UMass owns three Becton-Dickinson FCM instruments, (i) one called FACScan or the analyzer because it analyses only (does not sort); (ii) the FACStar Plus which analyses and sorts at high speed; and (iii) the FACSCalibur which analyses and sorts at low speed. The first two were purchased in 1988; the FACSCalibur was purchased in 1998 and is the most heavily used. Amherst College owns an analyzer from the Coulter Corporation, an Epics Profile.
Monodisperse cells. The cells must be in a single cell suspension with minimal aggregation (a "monodisperse suspension"). White blood cells are popular because they come monodisperse. Cells from solid tissues, such as a cancer biopsy, can be used if they can be dissociated and dispersed without breaking the cells.
Tagging. The monodisperse cells are labeled with a fluorescent tag. Typically, a monoclonal antibody to a cell surface glycoprotein receptor is used. The fluorescence can be direct or indirect. For direct fluorescence, fluorescein (or another fluorochrome) must be covalently attached to the primary monoclonal antibody. For indirect fluorescence, the primary receptor-specific antibody need not be fluorescent. It is bound to the cells first, and the excess rinsed away. Next, a secondary fluorescein-labeled anti-antibody is bound to the cell-bound primary antibody, and the excess rinsed away. The anti-antibody is purified from the serum of a rabbit or goat immunized with mouse antibody. If the primary antibody is a monoclonal mouse antibody, the secondary anti-antibody could be a rabbit anti- mouse antibody.
Data can be acquired in the flow cytofluorometer with the cells either alive, or fixed with paraformaldehyde. With fixed cells, the fluorescence and light scatter properties are stable, and data acquisition can be postponed for days or weeks.
In addition to surface receptors, intracellular components can be stained in fixed cells, such as DNA with propidium iodide. Quantitation of the total DNA per cell is very useful in analyzing the progress of cells through the cell cycle (G1 -> S -> G2 -> M). Fluorescent sequence-specific polynucleotide probes can be hybridized in situ to quantitate for example, the amount of mRNA for a particular gene. Fluorescent reporters are available for a wide variety of intracellular components, such as cytoplasmic [Ca++], membrane potential, cytoplasmic pH, or actin. Changes in Ca++ induced by a stimulus can be measured second-by-second in living cells.
In advanced procedures, it is possible to label cells with two, three, or even four different fluorochromes simultaneously (each fluorescing a different color) and quantitate a distinct fluorescence intensity for each.
Sheath Fluid and Fluidic Focusing. A suspension of tagged single cells (no clumps, please!) is fed into the instrument. While flowing rapidly through channels in the instrument, a small amount of cell suspension joins a larger amount of cell-free buffer (called "sheath fluid"). The fluids come together under laminar flow conditions so that they flow together evenly without mixing. The "sheath" fluid surrounds a thin core thread of sample. This spaces the cells out so that only one passes the laser beam at a time, and keeps the cells centered in the flowing stream so that they pass the laser beam optimally centered.
The FACSCalibur and FACScan (see terminology) pass the sample stream through a glass cuvette where it intersects the laser beam, then discard it into a disinfectant-containing waste tank, or into sort-receiving test tubes. This design eliminates much of the complexity of the fast sorter (see below), making the FACSCalibur and FACScan much easier and quicker to operate than the high-speed sorter. Analysis of potentially biohazardous samples is possible since the sample is contained at all times and can be disinfected easily.
The fast sorter (FACStar) shoots the sample stream through a nozzle (typically 70 micrometers in diameter) into the air, where the laser beam intersects the stream. This is required for high-speed sorting (see below), but is the Achilles heel of the instrument. Most technical problems result from debris clogging or sticking in the nozzle, causing the stream to deflect and become misaligned with the optics. Also, the resulting aerosolization of the sample prevents biohazardous samples (e.g. human blood cells potentially infected with HIV or hepatitis virus) from being sorted unless stringent precautions are taken.
Detection. The cells flow single-file past a laser beam (or in some more complicated instruments, two laser beams). The momentary pulse of fluorescence emitted as the cell crosses the beam is measured by photomultipliers at a 90 degree angle from the beam. Typically, 2-3 detectors are used with different wavelength bandpass filters, allowing the simultaneous detection of emissions at different wavelengths from different fluorochromes in a single cell.
In addition to fluorescence, two types of light scatter are measured. Low-angle forward scatter (often called simply "forward scatter") is roughly proportional to the diameter of the cell. Orthogonal, 90o or "side scatter" is proportional to the granularity: neutrophil granulocytes have higher side scatter than do lymphocytes, which are agranular. Thus, in the FACScan, each cell can provide up to five numbers: size, granularity, plus green, red, and far red fluorescence intensities.
Gating. Usually we want to see data only from single, viable cells. Typically one wishes to eliminate data from cell debris (particles smaller than cells), dead cells, and clumps of 2 or more cells. Subcellular debris and clumps can be distinguished from single cells by size (estimated by the intensity of low angle forward scatter). Dead cells have lower forward-scatter and higher side-scatter than living cells. These differences are accurately preserved following paraformaldehyde fixation (despite the fact that after fixation, all the cells are dead!).
The computer can be configured to display the fluorescence signals only from those particles with a specified set of scatter properties, namely, living single cells. This is called a scatter-gated fluorescence analysis. Actually, it is possible to "gate" on any set of signals. In some cases, it may be desirable to gate on a combination of fluorescence and scatter values. For example, the DNA of dead cells can be fluorescently stained under conditions which prevent staining of living cells. Then data from dead cells can be eliminated by gating out brightly-fluorescent cells.
|Fig. 2. Scatter dot plots distinguish single viable cells. Three dot plots of cultured mouse lymphocytes. Each dot represents a single cell; its position indicates its forward scatter (FSC) intensity value (cell size), and its side scatter (SSC) intensity value (cell granularity). Note that fluorescence is not required to distinguish living from dead cells, or aggregates from single cells.|
Histograms and dot plots. FCM data are most often represented as dot plots or histograms (see examples below). Basically, histograms quantitate intensities (of scatter or fluorescence) one parameter per histogram. In contrast dot plots quantitate percentages of cells with various properties. (In fact, dot plots also show intensities.) Both histograms and dot plots reveal whether there are discrete subpopulations of cells with different intensities. A two-peak histogram means there are dim cells and bright cells. Histograms reveal subpopulations for a single parameter (scatter or fluorescence), while dot plots show the relations of subpopulations for two parameters.
|Fig. 3. Scatter gating eliminates 'impossible' two-color cells. In these data from the 1994 immunology class, spleen cells were stained for MHC class II (green) and CD4 (red). No cell should express both markers, but the dots in region 3 (R3) appear to be double-stainers. These "impossible" cells were an artifact due to clumping (red cells stuck to green cells, appearing as 'double-stained cells'). Scatter gating eliminated the artifactual 'cells'.|
In a dot plot, each cell is represented by a dot, positioned on the X and Y scales according to the intensities detected for that cell. Scatter dot plots (X = forward scatter intensity; Y = side scatter intensity) are often informative (see examples below). Scatter scales are usually linear. Fluorescence dot plots typically plot X = green fluorescence intensity, Y = red fluorescence intensity. These two-color dot plots are often divided into four quadrants, the double negative cells, the green-only, red-only, and double positive cells. These are quantitated by giving the percentage of cells in each quadrant. Since fluorescence intensity often varies several orders of magnitude between cells, the scales are usually the logarithm of fluorescence intensity spanning four decades (a 10,000-fold range). See examples below.
In a histogram, the X axis is intensity (of scatter or fluorescence), and the Y axis shows how many cells had each intensity. Thus, histograms show the distribution of intensities for a single parameter, while dot plots show the correlated distribution for two parameters. The density of dots in a region of a dot plot shows the "number of cells", equivalent to the Y axis of a histogram. Indeed, dot plots are sometimes represented as pseudo-3D graphs where the Z axis is "number of cells".
|Fig. 4. The median is the most robust estimate of the intensity of a peak on an FCM histogram.|
The "middle" of a peak on a histogram quantitates the intensity (scatter or fluorescence) of the population of cells in that peak. However, an arithmetic average is a poor way to estimate the "middle" because it is too sensitive to the tail of the distribution. A few cells with extremely high intensities can pull the average well above the "middle" of a peak. The highest point on the histogram is the "mode". Often the mode is at the "middle" of a peak, but sometimes the highest point is sensitive to sampling error. More seriously, sometimes the mode can be away from the peak entirely, when there is a pileup of off-scale cells at either end of the histogram. Therefore the median is preferred as the most robust estimate of intensity characterizing a population of cells.
Autofluorescence. Even when no fluorescent tag or stain is added to cells, they fluoresce. This autofluorence comes from normal cell components which fluoresce, such as riboflavin and flavoproteins. To estimate the fluorescence intensity due to a tag, one must subtract the autofluorescence from the total fluorescence intensity measured. In order to quantitate autofluorescence, its value must be "on-scale", since if it is allowed to fall below the minimum scale value, it cannot be quantitated.
|Fig. 5. Scatter-gating eliminates artifacts from dead cells. Green fluorescence (FL1) histograms for ungated and scatter-gated cells stained with fluorescein-conjugated antibody to the T cell receptor (see text). Before gating (solid line), 4% of the cells were in a third, separate, bright peak (at the right). After gating out dead cells, this bright peak disappeared, showing that it represented nonspecific staining of dead cells. Only the two major peaks remain after gating; these represent two subpopulations of single viable cells with negative (left) and positive (right) staining.|
Analysis. Operation of the Becton-Dickinson FACSCalibur or FACScan very simple when only analysis is desired (without sorting). Calibration (setting of gains and cross-channel corrections) may be completely automated with a computer controlled process that uses a mixture of three types of cell-sized plastic beads: non-fluorescent, green, and red. These instruments are popular both for research and clinical diagnosis because they can inexpensively and quickly analyze a large number of samples (30-50 samples/hour). Typically, data are collected for 10,000 cells/sample, and the results are available as various computer-generated graphs as shown below.
Sorting. Instruments capable of sorting are much more complicated than are analysis-only intruments. The FACSCalibur uses a fluidic valve mechanism to sort cells. This mechanism is limited to a throughput of 300 cells/second. Hence, it is most useful when a few parent cells are needed to start clones. In order to process cells ten times faster (about 3,000 cells/second), the high-speed FACStar vibrates the sapphire nozzle at ultrasonic frequency, which causes the stream (in air, after passing the laser beam) to break up into droplets with constant size and spacing. Typically, cells are present in less than 1/3 of the drops. The circuitry analyzes the signals from each cell, and then charges the stream with the correct time delay so that as the droplet containing that cell breaks off the stream, it is properly charged. Charging is positive, neutral, or negative. The droplets then pass between plates which set up an electrostatic field. This directs each charged droplet to the right or the left; neutral droplets (typically rejected droplets such as dead cells, cell debris, double cells, etc.) fall into the middle waste collector. The operation and maintenance of the high-speed sorter are much more complicated than for the closed-flow instruments. While the experimenter can learn to use the latter for analysis in a single afternoon, a high-speed sorter is typically operated by a specially trained, professional technician, not by the experimenter.
Analysis. If one is studying a uniform (homogeneous) population of cells (such as a cloned cell line grown in culture) and needs only population median values for receptor densities, then FCM is not necessary. ELISA or RIA methods are equally quick and easy, less expensive, and allow processing of even larger numbers of samples/day than does FCM (which can nevertheless handle hundreds/day if necessary). Also, these non-FCM methods give you the single value you want, the average density. In contrast, FCM tends to overwhelm you with detailed data about individual cells and distributions which you may not need. Nevertheless, FCM is often used even for uniform cell populations, since it is guaranteed to reveal non-uniformity, should it occur, and is precise about eliminating dead cells, debris, and clumps from the final median values.
When studying heterogeneous populations of cells the FCM really shines (get it?). In a few minutes, the FCM can acquire data on all subpopulations of a sample. Not only can you tell the percentages of red cells vs. green cells, but the percentages of red-and-green doubly-labeled cells, and of cells which are neither red nor green. Moreover, there may be subpopulations of bright-green and dim-green, etc. No other method allows such rapid, quantitative, and detailed analysis of subpopulations.
Sorting. FCM sorters can purify an extremely small subpopulation (1 cell in 105), or a subpopulation defined by complex criteria of size and fluorescence properties. They have a very low error rate. However, FCM sorting has several disadvantages. The major disadvantage is the rate of throughput, and losses in the yield. For reliable sorting, even on a high-speed sorter, the flow rate cannot exceed a few thousand cells/second. When pairs of cells go by too close together to distinguish, they must both be discarded. For a subpopulation which is 20% of the starting population, the high-speed sorter will yield less than 106 cells/hour. Many experiments require far more cells than this. Running the high-speed sorter all day gets expensive, and may pose problems in keeping the cells in best condition during such a long sort. For a frame of reference, a single mouse can provide more than 108 lymphocytes, or 109 tumor cells.
A second problem arises when the sorted cells need to be kept gnotobiotic ("sterile"). The high-speed sorter can be used in a "sterile" mode, but this increases the complexity of the operation. Keeping the sample clean is much easier in the slow sorter, but it produces one tenth as many cells per hour. Finally there is the cost, which is hundreds of dollars per high-speed sort.
If only a few carefully selected cells are needed to grow up large clones, the FACSCalibur can sort the desired cells quickly and inexpensively while maintaining sterility. However, when large numbers of cells are needed, particularly if sterile, one is well advised to consider whether non-FCM methods would be more satisfactory than a high-speed sort. Most non-FCM methods can handle arbitrarily large numbers of cells, and often with simultaneous parallel purification of multiple samples. Examples are deletion with antibody + complement, panning with monoclonal antibody-coated dishes, affinity columns (which can be used with living cells quite successfully), and density (isopycnic) or sedimentation (isokinetic) gradient separations.
Research/UMass. FCM analysis is widely used in research on basic cellular and molecular mechanisms. Projects at UMass/Amherst have concerned the mechanisms by which white blood cells adhere to and kill infected cells, identification of the receptors used by viruses to infect cells, and the evolution of the genes which allow the immune system to distinguish self from foreign microorganisms. Virus projects have used fluorescent virus to measure the binding of virions to the cell surface. Our sorter has been used to sort slime mold cells expressing green fluorescent protein, bovine gamma-delta T cells, and human cell lines transfected with genes for integrin adhesion molecules. Analyses are done daily, but sorting is done rarely.
Clinical Applications. FCM analysis is widely used in hospitals in clinical research and diagnosis. A hospital or medical center typically has several FCM analyzers which are in constant use. FCM can distinguish subtle changes in the types of cells in the blood, and discriminate many different types of blood cell cancers (leukemias and lymphomas). FCM can quantitate the depletion of T lymphocytes during immunosuppression to prevent tissue graft rejection, or the depletion of the CD4 T-helper cells in AIDS. In cancer cells taken in surgery or biopsy, the distribution of DNA/cell reveals the stage of progression. FCM results are important for diagnosis, prognosis, and optimization of therapy.
||Fig. 6. Scatter dot plot for human leukocytes. Granulocytes have intense side-scatter (E=eosinophils, N=neutrophils). Granulocytes and monocytes are about the same size (forward scatter), but monocytes (M) are less granular. Lymphocytes (L) are smaller and agranular. Dead cells and red blood cells have been gated out. Note that these cell types can be distinguished without fluorescent tags. Clinical instruments used to determine 'complete blood counts' or CBC differentials are specialized flow cytometers which use scatter only.|
||Fig. 7. Calcium response by TCR. T cell receptors were cross-linked with anti-CD3. The human T cell line Jurkat responded with increased calcium concentrations in the cytoplasm, reported with indo-1. "FL4-R" is a ratio of intensities in two fluorescence channels, which is proportional to the cytoplasmic free calcium activity. The right panel shows the median FL4-R vs. time. Data kindly provided by Keith Kelley.|
Students interested in more depth should visit the UMass Flow Cytometry Facility website at http//:www.bio.umass.edu/mcbfacs/flowhome.html. You may also request "Basic Concepts in Flow Cytometry", a document prepared for training research personnel.