This article describes spectral cytometry, a new approach in flow cytometry that uses the shapes of emission spectra to distinguish fluorochromes. An algorithm replaces compensations and can treat auto-fluorescence as an independent parameter. This new approach allows for the proper analysis of cells isolated from solid organs.
Flow cytometry has been used for the past 40 years to define and analyze the phenotype of lymphoid and other hematopoietic cells. Initially restricted to the analysis of a few fluorochromes, currently there are dozens of different fluorescent dyes, and up to 14-18 different dyes can be combined at a time. However, several limitations still impair the analytical capabilities. Because of the multiplicity of fluorescent probes, data analysis has become increasingly complex due to the need of large, multi-parametric compensation matrices. Moreover, mutant mouse models carrying fluorescent proteins to detect and trace specific cell types in different tissues have become available, so the analysis (by flow cytometry) of auto-fluorescent cell suspensions obtained from solid organs is required. Spectral flow cytometry, which distinguishes the shapes of emission spectra along a wide range of continuous wavelengths, addresses some of these problems. The data is analyzed with an algorithm that replaces compensation matrices and treats auto-fluorescence as an independent parameter. Thus, spectral flow cytometry should be capable of discriminating fluorochromes with similar emission peaks and can provide a multi-parametric analysis without compensation requirements.
This protocol describes the spectral flow cytometry analysis, allowing for a 21-parameter (19 fluorescent probes) characterization and the management of an auto-fluorescent signal, providing high resolution in minor population detection. The results presented here show that spectral flow cytometry presents advantages in the analysis of cell populations from tissues difficult to characterize in conventional flow cytometry, such as the heart and the intestine. Spectral flow cytometry thus demonstrates the multi-parametric analytical capacity of high-performing conventional flow cytometry without the requirement for compensation and enables auto-fluorescence management.
In the last few decades, flow cytometry (FCM) became a widely available analytical method essential for cell phenotyping studies. There has been a substantial increase in the available fluorescent dyes, particularly fluorochromes excited by the violet laser (405 nm) (e.g., Brilliant Violet and new Q-dot dyes). However, the growth of available fluorescent dyes increases the risk of overlapping emissions and requires labor-intensive compensation matrices. FCM became widely used to analyze cell suspensions from solid tissue, but the presence of auto-fluorescent cells limits the discrimination of specifically labeled populations.
The basic principles of spectral FCM are reported in detail in Futamura et al.1,2. Briefly, the spectral FCM used here (see the table of materials) is equipped with 405, 488, and 638 nm lasers. The spectral FCM captures all the emitted fluorescence as spectra in a 32-channel linear array PMT (32ch PMT) for 500 nm to 800 nm and 2 independent PMTs for 420 nm to 440 nm and 450 nm to 469 nm, respectively, which replace the conventional band-pass filters. The 488 and the 405/638 nm laser spots are spatially separated, while the 405 nm and 638 nm laser spots are co-linear. For each individual particle, the spectral FCM measures up to 66 channels of fluorescence data excited by the 405 nm and 488 nm lasers. When cells are excited by the 638 nm laser, the spectral FCM measures 58 channels of fluorescence data because a mask is inserted to prevent the 638 nm laser from shining into the PMT. Spectral FCM analyzes the acquired full spectrum data with an algorithm based on the weighted least squares method (WLSM), which enables the separation of overlapping fluorescent spectra. Spectra derived from single-stained and unstained samples are recognized as the basic reference spectra. Multi-stained samples are mathematically fitted and unmixed, and the spectrum of a sample with mixed fluorescent labels is decomposed into a collection of its constituent spectra. The unmixing, in spectral technology3,4, replaces the compensation that removes the signal from all detectors except the one measuring a given dye.
In this study, we combined and tested 19 fluorochromes in a single, 21-parameter analysis that characterized the major hematopoietic subsets found in the mouse spleen. Additionally, we demonstrated that the spectral cytometry can manage auto-fluorescent signal, thus improving the characterization of intestinal intra-epithelial lymphocytes and of the embryonic heart. Indeed, in these tissues, auto-fluorescence management allowed for the assignment of specific fluorescence to cells that would be excluded from the analysis in conventional FCM.
All experiments were performed according to the Pasteur Institute Ethic Charter and the EU guidelines and were approved by the French Agriculture Ministry.
1. Cell Suspension Preparation from Adult Mouse Organs
2. Preparation of Single-staining for Each Fluorochrome on Commercial Compensation Bead Microspheres (e.g., UltraComp eBeads), Hereafter Referred to as Beads
3. Cell Suspension Preparation from Embryonic Mouse Hearts
4. Acquisition of Labeled Splenic, Intestinal, and Cardiac Cells with the Spectral Flow Cytometer
5. Analysis of the Data with Spectral FCM Software
21-parameter spectral FCM panel to analyze splenocytes
Figure 1 shows the results obtained with the 19-fluorescent-antibody panel applied to splenic cells comprising different antibodies recognizing subsets of T, B, NK, dendritic, and myeloid cells, while CD45 labels all hematopoietic nucleated cells. The panel also includes a viability dye (PI), as well as the size (FSC) and granularity (SSC) parameters. Figure 1A shows the reference fluorochrome spectra. After the elimination of dead cells in the CD45+ gate (Figure 1B), the analysis shows that CD3+ T cells express the TcRβ chain and have the expected CD4/CD8 distribution and proportion of CD25+ cells in CD4 T cells. CD8 T cells show a normal distribution of CD44- and Ly49D-expressing cells. CD19+ B cells co-express B220 and MHCII, as well as IgM and IgD, typical of mature splenic B cells. A subset of NK1.1+ NK cells represents a small subset of the splenic lymphocytes that co-express Ly49D. The remaining cells comprise a small subset of CD11c+ dendritic cells with high levels of MHC II, some of which were also CD11b+. F4/80+ cells characteristic of tissue-resident macrophages co-express CD11b and the highest levels of CD44 found in the spleen. The remaining cells (CD19–Nk1.1–CD3–F4/80–CD11c–) could be subdivided by the expression of CD11b and Gr1, where the small subset of double-positive cells also express high levels of CD44, likely corresponding to progenitors known to be present at low frequencies in the spleen. A large fraction was negative for the three markers. Figure 2 shows the same analysis before adjusting the compensation.
Figure 1: A 19-color Antibody Panel for the Analysis of Murine Spleen Cells. (A) Mouse splenocytes were stained with the previous antibody combination, to which CD45-evolve 655 was added. The reference spectra of all fluorochromes are shown. Fluorophore-labeled CD45 is marked by a red arrow. The fluorochromes are listed in Table 1. (B) The contour plots show the discriminative capacity of this multi-parametric analysis to separate the different populations of B, T, NK, dendritic, or myeloid cells. The analysis was done in conventional FCM software after unmixing deconvolution.
Figure 2: A 19-color Antibody Panel for the Analysis of Murine Spleen Cells. The same analysis as in Figure 1 was done before the adjustment of the compensation with the reference spectra adjuster.
Small intestine intra-epithelial lymphocytes
Several T-cell populations are found in the intestine within the epithelial cell layer5,6. Conventional isolation of these subsets includes a density gradient that separates lymphocytes from epithelial cells. However, this preparation is delicate, time consuming, and results in cell loss. To improve quantification and to facilitate the experimental procedure, the potential of spectral FCM to discriminate populations of intestinal lymphocytes within a large fraction of epithelial cells was tested. After the mechanical disruption of the epithelial tissue, cell suspensions were stained with antibodies that label subsets of αβ and γδ T cells usually found in the intestinal epithelium. The cells were analyzed in two conventional cytometers and in spectral FCM.
Figure 3 shows the results after following a similar gating strategy, with elimination of dead cells gating on PI-negative cells. The presence of auto-fluorescent cells is obvious in all conventional cytometers and in spectral cytometry when the auto-fluorescence manager is inactivated (top panels, red arrows), resulting in a poor discrimination of the living cells. In all plots, cells within the FSC-A/SSC-A lymphocyte gate are shown in blue and cells with larger scatters (epithelial cell gate) are shown in red. Both conventional instruments were unable to resolve the CD3+ T-cell subset (in blue) (turquoise arrow) from the remaining cells. Therefore, TcRβ– cells (green arrow) contaminated the CD3+TcRγδ– population, a biologically improbable subset never detected after the separation of the lymphocytes from epithelial cells. In contrast, in spectral FCM after auto-fluorescence management, the analysis of the different T-cell subsets was not impaired by the presence of epithelial cells, the populations were well-separated, and there were no TcR-negative cells in the CD3+ gate.
Figure 3: The Presence of Auto-fluorescent Cells Does not Impair the Detection of Intra-epithelial Lymphocytes in Spectral FCM. Small intestinal cells, including epithelial cells and lymphocytes, were stained with antibodies recognizing TcRδ, PI, TcRβ Cy7-APC CD3 (panels B), Vγ7 APC, Vδ4 Cy7-PE (panels C), and CD8 FITC (panels D). PI was added in the FACS buffer before analysis. The acquisition of the data was done sequentially in the three instruments after the appropriate quality control. Doublets were eliminated in FSC-H/FSC-W. The analysis was done in conventional FCM software. Plots show the different steps of the gating strategy. Cells within the FSC-A/SSC-A lymphocyte gate are labeled in blue, while cells within the intestinal epithelial cell gate are labeled in red. Arrows point to different population distortions and auto-fluorescence. Plots in C show the analysis of data obtained in spectral FCM without autofluorescence management in the unmixing, whereas D shows the same data after unmixing with autofluorescence management.
Embryonic heart
Conventional FCM showed a major population of auto-fluorescent cells (black arrow) that was excluded from the analysis because it fluoresced in the CD45 and TER119 dump channel (PE), marking hematopoietic cells. In contrast, this auto-fluorescent population was not present as such in spectral FCM and was comprised in the CD45–TER119– subset (Figure 4). To show that these cells were cardiac and not endothelial or hematopoietic, expressing low levels of CD45, TER119, or CD31, the expression of transcripts specific to cardiomyocytes were quantified on sorted cells. High levels of cardiac muscle troponin T (TNNT2)7 and atrial light chain-2 (MYL7)8 transcripts were found in the auto-fluorescent population.
Figure 4: Auto-fluorescent Cells in E17.5 Cardiac Cell Suspensions are Properly Included in the Negative Cell Fraction in Spectral FCM. (A) Cell suspensions from fetal hearts were stained with anti-TER119 and CD45-PE, Sca-1-PECy5, and CD31-APC antibodies and sequentially acquired in the three instruments. Black arrows show auto-fluorescent cells in the two conventional cytometers, while these cells are not detected as auto-fluorescent in spectral cytometry and can thus be analyzed for specific fluorescent staining. (B) Auto-fluorescent cells (within the elliptical gate in green), but not CD31+ cells (rectangular gate in blue, negative control), showed the expression of cardiac troponin (TNNT2)7 and atrial light chain-2 (MYL7)8 transcripts. a.u. – arbitrary units relative to the GAPDH house-keeping transcript. Data presented as mean value ± SEM.
Excitation laser | Fluorochrome | Specificity | Clone | |
488 nm | BB515 | CD8 | 53-6.7 | |
488 nm | FITC | Ly49D | 4E5 | |
488 nm | PE | F4/80 | 6F12 | |
488 nm | PE-CF594 | NK1.1 | PK136 | |
488 nm | PE-Cy5 | CD44 | IM7 | |
488 nm | PE-Cy7 | CD11b | M1/70 | |
488 nm | PerCP-Cy5.5 | IgM | R6-60.2 | |
488 nm | PerCP | Gr-1 | RB6-8CS | |
488 nm | AF532 | CD3 | 17A2 | |
488 nm | Propidium Iodide | Viability | ||
405 nm | V450 | IgD | 11-26c.2a | |
405 nm | BV421 | CD11c | HL3 | |
405 nm | BV510 | CD19 | 1D3 | |
405 nm | BV570 | TCRb | H57-597 | |
405 nm | BV605 | CD4 | RM4-5 | |
405 nm | BV650 | CD45R/B220 | RA3-6B2 | |
405 nm | BV711 | I-A/I-E | M5/114 | |
405 nm | BV786 | CD25 | PC61 | |
405 nm | eVolve 655 | CD45 | 30-F11 | |
N/A | purified | CD16/CD32 | 2.4G2 | Fc-receptor block |
List of Antibodies used in cardiac cell suspension | ||||
638 nm | Alexa Fluor 647 | CD31 | MEC13.3 | |
488 nm | PE | CD45 | 30-F11 | |
488 nm | PE-Cy5 | Sca-1 | D7 | |
488 nm | PE | Ter119 | TER-119 |
Table1: List of Antibodies Used in the 19-color Panel and in Cardiac Cell Suspension.
Conventional FCM is based on the detection of photons emitted after the excitation of fluorescent probes. The fluorescence emission of one fluorochrome detected in a detector designed to measure the signal from another fluorochrome induces physical overlap. This spillover among emission spectra needs to be corrected by compensations.
Spectral FCM and data processing by the unmixing deconvolution algorithm allows for the combination of fluorochromes with close emission peaks without additional compensation, provided that they have different spectral shapes. The algorithm used for the analysis treats auto-fluorescence as an independent parameter, subtracting non-specific fluorescence from solid tissue cell suspensions.
A panel of 19 fluorochromes was assembled to analyze immune cells using only two lasers (488 nm and 405 nm). Indeed, when the 638 nm laser was on, a section of the 32-channel PMT corresponding to the excitation wavelength of this laser was not available. To get the maximum detection capacity of the PMT, the red laser was switched off. Mouse splenic cells were analyzed, allowing for the detection of more than 12 different cell populations, some of which comprised 1% or less of the hematopoietic cells.
To test the capacity of spectral FCM to manage auto-fluorescence, two different situations that can impair the analysis were tested: one where lymphocytes were contaminated with epithelial cells, and one where auto-fluorescent cardiac cells were analyzed. In contrast with conventional FCM, spectral cytometry allows for the discrimination of all lymphocyte subsets, even when they represent minor subsets.
The controversy around the regenerative capacity of adult cardiomyocytes9,10,11 is in part due to the absence of a strategy to discriminate and isolate distinct subsets of viable cells in the heart. A multi-parametric analysis defining unique combinations of markers to discriminate the different cardiac subsets might enable the detection of rare subsets because of large numbers of cells that can be analyzed. Fetal hearts (E17.5) were analyzed, and a large fraction of auto-fluorescent cells, which would have escaped detection in conventional FCM, was integrated in the non-hematopoietic viable cardiac cell compartment in spectral FCM.
A recurrent problem of this method is the requirement for good-quality fluorescent labeled antibodies. This is particularly critical when using tandem dyes, where a component of the tandem is dominant. For example, BV786 and Cy7-PE can be difficult to distinguish from BV421 and PE, respectively.
FCM based on spectrometry provides analytical power higher than that obtained with conventional FCM, without the need for compensation matrices. This manuscript provides evidence that there is an alternative approach that demonstrates a high analytical capacity, the absence of complex compensation matrices, the unimpaired discrimination of cell subsets due to auto-fluorescence, and no additional financial commitments that still required for mass spectrometry FCM. Spectral FCM thus appears to be a method of choice for the analysis of cells obtained from a wide variety of solid tissues usually not amenable to conventional FCM and can be used in tumor, developmental, and stem cell biology.
The authors have nothing to disclose.
We acknowledge the technical and theoretical contributions in spectral cytometry of K. Futamura, who also critically revised the manuscript. We are also indebted to C. Ait-Mansour, P.-H. Commere, A. Bandeira, and P. Pereira for critically reading the manuscript and for endless technical advice and support. We also thank P. Pereira for the gift of the anti Vγ7-APC and Vδ4-Biotin labeled antibodies. We aknowledge the Centre d’Enseignements from Pasteur Institute for welcoming and supporting the filming logistics. This work was supported by the Pasteur Institute, Institut National de la Santé et de la Recherche Médicale (INSERM); the Swiss National Science Foundation and Pasteur Bourse Roux (S.S.); Fundação para a Ciência e a Tecnologia – SFRH/BD/74218/2010 (M.V.); and Université Paris Diderot, the Agence Nationale de la Recherche (ANR) project Twothyme, the ANR, and Program REVIVE (Investment for the Future) (A.C.).
Mice | Janvier Labs | CD45.2 | Source of cells |
Ethanol 70% | VWR | 83801.36 | Sterilization |
DPBS (Ca2+, Mg2+) | GIBCO ThermoFisher | 14040-174 | Embryos collection |
Hanks' Balanced Salt Solution (+Ca2+ +Mg2+) (HBSS+/+) | GIBCO Life ThermoFisher | 14025092 | Dissociation, digestion and staining solutions |
Hanks' Balanced Salt Solution (-Ca2+, -Mg2+) (HBSS-/-) | GIBCO Life ThermoFisher | 14175095 | Dissociation, digestion and staining solutions |
Fetal calf serum (FCS) | EUROBIO | CVFSVF00-0U | Dissociation, digestion and staining solutions |
Collagenase | Sigma | C2139 | Enzymatic solution |
90 x 15 mm, plastic tissue culture Petri dishes. |
TPP | T93100 | Embryos and hearts collection |
35 x 15 mm, plastic tissue culture Petri dishes. |
TPP | T9340 | Hearts collection |
Fine iris scissor | Fine Science Tools | 14090-09 | Dissection tools |
Gross forceps (narrow pattern forceps, curved 12 cm | Fine Science Tools | 11003-12 | Dissection tools |
Fine forceps (Dumont no. 7 forceps, | Fine Science Tools | 11272-30 | Dissection tools |
Scalpel | VWR | 21909-668 | Dissection tools |
LEICA MZ6 Dissection microscope | LEICA | MZ6 10445111 | Sampling; Occular W-Pl10x/23 |
Cold lamp source | SCHOTT VWR | KL1500 compact | Sampling; Two goose neck fibers adapted |
FACS tubes | VWR | 60819-310 | Digesting cells |
96 well plate (round bottom) | VWR | 10861-564 | Staining cells |
Nylon mesh bolting cloth sterilized, 50/50 mm pieces. |
SEFAR NITEX | SEFAR NITEX | Cell filtration |
UltrasComp eBeads | |||
Fluorescence labeled antibodies | Biolegend | Catalogue number see table below | Staining cells |