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.
Subscribe to Journal
Get full journal access for 1 year
only $17.42 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Rent or Buy article
Get time limited or full article access on ReadCube.
All prices are NET prices.
Geissmann, F. et al. Development of monocytes, macrophages, and dendritic cells. Science 327, 656–661 (2010).
Hildner, K. et al. Batf3 deficiency reveals a critical role for CD8α+ dendritic cells in cytotoxic T cell immunity. Science 322, 1097–1100 (2008).
Edelson, B.T. et al. Peripheral CD103+ dendritic cells form a unified subset developmentally related to CD8α+ conventional dendritic cells. J. Exp. Med. 207, 823–836 (2010).
Chitu, V. & Stanley, E.R. Colony-stimulating factor-1 in immunity and inflammation. Curr. Opin. Immunol. 18, 39–48 (2006).
Ginhoux, F. & Jung, S. Monocytes and macrophages: developmental pathways and tissue homeostasis. Nat. Rev. Immunol. 14, 392–404 (2014).
Merad, M., Sathe, P., Helft, J., Miller, J. & Mortha, A. The dendritic cell lineage: ontogeny and function of dendritic cells and their subsets in the steady state and the inflamed setting. Annu. Rev. Immunol. 31, 563–604 (2013).
Schulz, C. et al. A lineage of myeloid cells independent of Myb and hematopoietic stem cells. Science 336, 86–90 (2012).
Hashimoto, D., Miller, J. & Merad, M. Dendritic cell and macrophage heterogeneity in vivo. Immunity 35, 323–335 (2011).
Bogunovic, M. et al. Origin of the lamina propria dendritic cell network. Immunity 31, 513–525 (2009).
Yona, S. et al. Fate mapping reveals origins and dynamics of monocytes and tissue macrophages under homeostasis. Immunity 38, 79–91 (2013).
Zigmond, E. et al. Ly6Chi monocytes in the inflamed colon give rise to proinflammatory effector cells and migratory antigen-presenting cells. Immunity 37, 1076–1090 (2012).
Varol, C. et al. Intestinal lamina propria dendritic cell subsets have different origin and functions. Immunity 31, 502–512 (2009).
Serbina, N.V., Salazar-Mather, T.P., Biron, C.A., Kuziel, W.A. & Pamer, E.G. TNF/iNOS-producing dendritic cells mediate innate immune defense against bacterial infection. Immunity 19, 59–70 (2003).
Filipazzi, P., Huber, V. & Rivoltini, L. Phenotype, function and clinical implications of myeloid-derived suppressor cells in cancer patients. Cancer Immunol. Immunother. 61, 255–263 (2012).
Gabrilovich, D.I. & Nagaraj, S. Myeloid-derived suppressor cells as regulators of the immune system. Nat. Rev. Immunol. 9, 162–174 (2009).
Plantinga, M. et al. Conventional and monocyte-derived CD11b+ dendritic cells initiate and maintain T helper 2 cell-mediated immunity to house dust mite allergen. Immunity 38, 322–335 (2013).
Schlitzer, A. et al. IRF4 transcription factor-dependent CD11b+ dendritic cells in human and mouse control mucosal IL-17 cytokine responses. Immunity 38, 970–983 (2013).
Tamoutounour, S. et al. Origins and functional specialization of macrophages and of conventional and monocyte-derived dendritic cells in mouse skin. Immunity 39, 925–938 (2013).
Ornatsky, O., Baranov, V.I., Bandura, D.R., Tanner, S.D. & Dick, J. Multiple cellular antigen detection by ICP-MS. J. Immunol. Methods 308, 68–76 (2006).
Newell, E.W., Sigal, N., Bendall, S.C., Nolan, G.P. & Davis, M.M. Cytometry by time-of-flight shows combinatorial cytokine expression and virus-specific cell niches within a continuum of CD8+ T cell phenotypes. Immunity 36, 142–152 (2012).
Bendall, S.C. et al. Single-cell mass cytometry of differential immune and drug responses across a human hematopoietic continuum. Science 332, 687–696 (2011).
Amir, E.A. et al. viSNE enables visualization of high dimensional single-cell data and reveals phenotypic heterogeneity of leukemia. Nat. Biotechnol. 31, 545–552 (2013).
Shekhar, K., Brodin, P., Davis, M.M. & Chakraborty, A.K. Automatic classification of cellular expression by nonlinear stochastic embedding (ACCENSE). Proc. Natl. Acad. Sci. USA 111, 202–207 (2014).
Van der Maaten, L.& Hinton, G. Visualizing high-dimensional data using t-SNE. J. Mach. Learn. Res. 9, 2579–2605 (2008).
Bodenmiller, B. et al. Multiplexed mass cytometry profiling of cellular states perturbed by small-molecule regulators. Nat. Biotechnol. 30, 856–867 (2012).
Greter, M. et al. GM-CSF controls nonlymphoid tissue dendritic cell homeostasis but is dispensable for the differentiation of inflammatory dendritic cells. Immunity 36, 1031–1046 (2012).
Satake, S. et al. C/EBPβ is involved in the amplification of early granulocyte precursors during candidemia-induced “emergency” granulopoiesis. J. Immunol. 189, 4546–4555 (2012).
Fiedler, K. & Brunner, C. The role of transcription factors in the guidance of granulopoiesis. American journal of blood research 2, 57–65 (2012).
Theilgaard-Mönch, K. et al. The transcriptional program of terminal granulocytic differentiation. Blood 105, 1785–1796 (2005).
Throsby, M., Herbelin, A., Pleau, J.M. & Dardenne, M. CD11c+ eosinophils in the murine thymus: developmental regulation and recruitment upon MHC class I-restricted thymocyte deletion. J. Immunol. 165, 1965–1975 (2000).
Tenenbaum, J.B., de Silva, V. & Langford, J.C. A global geometric framework for nonlinear dimensionality reduction. Science 290, 2319–2323 (2000).
Codarri, L. et al. RORγt drives production of the cytokine GM-CSF in helper T cells, which is essential for the effector phase of autoimmune neuroinflammation. Nat. Immunol. 12, 560–567 (2011).
Codarri, L., Greter, M. & Becher, B. Communication between pathogenic T cells and myeloid cells in neuroinflammatory disease. Trends Immunol. 34, 114–119 (2013).
Hamilton, J.A. Colony-stimulating factors in inflammation and autoimmunity. Nat. Rev. Immunol. 8, 533–544 (2008).
Guilliams, M. et al. Alveolar macrophages develop from fetal monocytes that differentiate into long-lived cells in the first week of life via GM-CSF. J. Exp. Med. 210, 1977–1992 (2013).
King, I.L., Kroenke, M.A. & Segal, B.M. GM-CSF-dependent, CD103+ dermal dendritic cells play a critical role in Th effector cell differentiation after subcutaneous immunization. J. Exp. Med. 207, 953–961 (2010).
Nishinakamura, R. et al. Mice deficient for the IL-3/GM-CSF/IL-5 βc receptor exhibit lung pathology and impaired immune response, while βIL3 receptor-deficient mice are normal. Immunity 2, 211–222 (1995).
Gautier, E.L. et al. Gene-expression profiles and transcriptional regulatory pathways that underlie the identity and diversity of mouse tissue macrophages. Nat. Immunol. 13, 1118–1128 (2012).
Wynn, T.A., Chawla, A. & Pollard, J.W. Macrophage biology in development, homeostasis and disease. Nature 496, 445–455 (2013).
Spits, H. et al. Innate lymphoid cells—a proposal for uniform nomenclature. Nat. Rev. Immunol. 13, 145–149 (2013).
Ginhoux, F. et al. The origin and development of nonlymphoid tissue CD103+ DCs. J. Exp. Med. 206, 3115–3130 (2009).
Schlitzer, A. & Ginhoux, F. Organization of the mouse and human DC network. Curr. Opin. Immunol. 26, 90–99 (2014).
Jakubzick, C. et al. Minimal differentiation of classical monocytes as they survey steady-state tissues and transport antigen to lymph nodes. Immunity 39, 599–610 (2013).
Newell, E.W. et al. Combinatorial tetramer staining and mass cytometry analysis facilitate T-cell epitope mapping and characterization. Nat. Biotechnol. 31, 623–629 (2013).
Fienberg, H.G., Simonds, E.F., Fantl, W.J., Nolan, G.P. & Bodenmiller, B. A platinum-based covalent viability reagent for single-cell mass cytometry. Cytometry A 81, 467–475 (2012).
Parks, D.R., Roederer, M. & Moore, W.A. A new “Logicle” display method avoids deceptive effects of logarithmic scaling for low signals and compensated data. Cytometry A 69A, 541–551 (2006).
The authors thank the SIgN community, the SIgN Flow Cytometry Facility and members of E.W.N.'s, F.G.'s and B.B.'s labs, and L.G. Ng for helpful discussion. Some antibodies used for generating the mass cytometry panels were provided by R. Balderas and A. Tiong (Becton Dickinson). The antibodies generated by BioXcell were provided for this analysis free of charge. B.B. performed this work while on sabbatical at A*STAR/SIgN. Supported by A*STAR/SIgN (P.R.-C., F.G., M.P., E.W.N.) and the Swiss National Science Foundation (PP03P3_144781 (M.G.), 316030_150768, 310030_146130 and CRSII3_136203 (B.B.)), European Union FP7 project TargetBraIn, NeuroKine, Advanced T-cell Engineered for Cancer Therapy (ATECT) and the University Research Priority Project 'Translational Cancer Research' (B.B.).
The authors declare no competing financial interests.
Integrated supplementary information
To illustrate the functionality of each antibody probe, representative plots illustrating the expected staining profiles of each antibody are shown. Cells gated as shown (annotated above each plot) are represented as two parameter dot plots.
This flow chart describes how tSNE transformed data are clustered using the DensVM approach. Using the first two dimensions of tSNE map coordinates, a peak detection algorithm is applied that results in some cells assigned to clusters as previously described (see methods). These classified events together with their 38-antibody expression profiles are used to train Support Vector Machine (SVM) algorithm. Using remaining unclassified cells are then classified based on the SVM leaving an output with all cells classified.
Supplementary Figure 3 tSNE analysis of an independently collected data set comprising four mice and six tissues (spleen, liver, mesenteric lymph nodes, lung, bone marrow, brain).
Two C57Bl6 wild-type mice and two Csf2rb−/− mice were analyzed as described for the main dataset resulting in a new two-dimensional tSNE map of all myeloid cells derived from all tissues of these four mice. (a, b) As in Figure 2 manual traditional gating was applied to cell events to annotate the major clusters of cells derived from (a) lung and (b) bone marrow. For cluster 6-, 8- and 28-like cells, manual gating as illustrated in Figure 3a was used to annotate these cells. Note that the dominant clusters of cells display similar arrangements as in Fig. 2. (c) As in Figure 6a, the distributions and composition of lung cells are compared between two C57Bl6 wildtype mice and two Csf2rb−/− mice. Arrows point to clusters of cells (corresponding to alveolar macrophages in magenta, cluster 6-like eosinophils in orange, and cluster 28-like innate lymphoid cells in black) where reductions in frequencies were observed in Csf2rb−/− mice as described and further validated in Figure 6.
Supplementary Figure 4 tSNE analysis objectively delineates myeloid cell subsets of lymph node, thymus and liver.
As in figure 2, cells are plotted as tSNE composite dimensions 1 vs. 2. In the left panels, cells are colored according to traditional/biased definitions using gating strategies similar to that in Figure 1. gray color indicates cells that are remaining and unaccounted for by these definitions and predominant clusters are indicated with arrows. Cells in the right panels are colored according to their unbiased cluster designation (see results). Notable clusters including those that correspond to subsets not accounted for by traditional gating are labeled with cluster numbers. Frequencies as percentage of total CD45+CD90-CD19-CD3- are shown for each subset.
Supplementary Figure 5 Use of fluorescence flow cytometry–based cell sorting to analyze the morphology of ambiguous cell subsets delineated by higher-order analysis.
(a) For flow cytometric cell sorting of cells from lung tissue, matching characteristics of cluster 8 (eosinophils), cluster 6 (Ly6C- monocytes) and cluster 28 (NK cells) were gated and sorted at shown. Representative Geimsa stained images for each are shown in Figure 3. (b) For flow cytometric cell sorting of bone marrow, cells representative of cluster 4 (Ly6G+ neutrophils), cluster 11 and 12 (Ly6C+ monocytes, clusters 11 and 12 not distinguished by these fluorescent markers), and cluster 5 (Ly6G-Ly6G/C+Ly6Cint granulocyte precursors) were gated as shown. Clusters 5 and 11-12 differ in the expression of CD48 as shown in the histogram on the right. Representative Giemsa stained cells are shown for cluster 11-12 (Ly6C+ monocytes). Images of clusters 4 and 5 (neutrophils and granulocyte precursors) in are shown in Figure 4. (c) For flow cytometric cell sorting of thymus, cells representative of clusters 18-21 (DCs), cluster 11-12 (Ly6C+ monocytes) and cluster 7 (CD11c+ eosinophils) were gated as shown, sorted, mounted and subsequently Giemsa stained. Representative cells from cluster 18-21 (DCs) and clusters 11-12 (Ly6C+ monocytes) are shown. A representative cell from cluster 7 (eosinophil) is shown in Figure 3. (d) Additional fluorescence flow cytometry analysis of a thymus-derived ambiguous cell subset. Representative plots used to identify cluster 11-12 (Ly6C+ monocytes, red) and cluster 7 (CD11c+ eosinophils, green) are shown. Parallel staining of Siglec-F is also shown. Histograms of Siglec-F staining cells derived from red and green gates are also shown, which demonstrate strong Siglec-F staining on cells similar to those identified as cluster 7.
Supplementary Figure 6 tSNE analysis used to compare cellular composition of tissues derived from individual Csf2rb−/− and Csf2rb+/+ mice.
(a-c) As in figure 6a, bone marrow-, spleen-, and liver-derived cells color-coded by DensVM cluster number are plotted by tSNE scores. For each panel, the three upper plots represent three replicate C57Bl6 wild-type tissue cells and the three lower plots represent three Csf2rb−/− tissue cells run in parallel.
(a) Standard biaxial plots to were used to delineate lung alveolar macrophages (CD11c+MHC II-/low) and their quantities were compared between Csf2rb−/− vs. Csf2rb+/+ mice (Average ± S.E.M. frequencies are provided, n=3 mice each). (b) Based on the gate in (a) to exclude AM and DCs, frequencies of Ly6C-Ly6G- (non-monocyte, non-neutrophils) were quantified. These cells are composed of both eosinophils (cluster 6) and NK cells (cluster 28), both found to be reduced in Csf2rb−/− mice and here their frequencies are compared between Csf2rb−/− vs. Csf2rb+/+ mice (Average ± S.E.M. frequencies are provided, n=3 mice each).
Supplementary Figure 8 Comparison of granulocyte composition in various tissues derived from Csf2rb−/− and Csf2rb+/+ mice.
(a) As in figure 4, the five different populations of neutrophil-like cells and two different populations of eosinophil-like cells were used for ISOMAP dimensionality reduction analysis to compare their overall phenotypic relatedness. For each cell cluster (from Csf2rb+/+ tissues, left and Csf2rb−/− tissues, right), the median values in ISOMAP dimension 1 are plotted against the relative frequencies of each cluster for each tissue (expressed as percentage of total granulocytes for each tissue).
Supplementary Figure 9 Florescence flow cytometric gating strategy to quantify and validate changes in composition observed in tissues from Csf2rb−/− (bottom) and Csf2rb+/+ mice (top).
(a) Representative plots are shown as they were used to identify cells corresponding to cluster 16 (MHCIIlo/CD11c+: AM), cluster 28 was interrogated by further gating on CD45+MHCII−Ly6Clow/–Ly6G−CD11blow (NK cells) and clusters 6 and 8 were further gated on by F4/80+CD11b+ and specifically interrogated for CD11c (Cluster 8: Ly6Clow monocytes) and CD24 (Cluster 6: eosinophils). The same gating strategy was used to analyze (b) liver and (c) spleen. In panels b and c, for cluster 28, we further excluded more abundant CD3+ and B220+ cells.
Supplementary Figure 10 Comparison of DC composition in lung tissue derived from Csf2rb−/− and Csf2rb+/+ mice.
(a) To compare CD103 expression in the DC compartment of Csf2rb−/− vs. Csf2rb+/+ CD11c+MHCII+ cells are shown. Note, reduction of CD103 expression on CD11b- cells in Csf2rb-/- lung tissue. (b) Gating strategy to identify CD11b−/CD24+ (also referred to as CD103+ DCs red), CD11b+CD24+CD301+ DCs (green) and CD11b+CD24−CD301− interstitial macrophages (orange). (c) To specifically focus on DC phenotypic profiles, tSNE analysis was repeated using CD11c+MHCII+ cells derived from spleen, liver, bone marrow and lung. From this analysis two new tSNE dimensions were used to segregate lung DCs with better resolution. To annotate this representation, cells gated in (a) were color coded and overlaid on tSNE contour plots (CD11b-CD24+ DCs in red, CD11b+CD24-CD301+ DCs in green and CD11b+CD24-CD301- interstitial macrophages in orange). Note the prominence of the two DC subsets in Csf2rb+/+ mostly lost in Csf2rb−/− replaced by a prominent population of interstitial macrophage-like cells and an additional unknown population. (d) To compare the phenotypes of the four cell populations defined by zoomed-in tSNE analysis in (c), the median intensities of each marker for each population are summarized as a heat-plot.
About this article
Cite this article
Becher, B., Schlitzer, A., Chen, J. et al. High-dimensional analysis of the murine myeloid cell system. Nat Immunol 15, 1181–1189 (2014). https://doi.org/10.1038/ni.3006
American Journal of Respiratory and Critical Care Medicine (2021)
IEEE Transactions on Visualization and Computer Graphics (2020)
Host genetics and tumor environment determine the functional impact of neutrophils in mouse tumor models
Journal for ImmunoTherapy of Cancer (2020)
Frontiers in Immunology (2020)
Löfgren's syndrome sarcoidosis and Non-LS sarcoidosis prediction using 1d-Convolutional neural networks
Informatics in Medicine Unlocked (2020)