Since the initial commercialization of Flow Cytometry (FC) and Fluorescence Activated Cell Sorting (FACS) in 1968, they have undergone significant improvements. However, there remain numerous impediments, other than cost, to the further acceptance of the technology by many laboratories. Technical issues persist around the detection of low abundance molecules in intracellular compartments, the lack of "universal" cell permeabilizing chemistries, confounding effects from cell autofluorescence, the overlap of emission spectra between fluorochromes, and unavailability of reagents for targeting molecules of interest. Specifically for cell sorters, there are problems of cell survival after pressure changes during droplet formation and collection, dilution of the sorted cells before reanalysis or culture, and the time delay it takes to obtain a sufficient number of viable cells. Lastly, data analysis is tricky, particularly when dealing with low abundance targets. Cell sorting has been extended to sort cell organelles and fractions. For example, Hafner AS et al used PASS, fluorescence-activated synaptosome sorting, to study the local protein synthesis in neuronal pre- and postsynaptic compartments [1]. A recent publication describes an improved technique, "ghost cytometry," which may provide low-cost, fast, and accurate cell sorting [2]. Nitta N et altr described an optical-microfluidic-electrical-computational-mechanical system, named intelligent image-activated cell sorter (IACS), which integrated a three-dimensional on-chip hydrodynamic cell focuser, a frequency-division-multiplexed (FDM) microscope, a real-time intelligent image processor, and an on-chip dual-membrane push-pull cell sorter to achieve versatile and high-throughput single-cell acquisition [3].

There is a fine review of the theory behind flow cytometry that introduces the language of flow cytometry. In addition, another Labome.com article covers the major antibody suppliers that provide specific reagents for flow cytometric applications. Formal guidelines have also been published for specific applications [4]. This article focuses on 1) hardware choices and considerations, 2) practical aspects of flow cytometry and cell sorting, and 3) data analysis and software. The coverage of these topics, not exhaustive by any means, should allow the reader to understand the considerations that must be taken before embarking on a flow cytometric analysis of any of the applications listed in Table 1.

Cell Biologists need to understand their projected needs, as well as budget, when determining what instrument they will need. Table 2 will show the choices in the marketplace, but since these instruments are constantly being upgraded, it is recommended that the manufacturer"s web site be queried for the latest upgrades.

Companies that specialize in the analysis of small particles (JSAN, Partec, Accuri) target the individual lab or else the lab that needs to do regular quality analysis on their cell sorters. They tend to take advantage of the development of smaller diode lasers and improvements in microfluidics. The software is often not as robust as the software from larger companies, particularly when monitoring day-to-day performance or comparison of data from multiple instruments working on a large-scale project (e.g., a clinical trial). A recommendation would be to make sure the instrument produces an output file that is compatible with the post-collection analysis software you intend to use.

Researchers who sort objects that are unusually large (plankton, blast cells, early embryos), tiny (microparticles, bacteria, biological condensates) or highly asymmetric (sperm, myotubes), should first examine the intended instrument to ensure that it has been successfully used for that application in the past. Nozzle diameter, flow cell geometry, flow rate, droplet size, and detector placement can affect results.

Regenerative medicine and cancer therapies often require separation of cells under conditions that assure purity, viability, and ease of transport between the flow lab and the patient. Closed cartridges (Owl, Cytonome) appear to be one of the ways manufacturers are addressing this. Another is providing cell sorters with sterile hoods or designed to fit in standard-sized sterile hoods.

Some instruments improve on the fluid flow by acoustic focusing (Life Technologies) Iso-pressure fluidics (Stratedigm) microcapillaries (Millipore) or microfluidics (several in development). Most are technologies used successfully in other instruments, but they do not necessarily lend themselves to flow cytometry. For the most up-to-date feedback on the experience by users, I suggest readers of this article join the Purdue Cytometry Listserve. Many of the major thought leaders in flow cytometry regularly contribute to the discourse. Other blogs, like Google+, are becoming alternatives but they still lack the depth and breadth of the Purdue group.

Some exciting technologies are also making an impact on the field. First, pre-purification or depletion of target cells using magnetic beads by Miltenyi and others make the sorting step more efficient. Of course, many studies can be performed using one or multiple rounds of magnetic selection without FACS, but that is not the focus of this article. Amnis has combined the fluidics of flow cytometry with fluorescent microscopy and digital imaging. The result is a series of images of individual cells showing the distribution of fluorophores and not merely the mean fluorescent index (MFI). D Skokos et al analyzed immune synapses through Amnis Imagestream [22]. DVS Sciences has dramatically expanded the number of targets that can be analyzed by using Mass Spectroscopy (MS) to detect metals chelated to reporter antibodies. Currently, 33 different metals can be detected simultaneously. The instrument is called "Cytof" for Cytometry by Time of Flight [25]. The person considering the purchase of this instrument should be sure they need such in-depth cell profiling. Besides, the cells are vaporized in a plasma, so the cell that was analyzed cannot be recovered.

One place where the cell sorter manufacturers disagree is on whether laser interrogation is best done on droplets (stream in air) after the cells pass through the nozzle, or in a flow cell before droplet formation. Figure 1 shows (top) a stream-in-air system (Influx, but also found on the FACSCalibur and other instruments) and (below) a flow-cell system (BD FACS Aria III).

[enlarge]

Figure 1. Optical configuration of a stream-in-air (top) and a flow-cell (bottom) cell sorter. Reprinted with permission from BD Biosciences®.

Both have advantages and limitations. Stream-in-air sorters benefit from simpler hardware, a smaller distance between the interrogation point and the collection apparatus, and an increased number of lasers that can interrogate the sample. Flow cell sorters benefit from laminar flow and improved sample focusing through the cuvette. However, there are finite lengths for flow cells, and laser optics need sufficient spacing that cross-talk between channels does not occur. For sensitivity, the flow cell is typically superior, but stream-in-air does not have the issues associated with keeping the cell clean and can have additional lasers. It must be mentioned that some laser beams may, themselves, have harmful effects on cells.

Cells suspended in drops and then allowed to fall (or be pushed) undergo significant pressure fluctuations. For an image of how the drops are charged and deflected, see the other Labome article on flow cytometry Figure 7.

Besides, the type of media and the volume matter to the cell survival. The literature and the Purdue web site have numerous suggestions for media, but researchers often use fetal bovine serum (FBS). For collecting many cells into 11 x 75 mm or larger tubes, 0.5 - 1 mL of FBS is commonly used. For multiwall plates, half of the well volume should be FBS, because there is an additional concern with the wells drying out. For simply collecting nucleic acids, the same products used for cell lysis and nucleic acid isolation can be used. It is also recommended that cells be allowed to "rest" in rich growth media if they are to undergo further manipulation.

All of the FACS manufacturers have gone to great lengths to enhance the survival of sorted cells. Recent advances in microfluidics may make cell collection without droplet formation a reality. Whether the answer is in a capillary-based system or a microchannel based one, this will surely circumvent the stresses cells encounter when being sorted.

Unless you are creating diagnostic tests to be used on patient sera, there is no shortage of companies that sell antibodies for flow cytometry. See Labome article for a list. However, not all of these companies carry the same clones - and not all clones behave the same way in flow. If an article does not list the clone, and you want to run a similar experiment, contact the authors and find out which clone was used.

Most companies do not divulge antibody purification and conjugation methods, but if you are purchasing them from a reputable company, the same clone conjugated to the same fluorochrome will work, unless damaged in shipment, contaminated by a microorganism, or stored incorrectly. Follow the manufacturer"s recommendation for storage conditions and the antibody should work for at least one year.

Then you should decide if you want direct fluorochrome conjugates or use secondary antibodies, biotin-streptavidin, or other ligand-receptor pairs, for your experiment. Unless your lab is running on a shoestring budget, direct conjugates should be used. Direct conjugates benefit from a lower background, no interference from molecules like biotin released by lysed cells, and consistent performance.

Another option is to use a conjugation kit or have an antibody conjugated by a third party, like SoluLink or others. Many antibody manufacturers will do custom conjugation as well, but a company that builds its reputation on conjugation should be given serious consideration. The chemistries all have their advantages and disadvantages, so careful consideration of cross-linkers is appropriate (see Thermo Fisher Pierce practical guidelines).

There are five basic types of fluorochromes: small dyes (e.g., fluorescein isothiocyanate / FITC and Alexa dyes) protein dyes (Phycoerythrin [PE] Allophycocyanin [APC] GFP) Tandem dyes, where a protein-dye collects laser light, transfers it to a small dye, and the tandem emits at the wavelength of the smaller dye (e.g., APC-Cy7) quantum dots, and polymer dyes ( Brilliant Violet). All have advantages and disadvantages, but the protein and small molecule dyes have been used the longest in flow cytometry.

It is easy to see how conjugating a large protein (> 100 kd) or a hydrophobic dye or polymer would alter the activity of a monoclonal antibody. For an illustration of relative sizes, Figure 2 is an image of an antibody conjugated with a protein (in this case horseradish peroxidase "only" 44 kd). One can conclude that it is safest to use the same clone and fluorochrome combination used previously by others. However, there is no reason why one cannot create a unique antibody-conjugate combination and characterize its specificity using expressing and non-expressing cells. Then titrating the antibody to maximize the signal to noise ratio, and your lab has a new reagent.

[enlarge]

Figure 2. Reprinted with permission from SoluLink®.

The choice of fluorochromes to use in an experiment is based on 1) what lasers and filters are available on your flow cytometer or FACS, 2) relative abundance of the targets - brighter fluorochromes should be used on less abundant molecules, and 3) if any of the targets are intracellular. Since intracellular targets are buried within the cell, the brightest fluorochromes should be reserved for them. In addition, GolgiPlug containing brefeldin A and monensin can block the intracellular protein transport processes and accumulate cytokines and/or proteins in the Golgi complex to enhance the signals [26, 27].

Two laser instruments like the Guava® EasyCyte or BD® FACSCalibur can only use four fluorochromes, and so dye selection is as simple as setting up a matrix of available fluorochromes and targets to be interrogated. PE is typically the brightest, followed by APC, so they should be conjugated to antibodies to intracellular or low abundance targets. For the amount of emission light spillover, e.g. FITC into the PE channel, one can look at a Spectrum Viewer from Invitrogen or BD.

For more than two lasers / four colors, one has to consider automated vs. manual compensation [28, 29]. Fluorescence minus One (FMO) controls [30] instrument-specific strengths and weaknesses, and combinations of fluorochromes. For these issues, I suggest you read the articles referenced and others [31], and consult with your Flow Core Manager or in-house expert.

The antibody mixes can be stored. Laskowski TJ et al found that their 17-color flow cytometry panel antibody mixes remained stable for up to 15 days at 4°C while CyTOF antibody cocktails had a maximum storage time of 3 days at 4°C [31].

The fixative used almost exclusively is paraformaldehyde, PFA, or formaldehyde. All three names are used, but the active ingredient is the same. Fixation can be 1 - 4% in Phosphate-buffered Saline (PBS) and fixation is usually performed on ice. For kinases and phosphoproteins, where inactivation of phosphatases needs to be rapid, the PFA is often used at 37°C.

Formaldehyde reacts with primary amines, and since the antibody combining site often involves electrostatic interactions, modification of a lysine fixation can prevent some antibodies from binding fixed protein. That is another reason why following established protocols can save time and reagents. Fixation (and permeabilization) rapidly kills cells, so sorting of fixed cells is typically performed if the next step is cell fractionation or nucleic acid extraction.

There are different choices for cell permeabilization depending on whether the antibody target is to be secreted and therefore sequestered in the Golgi, a kinase or phosphoprotein that requires competitive inhibition of phosphatases, or a nuclear protein that requires partial dissolution of two membranes and relaxation of the DNA. All of these methods utilize detergents or alcohols, the detergents being the gentlest and the alcohols, particularly methanol, being the most aggressive.

The gentlest detergent, used primarily when the target is cytoplasmic, is saponin. Low concentrations, often 0.1%, are sufficient for allowing intracellular cytokines to be stained [32]. Some nonionic detergents, like Triton X-100, Tween 80 are used, but saponin is typically the detergent of choice. The commonly used BD Cytofix/Cytoperm™ Fixation/Permeabilization Kit, for example, for the intracellular cytokine staining of T cells [33, 34], uses saponin.

For kinases and phosphoproteins, the technology comes from the laboratory of Gary Nolan at Stanford [35, 36]. Phosphatases are inactivated with PFA, and cold methanol and the methanol also acts as a secondary fixative and strong permeabilizing agent. Papers suggest methanol:water mixtures from 70% - 100% methanol, but 70% and 90% are most common. There are also examples of staining before Permeabilization. A site that shows titrations of numerous antibodies in detergent and methanol-based protocols can be found at Cytobank, now part of Beckman Coulter. An example of the depth of analysis can be seen as figure 3. BD PhosFlow protocol can be followed [34].

[enlarge]

Figure 3. An example of the depth of analysis in Cytobank.

Nuclear targets are reached using methanol, or ethanol, as a permeabilizing agent. Ethanol is preferred for the proliferation-specific target Ki-67 [37], but methanol is more commonly used. Not all antibodies bind PFA-methanol-fixed cells, and so compatibility tables are provided by many monoclonal antibody providers. It is essential, however, that the methanol is completely removed before the cells are stained with antibodies if the fluor is a protein since many are sensitive to the dehydrating effects of alcohols. Neuronal nuclei can also be isolated through fluorescence-activated nuclear sorting (FANS) from isolated nuclei [38, 39].

Another victim of alcohol dehydration is the class of fluorescent proteins often expressed in cells undergoing flow cytometry. These include GFP, mCherry, and Cerulean. If the signal is destroyed with alcohol treatment, it may become necessary to use a fluorescent antibody to the denatured fluorescent protein.

Dyes, unconjugated, were the first reagents used to stain cells in the early days of flow cytometry. The types of dyes used covered below are divided into vital dyes, nucleic acid intercalating dyes, lineage dyes, cation and pH flux dyes, efflux dyes, and organelle dyes. Only the most popular, or first used dyes are discussed here; see Molecular Probes for an extensive resource. Note: Molecular Probes® has fifteen categories for dyes.

Trypan Blue was the first dye shown to be excluded from living cells but freely enter dead cells. PubMed has a reference for its use back to 1914. However, the first publication of Trypan as a vital dye dates back to 1980 [37]. Since then, propidium iodide (PI) and 7-amino-actinomycin D (7-AAD) have been the vital dyes of choice [40]. 7-AAD tends to show less spillover into adjacent channels in flow cytometry, and care must be taken when using a bright DNA dye with another fluorescent molecule that is less intense and has an overlapping emission spectrum. Other dyes can also be used. Aizarani N et al identifed viable cells in cell sorting through Zombie Green from BioLegend [41]. Similar dyes such as Zombie aqua [33] or FVS700 from BDBiosciences [42] have also been used.

Ethidium Bromide (EtBr) and PI bind the major groove of the DNA double helix. 4',6-diamidino-2-phenylindole (DAPI) and the Hoechst dyes (there are several commonly used) bind the minor groove. For example, Chopra S et al labeled mouse bone marrow–derived dendritic cells and paw single cell suspensions with 0.5 μg/ml DAPI from Thermo Fisher during flow cytometry and cell sorting [43]. Note, some dyes (e.g., PI) will also bind the grooves in hairpin RNA structures, so if quantification of DNA is desired, then often a mixture of dye and RNAse are used. Other dyes, such as 7-AAD, do not bind DNA in either groove.

DNA dyes are also used for cell cycle analøysis, ploidy, and genome size quantification. For these studies, the DNA dyes mentioned above are typically used, but the data interpretation is very demanding. Fortunately, there is an excellent article that outlines the process in great detail [44].

Apoptosis measurements utilize many of the unique qualities of flow cytometry. DNA binding, membrane impermeable dyes are used to identify late-stage apoptotic and necrotic cells. Antibodies to the cyclins, the cleaved form of Poly [ADP Ribose] Polymerase (PARP) and phosphorylated histone H3 are commercially available. The earliest marker of apoptosis is the appearance of Phosphatidyl Serine (PS) which is usually only found on the inner leaflet of the plasma membrane. Its appearance is detected, not by an antibody, but by a Ca²⁺ dependent binding protein, Annexin V. This molecule is commercially available conjugated to several different fluors and has become a mainstay of apoptosis research. Analysis of Annexin V is also detailed in [44].

Lineage dyes, for example, carboxyfluorescein diacetate succinimidyl ester (CFSE), bind protein by being membrane permeable and then reacting to nearby proteins via the succinimide ester. Once stained, the cells" progeny can be followed for several divisions, since the turnover of the stained protein is slow, compared to the time it takes rapidly proliferating cells to divide [45].

Calcium ions (Ca²⁺) are measured with the dyes Fura-2 and Indo-1 [46], and they require UV excitation, so many commonly used flow cytometers cannot be used with them. Magnesium Green and some Fura-derivatives have been used successfully to measure Mg²⁺ [47]. Some of these are excited in the visible spectrum. For pH determination, Fluoroscein dyes have been used since 1979 [48]. For collecting real-time data, the flow cytometer needs to be able to display data with a time axis, and the collection rate has to allow for multiple samplings during the event one wishes to measure.

Several normal and transformed cells can transport organic compounds out of the cell. This is of critical importance when the cell is cancerous and the molecules removed from the cells are chemotherapy agents. However, these transporters are also found in normal cells, including some stem cells. Some of the molecules that can be excreted are fluorescent, and cells with active transporters have a distinctive appearance, by flow cytometry, when in the presence of these dyes. The most commonly used dye is Hoechst 33342, and the cells with active transporters are referred to as side population (SP) cells [49].

There are three approaches to labeling organelles. First, one can transfect cells with a vector coding for a fluorescent protein and sufficient peptide information to cause the protein to segregate to the specific organelle. Second, there are antibodies to organelle - specific markers, but they require cell Permeabilization. Dyes tend to be simpler to use, are less labile, and fluoresce more brightly than most fluorescent proteins and some can be used on living cells. Molecular Probes® provides many of the dyes and a useful web site for their enumeration. Some commonly used dyes are MitoTracker red or green for mitochondria; Dihydrorhodamine 123 for the oxidative state of mitochondria; LysoTracker Green and Lysosensor Green; the DiOC dyes for endoplasmic reticulum; and BODIPY dyes for Golgi.

The software has eased the issues around data analysis slightly. However, most of the software combines instrument-specific controls and analytical methods in ways that are optimal for the hardware. Within manufacturers, there are major differences in the data, depending on whether the information gathered was analog or digital. Fortunately, there are some excellent third-party software packages, and one is even free, and they are nearly instrument-independent and allow most desired data analyses.

Even considering the difficulties listed above, one cannot minimize the breadth and depth of information gained by application of these instruments to questions of biological or medical importance. A search on ( PubMed) of review articles on "flow cytometry" OR "FACS" will yield numerous topics where either, or both, methods are utilized. Table 1 is a list of applications found in that way.

The software that comes with the flow cytometer or FACS should be used for controlling the instrument unless there is a compelling reason to use a third-party"s software. Usually, this involves the purchase of a used instrument where the manufacturer no longer supports it. Most machine-specific software allows for some analysis. A two-laser benchtop instrument may not have cutting edge compensation, standardization, or statistical packages built into it. Second, some software may have most features, but lack the one you need for a manuscript or presentation. Third, you may have a collaborator with a different instrument, but you need to share data with that person. Last, you may have a limited number of copies of the instrument software, but you also have numerous users that would like to manipulate their data at their computers, without corrupting the original data. Table 3 is a list of the major software packages and the instruments they were developed for.

First, there are considerations based on the type of instrument. Older instruments, like the BD FACS Calibur and FACScan, process data as analog information. Newer instruments, like most of those on the market today, process the data as digital information. The output of an analog instrument is an FCS 2.x file, which can be read by virtually any analysis package. Digital data can be exported as FCS 2.x or 3.x, and there are still some older analysis programs that cannot parse FCS3.0 data. One example is a free package, written by Joe Trotter and downloadable from WinMDI. It runs in Windows 95 and NT, and there is some discussion online that it may run on XP. There is also an Intel 80286 version for old MacIntosh computers running SoftWindows.

Unfortunately, even if two instruments produce FCS 3.0 data, it does not mean that one can freely share the data between instruments from different manufacturers. Usually, the issue is in the formatting of the text, and there may be an online resource for converting one instrument-specific format into another. The three instrument - independent packages either do that automatically or have small applications to do that for the subscriber.

What are the advantages of digital data? Most important, each data point (cell) contains associated data so that if you load the data into an analysis package, virtually all parameters can be modified, including compensation. FCS 2.x data cannot be brought back to a pre-compensated condition and re-analysis is limited.

There are also software packages available to fill perceived gaps in existing analysis or presentation options. Two examples are CytoPaint from ( Leukobyte) and Paint-A-Gate from BD Biosciences which allow for batch cluster analysis and display/ Another, that is useful for proliferation and DNA analysis is Phoenix Flow Systems.

By far, the best free package for analysis and data sharing is Cytobank, which was shown above for its FACSelect functionality. FACSelect is one example of its ability to share data with many researchers who are planning to run intracellular flow on signal transduction molecules. Uploaded data can be kept private, one can mark it public, or data can be shared only with specific collaborators. In that way, data can be restricted, completely unrestricted, or restricted to colleagues only. Cytobank has all of the essential tools: compensation, gating populations, setting hierarchies, displaying tables of subsets of events, creating contour plots and heat maps, and histogram overlays. Figure 4 is a heatmap drawn from Cytobank data.

Cytobank does offer a subscription-based set of features for high dimensional datasets. This software is called "SPADE".

[enlarge]

Figure 4. A representative heatmap drawn from Cytobank data.

FlowJo is currently the most often-used analysis package for flow cytometry data (as discussed in another Labome article on software programs in biomedical research, with a more detailed discussion on FlowJo). The program resides on individual computers, and TreeStar keeps up with current trends in data analysis and presentation. The FlowJo site offers a 30-day trial of Version 10 (9.6 for the Apple version). The trial is always recommended before purchase. Figure 5 is a series of stacked histograms generated from a batch analysis. Flow cytometry generates large amounts of data very quickly. The ability to analyze numerous samples as a batch and then display them in a meaningful way is an essential component of any cytometry analysis package.

[enlarge]

Figure 5. A representative stacked histograms generated from a batch analysis.

The alternative, also an excellent option, is the online package FCS Express from DeNovo Software. It also has an impressive features list and also produces publication-ready graphics. Figure 6 illustrates the power of this analysis package, and the data acquired when looking at near-instantaneous Calcium fluctuations inside of cells. Note that the X-axis is time and the more rapidly cells are interrogated, the more meaningful the data.

Note that FCS Express has adapted their software for imaging cytometry, and so the types of data are proliferating, therefore if one is generating data from a cutting-edge instrument, care must be taken to match the software to the data type.

[enlarge]

Figure 6. FCS Express from DeNovo Software.

To summarize the software section, flow cytometry generates an enormous amount of data. The FCS standards for exporting data has made working with data from different instruments straightforward. The instrument - specific and instrument - independent analysis packages have given new users a false sense of security. It is not difficult to generate a large amount of data, but as I have tried to illustrate, the number of experimental, acquisition, and analytical variables is endless. It takes time to develop the skills to select the correct reagents, treat the cells appropriately, set up the instrument to optimize for a particular application, compensate and run the proper controls, and analyze the data in a meaningful way. Unless you never plan to run more than four fluorochromes and all your staining is of the cell surface, it is best to work closely with an experienced user in order to bypass all of the common mistakes made by the self-taught. This is particularly important if you intend to sort and have live cells at the end.

Look at Table 1 with over 30 applications for flow cytometry. Find the one closest to your own interest and do a quick literature search in either PubMed or Google Scholar. Make sure the term "cytometry" is a keyword. Look at the way the data is shown in several recent papers, and get an idea of the number of reagents, the sophistication of the instrument used, and how the analyses are displayed. Then go back to the text of this article and look for all of the potential hazards I identified for new, and experienced, users. Make an estimate of the cost of cytometer time, reagents (and isotype controls, compensation beads, and other reagents and buffers). Please note that isotype control may not be necessary or optimal [53]. Isoclonic control, where an excess of identifical unconjugated antibody is mixed with a conjugated antibody, can also be considered. Then modify it to your specific application and run it by a handful of experienced users. That is the minimum amount of preparation I would recommend for someone entering the field or expanding from simple analyses to complex multicolor experiments. Once that is accomplished, Table 4 lists the web sites for the suppliers and manufacturers discussed in the text.