Edited by: Kenneth A Dorshkind, David Geffen School of Medicine at University of California at Los Angeles, USA
Reviewed by: Encarnacion Montecino-Rodriguez, University of California Los Angeles, USA; Gay Crooks, University of California Los Angeles, USA
*Correspondence: Thomas L. Rothstein, The Feinstein Institute for Medical Research, 350 Community Drive, Manhasset, NY 11030 USA. e-mail:
This article was submitted to Frontiers in B Cell Biology, a specialty of Frontiers in Immunology.
This is an open-access article distributed under the terms of the
Controversy over the frequency of human B1 cells in normal individuals has arisen as different labs have begun to employ non-uniform techniques to study this population. The phenotypic profile and relative paucity of circulating human B1 cells place constraints on methodology to identify and isolate this population. Multiple steps must be optimized to insure accurate enumeration and optimal purification. In the course of working with human B1 cells we have developed a successful strategy that provides consistent analysis of B1 cells for frequency determination and efficient isolation of B1 cells for functional studies. Here we discuss issues attendant to identifying human B1 cells and outline a carefully optimized approach that leads to uniform and reproducible data.
B1 cells constitute a distinct B cell lineage with a unique set of characteristics that includes constitutive production of natural antibody that is protective against microbial pathogens and that assists in the disposition of cellular debris. This antibody is broadly reactive, autoreactive, and repertoire skewed (Holodick et al.,
To elucidate markers that would better identify human B1 cells, functional criteria were established based on murine studies and then human B cells that fulfill those criteria were sought. Thus, human B cells that spontaneously secrete immunoglobulin, evidence tonic intracellular signaling, and efficiently stimulate T cells were targeted. Ultimately, a previously unrecognized population of CD20+ B cells within the CD27+ compartment, that is distinguished from memory B cells by expression of CD43, was identified (Griffin et al.,
Adult peripheral blood samples were obtained by venipuncture of adult volunteers after obtaining informed consent in accordance with the Declaration of Helsinki. Additional samples in the form of leukopacks were obtained from the New York Blood Center on the day of donation. This study was approved by, and all samples were obtained in accordance with, the Institutional Review Board of the North Shore-LIJ Health System.
All blood samples were treated in a similar manner and processed promptly upon receipt, except where the experimental design dictated otherwise. Peripheral blood and leukopack samples were initially diluted with RPMI 1640 medium (Cellgro) 1:1. Mononuclear cells were then obtained by density gradient centrifugation using lymphocyte separation medium (Cellgro) at 1,500 ×
For some experiments B cells were enriched by CD19 positive selection using the EasySep Human CD19+ B cell magnetic bead selection kit (StemCell Technologies) according to the manufacturer’s instructions. For some experiments B cells were enriched by CD20 positive selection using the EasySep Human CD20+ B cell magnetic bead selection kit (StemCell Technologies).
Cells were stained using a master mix of antibodies to maintain consistency among the samples. Cells for analysis and sorting were blocked with, and then stained in, dye-free RPMI 1640 supplemented with 10% FCS. For analysis, immunofluorescently stained cells suspended at 1 million/ml were examined at a flow rate of 1,000 cellular events per second or less on a BD Biosciences LSR-II and/or a Beckman-Coulter Gallios cytometer. Cells were vortexed briefly (1–2 s) directly before analysis. Doublet frequency was monitored during analysis in real time and stained cells were vortexed again if doublet frequency exceeded 2% or CD3+CD20+ events in the viable cell gate exceeded 1%. For sort purification, immunofluorescently stained cells (CD19 enriched B cells or PBMC) were suspended at a concentration below 10 million cells/ml in dye-free RPMI 1640/10% FCS supplemented with 2 mM EDTA and separated on an Influx instrument (BD Biosciences) at a flow rate not exceeding 5,000 cellular events per second unless otherwise noted, and a sheath core differential of 2 psi or less. Stained cells were periodically vortexed for 2 s and at times when flow rates dropped by greater than 20% or cells sedimented in sample tubes.
Cells (100 million or less) were pelleted and gently resuspended in 1 ml freezing solution containing 90% FCS and 10% DMSO at room temperature. Cells were quickly transferred to 2 ml cryovials and cooled by placement of these vials in freezing containers (Mr Frosty) in a −80°C freezer for a minimum of 48 h before transfer to liquid nitrogen for prolonged storage at −160°C. Cells were thawed by warming frozen vials in a 37°C water bath until ice was no longer visible, after which 1 ml of pre-warmed RPMI 1640 without supplementation was added to the now thawed cells in the cryovial. Cells were gently mixed to avoid excessive shear forces and transferred drop wise into 9 ml pre-warmed RPMI 1640. Cells were then pelleted and washed once with RPMI 1640 supplemented with 10% FCS. When necessary, cells were fixed by suspending them in 1.6% paraformaldehyde for 15 min at room temperature and were then washed and resuspended in PBS.
Fluorescently labeled mouse antibodies (anti-CD20-V450 cat#642274 clone L27 IgG1κ, anti-CD27-APC cat#337169 clone L128 IgG1κ, anti-CD43-FITC cat#555475 clone 1G10 IgG1κ, and anti-CD3-PE cat#555333 clone UCHT1 IgG1κ, V450-Isotype control cat#560373 clone MOPC-21 IgG1κ, APC-Isotype control cat#555751 clone MOPC-21 IgG1κ FITC-Isotype control cat#555748 clone MOPC-21 IgG1κ, and PE-Isotype control cat#340761 clone X40 isotype IgG1 κ), were obtained from BD Biosciences.
B1 cells represent a small subset residing within a dominant B cell population, as a result of which the ability to distinguish B1 cells from other lymphocytes is dependent on key parameters including: the selection, quality, and efficiency of immunofluorescent staining; the sensitivity and specificity of fluorescence excitation and detection; the specific application and hierarchy of gating strategies; and, the competency of computerized algorithms for cytometric data interpretation. Moreover, the phenotypic nature of human B1 cells, encompassing expression of CD20, CD27, and CD43, imposes certain limitations on methodology to identify and isolate this population; in particular, CD27 and CD43 expression is shared by most peripheral blood T cells, whose numbers far outweigh the number of B cells, so that B (CD20+)-T (CD27+CD43+) doublets could be mistaken for B1 cells. These and other issues discussed below dictate the need for careful review of technical considerations to optimize identification and recovery. Recommended procedures are discussed below in more or less chronological order leading from initial blood sample to analyzed/sorted B1 cell preparation.
The most accessible B1 cell-containing human tissue is blood, and typical samples arrive either as newly drawn adult peripheral blood, as recently processed adult blood leukopacks, or as freshly obtained umbilical cord blood. Access to the first is measured in minutes, whereas access to the latter two samples is typically measured in hours. In general, the more immediate the sample, the better the B1 cell recovery. From each of the sources noted above peripheral blood mononuclear cells (PBMC) are obtained by density gradient centrifugation after dilution as described in “Materials and Methods.” Although heparin is typically used for anticoagulation of newly drawn peripheral blood, some samples, such as leukopacks and umbilical cord blood samples, may rely on citrate anticoagulation which can be reversed with calcium-containing buffers, such as Hank’s balanced salt solution. This may lead to clot formation interfering with subsequent steps. Thus, dilution of citrate-containing samples with medium or buffer containing no calcium (e.g., PBS) is preferred.
Tissue samples such as spleen, tonsil, or lymph node are likely to become available with typically longer delays than noted above, such that changes occurring due to loss of viability prior to acquisition may be an issue. Analysis of such specimens is likely to benefit from viability staining to eliminate dead and dying cells. Viability staining may also be appropriate with blood samples if/when deterioration is suspected; however, it should be noted that annexin V staining is not suitable for recognizing apoptotic B1 cells (Dillon et al.,
Optimization of staining parameters facilitates recognition of human B1 cells. Mature B cells typically express both CD19 and CD20. Of note, however, as B cells differentiate to become plasmablasts and plasma cells, CD20 is lost but CD19 remains, as CD38/CD138 is acquired (Jego,
Cell subsets | Surface phenotype |
---|---|
Naïve B cells | CD3−CD19+CD20+CD27−CD43−CD69−CD70− |
Memory B cells | CD3−CD19+CD20+CD27+CD43−CD69−CD70− |
B1 cells | CD3−CD19+CD20+CD27+CD43+CD69−CD70− |
Plasmablasts/plasma cells | CD3−CD19lowCD20low/−CD27++CD43++CD138± |
CD43+ activated B cells | CD3−CD19+CD20+CD27±CD43+CD69++CD70++ |
T cells | CD3+CD19−CD20−CD27±CD43± |
Because expression by T cells of both CD27 and CD43 could complicate analysis of B1 cells as a result of B:T doublet formation (see below), anti-CD3 is added to exclude T cells from consideration, as displayed in the typical gating strategy for B1 cell identification shown in Figure
Among CD20+ B cells, B1 cells are recognized by expression of CD27 and CD43. These markers coincide in umbilical cord blood samples, as no memory B cells are present at birth, whereas in adult peripheral blood, CD43 separates B1 cells from the CD27+ B cell population which contains both B1 cells and true memory B cells (Griffin et al.,
The signal-to-noise ratio of antibody-fluorophore conjugates is only apparent following excitation and detection and thus instrument characteristics play a key role in determining the efficiency with which B1 cells are recognized. These features include laser intensity and stability; fiberoptic path length; filter number, bandwidth and efficiency; fluorescence detector sensitivity; and, coincident rejection rate; among others, all of which are beyond the scope of the present discussion. In aggregate, instrument variables alone can greatly affect recognition and enumeration of B1 cells. When the exact same immunofluorescently stained and fixed PBMC were evaluated on two different flow cytometers, CD43 expression by CD20+CD27+ B cells was more cleanly distinguished on one instrument (Gallios) than the other (LSR-II), as illustrated by contour display of the gated CD20+CD27+CD43+ B1 cells on the CD27 by CD43 plot (Figure
Existing machines can be optimized for visualization of antigens expressed at low levels. Replacement of the standard bandpass filters by filters with broader ranges will result in collection of more emitted light. This will result in improved frequency determinations for human B1 cells on the same instrument, dependent on the degree of optimization for detection of poorly expressed markers such as CD43. Adjusting the photomultiplier (pmt) voltage to achieve the greatest separation between positive and negative populations rather than arbitrary placement of the negative population in the first decade should also be considered.
The ultimate arbiter of frequency for, and/or isolation of, any phenotypically identified population lies in the way in which gating structures and boundaries are applied to fluorescence images thereby defining the cells of interest. Where populations overlap, slight changes in the location of the cutoff between positive and negative cells will greatly influence measured frequencies. This is exemplified by CD43 staining, discussed above, and is illustrated conceptually in Figure
A further illustration of CD43 discrimination at a single CD27 intensity is displayed in histogram plots of relative cell number vs CD43 expression for duplicate samples analyzed on the Gallios and LSR-II cytometers (Figure
Several methods have been advocated for delineating gate-setting that separates positive and negative staining values when clear separation is not achievable, with most focusing on generic approaches consisting of either no-staining controls, irrelevant-antibody-isotype-matched controls, and fluorescence-minus-one (FMO) controls (Perfetto et al.,
Because T cells are much more abundant than B cells in the peripheral circulation, any stochastic tendency of B cells to adhere to another lymphocyte will likely involve a B:T combination. It is not clear however that all doublets result from purely stochastic processes so the actual rather than predicted types of doublets involving B cells may vary for different B cell subpopulations. Because most T cells express CD27 and CD43, any B:T doublet is likely to phenotype as CD20+ (contributed by the B cell constituent), and CD27+CD43+ (contributed by the T cell constituent; Figure
Modern flow cytometer software incorporates algorithms that make it possible to detect and reject doublets and coincident events, based on area vs height or pulse-width considerations, thereby eliminating events that fall outside typical parameters for the size and roundness of single cells. There are three reasons why this approach can be problematical. (1) Two cells may line up one behind the other with respect to the light path so that a doublet masquerades as having a normal area/height configuration and fails to be rejected, leading to inadvertent overcounting; (2) Rejection of large numbers of cells presented for analysis may entail omission of many cells of interest that are tied up in doublets, leading to inadvertent undercounting; and, (3) Doublet discrimination parameters used to separate normal spherical cell shapes and sizes from outliers are essentially arbitrary and cannot incorporate perfect sensitivity and specificity and thus may omit unusually large or irregularly shaped cells (Bauer,
Following CD3-negative gating all the CD20+CD27+CD43+ events should correspond to B1 cells as naïve and memory B cells are negative for CD43 expression and consequently would not fall into this gate. However, it is conceivable that memory B cells might form doublets with activated B cells or with myeloid/monocyte cells, in each case producing CD20+CD27+CD43+ events. Doublet discrimination can then be used as a supplementary means of excluding doublets that may have persisted through the preparatory steps discussed above. When employing pulse-gating methods for doublet or aggregate discrimination, gating should be based on clearly defined singlet vs doublet populations to avoid arbitrary exclusion of singlet cells with larger size and/or lower sphericality. Since most doublet discrimination relies on the relationship between size and either area or width a perfectly round cell will fall in the center of a histogram or contour visualization of singlet cells. There will be a certain width to the histogram or contour visualization of single cells based on the mathematical principle of probabilistic aggregation formation, whereby events of a given population cluster about a mean value with a slight degree of random variation that results in a bell shaped curve (Sharpless and Melamed,
The locations of singlet events and doublet events will be clearly indicated even without gating on CD3+CD20+ events if the interrogated sample contains a significant number of doublets. However, if samples are vortexed, and if CD3 exclusion is employed, few, if any, doublets should be detected, as a result of which B1 cell frequencies determined with and without doublet discrimination should vary little (Figure
Unfortunately there may be times when the situation is less than optimal and cell preparations are not presented for analysis as true single cell suspensions. In such circumstances doublet discrimination can be used to computationally eliminate doublets that will both improve the reliability of data analysis and provide important information regarding the quality of the sample being analyzed. Thus, any significant difference in B1 cell frequency noted with/without doublet discrimination, or the presence of a large number of detected CD3+CD20+ events, would suggest improvements are needed in initial sample preparation to counteract doublet formation and/or persistence.
The effectiveness of doublet avoidance can be measured in several different ways. Monitoring DNA content after staining with Hoechst 33342, enumeration of CD20+CD3+ events after anti-CD3 staining, and determination of cells falling outside doublet discrimination settings, can each provide information on the relative abundance of singlet events that were evaluated and/or sorted.
The goal of interrogating dilute cell suspensions at modest events per second, to avoid doublets, is not always possible during sort purification of infrequent cell populations. In this situation enrichment of B cells with anti-CD19-magnetic bead conjugates substantially diminishes apparent B:T doublet formation (and in this respect Stem Cell Technology conjugates appear to yield higher post-enrichment B1 cell frequencies than Miltenyi products). In addition, suspension of cells in 2 mM EDTA counteracts doublet formation.
The characteristics and behaviors of human B1 cells suggest additional means to enhance identification and recovery. Although most B1 cells fall within typical lymphocyte gating for size (FSC) and granularity (SSC), some do not. For this reason a larger than usual cell gate will include more B1 cells (Figure
As noted above, the rapidity with which blood samples are processed following acquisition, the better the identification and isolation of B1 cells. The detection and/or recovery of B1 cells is diminished by any postponement of sample processing, regardless of condition. Samples retained overnight as heparinized whole blood or as separated PMBC, at room temperature or at 4°C, lose B1 cells (Figure
The peculiarities of human B1 cells – representing a small B cell population that shares surface antigens with memory B cells, T cells, and monocytes/macrophages – amplify any shortcomings in standard approaches to immunofluorescent staining and analysis. Failure to optimize protocols can easily lead to erroneous assignment and either undercounting or overcounting of B1 cell numbers, with similarly detrimental implications for functional studies of sort-purified B1 cell characteristics.
Because CD43 “carves out” B1 cells from among CD27+ B cells that have been characterized as memory B cells, it is now clear that much memory B cell work to date has been conducted on heterogeneous populations that included “true” memory B cells (CD27+CD43−) and B1 cells (CD27+CD43+). Some characteristics of memory B cells, such as efficient allogeneic T cell stimulation, appear to belong specifically to B1 cells (Griffin and Rothstein,
Human B1 cells do not express CD3, for which reason CD3 expression can be used to monitor the frequency with which B:T doublets have persisted and have become subject to analysis. Along the same lines, CD3 expression can be used as an exclusionary criterion to eliminate B:T doublets from consideration. However, it is important to emphasize our general bias against exclusionary criteria during identification of new lymphocyte populations lest B cells displaying unexpected surface antigens remain unrecognized. This principle is well illustrated by the phenotype of human B1 cells that includes determinants (e.g., CD27, CD43) better known as characteristics of other immune cell populations (Zhong et al.,
The relative loss of CD27 staining when PBMC processing is delayed leads to a conundrum regarding gating strategy because the topographical representation of CD27 then bridges the line separating CD27+ and CD27− B cells. This forces a choice between including the entire CD43+ contour, encompassing portions that are formally (by FMO) CD27−, or including just that portion of the contour that is clearly CD27+. This, though, may be a choice that need not be made. At the present time it appears that no other B cells beyond B1 cells constitutively express CD43, although naïve and memory B cells do so after activation. However, CD43-expressing activated naïve and memory B cells express CD69 and CD70, whereas CD43− expressing B1 cells do not, as noted elsewhere (Griffin et al.,
In our work to date we have documented extremely wide variation in the frequencies of B1 cells in the peripheral circulation of normal volunteers. Whereas there is a general trend toward fewer B1 cells in older individuals, at every age the range of “normal” B1 cell frequencies may span less than 1% to greater than 9% of circulating B cells. Some of these values may represent biological or statistical outliers, perhaps seemingly “normal” individuals who harbor an illness that is not yet apparent. Alternatively, a large reservoir of B1 cells may exist somewhere in the body (e.g., the peritoneal cavity, as in mice) such that despite variations in circulating B1 cells, the total body number of B1 cells is similar from individual to individual and for a given individual from time to time. Beyond these considerations, it is always necessary to apply appropriate statistical analysis before drawing conclusions that associate high or low frequencies of B1 cells with any particular disease state or normal parameter to be sure that adequate numbers of samples have been examined.
Summary of key approaches and recommendations:
Process biological samples urgently upon removal from the individual; immediately freeze, or stain and fix cells, if processing delays are unavoidable.
Utilize high affinity, high specificity antibodies for CD43 coupled to fluorophores with high signal-to-noise ratios (e.g., FITC, PE, APC) and analyze on a flow cytometer with optimal filters and high sensitivity.
Apply FMO controls in multiparametric analysis to set population-defining gates in the absence of clear separation between populations. It is possible to use isotype controls in setting gates, if four or fewer fluorophores are employed (consistently apply the chosen approach across all samples in a given experiment).
Prepare samples at appropriate concentrations for analysis (1 million cells/ml or less) and agitate to prevent/reverse doublet formation. Use EDTA (2 mM) during sort purification.
Acquire samples at rates that do not increase coincident events, such as 1,000 cellular events/s. Use inclusive cell gating whereby all cells are included and do not focus on a preconceived gate for lymphocytes.
Employ CD3 exclusion in the gating hierarchy to avoid B:T cell doublets, but do not use additional negative gating until it is proven appropriate to do so with the same degree of rigor used to establish CD3 as a negative gating marker.
Rely on doublet discrimination sparingly, keeping in mind that although it can counteract poor sample preparation and inherent cell–cell attraction, its use is accompanied by substantial collateral effects.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The authors thank Dr. Nichol E. Holodick for critical review of the manuscript. This work was supported by grants from the National Institutes of Health, the Lupus Research Institute, and the Zucker Family Foundation.