Allergology International. 2006;55:173-179 ORIGINAL ARTICLE Gene Expression Profiling of Human Mast Cell Subtypes: An In Silico Study Hirohisa Saito1,2, Kenji Matsumoto1, Shigeru Okumura2, Jun-ichi Kashiwakura2, Keisuke Oboki2, Hidenori Yokoi2, Naotomo Kambe3, Ken Ohta4 and Yoshimichi Okayama2 ABSTRACT Background: Human mast cells (MCs) were classified into at least two subtypes, i.e., tryptase- and chymase- positive MCs (MCTC) and tryptase-only-positive MCs (MCT). However, differences in global molecular expres- sion between these subtypes are unknown. Methods: We analyzed public microarray data of MC subtypes derived from various tissues and those of pe- ripheral blood granulocytes by using hierarchical clustering methods to understand the global gene expression profiles. Results: All the transcripts subjected to this clustering analysis were classified into two large clusters, i.e., MC- preferential or granulocyte-preferential. In the original works, MCs from tonsil, lung and skin had been cultured for more than several weeks to obtain highly viable and pure cell populations, and these MCs retained their typical profiles such as intensities of chymase protein expression. Most of the transcripts were commonly ex- pressed by these MC subtypes. However, tonsil-derived MCs and skin-derived MCs but not lung-derived MCs expressed high levels of chymase (CMA1) as expected for the properties of MCTC and MCT. These CMA1-high MCs and CMA1-low MCs respectively expressed distinct sets of transcripts as small gene clusters as well as CMA-1 even after being cultured in the absence of a tissue environment. Conclusions: The MC lineage seems to be far from the granulocyte lineages including basophils. CMA1-high MCs (MCTC) and CMA1-low MCs (MCT) can be regarded as differentiated MC subtypes. As such, importance of data analysis studies will be increasing along with the accumulation of global molecular data in the public da- tabase. KEY WORDS carboxypeptidase, chymase, human mast cells, microarray, transcriptome 1,2 INTRODUCTION ered to be the era of “systems biology”, aiming at the comprehension of the total function of a cell, an Due to the complete reading of the genome sequence organ, or the human body using bioinformatics and and rapid technical advances, microarrays are now functional genomics. In most of transcriptome analy- widely used in many laboratories especially for ana- ses for finding novel molecules, mixed populations of lyzing transcriptome, i.e., global gene expression . peripheral blood or tissues have been used as clinical Transcriptome assays may be categorized into two samples . However , these mixed cell populations types depending on the strategy. One is aimed at sometimes make interpretation of the results difficult. finding an important molecule whose function or Contamination by a very small population of different structure has not been fully described. Another is the 0cell types having a certain highly expressed tran- comprehension of the whole system controlling a cell script may cause an artifactual presence of the tran- and the simulation of human disease models in silico, script in the whole population even where the major i.e., in a computer. Indeed, the 21st century is consid- cell type lacks it.3 On the other hand, mRNA is unsta- 1Department of Allergy & Immunology, National Research Institute and Immunology, National Research Institute for Child Health & for Child Health & Development, 4Department of Medicine, Teikyo Development, 2−10−1 Okura, Setagaya-ku, Tokyo 157−8535, Ja- University School of Medicine, Tokyo, 2Research Unit for Allergy pan. Transcriptome, Research Center for Allergy & Immunology , Email: [email protected] RIKEN Yokohama Institute, Kanagawa and 3Department of Der- Received 6 September 2005. Accepted for publication 1 Decem- matology, Kyoto University Graduate School of Medicine, Kyoto, ber 2005. Japan. !2006 Japanese Society of Allergology Correspondence: Hirohisa Saito, MD, PhD, Department of Allergy Allergology International Vol 55, No2, 2006 www.jsaweb.jp! 173 Saito H et al. ble so that complicated procedures for purification of Genome U133A probe array (GeneChip, Affymetrix, a certain cell type should be avoided. In this context, Santa Clara, CA, USA), which contains the oligonu- computational identification of cell-type specificity of cleotide probe set for approximately 22,283 full-length a certain key role molecule found in crude tissues genes and expressed sequence tags (ESTs). Total may be preferable. For this purpose, we have estab- cellular RNA was immediately isolated from fresh hu- lished the cell-type specific transcriptome database. man MCs or other inflammatory cells with RNeasy This cell-type specific transcriptome database is also Mini Kits (Qiagen, Valencia, CA, USA) containing expected to be necessary for understanding the sys- DNase-treatment. These samples had been obtained tems biology. from volunteers and patients with written informed Mast cells (MCs) are known to play versatile roles consent using an explanatory document which was such as in evoking allergic reaction and in controlling approved by the Ethical Review Board in each hospi- physiological reactions such as innate immunity . 4 tal. The purity of RNA was assessed on the basis of Furthermore , MCs are classified into two distinct the A260!A280 ratio, and the integrity of RNA was subtypes in rodent tissues: connective tissue MCs verified by agarose gel electrophoresis. Total RNA preferentially located in tissues, such as skin, and ( 50 ! 100 ng) was extracted from approximately mucosal MCs dominantly found in mucosa, such as 5 × 105 cells. Double-stranded cDNA was synthesized the intestine. They are distinct in staining characteris- from DNase-treated total RNA, and the cDNA was tics , T-cell dependency , responses toward secre- subjected to in vitro transcription in the presence of tagogues or stabilizing drugs, and cytokine require- biotinylated nucleoside triphosphates, according to ment for development.5-8 For human MCs, two sub- the small sample protocol ( two-cycle in vitro tran- types have been recognized by the distribution of scription method). The biotinylated cRNA was hy- granular neutral proteases: MCs positive for tryptase bridized with a probe array for 16 hours at 45", and together with chymase, cathepsin G, and carboxypep- the hybridized biotinylated cRNA was stained with tidase (MCTC) and MCs positive for tryptase only streptavidin-phycoerythrin (PE), then scanned with a (MCT). MCT were found in nasal mucosal tissues and Hewlett-Packard Gene Array Scanner (Palo Alto, CA, lung, while MCTC were located in skin or tonsillar tis- USA). The fluorescence intensity of each probe was sues.9,10 Several functional and morphological proper- quantified using a computer program , GeneChip ties are proposed regarding differences between MC Analysis Suite 5.0 (Affymetrix). The expression level TC and MCT. Toru et al., 11 observed that IL-4 pro- of a single mRNA was determined as the average moted cell maturation with the expression of both fluorescence intensity among the intensities obtained tryptase and a high amount of chymase from by 11 paired (perfect-matched and single nucleotide- tryptase-single positive human cord blood-derived mismatched) probes consisting of 25-mer oligonu- MCs. On the other hand, it was recently shown that cleotides . If the intensities of mismatched probes skin-derived , 12 tonsil-derived 13 and lung-derived were very high, gene expression was judged to be ab- MCs 13,14 can retain their phonotypical characters sent even if a higher-than-average fluorescence was such as the intensity of chymase staining and respon- obtained with the GeneChip Analysis Suite 5.0 pro- siveness to substance P for more than several weeks gram. The level of gene expression was determined even in the absence of a tissue-specific environment as the average difference (AD) using the GeneChip or T cell cytokines . Nevertheless , differences in software. global molecular expression between the human MC subtypes are totally unknown. DATA ANALYSIS We have reported the results of several genome- All the data that had been treated with small sample wide gene expression (transcriptome) studies4,13,15-18 protocol 3 were downloaded from our website at using human mast cells and other allergic inflamma- http:!!www.nch.go.jp!imal!GeneChip!public.htm. tory cells. According to the microarray guideline,19 From the “human leukocytes” file, two basophil sam- we have opened the gene expression data in our web- ple, two eosinophil samples and two neutrophil sam- site. In the present paper, we collected these tran- ples were chosen for this study. Three lung-derived scriptome data including various types of MCs and MC samples, three tonsil-derived MC sample, three other cell types, and performed a hierarchical cluster- peripheral blood progenitor-derived MC samples and ing analysis to understand the comprehensive gene a cord blood-derived MC sample, were collected from expression profiles of human MC subtypes. “Human Mast Cells” No. 5 file and were normalized METHODS with the median value of 22,283 transcripts of every array sample. It should be noted that these MCs de- GENECHIP EXPRESSION ANALYSIS rived from several tissues were cultured for more All in vitro experiments were performed in the previ- than several weeks in the absence of the specific tis- ous studies15-18 but not in the present study. There- sue environment. 15-18 This is because it was neces- fore, only the method of genome-wide gene expres- sary to obtain highly viable and purified MC samples sion is briefly described . We used the Human for GeneChip analysis. We merged these 16 sample 174 Allergology International Vol 55, No2, 2006 www.jsaweb.jp! In Silico Study of Human Mast Cell Subtypes Bas2 Bas1 Eos2 Eos1 Neu2 Neu1 CMC PMC2 TMC1 TMC3 TMC2 PMC3 PMC1 LMC2 LMC1 LMC3 1 ADRB2 CPA3 1000 TPSB2 CMA1 HEY1 GPR43 2000 C5R1 GPR44 MS4A3 2639 Fig. 1 Gene Expression Profiles of 16 Sample Data of Mast Cell Subtypes, Basophils, Eosinophils and Neutrophils. Affymetrix GeneChip U133A microarray data regarding MCs and other inflammatory cells were collected from the website at http://www.