【共享】流式 Overview of Flow Cytometry

Flow cytometry, also called flow microfluorometry or flow cytofluorometry, employs instrumentation that scans single cells flowing past excitation sources in a liquid medium. The technology is unique in its ability to provide rapid, quantitative, multiparameter analyses on single living (or dead) cells. Measurement of visible and fluorescent light emission allows quantitation of antigenic, biochemical, and biophysical characteristics of individual cells. Flow cytometry technology can also separate distinct subpopulations of cells on the basis of these measured characteristics, usually by electrostatic deflection. This separation technology is called electronic cell sorting.

Flow cytometry has two general applications—quantitative analysis and cell separation. Cell separation is dependent upon the analytical capabilities of flow cytometry, in that cells of interest must be identified analytically in order to be separated. A graphical representation of flow cytometry analytical measurements made on individual cells is known as a histogram. Data plotted on a histogram include the number of cells and values of one or more measurements made on individual cells.

Performing flow cytometry experiments generally involves three distinct, interdependent phases. First is the pre–flow cytometry phase which involves reagent preparation, cell preparation, protocol design, and staining of cells with the fluorescent reagents. Second is the flow cytometry phase which involves processing the stained cells using flow cytometry instrumentation and collecting data for one or more measurements (parameters) made on each individual cell. Finally, the analysis phase involves analyzing the collected data. These three phases may all be performed by a single individual during the course of one day. Alternatively, each of the three phases may be performed by different persons and at different points in time. For example, one person might stain the cells and deliver them to a specialized flow cytometry facility for processing the next day. The data might then be analyzed by a third individual at a later time. Details on protocols for each of these phases of flow cytometry experiments are found throughout this chapter.

This introductory unit provides an overview of terminology and techniques unique to flow cytometry and is divided into two sections. The first section discusses technical aspects of flow cytometry which apply primarily to the instrumentation-oriented flow cytometry phase. The second section discusses electronic cell separation using flow cytometry. The flow cytometry phase for specific instruments is presented in UNIT 5.4.

UNIT 5.2 provides techniques for analysis of flow cytometry data, with emphasis on the analysis of immunofluorescence data. The pre– flow cytometry phase, including cell preparation and staining as well as fluorochrome-conjugated reagent preparation, is presented in UNIT 5.3. UNITS 5.2-5.4 emphasize cell surface immunofluorescence applications in flow cytometry, while UNIT 21.4 presents an alternative nonfluorescent technique for surface antigen analysis using conventional microscopy.

Additional specific applications of flow cytometry in immunobiology are presented in UNITS 5.5-5.7. UNIT 5.5 (intracellular ions) and UNIT 5.7 (DNA) focus on measurement of intracellular materials using flow cytometry. UNIT 5.6 presents a protocol which uses flow cytometry to measure cell:cell conjugates.

Flow cytometry is extremely diverse in its capacity for analytical measurements. Different levels of decision making, each of which influence the usefulness and quality of the data output, are required to exploit the potential of this technology. The terms explained below define operational decisions necessary to the execution of the flow cytometry phase of the experiments.

Single-Beam Versus Dual-Beam Excitation
Flow cytometers have one or more light sources for scanning and excitation of fluorescent probes attached to cells. Multiple distinct measurements can be taken from emissions resulting from each excitation beam. Each beam is operated at a single- or narrow-band wavelength, restricting its use to probes excitable at or near that wavelength. In addition, fluorescence emissions from different probes excitable at a single wavelength will generally have spectral overlap, requiring optical and electronic methods of separating emissions from different probes. A second excitation beam expands the number of probes that can be used simultaneously and also provides physical and kinetic separation between the signals from the independent excitation beams.

Electronic Compensation
As noted, multiple fluorescence emissions can result from a single-excitation wavelength and are likely to overlap spectrally. Current technology provides, in addition to color filtration, electronic procedures to compensate for this overlap. Electronic compensation involves subtraction of signal measured as fluorescence emission B as a function of the amount of the same signal measured as fluorescence emission A. This factor is predetermined for each experiment (and ideally for each reagent combination) with control samples that have been stained with a single probe. These single-stained samples are used to determine the amount of signal overlap between channels. For example, fluorescein isothiocyanate (FITC)- and R-phycoerythrin (R-PE)-conjugated antibodies are commonly used to analyze two antigenic determinants on the same cell population using a single excitation wavelength (488 nm). With this pair of fluorochromes, FITC is measured as green (530 nm) and R-PE as orange (575 nm) fluorescence. However, a significant amount of orange fluorescence is present in the FITC emission, while a relatively small amount of green fluorescence is present in the R-PE fluorescence emission. A sample of cells stained with the optimal concentration of FITC reagent is run and electronic compensation applied to force the orange fluorescence measurement of this sample to be identical to that of the unstained cells. A similar procedure is followed for a sample stained with the optimal concentration of R-PE reagent (UNIT 5.4). In practice, it is usually easiest to artificially mix unstained cells with FITC-only cells and R-PE-only cells.

A less precise method for selecting compensation values is the use of plastic beads coupled to fluorochromes. In this case, beads containing the same fluorochromes used in the actual experiment must be employed. In order to compensate using fluorochrome-conjugated beads, identical beads without fluorescent dye are required (just as unstained cells were required when using cells to compensate). If background fluorescence of the unstained beads is different from that of unstained cells of the type used in the experiment, or if the beads contain more or less of an individual fluorochrome than the stained cells used in the experiment, compensation values cannot be accurately selected.

It is important to note that appropriate compensation may not be achievable when pairs of reagents with extremes of fluorescence intensity (e.g., very bright green with very dim orange) are used. In such cases, it may not be possible to fully exclude the brighter channel signal in the opposite channel, resulting in false positives and/or artifactual histogram shapes. With proper controls using cells stained with each of the reagents alone, this problem is easily detected. A more insidious problem is caused by overcompensation in the channel used to measure the weaker fluorescence probe. In this case, false negatives and/ or artifactual histogram shapes may result. Since compensation is usually set using the brightest reagents, this problem may occur in experiments using multiple combinations of reagents. It is therefore critical to include controls (cells stained with only one reagent) for all reagent-cell combinations in which a weakly staining reagent has been used.

Detector Voltage
In a flow cytometer, emissions from fluorescent dyes bound to individual cells are detected by photomultiplier tubes (PMT). A PMT generates an electrical signal proportional to the amount of light detected. The amount of voltage used to operate the PMT determines the range of light intensity that can be measured (dynamic range).

On many flow cytometers, adjustment of this operating voltage is the only means of varying the relative sensitivity of the fluorescence measurements. If the voltage used is too low, negative cells will be off the lower end of the scale and low-level positives will not be detected. Conversely, if the voltage used is too high, negative cells will be falsely high on the intensity scale, resulting in a compression of small-intensity differences or resulting in positive cells being off the high end of the scale. Therefore, the ideal PMT voltage is one that allows both positive and negative cells to appear on a histogram (UNIT 5.4).

The proportionality relationship of light measured to electrical signal generated by the PMT is constant only for a range of selected voltages. Information regarding acceptable operating voltages is available from the instrument manufacturers. Fluorescence intensity measurements can only be compared when measurements are made with the same PMT voltage. Thus, it is important to record the voltage used. In addition, all voltage adjustments of the detector should be performed with electronic compensation set to zero; the latter should be set after all adjustments to the former have been made. Finally, matching voltage values as closely as possible between detectors will minimize difficulties in achieving proper electronic compensation.

Triggers and Thresholds
A flow cytometer is capable of analyzing individual cells because it takes measurements over short periods of time (usually 10 to 40 µsec). This requires that the instrument have some method of detecting when a cell (or other discrete particle such as a plastic bead) is at the point of light excitation so that measurement can begin.

One or more measured parameters alert or “trigger” the instrument to accept signals associated with a single particle for a discrete time period. Measurements of light scatter detect the amount of excitation light scattered by a cell in that beam of light. Forward light scatter, which is measured in front of the path of excitation light, is influenced by both the size and refractive index of the cell. Forward light scatter is the trigger parameter most commonly used for flow cytometry experiments involving immunofluorescence (UNIT 5.4).

A signal threshold value is required on the trigger parameter. (The lower limit cannot equal zero.) Therefore, experiments involving heterogeneous cell populations that are defined by fluorescence probes will require light-scatter triggers only. This is because setting a threshold value on fluorescence will result in falsely high frequency determinations due to negative cells which are below threshold and thus never seen. It should be noted that DNA analyses are often performed using fluorescence triggers. If submitting frequency determination experiments to a laboratory that usually performs DNA analysis, selection of trigger parameters should be discussed (see UNIT 5.7 for a discussion of DNA analysis).

Gates/Windows
Gates or windows set the upper and lower intensity values for parameters that define the cells of interest. When a histogram is displayed, lines may be drawn defining these intensity values. On a single-parameter histogram, a vertical line is drawn at the x value selected. On a dual-parameter contour histogram, lines drawn at the x and y values selected will form a box. This box, or gate, defines the area of the histogram or region of interest.

Gates may be defined electronically or by programming and are used to select cells of interest for analysis and cell separation. For example, analysis of cell-surface antigens using fluorochrome-conjugated antibodies is usually performed using forward light scatter and propidium\ iodide fluorescence gates to distinguish live cells from dead cells (UNIT 5.4). Analytical tools provided in the software permit mathematical calculations and display of additional parameters for cells that are within selected regions of interest (UNIT 5.2). Software gates may also be used for retrospective data analysis.

Amplification and Gain
These electronic processes raise and lower electrical output from detectors in order to produce signals in ranges appropriate for the instrument's electronic and data-handling devices. Amplification may be selected to operate in a linear or logarithmic fashion. When linear amplification is chosen, gain is used to define different factors of multiplication (e.g., gain 4 equals four times more signal amplification than gain 1; see UNIT 5.2).

Signal Processing
Each detector's measurement of each cell produces a pulse of data representing time versus intensity. Flow cytometers process these data in a variety of ways in order to reduce it to a single value. The processing involves either pulse-height, pulse-width, or pulse-area measurements. Pulse-height measurement is acceptable for spherical cells while pulse-area measurement may be used for nonspherical cells. Highly accurate detection of coincident cells (see below) including cell separation, rare-event detection, and DNA analysis will benefit from a combination of the pulse-processed parameters for individual measurements. For example, a two-cell clump will more often be correctly detected as two cells than incorrectly detected as one cell if pulse-area versus pulse-width is measured for a single fluorescence parameter, rather than pulse-height, -width, or -area alone.

Coincidence
As noted above, the flow cytometer is triggered to accept data pulses for discrete periods of time. Coincidence is defined as the appearance of more than one cell within this data-capture time. While commercial instruments are designed to detect and reject data from coincident cells, this electronic identification is not completely successful and may be dependent upon multiple biological and electronic variables. Undetected coincidence results in false positive data and is the major limitation in applications involving DNA analysis, rare-event detection (frequencies <1%), and cell separation. Precision of analysis and separation may be improved by multiple methods of signal processing (see above), additional probes used simultaneously, lowering of flow rate (cells/second), and cell preparation techniques designed to eliminate cell clumping.

Signal-to-Noise Resolution
Sensitivity is dependent upon the ability to distinguish “signal” (positive fluorescence) from noise (background) for any set of probes and target cells being studied. While a signal equivalent to 5000 molecules of fluorescein is easily resolvable on fresh mouse thymocytes (with low autofluorescence and low nonspecific binding), this same signal is impossible to detect using in vitro tumor cell lines which can have as much as 100 to 1000 times the background autofluorescence. Likewise, because frequency determinations are purely statistical and 100,000 or more cells can rapidly be analyzed, flow cytometry techniques can easily resolve minor subpopulations (0.1% total cells) when background staining in the region of interest is low (0.01%). On the other hand, if background staining in the region of interest is high (5%), it is probably impossible to detect subpopulations that constitute even 2% of the total cell population.

Autofluorescence
Background fluorescence of cells results from light emitted by naturally occurring intracellular materials excited at the wavelength used to excite chemically linked fluorescent probes. For immunologists, the most troublesome sources of autofluorescence are in hematopoietic cells excited at 488 nm. Because the peak wavelength of autofluorescence emission is near 560 nm (yellow), it significantly interferes with measurements of FITC and R-PE.

Autofluorescence is significantly higher (up to 3 logs) for in vitro cultured cells, cells with high granule contents, and tumor cells. For in vitro cell lines, the level of autofluorescence can be quite variable. Cells growing exponentially will generally exhibit not only a lower autofluorescence but also higher antigen expression than cells at low or high density. A variety of techniques employing signal measurement and software have been devised to correct for autofluorescence. These usually involve the sacrifice of one or more detectors. The most effective solution is to use red excitation (>585 nm) in conjunction with either Texas red or allophycocyanin as the fluorescent probe, as most hematopoietic cells (with the notable exception of eosinophils) exhibit little autofluorescence when excited by red light. This strategy is particularly useful for screening reagents directed against low-density target structures and for analysis of transfected cells expressing low amounts of the transfected gene product.

Data Storage Formats
Flow cytometry data are collected and stored either in list mode or in the form of a matrix (i.e., histogram). Measurements are stored in values proportional to the electrical signals generated by the photodetectors. Resolution depends upon the number of units (channel numbers) into which the values are stored—using either 32, 64, 128, 256, 512, or 1024 total channel numbers. Data are always stored in linear channel numbers, regardless of whether linear or logarithmic amplification was used to collect the data.

With list mode data storage, individual measurements of each parameter for each cell are serially stored in a list. This format has the advantage of allowing repetitive, retrospective analysis using multiple gates to define areas of interest. The disadvantage of list storage is the enormous volume of data generated. For example, a single experiment in which 50,000 cells were analyzed for three parameters and that contained 50 samples would generate 7.5 ×106 data points. An instrument could easily require >200 megabytes/day of storage capacity if all data were collected in list mode. Data of this magnitude precludes on-line storage and careful analysis. For this reason, list storage is usually inadequate for analyzing >5000 cells per sample, low-frequency measurements, or heterogeneous cell populations.

An alternative to list mode collection is data storage directly in matrices (or histogram format). This method employs hardware or software gating at the time of data collection (live gating) rather than retrospective gating as with list mode storage. A 64-channel by 64-channel matrix contains 4000 data points independent of the total number of cells measured because the number of cells for any correlated value within the matrix is the data value at that point. A single-column,1024 -row matrix contains 1024 datapoints. All list mode data are conv erted into matrix format for display; thus initial storage in matrix format avoids the need for later conversion and may speed analysis.

Data collection in matrix format with live gating is recommended for experiments involving complex data structure in two or less of the measured parameters. For example, an experiment that measures two colors of fluorescence on cells gated using forward light scatter intensity and propidium iodide exclusion (a third fluorescence measurement) is a four-parameter experiment. If list mode data were collected on 50,000 cells in each of 20 samples in this experiment, a total of 4 × 106 data points would be stored. If, on the other hand, data were stored as single-parameter histograms (1024 channels) for all four parameters and three dual-parameter histograms (64 channels), only 3.2 × 105 data points would require storage. Thus, data storage in histogram format requires less total storage space and is recommended when gating requirements can be accurately determined at the time of data collection or when correlation between more than two parameters simultaneously is not anticipated to be a requirement for data analysis. In addition, this method has the advantage of storage requirements that are independent of the total number of cells analyzed.

Electronic Cell Sorting

Cell separation using flow cytometry instrumentation involves scanning and sorting individual cells. However, one must have reasonable expectations when using electronic cell sorting (ECS). For example, a sorting experiment might involve selection of a fluorescence-positive subpopulation of cells representing 15% of total cells. At a flow rate of 3000 cells/sec, ∼6-8 × 106 selected cells (with ∼98.5% purity) may be recovered after 5 to 8 hr of sorting. While this recovery and purity will be adequate for a variety of purposes, it will be unacceptable in an experiment requiring 3 × 107 cells of 99.9% purity.

With ECS, expensive instrumentation is used to separate cells one cell at a time. The only way to increase cell yield/unit time is to increase the number of flow cytometers used. In contrast, bulk separation methods employ technology (tubes, columns, plates) which not only handles large cell numbers simultaneously, but also can be increased in size or number to increase yield.

Successive rounds of antibody and complement depletions or magnetic-bead depletions will generally yield a purity equivalent achieved with ECS (>99.5%; UNITS 3.4 & 7.4). Plate separations (panning) often achieve purity equivalent to ECS (98% to 99%; UNIT 3.5A). While bulk separation methods usually require pilot experiments to optimize parameters, once optimization is achieved, these separations can be successfully performed on a routine basis. For cloning of rare cells that can be identified by conventional methods such as ELISA and RIA, successive rounds of cloning by limiting dilution (UNIT 3.4) may be more efficient than cloning by ECS. Bulk separation methods may also be used to enrich cells to be sorted by ECS.

The discussion below addresses cell separation design and execution using flow cytometry. ECS should be used only when the cells of interest can be identified analytically and when the estimated cell recovery is adequate to answer the experimental question and when alternative methods of cell separation are inadequate. ECS is the method of choice when separation based upon quantity of antigenic determinants expressed by cells is desired.

Quantity of Antibody
ECS separates cells on the basis of quantitative differences in expression of ligands detectable with labeled probes. The technology is capable of separating cells with as low as 20% differences in measured fluorescence intensity. This degree of resolution will, however, require the use of linear rather than logarithmic amplification.

Estimation of Cell Recovery
The frequency of cells to be selected is estimated and used to calculate potential cell recovery based upon flow rate (cells/sec) and sorting time allowed. If the frequency of cells of interest is known to vary with the cell donor or some other variable, recovery should be estimated based upon the lowest predicted frequency. If potential cell recovery is not ≥1.5 times the minimum requirement for a meaningful experiment, enrichment methods such as antibody and complement depletions, plate separations, or magnetic bead depletions should be considered prior to sorting. For example, a two-fold enrichment of the cells of interest will double potential recovery.

Actual cell recovery of sorted subpopulations using ECS is generally in the range of 50% to 80% of potential recovery.

Selection of Reagents
It is critical to assess the effect of reagents on the experimental cell population, or at least on positive and negative control cells of the same type, prior to actual cell sorting (UNIT 5.3). If a population of 0.1% positive specificity is desired, but preanalysis indicates a background of 1.0% in the region of interest, it is unlikely that ECS will successfully enrich the cells of interest. In this case, ECS is likely to purify cells with high autofluorescence.

All reagents (including antibody preparations) must be membrane-filtered for aseptic ECS procedures. Many reagents contain bacterial contamination, which will become manifest after preservatives (usually sodium azide) are removed during sorting and washing. To preserve cell function, media without sodium azide should be used for antibody dilutions. The amount of sodium azide in antibody preparations usually does not affect cell function unless the antibody concentration is low and requires the use of undiluted reagent during the staining procedure. The saline used in the flow cytometer must not contain preservatives because the cells diluted in this sheath fluid during sorting usually remain in this solution for several hours.

One useful method for calculating the amount of antibody to be used in ECS is to determine the saturating concentration of the reagent needed for ananalytical run (UNIT 5.3). For 1 × 106 cells in a total reaction volume of 50 µl, ∼0.125 to 0.5 µg monoclonal antibody must be added to achieve saturation (final concentration in reaction is 2.5 to 10 µg/ml). For polyclonal reagents, 40 to 100 µg/ml are generally required to reach saturation.

Once the saturating concentration is determined, the total amount of antibody is calculated for staining the desired cell number in bulk by assuming the cells are being stained analytically in individual samples of 1 × 106 cells. One-half of this total amount of antibody is added to cells to bring the final cell concentration to 1-2 × 108 cells/ml. Many sources of labeled monoclonal antibodies to human cell-surface antigens are supplied at concentrations too low to achieve antibody saturation when performing bulk staining.

Aseptic ECS Procedures
A variety of protocols are used to eliminate most bacterial and fungal contamination in flow cytometers; however, cell separation by ECS is never a completely sterile procedure. For this reason, extreme care should be taken with aseptic procedures in cell preparation, media handling, and antibiotic use. The most common method of instrument decontamination involves extensive flushing with ethanol or with a 20% to 50% bleach solution. Some fluidic systems can be partially or completely autoclaved or gassed. Using this method, instruments used primarily for aseptic procedures will usually yield cell preparations which can be cultured in vitro without bacterial or fungal contamination. In contrast, ethanol flushing is inadequate for instruments used extensively with nonsterile rodent material. In any case, residual contamination is kept under control with antibiotics. Gentamicin is the antibiotic of choice for in vitro culture procedures after ECS and should also be included in media used for cell handling and cell collection during the sorting.

Cell Purity and Detection of Coincidence
ECS functions by interrogation and quantitative analysis in discrete time frames. The time when the cell of interest will be in an appropriate place in the sample saline stream is predicted, and at that time an electrical charge is delivered to that place. The stream is constantly oscillated sonically at high frequency to effect formation of droplets. These droplets are positively or negatively charged and are deflected when they fall past high-voltage plates. It should be noted that interrogation and cell-location predictions occur while the cells are still in the stream of saline (or with some instruments, in a cuvette), prior to droplet formation. The charging pulse is applied many microseconds later when the cell is, hopefully, just at the point where droplets are breaking off from the stream. Whether a droplet contains a single cell is determined by the behavior of spherical particles in flow and the presence of cell aggregates in the cell suspension. Because of the number of variables involved, bulk-sorted cells should always be reanalyzed and purity calculated with regard to the original selection criteria. When cloning, purity can be assessed only after regrowth, although microscopic examination can be used to assess whether appropriate cell numbers have been delivered to each well.

Different instruments have a variety of complex rules for selecting the cells to be sorted. Improper application of these sorting criteria may result in failure to select the cells of interest or in a reduction in purity and/or recovery. Most instruments have an error of <5% for fluorescence-intensity measurements. Therefore, only small ranges of fluorescence intensity between cells of interest need to be purposely discarded. If large ranges of fluorescence intensity between positive and negative cells are discarded, the purity and recovery may actually decrease.

As noted, the actual sort process takes place some time after an individual cell has been analyzed. The time between interrogation and sort depends on the machine's fluidic system. Some instruments handle this delay with an adjustable delay time, which is selected on the basis of visual measurement of the distance between interrogation and droplet breakoff. This method allows constant monitoring and minor adjustments of delay time. Other instruments require “mini-sorts” to determine proper delay time. With this method, test sorts are performed and sorted cells reanalyzed to check purity. This method also requires that sorted cells be checked occasionally during long sorts. Errors in delay time using either method will result in reduced cell recoveries (because the cell was elsewhere when the charge was applied) and in contamination of both positively and negatively selected cells. Even when delay times are properly selected, cells may arrive at the droplet breakoff point slightly before or after the predicted time. As most droplets do not contain cells, many instruments allow adjustment in the number of drops to be sorted for each cell. Three drops per cell is the most common setting and theoretically provides for an error in arrival time of plus or minus one drop. Selection of more than one drop per cell will improve final recovery of cell number, but may also lead to decreased purity of sorted populations.

Spherical particles such as cells do not arrive at the interrogation point in a flow cytometer at evenly spaced intervals. Since analysis is performed during discrete time periods, and the result of the analysis is individual data points, arrival of more than one cell during any of these separate time frames will constitute a coincidence error. To compensate for this error, instruments reduce the time of interrogation and abort the signalswhen twoindivid ual pulses, indicating two cells, are detected within one time frame. These instruments generally have a monitor that counts aborted signals. When the rate of aborted signals is compared to that of total signals, the rate of detected coincidence can be calculated. Not all cases of two cells in one interrogation time will be detected and sort errors will result. When the coincidence is due to two positive or two negative cells, there will be no impact as the cells will deflect correctly. If the coincidence is due to one positive cell and one negative cell, this signal will be seen as positive, and the negative cell will be incorrectly deflected into the positive pool (along with the positive cell). Thus, contamination due to undetected coincidence always results in more contamination of positive cells with negative ones than vice versa. Furthermore, this contamination is generally the most significant problem in cell purification by ECS. Since the probability of a positive cell being coincident with a negative cell will increase as the frequency of positive cells diminishes, this contamination is significant when selecting low-frequency positive populations.

Undetected coincidence is always a function of measured coincidence (abort rate). Specifically, there is a direct correlation between measured coincidence and undetected coincidence. Remember that a detected coincidence is due to at least two cells, and so an abort rate of 10% means that 20% of all cells are not being analyzed by the instrument. The most significant variables controlling all coincidence are the rate of cell flow and quality of the cell preparation. For sorts where 98% purity of positively selected cells is acceptable, sorting should be done at a rate where the measured coincidence (abort rate) is <10%. If higher purity is required, lower the flow rate until the abort rate is <5%. If the cells to be positively selected comprise 5% to 10% of all cells, it is unreasonable to expect purity >50% to 75% unless slow flow rates are used. The optimal rate for selection of more frequent subpopulations (>10%) will vary with the cells used but will range from 1500 to 5000 cells/ sec. For cloning, even slower rates (∼100 to 200 cells/sec) will be required.

The quality of the single-cell suspension is a significant factor in the purification and recovery of cell populations. It is important to prevent or eliminate cell clumping by performing gentle centrifugations and multiple resuspensions, keeping the media at cold temperature, and filtering the cell suspension through nylon mesh (UNIT 5.3). In addition, suspensions of cells from whole organs (spleen, lymph node, and thymus) should be prepared by teasing rather than crushing. Cell aggregation can also result from the use of staining reagents such as bivalent antibodies and it may be necessary to lower reagent concentrations to avoid this problem. Clumping can be detected either microscopically after staining or by comparing light-scatter profiles of stained versus unstained cells.

In addition to reduction of flow rate and elimination of cell aggregation, detection of coincidence can also be improved by pulse processing of signals (more than one measurement for a single parameter).

Cell Recovery
Since ECS functions by sorting one cell at a time, the total number of cells sorted depends on flow rate and frequency of the cells of interest. As discussed above, purity is increased by lowering flow rate which is at odds with the goal of recovering more cells. Therefore, the selection of a flow rate will always be a compromise between purity and total recovery. Increasing the flow rate will not necessarily increase total recovery because the measured abort rate will also increase and all detected coincidences will be discarded. For example, if the measured abort rate is 20%, at least 40% of all cells are being discarded before they are ever analyzed. In addition, abort rate often increases with flow rate. Some instruments include an option to disable this abort function for applications where purity can be sacrificed to improve cell recovery. Most instruments are capable of counting deflections as they occur. Based upon this number (how many cells the instrument thinks it has sorted), final recovery will range from 50% to 100% of the deflections counted. This recovery is dependent upon instrument variables already described, on the relative fragility of the cells used, and the methods used for collecting and harvesting the cells.

During sorting, cells are diluted in and remain in the instrument's sheath fluid (usually phosphate-buffered saline). Collection tubes should contain a cushion of tissue culture medium with 25% fetal calf serum. The volume of this cushion should be at least 25% of the total volume of the vessel used. The collection tubes may be frozen in dry ice prior to use as a convenient means of cooling the collected cells. Because sorted cells are extremely dilute, the volume of the collection vessel should be just enough to hold 2 × 106 cells. Vessel size will range from 5- to 15-ml, as the dilution factors vary between instruments.

Because of dilution factors and electrostatic charges, it is recommended that sorted cells be centrifuged while still in collection vessels. Recovery is usually improved by long, gentle centrifugation (30 to 40 min at 400 × g, 4°C).

Cell Viability
Immediately after sorting, cell morphology may be disrupted. A 30- to 60-min in vitro incubation encourages recovery. Cell viability after sorting is surprisingly good; this is probably attributable to selection of cells capable of withstanding the rigorous procedure so that sorted cells often display higher specific biological activity than unsorted cells. If sorted cells are to be used in a biological assay, it is recommended that a control sort be performed in which all cells are deflected for collection.

Cloning and Rare Event Selection
Commercial flow cytometers with ECS capabilities include options for sorting cells directly into the wells of tissue culture plates. This is extraordinarily useful for cloning cells with characteristics that are detectable by quantitative fluorescence analysis. Sorting options include the ability to select the number of cells delivered per well. While the number of cells delivered to each well will depend upon the cloning efficiency of the cells used, it is usually prudent to select a different number of cells/ well for each of several plates. As previously discussed, positive cloning selections need to be performed at low flow rates (<500 cells/ sec). The relative success of the cloning will depend upon the specificity of identification, the quality of the cell suspension, and the frequency of selected cells. For very rare populations (<1%), it may be more efficient to enrich cells by bulk sorting prior to cloning. The use of propidium iodide to exclude dead cells will improve cloning efficiency. Cells should be collected in wells containing a cushion of fetal calf serum and antibiotics (25% of the well volume), and the sorting medium should be replaced with culture medium containing antibiotics as soon as possible.