© 2014 Nature America, Inc. All rights reserved. NATURE IMMUNOLOGY ADVANCE ONLINE PUBLICATION 1 RESOURCE Myeloid cells in mammals are crucial in the formation and execu- tion of immune responses. They are able to recognize damage- and pathogen-associated molecular patterns through germline-encoded receptors and not only initiate but also amplify adaptive immune response through antigen presentation and cytokine production. In hematopoiesis, a myeloid cell is any white blood cell that is not a lymphocyte. By this definition, monocytes and granulocytes are the most common myeloid cell populations in the blood. But dendritic cells (DCs) and various populations of macrophages, monocytes and polymorphonucleated cells can be found in solid tissues 1 . Myeloid cells can be classified by cell surface–marker expression, ontogeny or differential dependence on lineage-defining transcription factors and growth factor signaling, which sharply distinguishes specific popula- tions 2–5 . However, recent studies have further highlighted the com- plex ontogeny of macrophages and DCs 5,6 . These advances highlight the unexpected complexity of the myeloid system and the limited capacity to identify the numerous cell types of myeloid origin with highly specialized functions and tissue distribution. Analysis of myeloid cells relies predominantly on historical nomen- clature and the use of traditional markers to define cell populations, but these do not encompass the full complexity of the myeloid sys- tem. For example, macrophages branded solely by the expression of the common myeloid marker CD11b and the macrophage marker F4/80 in adult mice are likely to be contaminated by eosinophils, monocytes or both 7 . CD11b is not expressed by alveolar macrophages, and many macrophage subpopulations do not express F4/80 (ref. 8). Monocyte-derived cells that have phenotypic and functional proper- ties shared by both DCs and macrophages are difficult to objectively distinguish 9–12 . Inflammation further complicates the picture when myeloid cells undergo phenotypic changes or other cell types not present in the steady state emerge. Such inflammatory DCs have been termed monocyte-derived DCs (moDCs) or tumor necrosis factor (TNF)- and iNOS-producing DCs (TIP-DCs) 13 . In addition, a number of subsets of myeloid-derived suppressor cells have been defined by association of functional and phenotypic properties in preclinical can- cer models, but their definitions often overlap with other described myeloid subsets, which leads to confusing nomenclature 14,15 . A series of reports identified additional cell surface markers on specific subsets within the myeloid compartment, which sharpened the distinction between DCs and macrophages while further increasing the number and complexity of myeloid subpopulation definitions 16–18 . Multiple populations of granulocytes, monocytes, monocyte-derived cells, resi- dent macrophages and DCs coexist in tissues, but their unambiguous identification is becoming increasingly difficult. Analysis of cellular lineages relies chiefly on fluorescence flow cytom- etry and informed biased gating strategies. Such technology, although far progressed, is bound by the number of independent parameters and spectral overlap. Fluorescence-based analysis of phagocytic myeloid cells is further complicated by their (often strong) autofluorescence at wave- lengths in the range of 500–600 nm, which contributes to potentially 1 Agency for Science, Technology and Research (A*STAR), Singapore Immunology Network (SIgN), Singapore. 2 Institute of Experimental Immunology, University of Zurich, Zurich, Switzerland. 3 School of Biological Sciences, Nanyang Technological University, Singapore. 4 Present address: Institute of Experimental Immunology, University of Zurich, Zurich, Switzerland. 5 These authors contributed equally to this work. Correspondence should be addressed to E.W.N. (evan_newell@immunol.a-star.edu.sg) or B.B. (becher@immunology.uzh.ch). Received 12 June; accepted 9 September; published online 12 October 2014; doi:10.1038/ni.3006. High-dimensional analysis of the murine myeloid cell system Burkhard Becher 1,4,5 , Andreas Schlitzer 1,5 , Jinmiao Chen 1,5 , Florian Mair 2 , Hermi R Sumatoh 1 , Karen Wei Weng Teng 1 , Donovan Low 1 , Christiane Ruedl 3 , Paola Riccardi-Castagnoli 1 , Michael Poidinger 1 , Melanie Greter 2 , Florent Ginhoux 1 & Evan W Newell 1 Advances in cell-fate mapping have revealed the complexity in phenotype, ontogeny and tissue distribution of the mammalian myeloid system. To capture this phenotypic diversity, we developed a 38-antibody panel for mass cytometry and used dimensionality reduction with machine learning–aided cluster analysis to build a composite of murine (mouse) myeloid cells in the steady state across lymphoid and nonlymphoid tissues. In addition to identifying all previously described myeloid populations, higher-order analysis allowed objective delineation of otherwise ambiguous subsets, including monocyte-macrophage intermediates and an array of granulocyte variants. Using mice that cannot sense granulocyte macrophage–colony stimulating factor GM-CSF (Csf2rb -/- ), which have discrete alterations in myeloid development, we confirmed differences in barrier tissue dendritic cells, lung macrophages and eosinophils. The methodology further identified variations in the monocyte and innate lymphoid cell compartment that were unexpected, which confirmed that this approach is a powerful tool for unambiguous and unbiased characterization of the myeloid system.