ABCB1 transporter discriminates human resting naive B cells from cycling transitional and memory B cells



The exact identification of B cell subsets is instrumental to understand their dynamics under physiological and pathological conditions. Human memory B cells are currently identified according to the expression of CD27, which is absent on naive B cells. We report here that the ATP-binding cassette (ABC)B1 transporter is exclusively present on mature CD27 naive B cells, while it is absent in CD27+ memory B cells and in a heterogeneous subset of CD27 cells that comprise both switch memory and transitional B cells. Thus, ABCB1 activity precisely discriminates naive from transitional and all memory B cells. Using this improved method to discriminate human B cell subsets, and Ki67 staining to identify recently divided cells, we show that in both cord blood and adult peripheral blood, mature naive B cells are quiescent while transitional B cells and memory B cells have a high in vivo turnover.


ATP-binding cassette


CpG ODN 2006


cyclosporine A


MitoTracker Green FM


MitoTracker Orange CMTRos


propidium iodide


Rhodamine 123


The identification of B cells at various stages of differentiation is relevant for understanding their physiology and immunopathology 1, 2. Two major B cell subsets can be identified in adult peripheral blood according to the expression of CD27 3, 4. CD27+ B cells comprise both IgM memory B cells 57 (IgMhi/IgDlo) and switch memory B cells carrying surface IgG or IgA. In addition, variable numbers of Ig-secreting plasma blasts expressing high levels of CD27 and CD38 may be present in normal peripheral blood under steady state conditions, and their number increases after immunization 8, 9. In contrast, CD27 B cells comprise naive B cells (which are IgMlo/IgDhi) and transitional B cells, which represent recently formed bone marrow emigrants and can be discriminated from naive B cells only by quantitative expression levels of several markers such as CD24, CD38, CD5, CD10, CD21 and CD23 10, 11. Thus, the current definition of human B cell subsets rests mainly on the relative expression of multiple markers, with the possible exception of CD27 that appears to reliably discriminate between naive and memory B cells.

A precise and feasible identification of B cell subsets would be instrumental to define their repertoire and understand their homeostasis 12. While the pool of naive B cells is replenished from transitional cells and is present in a quiescent state 1316, it has been proposed that memory B cells undergo spontaneous proliferation and differentiation, resulting in a sustained production of plasma cells that maintains serum antibody levels 17. Indeed, a recent study demonstrated that human CD27+ memory B cells had a fivefold higher incorporation of heavy glucose as compared to CD27 putatively naive B cells 18.

ATP-binding cassette (ABC) transporters are a large family of transmembrane proteins driving the transport of various molecules across cell membranes, whose prototypic member ABCB1 (P-glycoprotein, MDR-1) mediates multidrug resistance in cancer 19. A high activity of ABCB1 and ABCG2 is also a distinctive characteristic of stem cells 20, 21. ABC transporters are expressed in human lymphocytes and dendritic cells 22, 23 and influence their response to chemoattractants 24. The presence of functional ABC transporters can be detected by measuring the efflux of dyes such as Rhodamine 123 (R123), and the transporter involved can be identified using specific inhibitors.

In this study, we investigated the activity of ABC transporters in human B cells and report that ABCB1 expression is a hallmark of mature naive B cells, since it is up-regulated during the final differentiation from transitional B cells and is irreversibly lost on all memory B cells. Using ABCB1 to improve the discrimination of B cell subsets, we show that naive B cells are resting, while both memory and transitional B cells have a high spontaneous in vivo turnover.


A subset of peripheral blood B cells has functional ABCB1 transporters

The vital dye R123 freely permeates all cells and is extruded by ABC transporters. To assess the presence of ABC transporters on human B cells, we labeled PBMC with R123 and chased the cells at 37°C for 3 h (Fig. 1A). B lymphocytes, identified by CD19 expression, showed a heterogeneous staining pattern since only some cells extruded R123 while others retained the dye. As R123 is a substrate of several ABC transporters, we tested the effect of inhibitors specific for ABCC1 (MK571) and ABCB1 (PGP4008). Extrusion of R123 was inhibited only by the latter compound, implying that the transporter that mediates R123 extrusion in B cells is ABCB1 (Fig. 1B). Two additional dyes, MitoTracker Green FM (MTG) and MitoTracker Orange CMTRos (MTO), showed comparable patterns of extrusion and inhibition (Fig. 1B, C). These results indicate that a functional ABCB1 transporter is present in a subset of human peripheral blood B cells.

Figure 1.

Functional ABCB1 transporter is expressed in a subset of human peripheral blood B cells. (A) Total PBMC were pulsed with R123, chased for 3 h at 37°C and stained for CD19 expression. (B) CD19+ cells were labeled with R123, MTG or MTO as above in the absence (black line) or in the presence of ABCB1 inhibitor PGP4008 (filled grey) or ABCC1 inhibitor MK571 (dark gray line). (C) Percentage of inhibition of dye extrusion by specific and nonspecific inhibitors of ABC transporters.

ABCB1 activity is restricted to CD27 B cells and reveals heterogeneity within this population

To identify the B cell population expressing functional ABCB1, we stained peripheral blood B cells with R123 and an antibody to CD27 to mark memory B cells (Fig. 2A). CD27+ cells failed to extrude R123, suggesting that memory B cells lack ABCB1 activity. In contrast, the putatively naive CD27 B cells were heterogeneous, with a major population extruding R123 and a minor one retaining the dye. The three populations identified according to CD27 expression and ABCB1 activity were analyzed for additional surface markers (Fig. 2B). The CD27 ABCB1+ (R123lo) subset (I) showed the characteristic phenotype of mature naive B cells as far as cell size, surface Ig expression (IgMlo/IgDhi) and CD24/CD38 profile 10 were concerned. The CD27+ ABCB1 (R123hi) subset (II) showed the characteristic phenotype of memory B cells as far as cell size (larger than naive), surface isotypes (IgMhi/IgDhi/neg, IgG or IgA) and CD24/CD38 profile were concerned. Remarkably, the newly identified CD27 ABCB1 (R123hi) subset (III) was heterogeneous. It contained a minor population of switch memory B cells (mostly IgG+) with the characteristic memory CD24/CD38 profile. In contrast, the major population was IgDhi/IgMhi/+ and showed a distinct CD24hi/CD38hi profile that has been described to be characteristic of human transitional B cells, although a major fraction is overlapping with the profile of naive B cells.

Figure 2.

ABCB1 activity is restricted to CD27 B cells and reveals heterogeneity within this population. (A, B) PBMC were labeled with R123 and stained with antibodies to CD19, CD27, IgM, IgD, IgG, IgA, CD24 and CD38 in various combinations. (A) Three subsets were gated according to the expression of ABCB1 and CD27. (B) Size, isotype and CD24/38 expression profiles of the three gated subsets. (C) Surface expression of activated ABCB1 detected by specific mAb. PBMC were stained at 37°C in the presence of CsA with mAb UIC2 which recognizes the activated conformation of ABCB1. The histograms are gated on CD19+ cells and, as indicated, on CD27+ and CD27 cells.

To further confirm differential expression and activity of ABCB1 within CD27+ memory and CD27 B cells, we stained CD19+ cells with an antibody specific for the active conformation of ABCB1 25 (Fig. 2C). In line with the functional results, surface expression of active ABCB1 was restricted to CD27 B cells, while all CD27+ B cells were negative and a fraction of CD27 B cells stained low to negative compared to the isotype control.

We conclude that CD27 B cells are a heterogeneous population and that ABCB1 expression can be used to discriminate, within this population, naive B cells from putative switch memory B cells lacking CD27 expression and from a larger population comprising transitional B cells.

Characterization of CD27 memory B cells

To further characterize the putative CD27 memory B cell subset, we analyzed the distribution of switched isotypes within CD27+ and CD27 B cells (Fig. 3A, B). Approximately one third of IgGmath image cells and half of IgGmath image cells were CD27, while virtually all IgGmath image and IgA+ cells were CD27+.

Figure 3.

Characterization of CD27 switch memory B cells. (A) Expression of IgG subclasses and CD27 on IgG+ B cells in a representative donor. (B) Mean percentages and SD of switched isotypes in CD27+ and CD27 cells from six donors. (C) Concentration of IgG subclasses in culture supernatant of EBV-transformed CD27+ (empty histograms) and CD27 (filled histograms) IgG+ B cells. (D) Specific antibodies in culture supernatant measured by ELISA: bovine serum albumin (BSA; control); tetanus toxoid (TT); influenza virus (Flu); pneumococcal polysaccharide (pneumovax, PV).

CD27+ and CD27 IgG+ B cells were isolated by cell sorting and immortalized with EBV, as described 26. Total IgG, IgG subclasses and specific antibodies were measured in the 12-day culture supernatant by ELISA (Fig. 3C, D). The concentration of IgG1 and IgG3 was comparable in cultures generated from CD27+ and CD27 cells. In contrast, and consistent with the surface isotype expression, IgG2 was detected predominantly in CD27+ B cell cultures (Fig. 3C).

Furthermore, IgG antibodies to tetanus toxoid and influenza virus (which are mainly IgG1) were produced at comparable levels by CD27+ and CD27 B cells, while antibodies to pneumococcal polysaccharide (which are IgG2) were produced only by CD27+ cells.

As for IgA+ and IgGmath image B cells, comparison of sorted CD27+ IgMhi/IgDlo memory B cells with CD27 IgMhi/IgDlo B cells did not supply evidence for CD27 IgM+ memory B cells (data not shown). We conclude that a sizeable fraction of IgGmath image and IgGmath imagebona fide memory B cells is CD27 and can be discriminated from naive B cells according to ABCB1 expression.

Identification of human transitional B cells by the absence of ABCB1 activity

The IgD/IgM and CD24/CD38 phenotype of mature naive and putative transitional B cells discriminated by ABCB1 activity is in part overlapping, as shown in Fig. 2B. In the further analysis we included markers that had been proposed recently for the identification of human transitional B cells 11. Staining of CD19+ B cells showed that CD5 expression is limited to IgM+/++ B cells, while switch memory B cells (IgM) are entirely negative (Fig. 4A, left panel). Excluding all memory B cells (gating on CD27, CD38+/++, IgG/IgA cells) from further analysis (Fig. 4A, middle and right panel), we show for both CD5 and CD10 expression that the identified population of transitional B cells was mainly positive for these markers. However, while ABCB1 activity sharply discriminated both subsets, CD5 and CD10 were present at lower expression levels in naive B cells as well.

Figure 4.

Identification of human transitional B cells by absence of ABCB1 activity. (A) Total CD19+ B cells were stained in combinations for IgM, CD5, CD10 and ABCB1 activity (MTG), and all memory B cells were excluded gating on CD27 CD38+ IgG/IgA B cells (middle and right panel). (B) Differential requirements for survival of mature naive (ABCB1+) and transitional B cells (ABCB1 CD38+ CD27 IgG/IgA). Shown are the percentages of PI annexin V cells cultured in complete medium (left panel) after cell sorting, and on day 4 of cultures supplemented with F(ab’)2 fragments of anti-Ig antibodies or IL-4 (middle and right panel, respectively). One representative experiment out of three. (C) CD19+ B cells were stained for ABCB1 activity, CD27 and CD124 (IL-4Rα) expression, and gated on all B cells (left panel), mature naive B cells (ABCB1+, middle) and ABCB1 B cells, comprising transitional and all memory B cells. One representative experiment out of five.

Functional analysis of mature naive and transitional B cells demonstrated comparable rates of spontaneous apoptosis when cultured in completed medium after cell sorting (Fig. 4B, left panel). However, ABCB1+, mature naive B cells where in part rescued from apoptosis by BCR stimulation, while no effects where seen on the ABCB1 transitional B cells (Fig. 4B, middle panel). In contrast, both subsets, mature more than transitional, where rescued from apoptosis by low doses of IL-4 (Fig. 4B, right panel). Further analysis of IL-4 receptor alpha (IL-4Rα, CD124) expression on all B cells (Fig. 4C) revealed homogenous high expression on ABCB1+ mature naive B cells, lack of IL-4Rα on CD27+ (ABCB1) memory B cells and low expression on ABCB1 CD27 B cells, comprising transitional B cells and the above described subset of CD27 switch memory B cells.

We conclude that ABCB1 activity sharply discriminates mature naive from transitional B cells, in contrast to all other useful markers, which were largely overlapping with the phenotype of naive B cells. Functional data confirmed the proposed assignment, with IL-4 as a survival factor more for naive than for transitional B cells, according to the differential expression of IL-4Rα (CD124).

Loss of ABCB1 activity as a function of cell division and differentiation

The above results suggest that ABCB1 function is a hallmark of mature naive B cells and is lost upon differentiation to memory B cells. We sorted mature naive B cells (ABCB1+), labeled with CFSE and stimulated polyclonally to follow division-dependent differentiation (Fig. 5). On day 5 of culture, non-divided cells maintained their ABCB1+ phenotype, while cells undergoing division gradually lost ABCB1 activity, which was virtually absent beyond the fourth division (Fig. 5A). Accordingly, ABCB1 was absent in all B cells gaining CD27 expression in culture or expressing the switch isotypes IgG/IgA (Fig. 5B). Loss of ABCB1 from mature naive B cells was thus a cell division- and differentiation-dependent process.

Figure 5.

Loss of ABCB1 activity as a function of cell division and differentiation. (A) Purified mature naive B cells (ABCB1+) were labeled with CFSE and stimulated polyclonally (anti-Ig F(ab)2 fragments, CpG-B, IL-2, IL-10). ABCB1 activity was determined on day 5 as a function of cell division. (B) Division-dependent expression of differentiation markers CD27 and switch isotypes, as in (A), and ABCB1 activity determined by MTO staining.

Ki67 expression is restricted to memory and transitional B cells identified by ABCB1

To estimate the in vivo turnover of human B cell subsets, we analyzed the expression of Ki67, a nuclear antigen that identifies recently divided cells 27, 28. In a first series of experiments, we stained B cells for CD27 and CD38 or intracellular Ig. The plots in Fig. 6A display all B cells (left) and Ki67+ gated B cells (right) in a representative donor. Of Ki67+ B cells, 17% were plasma blasts (characterized by high levels of CD27, CD38 and intracellular Ig (icIg), while ∼60% were present within CD27+ memory cells. However ∼15% of Ki67+ cells were CD27, although they expressed somewhat higher CD38 levels. This finding raised the question of whether the CD27 Ki67+ cells were indeed mature naive B cells or belonged to the transitional or memory subsets. As the turnover of all memory B cells was restricted to CD27+ cells (data not shown) and as virtually all Ki67+ cells were ABCB1 (Fig. 6B), the remaining turnover was attributed to transitional B cells.

Figure 6.

Turnover within transitional B cells, memory B cells and plasma blasts, and quiescence of naive B cells. (A) CD19+ B cells were stained with antibodies to CD27 and CD38, permeabilized and stained with anti-Ki67 antibody and for intracellular Ig. The left panels display all B cells, the right panels Ki67+ B cells only. (B) CD19+ B cells were labeled with the fixable ABCB1 substrate MTO or with anti-CD27 antibody, fixed, permeabilized and stained with anti-Ki67 antibody. (C) CD19+ B cells were surface stained for Ig isotypes followed by Ki67. (D) Percentages of Ki67+ cells within each individual subset gated for transitional (ABCB1, CD27, CD38+, IgG/IgA), mature naive (ABCB1+), total memory (ABCB1, CD27+ and CD27/CD38) B cells, and their respective isotype controls (IgM/IgG/IgA), as well as plasma blasts (CD27++/icIg++) (+ SD, n = 8–20). (E) Correlation between the percentage of Ki67+ plasma blasts relative to total B cells and the percentage of Ki67+ memory B cells relative to total B cells.

The above results indicate that spontaneous turnover is limited to plasma blasts, transitional and memory B cells, while mature naive B cells are entirely quiescent. The percentages of Ki67+ cells in various subsets, including results from cell-sorted populations followed by Ki67 staining, are summarized in Fig. 6D. Ki67+ cells were approximately 2% in transitional B cells, virtually absent in mature naive B cells and were present in variable proportions in memory B cell subsets (ranging from 1.5% in IgM memory to 5% in IgA memory cells; Fig. 6C, D). The highest percentage of Ki67+ cells (∼40%) was found within the small population of circulating plasma blasts. Remarkably, in 13 individuals analyzed there was a strong correlation between the absolute numbers of Ki67+ memory B cells and Ki67+ plasma blasts (Fig. 6E). These results indicate that memory B cells spontaneously divide at a high rate and that this is proportional to the rate of generation of new plasma blasts.

High frequency and turnover of transitional B cells in cord blood

To validate ABCB1 activity as a way to discriminate transitional from mature naive B cells, we analyzed the distribution and turnover of B cell subsets in cord blood (Fig. 7A, B). Most cord blood B cells resembled the transitional B cells of the adults being ABCB1 IgM/IgDhi and CD24+/hi CD38hi. The few ABCB1+ B cells had a mature phenotype being IgM+/lo IgD+, CD24lo and CD38hi. As expected, CD27+ and switch memory B cells were not detectable (data not shown). Remarkably, Ki67 expression was relatively high (5%) and was limited to ABCB1 cells (Fig. 7C, D). These results indicate that a high proportion of cord blood B cells have the characteristics of transitional B cells with high spontaneous turnover.

Figure 7.

High levels and turnover of transitional B cells in cord blood. (A) Cord blood CD19+ cells were labeled with MTG, chased in the absence (dark grey line) or in the presence (filled grey) of specific ABCB1 inhibitor PGP-4008. (B) Expression of CD24/CD38 and IgD/IgM on cord blood CD19+ cells, gated for ABCB1 activity as above (gates a and b). (C) Total cord blood B cells were labeled with MTO, fixed, permeabilized and stained for Ki67. (D) Mean percentages of mature naive B cells (empty bars) and transitional B cells (grey bars) in cord and adult blood (+ SD, n = 6).


We have shown that within the human B cell lineage, ABCB1 activity is restricted to naive B cells and can therefore be used to precisely distinguish mature naive from transitional and all memory cells. By combining this new marker with CD27 and Ig isotypes, it is possible to obtain a better definition of human B cell subsets and to estimate their spontaneous turnover.

CD27 has been a very useful marker to positively identify memory B cells including IgM memory cells 57. However, we have shown here that some bona fide memory cells expressing selected IgG subclasses (IgG1 and IgG3) are CD27 and can be easily discriminated from naive B cells using ABCB1. Nature and functional differences of these CD27 B cells are not yet evident, although their emergence has possibly been observed during rotavirus infection 29. B cell differentiation has been demonstrated to be a division-dependent process 30, and accordingly, we observed loss of ABCB1 activity as a function of division and differentiation.

Discrimination of human transitional B cells from mature naive B cells relied on gradual differences in expression of multiple surface markers in analogy to their murine counterparts 10. However, a large proportion of transitional B cells has an overlapping phenotype with mature B cells, which was thus termed “intermediate” 11. This phenotypic overlap could be resolved using ABCB1 activity. The so identified ABCB1 population had distinct survival requirements and, importantly, recent turnover in vivo was restricted to these cells, while ABCB1+ mature naive B cells were highly quiescent.

Recent studies using in vivo labeling with deuterated glucose indicate that human CD27+ B cells have an in vivo turnover that is five times higher than that of CD27 putatively naive B cells 18. Our estimate of turnover for memory cells, using Ki67, is similar to that reported in the above study, considering the rate of Ki67 decay reported previously 27. In addition, we were able to show that naive B cells are completely quiescent, since the Ki67+ cells found within the CD27 fraction represented newly formed transitional B cells.

The high turnover of memory B cells observed in this and in a previous study 18 is consistent with a model where homeostatic proliferation maintains a continuous production of plasma blasts that sustains serum antibody levels 17. In this study, we provide further supporting evidence by showing that, under steady state conditions, there is a strong correlation between proliferating memory cells and plasma blasts. A similar population of pre-plasma cells that divides spontaneously in the absence of antigen and spills out mature plasma cells has been described in the mouse 31.

Finally, a relevant question is whether ABCB1 plays a specific function in naive B cells. The role of ABC transporters in dendritic cell chemotaxis 24 suggests that ABCB1 expression may influence the migratory capacity and positioning of naive B cells relative to transitional and memory B cells. Furthermore, the function of ABCB1 in cholesterol transport 32 and its association with rafts 25 point to a possible role in regulating the abundance and composition of rafts that are critical for B cell signaling 33.

Materials and methods

Cell culture and phenotyping

PBMC were isolated from healthy donors, cord blood mononuclear cells from term newborn. Four- and five-color staining for immune phenotyping was done with the following antibodies in appropriate combinations and controls: anti-CD19 [HB19 (allophycocyanin (APC)); SJ25C1 (PE-Cy7); BD Biosciences]; anti-CD27 [O323 (biotinylated), e-bioscience; M-T271 (PE), BD Biosciences]; goat anti-human IgG, IgA and IgM (Cy5), IgD (PE) (Jackson Immunoresearch); anti-CD38 [LS198-4-3 (PC5)], anti-CD24 [ALB9 (PE)] (both Coulter Immunotech); anti-CD5 [BL1a (PE), Coulter; UCHT2 (APC), Beckton Dickinson]; anti-CD124 [S456C9 (PE), Coulter]. IgG subclasses were analyzed with mAb against IgG1 (MH161-1), IgG2 (MH162-1), IgG3 (MH163-1) (all from Sanquin, The Netherlands) with appropriate secondary antibodies. Intracellular Ig staining was performed ex vivo after surface staining and fixation, followed by permeabilization. Samples were acquired with a FACS Canto (BD Biosciences) and data analyzed with FlowJo software (Tree Star).

For in vitro assays, B cells were enriched using CD19 magnetic beads (Miltenyi Biotec) and B cell subsets were sorted using a FACS Aria (BD Biosciences). Cells were cultured in complete medium 17 containing 10% FCS (Hyclone). Survival of sorted B cell subsets was determined by propidium iodide (PI) and annexin V staining and quantified with counting beads. For proliferation and differentiation assays, sorted B cells were labeled with carboxyfluorescein succimidyl ester (CFSE; Molecular Probes) and were stimulated with 2 µg/mL F(ab’)2 (Jackson Immunoresearch), 2.5 µg/mL CpG ODN 2006 (CpG-B) (Microsynth), IL-2 and IL-10 (both 10 ng/mL; R&D Systems). EBV transformation of sorted subsets and quantification of total IgG, IgG subclasses and antigen-specific antibodies was done as described 17, 26.

ABC transporter expression and activity

PBMC were stained in culture medium at 37°C with R123 at 12.5 µM for 10 min, with MTG at 100 nM for 30 min, and chased for 3 h and 20 min, respectively. Staining with the fixable dye MTO was at 50 nM for 20 min without chase (all Molecular Probes). As unspecific ABC transporter inhibitors, cyclosporine A (CsA; 25 µM) and verapamil (50 µM) were added during staining and chase; for specific inhibition, MK571 (25 nM, for ABCC1 inhibition) and PGP-4008 (10 nM, for ABCB1 inhibition) (all from Alexis) were added. For surface staining of ABCB1, PBMC were incubated for 20 min at 37°C with 25 µM CsA (exposes epitope of the activated conformation) together with an anti-ABCB1 antibody (PE-labeled UIC2; Coulter-Immunotech), followed by regular surface staining at 4°C.

Ex vivo Ki67 staining

CD19-enriched cells, stained for ABCB1 activity with MTO and surface labeled, were fixed with 1% paraformaldehyde, permeabilized with 0.5% saponin and stained for the cell cycle-associated antigen Ki67 (FITC-labeled B56; BD Biosciences).


This work was supported in part by the NIH (grant no. U19AI057266/01) and the Swiss National Science Foundation (grant no. 31-63885) and by the European Community (MUVAPRED no. LSHP-CT-2003–503240). S.W. is supported by a Mildred-Scheel grant of the Deutsche Krebshilfe. We thank Federica Sallusto and Jens Geginat for critical reading and comments, and David Jarrossay for cell sorting.


  1. 1


  2. 2


  3. 3


  4. 4


  5. 5


  6. 6


  7. 7