Flow Cytometry, Amped Up

See allHide authors and affiliations

Science  06 May 2011:
Vol. 332, Issue 6030, pp. 677-678
DOI: 10.1126/science.1206351

In multicellular organisms, cells carry out a diverse array of complex, specialized functions. This specialization occurs mostly through the expression of cell type—specific genes and proteins that generate the appropriate structures and molecular networks. A central challenge in the biomedical sciences, however, has been to identify the distinct lineages and phenotypes of the specialized cells in organ systems, and track their molecular evolution during differentiation. On page 687 of this issue, Bendall et al. (1) offer a brilliant proof of principle for a novel technology—mass cytometry—and provide a uniquely detailed view of cell differentiation in the human hematopoietic system. They used this technology to simultaneously examine 34 attributes of human bone marrow cells and then create a superimposed map showing the complex interactions of cell signaling molecules, all at an unprecedented level of resolution. This opens a new chapter in single-cell biology.

Since the 1970s, fluorescence-based flow cytometry has been the leading technique for studying and sorting cell populations (2). It involves passing cells through flow chambers at high rates (>20,000 cells/s) and using lasers to excite fluorescent tags (“fluorochromes”) that are usually attached to antibodies; different antibodies are tagged with different colors, enabling researchers to quantify molecules that define cell subtypes or reflect activation of specific pathways. Progress in instrument design, multilaser combinations, and novel fluorochromes has led to experimental configurations that simultaneously measure up to 15 markers. This has enabled very detailed description of cell subtypes, perhaps most extensively in the immune system, where the Immunological Genome Project is profiling >200 distinct cell types. Fluorescence cytometry seems to have reached a technical plateau, however: In practice, researchers typically measure only 6 to 10 cell markers because they are limited by the spectral overlap between fluorochromes (see the figure).

To escape this plateau, a group led by Scott Tanner of the University of Toronto in Canada devised the radically new approach of mass cytometry (CyTOF) (3), and teamed with a group led by Garry Nolan of Stanford University in California, who has long been a leader in developing higher-dimensional multicolor flow cytometry, especially for studying intracellular signaling components (4). In mass cytometry, fluorochrome tags are replaced by a series of rare earth elements (e.g., lanthanides), which are attached to antibodies through metal-chelator coupling reagents. Cells are labeled by incubation in a cocktail of tagged antibodies; as the cells flow through the instrument, they are vaporized at 5500 K, and the released tags are identified and quantified by time-of-flight mass spectrometry (MS). Rates are reasonable, at 1000 cells/s. The beauty of the approach stems from three factors: the precision of MS detection, which eliminates overlap between tags (a dream for any investigator who has battled this problem, known as fluorescence compensation); the number of detectable markers (34 here, but easily more); and the absence of background noise (because rare earth elements are essentially absent from biological materials, there is no equivalent of “autofluorescence”). Because the software tools commonly used for flow cytometry data would be woefully inadequate for analyzing dozens of dimensions, Bendall et al. used software that clusters cell populations into “minimum-spanning trees” that reproduce known hematopoietic differentiation, but with much finer granularity. As a result, cells that once would have been grouped into one population are now divided into many more; for example, naïve CD4+ T lymphocytes, a priori considered a homogeneous population, are now split into more than 10 subsets.

Fluorescence versus mass cytometry.

Mass-tagging of antibodies allows for unfettered resolution of more labels (bottom, left) when compared with conventional detection of fluorescent antibody tags (top, left), which is hampered by overlap between emission spectra. Mass cytometry thus increases the discrimination power for cell subset analysis (top and bottom, center) and allows a far more comprehensive perspective on intracellular signaling pathways (top and bottom, right). m/z, mass/charge ratio.


Like any new technology, mass cytometry has limitations. Most obviously, cells vaporized in a CyTOF cannot be recovered for further analysis or growth, as with conventional flow cytometry. A combination of techniques may evolve, with researchers using mass cytometry for broad initial analyses, from which they will define a minimal set of markers for fluorescence-based sorting. In addition, the procurement of labeled reagents will undoubtedly be a hurdle. Assembling and testing a set of labeled antibodies for a 35-plex experiment is no small feat. Unlike nucleic acid probes, which have reasonably uniform properties and can be synthesized cheaply in massively parallel formats, monoclonal antibodies are expensive, and each has a different affinity, stability, and resistance to conjugation chemistries. The range of antibodies that bind to specific proteins also is limited, in particular for their posttranslational, modified forms. Hopefully, researchers and commercial suppliers, with support from funding agencies, will tackle the challenge of developing panels of reagents. Standardized sets of conjugated antibodies that bind with surface molecules and intracellular signaling intermediates (across many known pathways) would be invaluable for studies of human inflammatory diseases.

Why is this technology a game-changer? One might argue that increasing from 15 to 35 simultaneous parameters is not in itself a dramatic conceptual leap. It is conceivable that the fine parsing of lymphocyte subsets by Bendall et al. could have been achieved by iterative application of conventional flow cytometry. Yet there is no doubt that the broader perspective given by this single CyTOF experiment greatly accelerated this parsing. It will be important, however, to determine whether the finer and finer cell subsets identified by mass cytometry merely represent an artifact of the clustering algorithm, or are both “real” (reproducible) and have functionally relevant differences. Although the physiological functions of different cell subsets may take a long time to work out, CyTOF can be used to test whether signaling pathways behave differently in each subset. After placing cells in different contexts (e.g., a range of stimuli), CyTOF can measure tens of signaling intermediates and determine if each cell subset uses subtly distinct cascades of signaling events. For example, differences in NF-κB activation were detected by Bendall et al. Quantitation of RNA or other molecular species, which should also be possible by mass spectrometry, may further help to anchor these subphenotypes.

It is perhaps in the emerging field of “single-cell biology” that CyTOF may make a unique contribution to dissecting intracellular networks. Much of biochemistry and molecular biology rests on the assumption that the behavior of cells in a population (in a culture, in an organ) can be reliably approximated by the population average that results when cells are lysed and their molecules analyzed as a pool. Increasingly, however, investigators realize that stochastic fluctuations in gene or protein expression, between cells of an otherwise identical group, can lead to major differences in their behavior. This “noise ” in gene expression (5, 6) can have profound consequences for cell differentiation (7), the response to apoptosis-inducing stimuli (8), or T lymphocyte triggering at the initiation of immune responses (9).

Mass cytometry can help researchers both take this stochasticity into account, and benefit from it. The secret is that, in flow cytometry, every cell acts as an independent “test tube.” The ability to generate millions of independent datapoints, simultaneously measuring the activation of many signaling nodes in each cell, together with expression levels of key sensors or controlling factors, will enable researchers to infer connections within these biochemical networks (10).

It is easy to see how this technology will be used to analyze disease states or an individual's response to therapeutics. A problem in disease exploration is often to study the “correct” cell subtype or pathway. Most often, RNA or protein is analyzed from pooled cells, obscuring disease-related signals that might show up in specific cells or pathways. Mass cytometry is poised to revolutionize our studies of disorders in the human immune system by probing multiple critical parameters in parallel, across a broad range of cells and pathways.


View Abstract

Stay Connected to Science

Navigate This Article