The French Association of Cytometry (AFC) had initiated a particularly important program of one-day workshops, each one dedicated to a specific question regarding the cytomery practice. One of these workshops per year was free of charges for our members and was especially organised as a Prestige day. This year we have planned to talk about the fluorescence compensation settings in flow cytometry. Because of the increased number of fluorochromes and reagents available and to the greater sophistication of the instruments, the applications in clinical practice are becoming more and more numerous. They are mainly multicolor analysis (four to six and even nine colors) to detect rare events such as minimal residual disease (MRD) or scarce cells populations in biological samples. These applications are much more complex and need special attention from the scientist to avoid misinterpretation of the data observed. One of the main problems is the spectral overlap observed between the different fluorochromes used in combination that has to be corrected by more and more complex mathematic compensation settings. These settings are not always so easy, and strict procedures are needed to get the most reliable data. Different speakers were invited to make a review on fluorescence phenomena, to present the compensation settings procedure and the few observed confunding situations to be aware of. A summary of their presentations was made here.
This theme was particurlarly appropriate to make a special tribute to our late lamented past President, Jean Luc D'Hautcourt, who passionately contributed to improve wide and good practice in flow cytometry (1, 2). This special 2006 Prestige workshop was dedicated to his memory.
FLUORESCENCE AND FLUOROCHROMES
Fluorescence is the property of molecules which are able to absorb the energy of light in a range of specific wavelength (excitation), and almost instantaneously to reemit the absorbed energy as a light with a weaker energy, that means a higher wavelength (emission). The use of such molecules in the field of biological research has been a real revolution, by giving access to molecular information at the cellular or sub cellular level with a very high sensitivity. At biologists disposal are different type of fluorescent probes, and regularly new probes or ways of using them are developed.
The most common and easy to use are the chemical probes specific of a particular type of molecule. A typical example is given by the poly-nucleotides specific dyes, that give access to the DNA content in a very precise way, allowing an accurate measurement of the cell cycle of a proliferating cell population.
But it is no doubt that the more powerful tools in cell studies are the mono or polyclonal antibodies coupled (directly or not) to a fluorescent molecule. Thanks to their high specificity, they permit a quantitative as well as qualitative study of the proteins they are directed to. This type of approach is powered by the wide variety of existing fluorophores, which authorized the simultaneous measurement of different molecules, each of them being revealed by a specific colour. Today, the technological developments of flow cytometry, associated to the new designed fluorescent probes, allow to detect and to quantify in a single measurement up to 12 different molecules.
Nevertheless, it is important to notice, for judicious and correct use of them, that the enormous potential of these fluorescent molecules, may be limited by the conditions of use. Indeed, the properties of these compounds, like spectral characteristics or quantum yield, can be disturbed for example by the conditions of pH. Besides, this phenomenon is turned to good account in the case of the ratiometric probes precisely used to measure the pH.
When studies on living cells are needed, it is important to know that some dyes, despite the fact they are known as “vital dyes”, may have toxic effects and so, may influence the cellular metabolism.
In the case of multi colour analysis, it is important to take into account the possible overlap of the emission spectra, susceptible to induce incorrect conclusions. This crucial problem can be corrected thanks to the fluorescence compensation systems which have been developed on flow cytometer softwares. In the same order of idea, it is important to keep in mind the possible auto-fluorescence of the studied biological objects, especially when working with plants, but which can also be originated by fixation protocols.
Among the problems encountered with fluorochromes, and which have to be mentioned, are the quenching and the energy transfer. The quenching corresponds to an absorption of the fluorescence emitted by the fluorophore by an other molecule, inducing a loss of fluorescence emission, as do BrdU with Hoechst fluorescence (3, 4). The energy transfer is an equivalent phenomenon, but in this case, the fluorescence absorbed by the second molecule, is reemitted as light with a different wavelength. It is possible to take advantage of this process for fluorochrome association, or for fluorescence resonance energy transfer (FRET) experiments.
A limiting property of the standard organic fluorophores is their sensitivity to bleaching, a progressive extinction of the fluorescence potential because of the excitation illumination. But today, new luminescent probes developed from inorganic compounds are insensitive to bleaching. These nano-objects, called quantum dots (Q-dots), are semiconductor beads. In addition to their resistance to light, the Q-dots, which mean size is around 5 nm, emit a fluorescence related to their size. The result is that for a unique excitation wavelength, you may have at your disposal various probes with specific and narrow emission spectra. The smaller Q-dots will emit in the violet or blue range, while the bigger will emit in the red. Still under development, Q-dots are very promising probes for multicolour analysis (5).
Another new class of fluorescent tools has been developed recently, which are the fluorescent proteins. They are used under the form of recombinant protein, by association with a protein of interest, of which the expression, localization, and even the dynamics in living cells, can be followed thanks to this fluorescent tag. The first of these fluorescent proteins is the green fluorescent protein (GFP) extracted from the jellyfish Aequorea victoria. From this molecule, mutated forms have been designed to modify the absorption and emission spectra. Similarly, from a coral species (Discosoma) a red fluorescent protein (RFP or DsRed) has been purified and cloned, and mutants were engineered. Presently, biologists may use more than 15 variants of fluorescent proteins covering the wide visible spectrum, and allowing the simultaneous tracking of various proteins (6, 7).
From their first use in biology until today, fluorescent probes have been the object of constant developments, and scientists for which these tools are essential must be vigilant about it.
MULTICOLOR IMMUNOPHENOTYPING AND COMPENSATION OF FLUORESCENCE
Peripheral blood lymphocytes represent a highly heterogeneous population of cells involved in diverse immune functions that are associated with expression of a variety of cell surface proteins. The principal populations of lymphocytes, defined by their levels of expression of relevant cell surface proteins, are T-cells, natural killer cells (NK-cells), and B-cells. Further sub-divisions can be made into helper/inducer (T4) and cytotoxic/suppressor (T8) T-cells, and by defining their respective states of immunological experience (effector, effector memory, central memory, and naïve). Their phenotypic characterization requires coordinated analysis of the expression of several cell surface markers, and thus the simultaneous application of multiple antibody markers. Flow cytometric equipment is presently capable to simultaneously quantify 10 or more different fluorochromes, and a wide array of fluorochrome-conjugated antibodies, as well as software capable of assisting multiparameter analysis and sorting has become readily available. The phenotypic diversity in, for instance, the immune system, can only be respected by the application of multicolor immunophenotyping, allowing non-deductive, positive identification of cell types at study.
Multicolor immunophenotyping becomes relatively straightforward if a few basic principles are understood and applied. First, proper attention should be given to machine-related issues, such as the optimization of the optical system (dichroic- and bandpass interference filters; PMT choice), as well as the daily applied alignment/calibration routines. The latter should be aimed at reproducibility of machine response, rather than be based on pre-established machine settings. The optical system of present day flow cytometers is perfectly capable of deriving at least five colors off a 488 nm laser line (e.g. argon-ion- or solid state), and at least two each off a 638 and 407 nm diode laser. Spectral overlaps between fluorochromes that can be used are multiple, and cannot be avoided (8). To correctly deal with them, the proper controls will have to be devised, in order to assess the extent of fluorochrome spill. The most rapid, flexible, and reliable way of doing so is by performing single color staining of antibody capture beads. They present the advantage of allowing precise assessment of spill characteristics of the exact conjugate that is being used in the experimental panel. Also, they will provide a single population of intensely labeled events, improving the precision of spill calculation. Staining conditions for each fluorochrome conjugated monoclonal antibody should be determined in titration assays. In most cases, the actual amount of target (cells or beads) in the staining volume does not affect the amount of antibody to be used (Fig. 1).
Proper annotation of data files during acquisition is extremely inportant. Ideally, information should include laser power, PMT voltage, optical filters, and stain (provider, clone, and lot number). Acquisition should include a sufficient number of events, to allow for a significant number of observations, even in minor subpopulations. Data analysis and interpretation should be performed using any of the modern software capable of calculating compensation matrices based on the single-color control samples provided (based on antibody capture beads). Also, software should provide data space transformation, facilitating visual appreciation of compensation, and allow the user to manually interfere with individual values in the compensation matrix (9).
More than ever, and because flow cytometric data are used to make clinical decisions, instrument settings have to be carefully adjusted to allow good reproducibility and sensitivity of the cytometer. Several parameters must be monitored with awareness. This includes: monitoring of the instrument performance, settings of compensations, and achievement of controls.
Performances of the Flow Cytometer
Most of the cytometers used in a clinical laboratory have fixed optical systems: the relative position of the flow cell with respect to the optical elements is fixed. In such systems, the instrument operator cannot optimize alignment but must verify that the instrument is in order to ensure identical conditions of measurement on a daily basis. It should be checked with standardized fluorescent beads, which are known to have stable characteristics over time if analyzed in the same conditions. Results of daily monitoring of instrument performance should be depicted and reported on a Levy Jenning's plot to detect variations in the instrument resolution and sensitivity. Daily monitoring of optical alignment ensures that the cytometer gives acceptably bright fluorescence measurements and that peaks with a low coefficient of variation are produced for all parameters. Moreover, an annual inspection of the optical filters should be done by spectroscopy. It allows the following of the ageing of the filters associated with the decrease of their transmission efficiency. This step also permits to verify the filter's characteristics: transmission and blocking efficiency, transmission and blocking range.
For a typical six-colors stain, the operator would potentially set up 30 compensation values to fully compensate all colors. This operation may be very time consuming and may be source of errors if it is not properly performed. An inappropriate setting of compensations may induce over or under-estimation of a cell population and incorrect conclusions about data. Because it is not a question of estheticism, compensation adjustment should never be done by eyes. In general, a visual adjustment of compensation results in an over-compensation of data. Compensation is better set using software and in a very short time. However, if these settings have to be input manually, better results will be obtained by the comparison of the median values of both positive and negative populations. When logarithmic axes are used, the median is a better representative of the fluorescence intensity of a cell population than its mean. An additional crucial point in compensation setting is the choice of the reagent that may be used. Ideally, the reagent used in a single color staining to set the compensation should be the same as that used in the staining panel of fluorochrome labelled antibodies. However, for non-tandem dyes (FITC, PE, APC), a reagent that stains a large number of cells with an high intensity can be used instead of a reagent that might only stains rare cells or poorly expressed antigens. In contrast, the tandem dyes (PE-Cy5, PE-Cy7, APC-Cy7, etc.) may exhibit lot-to-lot variations and a modification of their spectral properties when ageing. That is the reason why each tandem-dye must be used as its own compensation control. This may be a problem if the antigen detected is expressed at a low level or by a poor subpopulation of cells. In this case the use of antibody-capture beads may be useful. Thus, the fluorochrome used for the spectral overlap compensation is exactly the same than the one used in the antibody cocktail.
The combination of all the isotype controls antibodies in the negative control tube may account for inaccuracies in the identification of specific antigen-expressing cell subpopulations. The best control is the fluorescence minus one control (FMO control) described by Baumgarth and Roederer (10) where all the reagents of the complete stain are present but one. This kind of control allows an accurate placement of gates to quantify the subpopulations of interest. These controls are also powerful for the analysis of samples with numerous types of cells that exhibits different levels of background or for the identification of rare cells whose fluorescence may be hidden by the one of others cells composing the sample (in the monitoring of the residual disease for example).
CONFOUNDING PHENOMENA OBSERVED IN MULTICOLOR ANALYSIS
The availability of new five to six color flowcytometer for routine analysis has brought considerable extent in potential applications by improving the precision of information because of combination of simultaneous labeling (11, 12). However, mixing fluorochromes brings more possible artifacts because of interferences between fluorescent molecules, according to their adsorption spectra and excitation wave lengths (488 and 633 nm) available on routine instruments. In the present section, we intend to list some artifacts that could be observed in routine practice with such systems.
1Labeling intensity is usually reproducible but could be occasionally reduced with a shift of the peak of positive cells (Figs. 2a and 2c). The main reason is due to an insufficient amount of antibody related to the number of cells in the sample (ie : a very high number of cells or a insufficient amount of antibody due to hazardous pipetting error). This artifact is easily corrected by sample dilution (first instance) or addition of antibody (both instances; Figs. 2b and 2d). On the other hand, lymphoproliferative disorders could produce cells with reduced differentiation status. This has been described for CD45 (Fig. 2a) but can also be observed for CD19, CD3 (Fig. 2c), CD4 or CD8. Finally, long exposure to daylight could induce labeling decrease because of photobleaching or tandem dissociation, specially with Cyanin 7 tandems (Fig. 3).
2Interlaser compensation should be low with separated lasers because the delay between laser excitation is much longer than fluorescence decay duration. However, interlaser overlap could be observed, specially around red light. The main artifact is observed with PE-cyanin tandems where the cyanin5 could be directly activated by the red laser. Thus true fluorescence emission could be detected on red laser by PE-Cy5 labeled cells. PerCP is recommended instead of PE.
3Physical mechanism: Fluorochromes could be unexpectedly excited by another fluorochrome because of FRET. JL D'hautcourt has shown, in the 2004 AFC meeting, that labeling of TCRγδ -PE could induce fluorescence emission from the very close CD3-APC. A signal was then detected on the detector that was dedicated to the CD4-PerCP.Cy5 (Fig. 2e). Thus γδT cells were unexpectedly CD4+. Less clear was a phenomenon of interference of ALEXA647 that could be excited by a close PE-Cy5 and not PE. This was not clearly observed with APC.
4False interferences could be due to the sample itself, specially when using total blood without any washing (13–15). A large spectrum of non specific interferences could be observed in patients with high levels of auto-fluorescent biliary salts (16) during acute hepatitis (Fig. 2f). The artifact was present in all wave lengths and superimposed with the normal staining. The noise can be deleted electronically or by washing the cell before labeling. On the other hand, a cell population (i.e. CD3+, CD8+) could express unexpected fluorescence probably because of agglutination of both fluorochromes appearing as a default of compensation (Fig. 2g). This can be observed for tandems containing cyanins or, less frequently alexa 750. The mechanisms are not clearly elucidated yet. All these artifacts could be eliminated if the sample is washed before labeling (Fig. 2h) and could be reconstituted with serum transfer to another cells. The patient serum and both fluorochromes are required during the incubation.
Using simultaneously five to six fluorochromes induce more risks for unexpected interferences. Physical systematic interferences could be anticipated and prevented during experiment settings, but the poor panel of fluorochomes used in commercially available conjugated antibodies limits the alternatives. On the other hand, hazardous interferences could be (rarely) observed in some patients with unexplained but possible clinical significance. Dotplot examination usually is sufficient to detect them. Simple procedures can eliminate the artifact (ie: cell washing, electronic exclusion). Beads coated with antibodies are very convenient tools to confirm and characterize the interaction.
The past 5 years have seen a lot of advances in the area of multicolor analysis by flow cytometry. The appearance of solid state lasers, of new fluorochromes and dyes, the improvement of signal digitalization and other evolutions have led to the democratization of up to six color instruments. In the past, the multicolor analysis was the play-ground of fundamental researchers, but this field of flow cytometry is now shared with routine and clinical applications. However, the only component that has never been improved is the cytometrist who is responsible for the instrument settings and for the data analysis. While the number of parameters was limited, settings and analysis were easily performed. The increase in the number of analyzed parameters has made the interpretation of six colors (eight parameters) data more complex and numerous subpopulations could now be identified. In addition, the multiplication of fluorochromes used in one single experiment has extended the compensation matrix. Its settings could sometimes be compared to the solving of a sudoku grid. Unfortunately, numerous factors are responsible for variations in analytical conditions and may affect results obtained by flow cytometry. Despite this, multicolor flow cytometry is a valuable tool for the identification and the quantification of specific cell subpopulations that help the clinician to take a decision. Therefore, it requires rigor to avoid pitfalls inherent to the instrument settings and to produce reliable and reproducible data. Jean-Luc D'Hautcourt recognized earlier the importance of quality control and insisted on the correct achievement of flow cytometric analysis (1, 2).