to a variety of exogenous ligands, such as bacteria (e.g., Myco- ators produced as a result of inflammatory stimulation, e.g., bacterium tuberculosis), viruses (e.g., HIV), and fungi (e.g., Can- CMKLR1 CXCL8 IL-8 dida albicans), as reviewed in Refs. 11 and 17. Even though (22) and ( ) (25). MRC-2 (Endo180) is a member of the mannose receptor family, it The differential expression of cytokines/cytokine receptors by by guest on September 26, 2021 seems that MRC-1 and MRC-2 have a distinct set of glycoprotein blood and tonsillar DCs may also provide information about their ligands (18). Expression of MRC-2 by DCs has, to our knowledge, innate functions and ability to polarize the immune response. Sim- not been reported previously. The fact that we identified MRC-2 as ilar to the chemokines, a majority of the differentially expressed preferentially expressed by BDCA3ϩ DCs in blood and tonsils, ILs/IL receptors was found in the myeloid DCs isolated from ton- suggests that this subset is specialized to recognize different gly- sils. IL-18 was recently shown to attract plasmacytoid DCs coproteins compared with the MRC-1-expressing BDCA1ϩ DCs. through IL-18R and promote Th1 induction (26), suggesting a role DCIR and BDCA2 (CLECSF7), selectively expressed by BDCA1ϩ for IL-18 in the recruitment of plasmacytoid DCs to sites of in- DCs and CD123ϩ DCs, respectively, belongs to the DCIR family flammation. IL-18R is also expressed selectively by Th1, and not of C-type lectins, whose biological ligands and function are largely by Th2 cells, and IL-12 and IL-18 synergize in the induction of unknown (19). Furthermore, we identified expression of CLEC1, IFN-␥ production by Th1 cells (27). Interestingly, IL18 and IL12B CLECSF13, and CLECSF14, not previously described in primary were both expressed mainly by the myeloid tonsillar DC subsets, DCs. The function and ligands of these lectins remains to be which suggests ongoing Th1 priming in the lymphoid compart- determined. ment. IL-18 is related to the IL-1 family and shares many functions Plasmacytoid DCs expressed a similar profile of chemokine re- with IL-1␤, such as induction of proinflammatory cytokines and ceptor expression (CCR2, CCR3, CXCR3, and CMKLR1) in both protection against infections (28). Interestingly, transcription of tonsils and blood. CXCR3 is expressed by blood plasmacytoid IL1R1, IL1R2, and IL1B was also detected in tonsillar DCs (20) and is required for migration of these cells into inflamed myeloid DCs. lymph nodes (21). Unexpectedly, we detected the CXCR3 ligands In conclusion, we have identified large clusters of DC subset- CXCL9 and CXCL10 in BDCA1ϩ DCs and BDCA3ϩ DCs in ton- specific gene expression, which suggests that the isolated DC sub- sils. Recently, blood plasmacytoid DCs, but not myeloid DCs, sets are highly specialized and respond to different spectra of were shown to express CMKLR1 and migrate in response to the pathogens. Furthermore, the transcriptional activity in the tonsillar CMKLR1-ligand chemerin, which is activated during blood coag- ϩ ϩ ulation (22). This may enable plasmacytoid DCs to migrate to sites BDCA1 DCs and BDCA3 DCs populations was more pro- of bleeding, tissue damage, and inflammation. Interestingly, we nounced and these DC subsets displayed a switch in expression of detected CMKLR1 mRNA in tonsillar plasmacytoid DCs in addi- chemokine/chemokine receptors and IL/IL receptors in tonsils, tion to blood plasmacytoid DCs, which suggests that the receptor compared with blood. Because no exogenous trigger was included may be involved in migration to lymphoid tissues with high in- in this study, we depended solely on the tonsil environment to fectional load. Furthermore, subset-specific transcription of che- affect the isolated DC populations. In contrast, plasmacytoid DCs mokine receptors was detected in blood DCs, because CD16ϩ DCs in blood and tonsils expressed a similar set of chemokine recep- expressed CX3CR1 and HM74, whereas BDCA3ϩ DCs expressed tors, IL receptors, and PPRs, which suggests that plasmacytoid 4846 TRANSCRIPTIONAL ANALYSIS OF HUMAN BLOOD AND TONSILLAR DCs

DCs are less prone to switch their functional repertoire in the sec- 13. Heil, F., H. Hemmi, H. Hochrein, F. Ampenberger, C. Kirschning, S. Akira, ondary lymphoid organs and may not be as sensitive to the envi- G. Lipford, H. Wagner, and S. Bauer. 2004. Species-specific recognition of single-stranded RNA via Toll-like receptor 7 and 8. Science 303: 1526–1529. ronment in tonsils as the myeloid DCs. 14. Diebold, S. S., T. Kaisho, H. Hemmi, S. Akira, and C. Reis e Sousa. 2004. Innate antiviral responses by means of TLR7-mediated recognition of single-stranded RNA. Science 303: 1529–1531. Acknowledgments 15. Gorden, K. B., K. S. Gorski, S. J. Gibson, R. M. Kedl, W. C. Kieper, X. Qiu, We thank Ann-Charlotte Olsson for expert laboratory assistance and M. A. Tomai, S. S. Alkan, and J. P. Vasilakos. 2005. Synthetic TLR Agonists Carl-Magnus Ho¨gerkorp for advice regarding cell sorting. Reveal Functional Differences between Human TLR7 and TLR8. J. Immunol. 174: 1259–1268. 16. Herre, J., S. Gordon, and G. D. Brown. 2004. Dectin-1 and its role in the rec- Disclosures ognition of ␤-glucans by macrophages. Mol. Immunol. 40: 869–876. The authors have no financial conflict of interest. 17. Taylor, P. R., S. Gordon, and L. Martinez-Pomares. 2005. The mannose receptor: linking homeostasis and immunity through sugar recognition. Trends Immunol. 26: 104–110. References 18. East, L., S. Rushton, M. E. Taylor, and C. M. Isacke. 2002. Characterization of 1. Steinman, R. M. 2003. Some interfaces of dendritic cell biology. APMIS 111: sugar binding by the mannose receptor family member, Endo180. J. Biol. Chem. 675–697. 277: 50469–50475. 2. Liu, Y. J. 2001. Dendritic cell subsets and lineages, and their functions in innate 19. Kanazawa, N., K. Tashiro, and Y. Miyachi. 2004. Signaling and immune regu- and adaptive immunity. Cell 106: 259–262. latory role of the dendritic cell immunoreceptor (DCIR) family lectins: DCIR, 3. Shigematsu, H., B. Reizis, H. Iwasaki, S. Mizuno, D. Hu, D. Traver, P. Leder, DCAR, dectin-2 and BDCA-2. Immunobiology 209: 179–190. N. Sakaguchi, and K. Akashi. 2004. Plasmacytoid dendritic cells activate lym- 20. Penna, G., S. Sozzani, and L. Adorini. 2001. Cutting edge: selective usage of phoid-specific genetic programs irrespective of their cellular origin. Immunity 21: chemokine receptors by plasmacytoid dendritic cells. J. Immunol. 167: 43–53. 1862–1866. 4. Bendriss-Vermare, N., C. Barthelemy, I. Durand, C. Bruand, C. Dezutter- 21. Yoneyama, H., K. Matsuno, Y. Zhang, T. Nishiwaki, M. Kitabatake, S. Ueha, Downloaded from Dambuyant, N. Moulian, S. Berrih-Aknin, C. Caux, G. Trinchieri, and F. Briere. S. Narumi, S. Morikawa, T. Ezaki, B. Lu, et al. 2004. Evidence for recruitment 2001. Human thymus contains IFN-␣-producing CD11cϪ, myeloid CD11cϩ, and of plasmacytoid dendritic cell precursors to inflamed lymph nodes through high mature interdigitating dendritic cells. J. Clin. Invest. 107: 835–844. endothelial venules. Int. Immunol. 16: 915–928. 5. Vandenabeele, S., H. Hochrein, N. Mavaddat, K. Winkel, and K. Shortman. 2001. 22. Zabel, B. A., A. M. Silverio, and E. C. Butcher. 2005. Chemokine-like receptor Human thymus contains 2 distinct dendritic cell populations. Blood 97: 1 expression and chemerin-directed chemotaxis distinguish plasmacytoid from 1733–1741. myeloid dendritic cells in human blood. J. Immunol. 174: 244–251. 6. McIlroy, D., C. Troadec, F. Grassi, A. Samri, B. Barrou, B. Autran, P. Debre, 23. Sallusto, F., P. Schaerli, P. Loetscher, C. Schaniel, D. Lenig, C. R. Mackay, J. Feuillard, and A. Hosmalin. 2001. Investigation of human spleen dendritic cell S. Qin, and A. Lanzavecchia. 1998. Rapid and coordinated switch in chemokine

phenotype and distribution reveals evidence of in vivo activation in a subset of receptor expression during dendritic cell maturation. Eur. J. Immunol. 28: http://www.jimmunol.org/ organ donors. Blood 97: 3470–3477. 2760–2769. 7. Summers, K. L., B. D. Hock, J. L. McKenzie, and D. N. Hart. 2001. Phenotypic 24. Shimaoka, T., T. Nakayama, N. Fukumoto, N. Kume, S. Takahashi, characterization of five dendritic cell subsets in human tonsils. Am. J. Pathol. J. Yamaguchi, M. Minami, K. Hayashida, T. Kita, J. Ohsumi, et al. 2004. Cell 159: 285–295. surface-anchored SR-PSOX/CXC chemokine ligand 16 mediates firm adhesion 8. Dzionek, A., A. Fuchs, P. Schmidt, S. Cremer, M. Zysk, S. Miltenyi, D. W. Buck, of CXC chemokine receptor 6-expressing cells. J. Leukocyte Biol. 75: 267–274. and J. Schmitz. 2000. BDCA-2, BDCA-3, and BDCA-4: three markers for dis- 25. Lindstedt, M., B. Johansson-Lindbom, and C. A. Borrebaeck. 2002. Global re- tinct subsets of dendritic cells in human peripheral blood. J. Immunol. 165: programming of dendritic cells in response to a concerted action of inflammatory 6037–6046. mediators. Int. Immunol. 14: 1203–1213. 9. MacDonald, K. P., D. J. Munster, G. J. Clark, A. Dzionek, J. Schmitz, and 26. Kaser, A., S. Kaser, N. C. Kaneider, B. Enrich, C. J. Wiedermann, and H. Tilg. D. N. Hart. 2002. Characterization of human blood dendritic cell subsets. Blood 2004. Interleukin-18 attracts plasmacytoid dendritic cells (DC2s) and promotes 100: 4512–4520. Th1 induction by DC2s through IL-18 receptor expression. Blood 103: 648–655. 10. Reis e Sousa, C. 2004. Toll-like receptors and dendritic cells: for whom the bug 27. Xu, D., W. L. Chan, B. P. Leung, D. Hunter, K. Schulz, R. W. Carter, by guest on September 26, 2021 tolls. Semin. Immunol. 16: 27–34. I. B. McInnes, J. H. Robinson, and F. Y. Liew. 1998. Selective expression and 11. McGreal, E. P., J. L. Miller, and S. Gordon. 2005. Ligand recognition by antigen- functions of interleukin 18 receptor on T helper (Th) type 1 but not Th2 cells. presenting cell C-type lectin receptors. Curr. Opin. Immunol. 17: 18–24. J. Exp. Med. 188: 1485–1492. 12. Jarrossay, D., G. Napolitani, M. Colonna, F. Sallusto, and A. Lanzavecchia. 2001. 28. Biet, F., C. Locht, and L. Kremer. 2002. Immunoregulatory functions of inter- Specialization and complementarity in microbial molecule recognition by human leukin 18 and its role in defense against bacterial pathogens. J. Mol. Med. 80: myeloid and plasmacytoid dendritic cells. Eur. J. Immunol. 31: 3388–3393. 147–162.