Evaluation of intranuclear BrdU detection procedures for use in multicolor flow cytometry




Measurement of cell proliferation via BrdU incorporation in combination with multicolor cell surface staining would facilitate studies on cell subsets that require multiple markers for their identification. However, the extent to which the often harsh cell preparation procedures required affect the staining quality of more recently developed fluorescent dyes has not been assessed.


Three cell preparation protocols for BrdU measurement were compared for their ability to maintain fluorescent surface staining and scatter parameters of in vivo BrdU-labeled cells by flow cytometry. A 10-color fluorescent panel was developed to test the quality of surface staining, following cell treatment and the ability to perform BrdU measurements on even small B lymphocyte subsets.


All cell preparation procedures affected the quality of fluorescent and/or scatter parameters to varying degrees. Paraformaldehyde/saponin-based procedures preserved sufficient fluorescent surface staining to determine BrdU incorporation rates among all splenic B cell subsets, including B-1a cells, which constitute roughly 0.5% of cells. Turnover rates of B-1a cells were similar to immature B cells and higher than those of the other mature B cell subsets.


Paraformaldehyde/saponin-based cell preparation procedures facilitate detailed cell turnover studies on small cell subsets in vivo, revealing new functional information on rare cell populations. © 2006 International Society for Analytical Cytology

Flow cytometry is the method of choice for identifying and enumerating even small lymphocyte subsets (1, 2), because many measurements can be performed simultaneously. All currently known splenic B lymphocyte subsets can be identified using a single cocktail of mAb-conjugates to eight cell surface receptors (3). Splenic B cell subsets include follicular B cells, the majority B cell population in the spleen (roughly 75–80%); “transitional” B cells, recent bone marrow emigrants, which are further divided by their levels of CD23 expression into T1 and T2 (5–15%) (4); marginal zone (MZ) B cells (5–10%); and a small B cell subset termed “B-1” (1%), which are further distinguished into CD5+ “B-1a” and CD5- “B-1b” cells (5, 6).

In the spleen, B-1a cells are characterized by the expression of CD5, a surface marker otherwise expressed exclusively on T cells, low-level expression of IgD and the pan-B cell marker B220 (CD45R), as well as relatively high expression of IgM and CD19 (3, 5, 6). Additional markers include CD9 (7) and CD43 (8). Since none of these markers exclusively stain B-1 cells, combinations of multiple antibody-conjugates used in conjunction with multicolor flow cytometry are an important tool for the analysis of this B cell subset. It is undisputed that B-1 cells generate most circulating “natural” IgM, thereby providing an important role early in immunity to infections (9, 10). Much controversy, however, has arisen over the developmental origins of these cells. B-1 cells are thought to arrive either from distinct precursors expressed mainly in the fetal liver and spleen or by selection events from bone marrow precursors (reviewed in (6, 11)). After selection, this cell population is thought to be maintained through slow cell turnover or “self-renewal” (5, 6, 12). Comparative analysis of the dynamics of all B cell subsets in the spleen, including that of B-1 cells, might greatly aid our understanding of the development and regulation of B cells in the adult, and thus help to resolve the current controversies regarding the B-1 cell subset.

Lymphocyte dynamics in vivo have been measured successfully using BrdU-labeling, followed by 3 and 4-color flow cytometric or microscopic analysis (13–19). This assay relies on the identification of BrdU incorporated into the nuclear DNA of dividing cells. The cell fixation and permeabilization regimes required to gain access to the cell nucleus for BrdU labeling of DNA could significantly limit the usefulness of this approach in multicolor flow cytometry. Harsh fixation and permeabilization procedures might alter or destroy the cell surface receptors used for cell identification, or the fluorochromes with which antibodies to these receptors are tagged (18, 20, 21). Therefore, we have sought to evaluate the applicability of various BrdU staining procedures for multicolor flow cytometry. We report here on the identification of a cell fixation/permeabilization method for detection of intranuclear BrdU that can be used in conjunction with 13 distinct fluorochromes most widely used for surface staining in multicolor flow cytometry. Using this technique, we identify the splenic B-1a cell subset as a B cell population, with surprisingly high turnover rates.


Mice and In Vivo Labeling with Bromodeoxyuridine (BrdU)

Eight to fifteen week old female BALB/c mice (Charles River, Maine) were used for all experiments. Mice were kept under conventional housing conditions at the animal facility at UC Davis, Davis, California. All experiments were performed in accordance with approved UC Davis Animal Use and Care protocols. In vivo BrdU labeling of thymocytes was done by intraperitoneal (i.p.) injection of BrdU (Sigma-Aldrich, Dallas, TX; 1 mg/100 μl of PBS per mouse) for 1 h prior to tissue isolation. Other cell populations were labeled by administering BrdU via drinking water (1 mg/ml water given ad lib) for up to 28 days. The water was sterile filtered, protected from light, and exchanged every 3–4 days (14).

Cell Preparation and Surface Staining

All cell preparation procedures, including cell washes were carried out in “staining medium” (buffered saline solution (BSS): 0.168 M NaCl, 0.168 M KCl, 0.112 CaCl2, 0.168 M MgSO4, 0.168 M KH2PO4, 0.112 M K2HPO4, 0.336 M HEPES, 0.336 M NaOH, containing 3.5% heat inactivated, filtered newborn calf serum, 1 mM EDTA, and 0.02% sodium azide) on ice. Single cell suspensions were prepared by pressing tissues between the frosted ends of two glass slides. Cell debris was removed by filtering the ruptured tissues through nylon mesh (50 μm pore size). Erythrocytes were lysed by incubation on ice in ammonium chloride lysing buffer (150 mM NH4Cl, 10 mM KHCO3, and 10 mM EDTA, pH 7.2–7.4) for 1 min. Cells were washed and filtered twice before Fc-receptor blocking via 15 min incubation with anti-CD16/32 mAb (clone: 2.4G2). Cells were then stained for 20 min with cocktails of various anti-murine antibody conjugates. For indirect staining using biotinylated antibody-conjugates, cells were washed followed by staining with SA-fluorochrome conjugate for 15 min. Cells were filtered once more when data were acquired on a FACSAria (Becton Dickinson, San Jose, CA).

Antibodies and Conjugates

Commercial reagents.

SA-QDOT605 (Quantum Dot Corporation, Hayward, CA); SA-CasBlue and PacBlue (Molecular Probes, Eugene, OR); anti –I-A/I-E-FITC, anti-CD3 (2C11)-Biotin, anti-BrdU (B44)-FITC, anti-CD19 (1D3)-PE, and anti-CD4 (RM4-5)-APC (BD Biosciences/Pharmingen, San Jose, CA); anti-BrdU (3D4)-FITC, anti-CD45R (RA3-6B2) and anti-CD8a (5H10)- Alexa610PE, anti-CD19 (6D5)-Cy5.5APC, and Cy5.5PE, SA-Cy7APC (Caltag, Burlingame, CA); anti-Pan NK-Biotin (DX-5), SA-Cy5PE, anti-CD4 (GK1.5)-Cy5PE, anti-CD4 (RM4-5)-Cy5.5PE and Cy7APC, anti-CD19 (MB19-1)-FITC and Cy5PE and APC, (eBioscience, San Diego, CA).

In-house generated conjugates.

Anti-CD4 (GK1.5) and anti IgD (1126)-CasYellow; anti-CD45R (RA3-6B2)-PacBlue; anti-CD43 (S7)-Pacific Blue (PacB); anti-CD4 (GK1.5)-FITC; anti-CD4 (GK1.5) and anti-CD5 (53–7.3)–PE, and anti-macrophage (F4/80)-, anti-granulocyte (GR-1)-, anti-CD4 (GK1.5)-, anti-CD8 (53.6.7), anti-CD19 (1D3)– Biotin; anti-CD11b (M1/70) and anti-CD21 (7G6)-Cy5.5PE; anti-IgD (1126) and anti-CD4 (GK1.5) –Cy7PE; anti-CD23 (B3B4); anti-CD8 (53.6.7) and anti CD45R (RA3-6B2)-APC; anti-CD4 (GK1.5) – Cy5.5APC; anti-IgM (331) and anti-CD45R -Cy7APC. Antibody-conjugates were prepared by growing hybridomas in serum-free HB101 media (Basal Medium, Irvine Scientific, Santa Ana, CA). Hybridoma supernatants were concentrated using an ultra filtration system (Millipore, Billerica, MA). After purifying the antibodies using a “HiTrap Protein G HP”, (Amersham Biosciences/GE Healthcare, Piscataway, NJ) affinity column, they were conjugated to various fluorochromes according to published protocols (www.drmr.com).

Cell Preparation Procedures to Measure Intranuclear BrdU

Protocol 1A and 1B.

For Protocol 1A solutions and fixative were prepared as described (20). Briefly, 4% PFA/1% saponin (w/v) fixation/permeabilization solution (FixPerm) was prepared by dissolving the saponin in fixative (4% PFA) solution and adjusting the pH to 7.4–7.6. The permeabilization (Perm) solution (here 1% saponin w/v) was prepared by first making a 10% (w/v) saponin stock solution in Earle's balanced salt solution (EBSS) without phenol red (HyClone, Logan, UT), and then diluting this stock at 1 to 10 in 0.01 M HEPES. The wash solution consisted of 0.5% saponin in 0.01 M HEPES, prepared from the 10% saponin stock solution. The freezing media contained 10% DMSO and 90% heat inactivated fetal bovine serum (FBS). For Protocol 1B, 10× Perm/Wash buffer and Cytofix/Cytoperm solution were obtained commercially (BD Biosciences Pharmingen, San Jose, CA).

Surface-stained cells were fixed and permeabilized with Fixperm solution (1A) or Cytofix/Cytoperm (1B) for 30 min on ice. The surface staining as well as the first fixation/permeabilization procedures were carried out in 96-well plates. Twenty-five microliters of cell suspension at 2.5 × 107/ml were mixed with 75 μl of fixation/permeabilization solution. Cells were washed three times with 200 μl of wash solution (1A) or 1× Perm/Wash buffer (1B). For protocol 1A, the cells were further permeabilized in 75 μl of Perm solution for 10 min prior to washing in wash solution. After washing in staining media, the cells were frozen to −80°C in freezing media to permeabilize the nuclear membrane (22). For this, cells were transferred from plates into 5-ml round bottom tubes, and frozen with 1 ml of media for at least 12 h. After quick thawing of the cells at 37°C (water bath), the cells were washed with 1 ml of staining media. Cells were resuspended in corresponding fixation/permeabilization solutions and incubated for 10 min on ice and washed with 1 ml wash solution. For digestion of cellular DNA, cells were incubated in a 37°C water bath for 1 h with freshly prepared DNase I (Worthington lyophilized DNase I, source: bovine pancreas) at 2.5 U/ml KDS-BSS. Optimal BrdU staining was achieved with 4 units DNase/ml KDS-BSS per 106 cells (Table 1, Fig. 2B and data not shown).

Figure 2.

Evaluation of BrdU staining protocols on fluorescence and scatter parameters. A: Shown are 5% contour plots with outliers depicting the effects of 4 different fixation/permeabilization procedures as outlined in “Materials and Methods” and summarized in Table 1 (from top to bottom protocols 1A, 1B, 2, and 3). Two mouse thymi were pooled after 1 h pulse of BrdU (1 mg BrdU/100 μl PBS per mouse) through i.p. injection. Samples were run in quadruplicates. Data were acquired over three different days with a 4-color FACS Calibur (Becton Dickinson), which was calibrated prior to each run as outlined in “Materials and Methods.” Arrows indicate the flow of the data analysis. The scatter profiles are shown in the left panels. Lymphocytes were selected from those plots, CD4 CD8 “double positive” T cells identified (middle panels), and average (n = 4) % BrdU+ frequencies determined (right panels). B: Shown are 5% contour plots with outliers of thymocytes selected for expression of CD4 and CD8. Cells were treated according to protocol 1B with indicated concentrations (Kunitz U DNAse/106 cells) prior to staining with anti-BrdU antibody. Results shown in the top and bottom row are from two independent experiments. Separation indices (S.I.) for BrdU staining are indicated for each DNAse concentration.

Table 1. Summary of Cell Preparation Procedures for Detection of Intranuclear BrdU
 aProtocol 1AProtocol 1BProtocol 2Protocol 3
  • a

    Treatment protocols are outlined in more detail in “Materials and Methods.”

  • b

    Separation index for BrdU staining calculated from mean values obtained from samples stained in quadruplicate. Results are from at least 3 independent experiments that yielded similar results.

Fixation and permeabilization of surface-stained cells1. Fix and permeabilize in 4% PFA (w/v), 30 min. on ice1. Fix and permeabilize in BD Cytofix/Cytoperm™ 30 min on icePerm solution (PBS/1% formaldehyde/0.0005% NP40) at 4°C overnight1. Ice-cold 95% ethanol 30 min
2. Permeabilize in 1% saponin for 10 min2. Freeze cells in 10% DMSO/90% FBS at −80°C overnight2. PFA (1%) (w/v)/0.01 % Tween in PBS 30 min at room temperature
3. Freeze cells in 10% DMSO/90% FBS at −80°C overnight3. Refixation/permeabilization in BD Cytofix/Cytoperm™, 10 min on ice 
4. Refixation/permeabilization in 4% PFA (w/v), 10 min on ice  
Cells are washed in 0.5% saponinin BSSCells are washed in 1× BD Perm/wash bufferCells are washed in PBS and NP40/PBSCells are washed in PBS
DNase I treatment4 Kunitz units/ml DNase for 1 × 106cells 37°C for 1 h4 Kunitz units/ml DNase for 1 × 106cells37°C for 1 h50 Kunitz units/ml DNase 37°C for 30 min50 Kunitz units/ml DNase 37°C for 30 min
BrdU labelingAnti-BrdU mAb in BSS/0.5% saponinAnti-BrdU mAb in 1 × BD Perm/wash bufferAnti-BrdU mAb in NP40/PBS solutionAnti-BrdU mAb in PBS
Advantages of protocolPE and APC signal maintainedPE and APC signal maintainedPE and APC signal maintainedPE signal maintained
No “cell clumping” observed by forward angle light scatteringNo “cell clumping” observed by forward angle light scattering No “cell clumping” observed by forwardangle light scattering
Disadvantages of protocolIncreased autofluorescenceIncreased autofluorescenceIncreased autofluorescenceIncreased autofluorescence
Reduction in quality of surface stainingReduction in quality of surface stainingReduction in quality of surface stainingTotal loss of surface staining with APC
Strong changes in forward angle light scatter signals suggesting excessive cell clumping
bSeparation Index BrdU stain4.7 ± 1.66.2 ± 1.110.2 ± 2.534.9 ± 19.8

Protocol 2.

Fixperm solution was 0.0005% NP40/1% formaldehyde (v/v). The wash solution was 0.5% NP40/PBS. Surface stained cells were transferred into 5-ml round bottom tubes, resuspended in 1 ml of Fixperm solution, and stored at 4°C overnight. Cells were washed in 1 ml of PBS, and cellular DNA was digested by incubation with freshly prepared DNase I at 50 U/ml KDS-BSS in a 37°C water bath for 30 min.

Protocol 3.

Reagents were prepared as described (23). The Fixperm solution consisted of 95% ethanol/1% PFA/0.01% Tween 20. After surface staining, protocols outlined by Tough and Sprent (23) were followed exactly as described. DNAse treatment was as for protocol 2.

Independent of the cell preparation procedures used, cells were washed following DNAse treatment in 1 ml of wash solution. Anti-BrdU mAb (B44) was added at a previously determined optimal concentration in wash solution (protocol 1A, 1B, and 2) or in PBS (protocol 3). Cells were incubated for 20 min (protocol1A, 1B), 45 min (protocol 2), or for 30 min (protocol 3) at room temperature or on ice (protocol 2). Cells were washed with 1 ml of the appropriate wash solution and resuspended in 200 μl of staining medium for analysis on a flow cytometer.

Flow Cytometric Analysis

Four-color measurements of thymocytes were performed using a FACSCalibur (BD Biosciences, San Jose, CA). For multicolor analyses of splenocytes, a FACS-Aria (BD Biosciences) was used, equipped with three lasers (100 mW, 488 nm; 20 mW, 633 nm; and 15 mW, 407 nm excitation wavelengths, respectively), custom-ordered dichroics and bandpass filters, and DIVA-software for data acquisition. The optical path configuration of the FACS Aria is outlined in Figure 1. Both flow cytometers were calibrated prior to each run, using “rainbow fluorescent particles” (Spherotech, Libertyville, IL) to allow direct comparisons of data collected over multiple experiments and to ensure the ability to measure and properly compensate all parameters. For calibration, voltage settings for each parameter were chosen so that the bead signals would fall within narrowly defined optimal channel numbers, as previously determined with stained murine splenocytes. At least 200,000 events per sample were collected for multicolor analyses of splenocytes, and 100,000 events for thymocyte stains. Data were analyzed using FlowJo software (Treestar, Ashland, OR).

Figure 1.

FACS-Aria optical set-up. Shown are the excitation and emission optics of a FACS-Aria (Becton Dickinson) equipped with three lasers, a 100 mW blue laser, a 20 mW red laser, and a 15 mW violet laser, which emit a 488 nm (left), 633 nm (middle), and a 407 nm (right) laser line, respectively. Custom-ordered band pass filters (large rectangles) and dichroic mirrors (smaller rectangles) (Chroma Technologies, Fort Collins, CO) were added to allow for the simultaneous detection of 12 colors and 2 scatter parameters. The “filtered” light signal is detected by photomultiplier tubes (PMTs, shown as circles): 7 PMTs detect light from the blue laser signal, 3 from the red laser, and 3 from the violet laser signal.

Assessing the Quality of a Stain

To evaluate the quality of a particular FACS stain and the effects of the different treatment protocols on staining, we calculated a “separation index”: [Median fluorescence intensity (MFI, arbitrary units) positive (stained) – MFI background (unstained)]/2× Standard Deviation (SD) of the fluorescence intensity (background) (24). MFI and SD were obtained using the FlowJo software.

Statistical Analysis

Statistical analysis was done using a two-tailed paired t-test with help of the Prism 4 software (GraphPad Software, San Diego, CA, USA).


Treatment with Paraformaldehyde/Saponin Best Preserves Scatter and Fluorescent Parameters

We sought to identify a BrdU staining protocol that could be combined with multicolor flow cytometry for the study of turnover rates among small B cell subsets. Four anti-BrdU staining procedures (summarized in Table 1) were compared. Protocols 1A and 1B used PFA/saponin-based fixation procedures. These protocols differed only in the source of the reagents used for fixation and permeabilization (in-house generated versus commercial, respectively). Both protocols were modified from those published as the “option 2” pathway (22). Protocol 2 used formaldehyde/NP40 as a fixation and permeabiilization solution and protocol 3 used ethanol/PFA/Tween 20.

First, we aimed to determine the effects of fixation, permeabilization, and DNase I treatment procedures on the maintenance of scatter and fluorescent parameters of surface-stained cells. For the initial analysis, thymocytes from two mice injected 1 h prior to tissue harvesting with BrdU were isolated. The thymus was chosen because this tissue contains CD8 and CD4 “double positive” T cell precursors that are known to rapidly divide (Fig. 2A). Since APC and PE are commonly used as fluorochromes for antibody conjugates and as part of a number of frequently used tandem dyes (such as Cy5-, Cy5.5-, Cy7- PE or -APC), we first tested whether fixation/permeabilization procedures would compromise stability of those two dyes.

As shown in Figure 2A, surface staining for APC and PE was maintained when PFA/saponin (protocol 1A and 1B) and formaldehyde/NP40 (protocol 2) based fixation/permeabilization methods were used. In contrast, APC but not PE signals were lost when cells were fixed/permeabilized with ethanol/PFA/Tween (protocol 3). With regard to the scatter parameters, extensive changes were noted only for protocol 2 (Fig. 2A). The observed strong increases in the scatter measurements appeared to be due to excessive cell clumping.

A “separation index (S.I.)” was calculated from the mean of MFI of quadruplicate samples of the same pool of thymocytes treated with the various protocols to assess the quality of BrdU staining (24). While protocols 2 and 3 showed a greater S.I., i.e. better separation of stained from nonstained cells compared with protocols 1A and 1B (Table 1 and Fig. 2A), the cell clumping observed with protocol 2 and the lack of an APC signal with protocol 3 precluded their use in multicolor flow cytometry.

Permeabilization of the nuclear membrane is crucial for optimal access of the anti-BrdU antibody to the DNA. This is achieved in protocols 1A and 1B by freezing cells in DMSO prior to DNase treatment (22) or by using the Cytoperm Plus Buffer (protocol 1B) provided with the BrdU Flow Kit. The quality of BrdU stain with protocols 1A and 1B was strongly affected by the concentration of the DNase used for DNA digestion procedures (Fig. 2B); this step is required to generate single-stranded DNA that can be accessed by anti-BrdU mAb (19). We noted that a very narrow range of DNAse concentrations would give adequate staining (S.I. > 2.0) (Fig. 2B). Careful titration showed that 4 U DNase/106 cells in 1 ml KDS-BSS gave the highest quality, i.e. largest S.I. (Fig. 2B and data not shown). When larger numbers of cells were stained, the concentration of DNAse was adjusted to maintain the same ratio of cells/DNAse while maintaining the staining volume.

Comparison of in-house-generated fixperm solutions to commercial reagents (protocol 1A vs. 1B) showed a reduced separation of positive and negative events with in-house reagents (S.I. of 4.7 ± 1.6 and 6.2 ± 1.1, respectively). This was due to both reduced positive BrdU staining and increased background. However, staining with both protocols showed sufficient separation between BrdU positive and negative populations (S.I. > 4.0, Fig. 2A). Further optimization of in-house reagents, which was not attempted here, would likely improve the staining obtained with these reagents. Since the PFA/saponin-based intracellular staining procedures were superior at maintaining fluorescent (APC, PE, FITC) and scatter parameters, all subsequent experiments were carried out following protocol 1.

Effects of Fixation/Permeabilization Regimenson Mouse Splenocytes Stained Individually with 13 Different Fluorochromes

To determine how the PFA/saponin-based fixation/permeabilization procedures (protocol 1) affect the stability of fluorochromes frequently used in multicolor flow cytometry, mouse splenocytes were surface-stained prior to fixation and permeabilization with one T cell- and one B cell-specific antibody conjugated to 13 different fluorochromes (Fig. 3). We compared untreated (live) surface stained cells with treated (fixed/permeabilized) splenocytes, using the same cell sample and staining panels for both groups. Cells were analyzed on a multicolor flow cytometer with the optical set-up depicted in Figure 1. As shown in Figure 3, all fluorochromes, with the exception of Cascade Yellow, resulted in clearly distinct staining for a subset of cells. Staining with Cascade Yellow was not found to be useable following any of the fixation procedures (Fig. 3 and data not shown). Both in-house and commercial reagents gave similar results (data not shown).

Figure 3.

Effects of PFA/saponin-based cell preparation procedures on surface staining. Thirteen fluorochromes conjugated to a B cell marker (red) and a T cell marker (blue) were used to compare surface staining on nontreated/live splenocytes (solid line) with that on cells stained according to protocol 1B (dashed line). Shown are overlay histogram profiles after selection for lymphocyte forward and side scatter. B cells were either stained with direct conjugates to CD19 FITC/PE/PE-Cy5/PE-Cy5.5/APC/APC-Cy5.5, CD45R-PE-Alexa610/APC-Cy7/PacB, and IgD-PE-Cy7/CasY or indirectly using biotinylated CD19 plus streptavidin (SA) CasB (not shown) and SA-QDOT605™. T cells were stained with direct conjugates to CD4 FITC/PE/PE-Cy5/PE-Cy5.5/PE-Cy7/APC/APC-Cy7/APC-Cy5.5/CasY, and to CD8-PE-Alexa610. Anti-CD3-biotin was used for indirect staining with SA-CasB/PacB and SA-QDOT605™. Excitation wavelengths are noted on the left. [Color figure can be viewed in the online issue, which is available at www.interscience. wiley.com.]

The quality of the analyzed surface stains was compromised by the fixation and permeabilization procedures for most of the fluorochromes tested (reduction in S.I. following treatment, Table 2). This was due mainly to an increase in background staining (Table 2). Changes in background staining were similar for T and B cells (data not shown). In addition, some fluorochromes, such as PE and most of its tandem-dyes as well as PacBlue and Qdot605, showed reduced staining of the positive cells (Table 2), independent on the specificity of the antibody (data not shown). For the 633 nm-excited dyes, degradation was due mostly to increased background staining.

Table 2. Effects of PFA/Saponin Treatment on Quality of Cell Surface Stain
 SpecificityaMedian positiveaMedian backgroundbSD backgroundcSeparation index
  • a

    Median fluorescence intensity (arbitrary units) of positive (stained) and background (unstained) mouse spleen cells. Mean values of stains performed in triplicate.

  • b

    SD standard deviation of background (unstained) fluorescence intensity (arbitrary units). Mean values of stains performed in triplicate.

  • c

    Separation Index for BrdU staining: (Medianpositive − medianbackground)/(2 × SDbackground) (24).

  • d

    Values obtained from single stain.

  • e

    n.d., not determined; Separation of negative and positive stained cells was lost upon treatment.

PE-Alexa 610CD45R    
QDOT 605™Streptavidin(CD3-biotin)    

In vivo BrdU Labeling in Conjunction with 10-Color Flow Cytometry to Phenotypically Characterize Small Proliferating Lymphocyte Populations

Because of spectral overlaps and necessary compensation, degradation of staining quality is a complication in multicolor flow cytometry (2). We therefore sought to determine whether BrdU-detection procedures could be combined with multicolor surface staining, despite the noted degradation of the individual stained cells. We applied a 9-color staining panel that distinguishes the known murine B lymphocyte subsets in combination with measurements of BrdU (Fig. 4A). The staining cocktail and subsequent analyses relied on a “dump” channel (Cy5-PE: CD3, 4, 8, F4/80 GR-1) and a pan-B cell marker (Cy5.5-APC: CD19) to cleanly identify B cells in a first step. This was followed by separating cells according to a number of surface markers known to be differentially expressed on the various B cell populations (CD21, CD23, CD43 and CD5, Fig. 4A and B220, not shown). Finally, staining for surface immunoglobulins IgM and IgD (Fig. 4A) confirmed the expected phenotype. These were Transitional B cells, CD23−/lo, CD21−/lo, CD5−, CD43-IgMintm/hi, IgD−/lo; B-1a cells, CD23−, CD21lo, CD43+, CD5+, IgMint/hi, IgDlo; Marginal zone B cells, CD23−, CD21hi, CD5 CD43, mostly IgMhi and IgD+/hi; Follicular B cells, CD23+, CD21int, CD43 CD5 mostly IgMlo, and IgDhi.

Figure 4.

Analysis of B cell subset proliferation by 10-color flow cytometry. A: Shown are 5% contour plots of spleen cells from one representative mouse (of 4 analyzed) continuously fed for 14 days with BrdU via drinking water. Gating strategies for the exclusion of nonB cells followed by positive identification of B cells via B220 and CD19 are indicated. B cells were further divided into follicular B cells (foll. B) and marginal zone (MZ) B cells according to differences in CD21 and CD23 expression. CD21 and CD23 double negative B cells were further divided into transitional (trans. B) and B-1a cells according to their expression of CD5 and CD43. Numbers indicate frequencies (%) of cells within the indicated gate. B: Shown are histogram plots of BrdU-labeling among the indicated B cell subsets. Horizontal line indicates gating of BrdU-positive cells. Numbers are frequencies of BrdU positive cells within the B cell subset. C: Left panel shows the frequencies (Mean ± SD, per timepoint n = 4 mice) of BrdU-positive events among each of four splenic B cell subsets (transitional B cells, open diamond; B-1a cells, stars; marginal zone B cells, closed triangles; and follicular B cells, open squares), identified as shown in A, over time of BrdU treatment. Right panel shows turnover rates calculated as % BrdU-labeled B cell subsets (left panel) divided by the number of days the mice (n = 16) were fed BrdU. Horizontal line indicates mean turnover rates. All pair-wise comparisons of turnover rates of the different B cell subsets showed highly significant differences. Note the strong BrdU uptake of B-1a cells compared with marginal zone and follicular B cells (P < 0.0001).

Our data show that the cell preparation procedures in protocol 1 sufficiently maintained surface staining so that known splenic B cell subsets could be separated cleanly (Fig. 4A). Furthermore, multicolor flow cytometric analysis of individual mice (n = 20) resulted in frequencies for each B cell subset that were similar to those previously published (3): transitional B cells (6.7 ± 2.3)%; B-1a (0.8 ± 0.2)%; marginal zone B cells (10 ± 1.4)%; and follicular B cells (69 ± 5.1)%. Thus, the identified fixation/permeabilization regimen can be used for accurate measurement of even small cell subsets by multicolor flow cytometry.

B Cell Subset Turnover Rates in the Spleen

We next combined this 9-color surface-staining panel with anti-BrdU-FITC staining to assess the turnover rates of all splenic B lymphocyte populations (Figs. 4B and 4C).

We provided groups of four mice with BrdU via drinking water for up to 28 days. Tissue sampling and analysis were done in weekly intervals over a 4-week time period (Fig. 4C, left panel). At the time of BrdU application, tissues were also isolated from a control group that did not receive BrdU. This analysis confirmed the absence of FITC-staining prior to BrdU-application and was used to establish the levels of FITC signal for negative samples (data not shown).

As expected, the majority of recent bone marrow emigrants (transitional B cells), with a reported lifespan of 3–4 days (25), had incorporated BrdU within 1 week of BrdU application (Figs. 4B and 4C). In contrast, the mature follicular and marginal zone B cells showed the expected low-levels of BrdU uptake. Somewhat surprisingly, the levels of BrdU incorporation among splenic B-1a cells were very high. Roughly 50% of B-1a cells were positive for BrdU within 7 days post BrdU application (Fig. 4C). These levels of BrdU-uptake were closer to those of the transitional B cells than to those of the mature B cell subsets (marginal zone and follicular B). Significant differences were found between all pair-wise comparisons of turn-over rates among the tested B cell population (Fig. 4C, right panel). Strikingly, the turnover rates calculated from the 16 mice analyzed in this timecourse study showed that splenic B-1a cells had turnover rates of 4–5%/day (Fig. 4C, right panel), markedly higher than the approximate rate of 1.4% previously reported for peritoneal cavity B-1 cells (12).


In this study, we evaluated the effects on scatter and cell surface fluorescent parameters of various cell fixation and permeabilization procedures and DNase treatments aimed at facilitating antibody-mediated labeling of intranuclear BrdU. All procedures tested were found to compromise these parameters to varying degrees. Nonetheless, procedures that relied on PFA fixation followed by saponin and DMSO-mediated membrane permeabilization caused the least effects on scatter and fluorescent parameters and were successfully applied for use in 10-color flow cytometry. The usefulness of the approach is highlighted by our data identifying a small splenic B cell population as a population with unexpectedly high turnover rates.

Multicolor flow cytometry requires the use of high-quality individual staining reagents. “Compensation” of data from cells stained with large numbers of fluorochromes must be done to correct the significant spectral overlaps among the emission profiles of various currently used dyes (26). This can greatly reduce the usefulness of a dye or a particular conjugate (2). Fixation and permeabilization procedures are well known to compromise the integrity of cell membranes and their surface antigens (20), potentially degrading staining intensities. Furthermore, not all fluorochromes commonly used in multicolor flow cytometry are stable when cells are fixed, permeabilized, and DNase treated for labeling with anti-BrdU (Fig. 3 and Table 2 and (17)). To identify cell preparation procedures that can be used for the detection of BrdU in conjunction with staining for multiple surface receptors, and to study their effects on some of the more recently developed dyes, we systematically compared four different fixation/permeabilization regimens (Table 1). Some of them are also used for intracytoplasmic cytokine staining (20). As shown in Figures 2A and 3, and summarized in Table 1, PFA/saponin-based techniques best preserved both scatter and fluorescent parameters. This confirms and expands earlier reports that this treatment strategy better retains antigenicity compared with glutaraldehyde-based fixation methods (20, 27–29). Our data (Fig. 2A and table 1) also show that detergents, such as NP40 and organic solvents, such as 95% ethanol, commonly used to permeabilize cell membranes (20) cause cell clumping and loss of APC signals. The loss of the APC signal with ethanol fixation was previously noted (17).

Evaluation of 13 dyes commonly used in flow cytometry showed that PFA/saponin-based fixation and permeabilization procedures affect the quality of separation to various degrees when compared with nontreated samples (Table 2). This loss of separation of stained and unstained events seems to be mainly the result of increased background staining, with lesser effects due to a broadening of the background and reduction of positively stained cells (Table 2). Dyes like APC, QDOT605™, and FITC were least affected by increased background staining. Comparing separation indices, fluorochromes excited by the red laser line were least affected and dyes excited by the violet laser line were affected the most (except QDOT605™), with CasY being lost entirely and PacB showing the most profound decrease of all fluorochromes tested. There is also a reduction of positively stained cells, mostly seen in 488 and 407 nm excited dyes (Table 2).

By providing a protocol that relies entirely on in-house generated reagents (protocol 1A), we also provide a cost-effective means of performing studies on cell turnover rates that often require analysis of large sample sizes. Comparing in-house generated PFA/saponin-based fixation/permeabilization regimens to commercially available reagents, we found that both reagent sets resulted in adequate staining, although the quality of the stains done with commercial fixperm reagents was better than that with in-house generated reagents (Fig. 2A, Table 1). It is likely that further optimization of in-house reagents would result in even better separation of BrdU-labeled from nonlabeled cells.

Using the staining protocol developed here, we determined the cell turnover rate of the known B cell subsets in the spleen. While we confirmed previous reports on rapid BrdU-uptake amongst transitional B cells and slow uptake on the mature long-lived marginal and follicular B cell pools (13, 14), we also found an unexpectedly high uptake among the splenic B-1a cell population (Figs. 4B and 4C). Previous reports on B-1 cell proliferation in the peritoneal cavity showed a rate of ˜1.4%/day (12), significantly slower than the 4–5% measured in this study (Fig. 4). The frequencies of B cell subsets determined in this study, including that of B-1a cells (0.8 ± 0.2)%, were similar to those previously published (30, 31), indicating that our results are not due to inaccurate identification of B-1a cells. Our data open up the possibility that splenic and peritoneal cavity B-1 cells are comprised of distinct subsets of B cells that are maintained by different mechanisms. In fact, differences in the repertoire of splenic and peritoneal B-1 cells have been reported over 15 years ago (32). More recently, others using gene expression profiling and functional analyses concluded that B-1 cells in peritoneal cavity and spleen have different developmental and functional traits (33, 34).

Thus, our data could suggest that B-1a cells in the peritoneal cavity might be derived mainly from fetal precursors maintained by “self-replenishment”, i.e. slow proliferation as proposed by Kantor and Herzenberg (5) and Herzenberg (11). In contrast, at least some of the splenic B-1 cells might originate from precursors in the bone marrow and are continuously replenished through newly emerging cells, as proposed by Wortis, Rajewsky, and others (reviewed in (6)). Our data are more difficult to reconcile with studies in splenectomized mice that showed a dependence of the peritoneal B-1 cell pool on the spleen (35). Those studies have recently been challenged by others that showed the maintenance of peritoneal cavity B-1 cells, despite the absence of splenic B-1 cells in L2 transgenic mice (36). The data from the current study might indicate that a requirement for the spleen in the maintenance of the peritoneal B-1 cell pool could be due to the need for peritoneal B-1a cells to circulate through the spleen. They are not consistent with a model in which splenic B-1 cells replenish the peritoneal B-1 cell pool.

Further work is required to directly compare turnover rates of B-1 cells in different tissues simultaneously and to perform pulse-chase experiments to study the fate of splenic B-1a cells. The cell preparation procedure identified by this study will provide a powerful tool to further address this and other issues.


We are grateful to Paul Milllman (Chroma Technologies, Fort Collins, CO) for providing custom-made filters for the FACS Aria and Adam Treestar (Treestar, Ashland, OR) for Flow Jo software. We also like to thank Marty Bigos (The J. David Gladstone Institute of Virology and Immunology, San Francisco, CA) for discussions on the Separation Index, Randy R. Hardy (Fox Chase Cancer Center, Philadelphia, PA) for information on the choice of optical filters for QDOT605™; Abigail Spinner for help with the FACS Aria; Andy Fell for editing; Dr. Robert Zucker (EPA) for pointing us to the origins of the BrdU-staining procedures; and the individual who generously provided us with the detailed procedures of what is described here as “protocol 3”.