Clonal dissection of the human memory B cell repertoire following infection and vaccination

Summary

The analysis of the human memory B cell repertoire is of both fundamental and practical significance. We developed a simple method for the selective activation of memory B cells in total fresh or frozen PBMC using a combination of R848 and IL-2. In these conditions 30–40% of memory B cells generated clones producing on average 200 ng IgG in 10 days. This method was used to measure the frequency of antigen specific memory B cells as well as the fine specificity, crossreactivity and neutralizing activity of the secreted antibodies. Following influenza vaccination, specific B cells expanded dramatically, reaching up to 50% of total clonable memory B cells on day 14. Specific B cell expansions were detected also in individuals that did not show a significant serological response. Dynamic changes and persistence of B cells specific for a variety of pathogens were documented in serial PBMC samples collected over almost two decades. These results reveal novel aspects of memory B cell kinetics and provide a powerful tool to monitor immune responses following infection and vaccination.

INTRODUCTION

In response to antigenic stimulation, specific B lymphocytes undergo clonal expansion, class switch and somatic hypermutation leading to the selection of antibodies with increased affinity [1]. Of the clonally expanded B cells, some differentiate to plasma cells that secrete antibodies at high rate and persist in niches in the bone marrow, while others become memory B cells [2–5]. The latter can rapidly respond to antigenic restimulation and it has been suggested that they may contribute to maintaining the plasma cell pool and therefore serum antibody levels over prolonged periods of time [6]. Human memory B cells comprise an IgM⁺ population that is related to mouse marginal zone B cells and carries a prediversified Ig repertoire [7], as well as switched populations of B cells expressing surface IgG or IgA. Human memory B cells can be identified according to the expression of a variety of markers, including CD27 and the ABCB1 transporter, which is characteristic of mature naïve B cells [8, 9].

Memory B cells are known to persist for a lifetime and therefore carry the “history” of the individual's immune response [10, 11]. While the persistence of memory B cells is well documented, there is much less information as to the homing and to the short and medium term dynamics of memory B cells following infection or vaccination [12, 13]. There is therefore a considerable interest in developing simple methods to analyse the composition of the human memory B cell repertoire. This approach would considerably extend the classical serological analysis by addressing function, fine specificity and cross-reactivity of the antibody response at the clonal level. Such methods would also be suitable to study the dynamics of specific memory B cells that may not be reflected by the dynamics of serum antibodies.

Specific memory B cells can be identified and isolated by flow cytometry using fluorescent antigen [14–16]. This method, which has been used to measure increases of memory B cells following vaccination, requires a highly purified labelled antigen and cannot be readily applied to study the response to complex or multiple antigens. In addition, it does not provide information on functional properties of the antibodies such as for instance viral neutralization. An alternative approach that has been followed by several laboratories has been to activate all B cells polyclonally and measure the specificities of the antibodies secreted by B cells in the culture supernatant. Using alloreactive T helper clones we demonstrated that approximately 20% of circulating B cells could be clonally expanded resulting in clonal production of 200 ng IgG [17, 18]. In a subsequent study CD40L⁺ EL4 thymoma cells and IL-4 were used to activate total B cells with high efficiency but lower clonal Ig production [19]. More recently we reported that memory B cells can be selectively triggered by TLR agonists and used the TLR9 agonist CpG and IL-2 to drive polyclonal memory B cell activation [20]. Two recent studies combined TLR and stimulation with additional stimuli such as staphylococcus aureus Cowan (SAC) and Pokeweed mitogen (PWM) and a cytokine cocktail. The first used total PBMC stimulated with PWM, SAC and CpG and as a readout an ELISPOT assay [21]. Frequencies of specific B cell were estimated from the ratio of cells secreting specific IgG antibodies versus IgG-secreting cells. The second study used human purified B cells stimulated by SAC, PWM, CpG, LPS, IL-2, IL-6, IL-10 in the presence of NIH/3T3 feeder in limiting dilution assays to measure frequencies of B cells specific for several antigens [14].

In this study we describe a simple and reliable method for the selective activation of memory B cells using fresh or cryopreserved PBMC. We used this method to analyse the memory B cell repertoire dynamics following infection or vaccination.

RESULTS

Identification of conditions for selective expansion and differentiation of memory B cells in whole PBMC cultures

To establish a simple and reliable method for the quantitative assessment of the human memory B cell repertoire at the clonal level, we searched for culture conditions that would drive the selective expansion and differentiation of human memory B cells. Naïve (CD19⁺CD27^‒IgM⁺ABCB1⁺) and IgG⁺ memory B cells (CD19⁺CD27⁺) cells were isolated from peripheral blood of healthy donors, labelled with CFSE and seeded at 10³ cells/well in 96 U-bottom microplates in the presence of autologous irradiated PBMC. The cultures were stimulated with different combinations of TLR agonists (CpG, R848, 3M001, 3M002 and LPS), cytokines (IL-2, IL-6, IL-10 and IL-4) and CD40L-transfected cells. B cell proliferation was measured by CFSE dilution on day 6 and total IgG production was measured by ELISA in the 10 day culture supernatant.

Out of the 60 conditions tested, the combination of R848 (a TLR7 and TLR8 agonist) and high doses of IL-2 was the minimal and most effective for selective proliferation of memory B cells and induction of IgG secretion (Fig. 1A to B). 3M001 (a selective TLR7 agonist) and 3M002 (a selective TLR8 agonist) were also effective, whereas CpG (TLR9 agonist) was less efficient, and LPS (a TLR4 agonist) was virtually uneffective. While IL-2 strongly promoted memory B cell proliferation and differentiation, IL-6 and IL-10 had only minor effects; IL-4 boosted naïve B cell proliferation and class switching, but inhibited memory B cell proliferation and IgG production (Figure 1B and Supplementary Figure 1). Finally, addition of CD40L-transfected cells to cultures stimulated with R848 and IL-2 induced proliferation of naïve B cells, but did not increase proliferation and moreover inhibited IgG production by memory B cells (Fig. 1A and B and Supplementary Figure 1). PWM and SAC provided only a modest stimulation and actually inhibited memory B cell proliferation and IgG production induced by R848 and IL-2 (data not shown).

Figure 1.

A combination of R848 and IL-2 triggers optimal and selective proliferation and differentiation of human memory B cells. (A) Highly purified naïve (CD19⁺CD27⁻ABCB1⁺) and IgG⁺ memory B cells (CD19⁺CD27⁺) were labelled with CFSE and stimulated with R848 and IL-2 in the presence or absence of CD40L-transfected cells. Shown is the CFSE profile on day 6. (B) 10³ purified memory B cells (CD19⁺CD27⁺) were cultured in 96 U-bottom plates in the presence of irradiated (12 Gy) autologous PBMC and stimulated with different combinations of TLR agonists and cytokines in the absence or presence of CD40L-transfected cells. Shown is the IgG concentration in the 10 day culture supernatant measured by ELISA. Results are representative of 4 experiments performed. (C-E) Total unfractionated PBMC were seeded at 3×10⁴/well in replicate cultures in 96 U-bottom plates and stimulated with R848 and IL-2. (C) Expression of CD19 and CD20 on day 0 and day 7. (D) Percentage of ASC on on day 7. (E) Ig concentrations of 96 pooled supernatants. Results in D and E indicate mean ± SD of 3 experiments performed.

In the above experiment, highly purified memory B cells were stimulated in the presence of irradiated autologous PBMC acting as feeder cells. In order to develop a simple method for memory B cell repertoire analysis, we asked whether the combination of R848 and IL-2 could be used for activation and expansion of memory B cells in cultures of total PBMC. When a PBMC sample (containing 6.5% total B cells) was cultured at low density (3×10⁴ cells/culture) in the presence of R848 and IL-2, the total cell number increased by 10 fold by day 7 and up to 50% of the cells recovered were CD19⁺ B cells (Fig. 1C). Furthermore, most of the CD19⁺ cells expressed reduced levels of CD20, which is characteristic of differentiated ASC. On day 7, the percentages of IgM, IgG and IgA ASC in the same culture conditions accounted for up to 20% of the total cells (Fig 1D). High amounts of IgM, IgG and IgA were present in the day 10 culture supernatants in proportions that roughly reflected the frequency of memory B cells expressing the corresponding surface Ig isotype in PBMC (Fig. 1E). Of note, comparable results were obtained using freshly isolated and cryopreserved PBMC (data not shown). We conclude that stimulation of total PBMC with R848 and IL-2 provides a selective, efficient and simple method for clonal expansion and differentiation of human memory B cells, which can be used with both freshly isolated and cryopreserved samples of total PBMC.

Cloning efficiency of memory B cells and burst size

Memory lymphocytes are known to be heterogeneous and have reduced proliferative capacity as compared to naïve cells. It was therefore important to establish the frequency of memory B cells capable of generating a clone of ASC as well as its size in the above culture conditions. In the experiment reported in Figure 2A, sorted IgG⁺ memory B cells were seeded at 1 or 500 cells/culture in 96 replicates in the presence of lightly irradiated (12 Gy) autologous PBMC as feeder cells and stimulated with R848 and IL-2. Thirty three out of 96 cultures receiving an average input of 1 B cell/well contained on day 10 IgG in amounts ranging from 20 to 1500 ng/culture. These data indicate a cloning efficiency of 37% and an average burst size of 200 ng IgG. In different experiments using sorted IgG⁺ memory B cells, the cloning efficiency was 32 ± 5.5% (range 26–40%, n=5) and the IgG burst size was 205 ± 100 ng/culture (range 70–370 ng, n=5) (Fig. 2B to C). Of note, all cultures containing 500 B cells produced on average 35 μg IgG (Fig. 2A), an amount which is consistent with ~35% of input B cells producing 200 ng IgG. In limiting dilution experiments using total PBMC containing known amounts of IgG⁺ memory B cells stimulated with R848 and IL-2 the cloning efficiency was 42 ± 15.4% (range 29–64%, n=5; Fig. 2D to E). The cloning efficiency calculated with total PBMC or sorted IgG⁺ memory B cells was non significantly different (p=0.22, n=5; Fig. 2B to E). To estimate the rate of Ig synthesis by single ASC, we harvested cells on day 7, washed and plated them in graded numbers in ELISPOT or conventional culture plates. By comparing the number of IgG spots formed and the amount of IgG released over 6 hours we could estimate that ASC secreted on average 26 pg/cell/day.

Figure 2.

Cloning efficiency of memory B cells is approximately 30%. (A–B) IgG⁺ memory B cells isolated by negative cell sorting were plated at 1 cell/well or 500 cells/well in the presence of autologous 12 Gy irradiated PBMC and stimulated with R848 and IL-2. Control wells did not receive any B cells. (A) IgG concentration in individual culture supernatants on day 10 in one representative experiment. The number of positive cultures is indicated (B–C) Mean ± SD cloning efficiency and IgG production as in A in 5 independent experiments with different PBMC donors. (D) Different numbers of total PBMC from two blood donors (A and B, containing 0.53 and 0.51% IgG⁺ memory B cells, respectively) were stimulated with R848 and IL-2 in the presence of autologous 12 Gy irradiated PBMC. The frequency of clonable IgG⁺ B cells was calculated, according to the Poisson distribution, as the seeded PBMC number at which 37% of cultures were negative (shown by the horizontal dotted line). In this case the frequency was 1 in 748 PBMC (donor A) and 1 in 622 PBMC (donor B), corresponding to 1 in 4 and 1 in 3.2 IgG⁺ B in B cells, respectively. (E) Mean ± SD cloning efficiency as in (D) in 5 independent experiments with different PBMC donors. The cloning efficiency measured on sorted B cells or total PBMC was not significantly different (p= 0.22, n=5).

Clonal analysis of antigen specific memory B cells

We next assessed whether the above method can be used for a quantitative and qualitative analysis of the memory B cell repertoire. PBMC from immune donors were stimulated with R848 and IL-2 in 96 well U-bottom plates. In order to detect clonal antibody production, different numbers of PBMC (ranging from 10⁴ to 10⁵ cells/well) were stimulated in replicates cultures. The percentage of IgG⁺ B cells in PBMC was measured at the beginning of the experiment accordingly to the surface expression of CD19 and IgG. IgG antibodies in the 10 day culture supernatants were analyzed by ELISA on plates coated with tetanus toxoid (TT), measles virus antigens (MV), Varicella Zoster virus antigens (VZV) or control uncoated plates. Cultures producing specific antibodies could be detected for all three antigens and the frequency of specific memory B cells could be estimated by limiting dilution analysis. Notably, in some donors, cultures containing sticky IgG antibodies (that may be related to polyspecific antibodies), were detected with a frequency ranging from 0.1% to undetectable levels. These antibodies bound to control uncoated plates, as well as to all antigen-coated plates, were excluded from the analysis. In the experiment shown in Fig. 3A and B, the frequency of MV specific IgG⁺ memory B cells was 1 in 10000 total PBMC, corresponding to 1 in 50 IgG⁺ memory B cells, while the frequency of TT- and VZV-specific B cells was lower (1 in 80,000 and 1 in 117,000 PBMC corresponding to 1 in 400 and 1 in 585 memory B cells, respectively). Cultures containing antibodies specific for these unrelated antigens segregated independently, as determined by the Chi-square-based correlation test. Comparable figures were obtained by testing frozen and fresh PBMC samples collected at the same time or a few days apart (data not shown).

Figure 3.

Limiting dilution analysis of antigen specific IgG memory B cells. Total PBMC were plated in different numbers (from 10⁴ to 10⁵) in 96 replicate cultures and stimulated with R848 and IL-2. The 10 day culture supernatants were analyzed by ELISA for the presence of IgG antibodies that bind to tetanus toxoid (TT), measles virus antigens (MV), varicella zoster virus antigens (VZV) or uncoated control plates. Cultures showing binding to uncoated plates were excluded from the analysis. (A) Shown are ELISA OD values of individual cultures. Cut off values are indicated by a dotted line. (B) Frequency of TT- MV- and VZV-specific B cells determined by limiting dilution analysis in one representative donor. (C) The frequency of memory B cells producing antibodies able either to bind or to neutralize MV was calculated in 3 different donors 49 (ALA), 24 (DCO) and 30 years (AMA) after MV infection. Of note, AMA was boosted with a MV-vaccine 22 years after MV infection.

To understand the relationship between binding and neutralizing antibodies we analyzed PBMC from three MV-immune donors and measured, on the same culture supernatants, antibodies that bound to MV antigens in ELISA and antibodies that neutralized MV infection of Vero cells (Fig. 3C). Cultures containing MV-neutralizing antibodies were in all cases a small fraction of those containing MV-binding antibodies (14.8%, 4.5% and 19.7% in the three donors analyzed), independently of the size of the memory B cell pool that varied considerable among the individuals tested.

In the above experiments the frequency of antigen-specific IgG⁺ memory B cells was expressed relative to the total number of IgG⁺ memory B cells plated. This figure is underestimated since only ~30–40% of IgG⁺ memory B cells are capable of clonal antibody production in these culture conditions. Indeed, it would be more correct to express the frequency of specific B cells relative to clonable rather than total B cells. In order to establish memory B cells frequencies using an independent method, we isolated a large number of IgG⁺ EBV-immortalized B cell clones as described [22] and measured among them the frequency of B cells specific for MV, influenza HA and influenza vaccine virus antigens. On the same PBMC sample we used the polyclonal activation method to determine memory B cell frequencies and cloning efficiency. The results in Table 1 indicate that the frequencies relative to clonable B cells are in good agreement with those estimated on random EBV-B cell clones. These findings indicate that it is important to take into account the cloning efficiency when estimating frequencies of specific memory B cells using methods based on activation and clonal expansion of B cells. We will subsequently refer all the calculated frequencies to the memory B cell clonable fraction.

Table 1.

Antigen specific memory B cell frequencies estimated using polyclonal activation and EBV-B cell immortalization.

	% of Ag-specific B cells relative to total IgG+B cells	% of Ag-specific B cells relative to clonable IgG+ B cells*	% of Ag-specific IgG+ EBV-B cell clones**
MV neutralization	0.4	1.2	1.01
MV ELISA	1.4	4.2	5.06
HA ELISA	0.12	0.36	0.29
Vaccine ELISA	12	36	23

The table shows for four different antibody specificities the memory B cell frequencies obtained by polyclonal B cell stimulation before and after correction for plating efficiency and the absolute frequency among 800 IgG⁺ B cell clones isolated from the same individual using and improved method of EBV immortalization.

^*calculated cloning efficiency in these experiments was 32%.

^**absolute frequency was calculated among 800 IgG⁺ EBV-B cell clones isolated using the EBV immortalization method.

Kinetics of influenza-specific memory B cells upon booster immunization

While it is appreciated that circulating ASC rise rapidly and transiently following booster immunization concomitant with a rise in serum antibody levels, much less is known concerning the kinetics of antigen-specific memory B cells. We used the above method to measure in one individual the frequency of vaccine-specific B cells before and at different times after immunization with the 2006–2007 seasonal influenza vaccine containing A/New Caledonia/20/99 antigens (Fig. 4A). After polyclonal activation we measured in the same culture supernatants antibodies that bind A/New Caledonia/20/99 hemagglutinin (HA) or neutralize A/New Caledonia/20/99 virus infection of MDCK cells. Specific B cells producing binding and neutralizing antibodies showed a marked increase (up to 50 fold) that peaked on day 14, while plasmablasts and serum antibodies had reached their peak level on day 6–8 (Figure 4A and data not shown), indicating that memory B cells and plasmablasts have different kinetics in peripheral blood [6][23]. All neutralizing antibodies also bound HA, while several HA-binding antibodies failed to neutralize the virus. Frequencies of B cells making binding or virus neutralizing antibodies decreased of two fold over the following 16 days.

Figure 4.

Kinetics of memory B cell frequencies following influenza vaccination. (A) PBMC were collected before and at different times after vaccination of a healthy volunteer with seasonal influenza vaccine and stored frozen. The cells were thawed and stimulated with R848 and IL-2 and day 10 culture supernatants were analyzed for the presence of antibodies binding to recombinant A/New Caledonia/20/99 HA (filled squares) or neutralizing A/New Caledonia/20/99 virus (empty circles). Shown is the kinetics of memory B cell frequencies calculated according to the Poisson distribution expressed as per thousand clonable IgG⁺ memory B cells over a 75 days period after influenza vaccine boost. (B) Frequency of B cells producing IgG antibodies to influenza vaccine antigens (a mixture of H1N1, H3N2 and B monovalent particles[41]) measured in 14 volunteers (A–N) before (grey bars) and 14 days after vaccination with the 2007–2008 seasonal influenza vaccine (black bars). (C) Antibody levels measured using the same ELISA in the serum before and 14 day after vaccination expressed as the reciprocal EC50 value. (D) Linear correlation between the increase in the frequency of vaccine-specific B cells and the increase in serum antibody levels after vaccination. Dotted lines indicate the 95% confidence interval of the linear regression.

To investigate the relationship between the memory B cell response and the serological response, 14 volunteers were immunized with the 2007–2008 seasonal influenza vaccine and measured the frequency of total vaccine-specific IgG⁺ B cells and specific serum IgG antibody levels before and 14 days after vaccination (Fig. 4B). All individuals showed an increase in vaccine-specific B cells ranging from 5 to 200 fold. In 8 out of 14 donors vaccine specific cells exceeded 10% of all clonable B cells with individual donors showing as many as 50% of clonable IgG⁺ memory B cells being vaccine-specific. With three exceptions (C, F and M), IgG antibody levels increased in serum after vaccination up to 30 fold (Fig. 4C). Overall there was a correlation between the increase of serum antibodies and memory B cells, with few remarkable exceptions (Fig. 4D). In particular, two donors (C and M) did not show a serological response in spite of a sharp increase in memory B cells. In other cases (I, D and J), the serological response was higher than the increase in specific B cells.

These results indicate that the kinetics of vaccine-induced B cell expansion is remarkably different from that reported for circulating ASC [6, 23] since it peaks approximately one week later and decreased slowly over the following weeks. In addition the expansion of specific B cells can account for as many as 50% of all clonable B cells. Finally, the magnitude of the B cell response does not necessarily reflect the magnitude of the antibody response.

Kinetics of memory B cell frequencies over a 17 year time window

To follow the kinetics of memory B cells over an extended period of time we took advantage of the fact that the method described can be effectively used on cryopreserved samples of just a few million cells and that multiple assays can be run on the same culture supernatant. We therefore used 4 samples of cryopreserved PBMC collected from the same donor over a period of 17 years and measured the frequency of memory B cells against five microbial antigens and, for two of them, analyzed fine specificity and function of the produced antibodies (Fig. 5 and Table 2). VZV-specific memory B cells were present at very high frequency (24% of clonable B cells) in the 1989 sample which was collected 6 months after varicella infection and progressively decreased over the 17 years period to reach 1.4% in the 2006 sample. Toxoplasma Gondii (TG)-specific IgG memory B cells were absent in the 1989 sample, while they were present at very high frequencies (8.3% of clonable B cells) in the 2002 sample, one year after a primary TG infection and decreased to 0.8% in the 2006 sample. TT-specific memory B cells were present at a relatively constant frequency in all the samples and increased about two fold following a booster immunization in 1999. Remarkably, following an MV infection in 1956 this donor showed constant and high levels of both binding and neutralizing MV-specific memory B cells (ranging from 2.9% to 5.8%) throughout the period of observation.

Figure 5.

Kinetics of memory B cell frequencies over a period of 17 years. Frozen PBMC collected from a healthy donor over a 17 year period were stimulated as in Figure 4A. The cultures containing antibodies specific for VZV, TT, MV, Toxoplasma gondii (TG) were determined by ELISA in the 10 day culture supernatant. Cultures containing MV-neutralizing antibodies were identified using a viral neutralization assay. The pie charts show the composition of the memory IgG⁺ B cell repertoire of this individual at the 4 time points analysed. The frequencies of specific memory B cells were calculated as in Figure 4A and expressed as percentage of clonable IgG⁺ memory B cells.

Table 2.

Cross-reactivity of influenza HA-specific IgG⁺ memory B cell frequencies over a 13-years period

	% of clonable antigen-specific IgG+ B cells
Year	A/Texas/36/91 HA	A/New Caledonia /20/99 HA	Texas and New Caledonia HA	Tetanus toxoid
1993	0.06	<0.01	0.10	0.78
2002	0.04	0.34	0.20	2.60
2006	<0.01	0.41	0.12	2.27

Culture supernatants from polyclonally stimulated PBMC of the experiment in Figure 5 were screened by ELISA for the presence of IgG antibodies that are specific or cross-reactive with different H1N1 hemagglutinins (A/Texas/36/91, A/New Caledonia/20/99) and for TT (used as a control). Frequencies of specific memory B cells were calculated as in Figure 4 A and expressed as percentage of total clonable IgG⁺ B cells.

In the same samples we analysed memory B cells specific for recombinant HA of different influenza virus isolates (Table 2). B cells specific for A/Texas/36/91 HA were present at low frequencies in the 1993 and 2002 samples and decreased to an undetectable level in the 2006 sample. B cells specific for A/New Caledonia/20/99 HA were present only in the 2002 and 2006 samples. Interestingly memory B cells producing antibodies that cross-reacted with both A/Texas/36/91 and A/New Caledonia/20/99 HA were already present in the 1993 sample and increased in the samples collected after 1999 most likely as a result of selective stimulation by the A/New Caledonia/20/99 virus.

Taken together these results indicate that the human B cell repertoire shows a very dynamic behaviour over a period of almost two decades. A large fraction of the clonable total repertoire can be occupied by cells of a given specificity and high frequencies of memory B cells can be maintained for half a century.

DISCUSSION

In this study we describe a simple method for the clonal analysis of the human memory B cell repertoire which is efficient, reproducible and can be used with total cryopreserved PBMC. We used this method to determine the fine specificity of virus-specific memory B cells, the relationship between viral binding and neutralizing antibodies and the kinetics of antigen specific B cells following infection or vaccination.

The method developed is simple since it uses total unfractionated PBMC and only two stimuli: R848, a TLR7 and TLR8 agonist [24], and high doses of IL-2. These stimuli were identified as the most efficient and selective for memory B cell activation among the 60 conditions tested that included several TLR agonists, cytokines and CD40L. In total PBMC, R848 stimulates directly memory B cells, that constitutively express TLR7 [20], as well as myeloid cells that express TLR8 and produce cytokines that further support B cell proliferation and differentiation. Indeed, comparable results were obtained using combinations of TLR agonists selective for B and myeloid cells, for instance CpG and LPS. IL-2 is known to be a growth factor not only for T and NK cells, but also for activated B cells [6, 25]. Remarkably, in spite of the high doses of IL-2 used, we did not observe outgrowth of T or NK cells in the 10 day culture period. Additional polyclonal stimuli, such as CD40L, IL-4 and CpG, which have been used by other authors [14, 26–28], were found to increase naïve but not memory B cell activation. The current results are consistent with previous reports indicating distinct requirements for activation of naïve and memory B cells [20].

Using limiting dilution experiments, we established that in cultures stimulated with R848 and IL-2, memory B cells are selectively activated with a cloning efficiency ranging from 26 to 64% and each activated B cell clone produces on average ~200 ng IgG with individual ASC secreting 26 pg IgG/cell/day. Although the present study was focused on IgG⁺ memory B cells, all isotypes, including IgG subclasses, IgM and IgA, were produced in proportion that reflected the frequency of B cells in the memory repertoire. This method is therefore suitable to investigate the IgA response following mucosal infection or vaccination [29, 30], as well as the response of IgM-producing human MZ-like B cells [7, 31, 32]. The finding that only a fraction of memory B cells shows in vitro expansion potential has a precedent in the T cell field. Indeed it has been shown that cloning capacity and expansion potential is maximal for naïve T cells, and progressively reduced in central memory T cells and effector memory T cells [33]. These results are consistent with the notion that the expansion potential of B cells is reduced as the cells enter the memory pool; this is also supported by the observation that, using an improved EBV immortalization method [22], naïve B cells are immortalized with very high efficiency (close to 100%), while memory B cells show efficiencies ranging from 10 to 40% (D.C., unpublished).

The polyclonal activation method was used to monitor antigen-specific B cells in immune donors and the response to an influenza vaccine. Using this method it was possible to measure simultaneously the frequency of memory B cells relative to the clonable B cells and to define the fine specificity, cross-reactivity and neutralizing capacity of the antibodies produced, while excluding artefacts due to the presence of polyspecific antibodies [34]. Several observations were made. First, in the MV- and influenza-immune donors analyzed, the memory B cells that produced virus-neutralizing antibodies were a small fraction (4–25%) of those producing antibodies binding to viral antigens detected by ELISA. This difference is particularly striking if we consider that the neutralization assay detects by definition the entire spectrum of neutralizing specificities while the ELISA assay may underestimate the frequency of B cells making binding antibodies due to the limited representation of some viral antigens in their native or denatured conformation. It is possible that some neutralizing antibodies may have been undetected because of insufficient antibody concentration. The fact that similar proportions of neutralizing memory B cells were obtained in parallel experiments using polyclonal stimulation of PBMC and isolation of EBV-immortalized B cell clones (DC, data not shown) suggests that the polyclonal stimulation method is suitable to detect most in vitro neutralizing antibodies. It is perhaps not surprising that only a fraction of antigen-specific memory B cells makes neutralizing antibodies if one considers that the neutralizing epitopes are only a fraction of all viral epitopes. The role of memory B cells that produce non-neutralizing antibodies remains to be established. While it is possible that some of these antibodies may restrict viral replication by indirect mechanisms [35, 36], it is also possible that they may play no role or even a detrimental role by enhancing immune pathology [37, 38].

The second observation made relates to vaccination. In response to a seasonal influenza vaccine, specific B cells peaked on day 14, i.e. at a time point when specific ASC have virtually decreased to background levels [6] and remained elevated for several weeks. Remarkably, in 8 out of 14 donors antigen-specific B cells accounted for more than 10% of all clonable B cells and in some reached up to 50%. The increase in specific B cells was detected also in those individuals that did not developed increased serum antibody levels, indicating a substantial dissociation in the magnitude of plasma cell and memory B cell responses.

Using PBMC samples of the same donor collected over the last 17 years, we have been able to follow the dynamics of antigen-specific memory B cells. A surprising finding was that several months after infection with VZV or TG, a large proportion of clonable memory B cells was still specific for VZV or TG antigens and their frequency decreased progressively over a period of several years. Intriguingly, this donor maintained a very high frequency of MV-specific memory B cells in all samples analyzed, in spite of having been infected by this virus in 1957. In the same individual memory B cells specific for influenza viruses were present at much lower frequency. In this case the clonal analysis provided relevant information as to the selection of B cells following exposure to different viruses. B cells producing antibodies specific for a 1991 influenza virus (A/Texas/36/91) were present at low levels only in the 1993 and 2002 samples and decreased to undetectable levels in the most recent sample (2006), whereas those specific for a 1999 virus (A/New Caledonia/20/99) were found only in the 2002 and 2006 samples. Remarkably, memory B cells producing antibodies that reacted with both viruses were already present in samples collected before 1999 and increased in the samples collected thereafter. This example illustrates the possibility of predicting the fraction of memory B cells capable of recognizing the new yearly variant influenza viruses, which is relevant to understand the cellular basis of the original antigenic sin [23].

In a broader context, the method described provides an easy tool to monitor at the single cell level the B cell response response to vaccination or infection and to address fundamental questions as to the fine specificity such as the cross-reactivities of the B cell responses.

MATERIALS AND METHODS

Cells and media

The medium used throughout was RPMI-1640 (Invitrogen) supplemented with glutamine, non-essential aminoacids and 10% fetal bovine serum (FBS; Hyclone). Fresh or cryopreserved PBMC were obtained from healthy volunteers. All donors gave written informed consent for research use of blood samples, following approval by the Cantonal Ethical Committee of Cantone Ticino. B lymphocytes were isolated using CD22 microbeads (Miltenyi) and purified by sorting using a FACSAria (BD Biosciences). To isolate IgG⁺ memory B cells CD22⁺ cells were stained with antibodies to CD27 (Pharmingen), CD14, CD3, CD16 (Immunotech), IgD (BD Biosciences), IgM, IgA (Jackson Immunoresearch). To isolate highly purified naïve B cells, CD22⁺ cells were labeled with 25 nM MitoTraker Green FM (Molecular Probes) for 25 min at 37°C [8], washed and stained with CD27, CD14, CD3, CD16, IgG, IgA and IgD antibodies. IgD⁺ cells that were negative for all other markers were used as naïve B cells. In some experiments purified B cells were labelled with 0.5 μM CFSE (Molecular Probes).

Cell cultures

Replicate cultures of total PBMC were set at different cell densities (10⁴ to 10⁵ cells/culture) in 96 U-bottom plates (Costar). In experiments where low numbers of PBMC or sorted B cells were analyzed, lightly irradiated (12 Gy) autologous PBMC (3×10⁴ cells/culture) were used as feeder cells. This dose of irradiation was found to be sufficient to completely abolish B cell activation and differentiation while preserving the feeder effect. Cultures were stimulated with 2.5 μg/ml R848 (3M) and 1000 U/ml human recombinant IL-2. In some experiments additional stimuli and cytokines were tested in different combinations. These include IL-6, IL-10 and IL-4 (10 ng/ml; R&D Systems), pokeweed mitogen (PWM, 5 μg/ml; Sigma), Staphylococcus aureus Cowan (SAC, 1/10000; Sigma), the TLR agonists CpG-2006 (Microsynth) 2.5 μg/ml, R848 2.5 μg/ml, 3M001 2.5 μg/ml, 3M002 2.5 μg/ml (3M), TLR-grade LPS 10 μg/ml (Invivogen) and CD40L transfected J558 cells (4500 cells/well).

ELISPOT

Filter plates (Millipore) were coated with isotype-specific goat anti-human IgG, IgA or IgM (Southern Biotechnology) in PBS buffer (Invitrogen) and blocked with 1% BSA-PBS. Serial dilutions of cells were added to the plates and incubated overnight at 37°C. Plates were then washed and incubated with biotinylated anti-human IgG, IgA or IgM (Southern Biotechnology) followed by streptavidin-horseradish peroxidase (Sigma). The assay was developed with 3-amino-9-ethylcarbazole (Sigma) as chromogenic substrate.

Viral neutralization

Measles Virus neutralization: Vero cells (ATCC) cultured in DMEM 5% FBS were plated at 7500/well in 96 flat bottom microplates. The day after Measles virus (MOI=1) was incubated with culture supernatants at 4°C for 4 hours and added to VERO cells. Three days later cells were stained with 1% Crystal Violet (Sigma) in 70% methanol. Neutralizing antibodies were detected for their capacity to protect the cell monolayer. Influenza virus neutralization: MDCK cells (ATCC) were seeded at 4 × 10⁴/well the day before infection in EMEM 10% FBS. Influenza virus H1N1 A/New Caledonia/20/99 (MOI=0.1) and supernatants were mixed in FBS-free EMEM supplemented with 2 μg/ml of TPCK-treated Trypsin (Worthington Biochemical) (virus growth medium) and added to cells after 30 min at room temperature. Before infection cells were washed with PBS. After 30 min at 37°C virus growth medium was added to the culture and 3 hours later the supernatant was removed and replenished with fresh virus growth medium. Three days after infection cells were stained with 1% Crystal Violet in 70% methanol [39].

ELISA

Total Ig produced in culture supernatants were measured using IgG, IgA-, IgM-, IgG1-, IgG2-, IgG3- and IgG4-specific ELISA. The capture antibodies were from Southern Biotech (IgG, IgA and IgM), Biogenesis (IgG1 and IgG3) and BD (IgG2 and IgG4). Certified Reference Material 470 (ERM®-DA470, Institute for Reference Materials and Measurements [40]) were used as standard material for Ig- subclass quantification. Antigen specific antibodies in culture supernatants were identified by ELISA using half area plates (Corning) coated with appropriate concentrations of the following antigens: Tetanus Toxoid (Chiron); T. Gondii extract (Sorin Biomedica); Measles and Varicella Zoster (Biodesign); recombinant HAs from A/New Caledonia/20/99 and A/Texas/36/91 (Prospec-Tany Technogene); Inflexal (Berna Biotech) or Influvac (Solvay Pharma) vaccine preparations. Cultures with OD 3-fold over the average of the background of four negative control wells were scored as positive.

Immortalization of Memory B cells

PBMC were stained directly with labelled antibodies to CD22 (Pharmingen) and to immunoglobulin IgM, IgD, and IgA. CD22⁺IgM⁻IgD⁻IgA⁻ B cells were isolated using FACSAria and immortalized at 5 cells/well in replicate cultures using EBV in the presence of CpG 2006 and irradiated allogeneic PBMC, as previously described [22].

Statistical analysis

The frequency of B cells precursors specific for a given antigen was calculated accordingly to the Poisson distribution with the following equation: % of Ag-specific B cells = −100 (LN(number of negative wells / number of total seeded wells)) / number of precursors per well. The single hit hypothesis was tested by the χ² test. The data were analyzed using GraphPad Prism software package. The results are expressed as the mean ± 1 SD.

Abbreviations

ASC

antibody secreting cell