Maintenance of Human Germinal Center B Cells In Vitro

THIS STUDY was instigated to identify a culture system for maintaining human germinal center (GC) B cells in vitro. Germinal centers develop in the B-cell follicles of secondary lymphoid tissues during TD antibody responses. They are sites of clonal proliferation, differentiation, and selection of B cells that have been activated in T zones. GC formation is believed to be essential for affinity maturation of antibody responses and the generation of B-cell memory.1 Two major types of B cells can be distinguished morphologically in the GC: centroblasts that form the histologically distinct dark zone and are the likely targets of Ig variable (v)-region gene hypermutation and their nondividing progeny, centrocytes that occupy the follicular dendritic cell (FDC)-rich network of the light zone and express mutated (and at least in tonsil) predominantly isotype-switched Ig.2 The centrocyte population is the subject of selection: these cells die by apoptosis in situ unless they are positively selected. Rescued centrocytes that leave the GC either differentiate to plasma cells or are recruited to the memory pool.2 

Initial selection of centrocytes is believed to occur on the basis of the ability of their mutated sIg to bind native Ag held on FDC.2 Ligation of sIg by immobilized polyspecific antibody can extend the viability of isolated GC B cells by 12 to 24 hours.3 This is believed to be sufficient to allow selected centrocytes to respond to further signals which in vivo are probably provided by GC T cells and FDC. The most potent signals for rescue of GC B cells in vitro are delivered via CD40, a natural counterstructure for which (CD40L) is expressed on activated T cells.2 Engagement of CD40 can also stimulate DNA synthesis in GC B cells in short-term (3 day) culture4 but additional signals are probably required for longer term maintenance of GC B-cell proliferation. Previous studies suggest that these are distinct from those which costimulate CD40-dependent proliferative responses of resting B cells: interleukin-4 (IL-4), for example, provides only a weak costimulus for GC B cells.5 6 

Interestingly, GC B cells have some similarities to pre-B cells: for example, both express CD10 and CD38, show a ready propensity to undergo apoptosis,7 and exhibit a high dependence on a stromal support for their culture which requires both direct cell contact and cytokines.8,9 Recently, Saeland et al10 reported that human bone marrow B-cell precursors can be stimulated to long-term CD40-dependent DNA synthesis by a combination of IL-3, IL-7, and IL-10. A feature of their culture system was the use of a mouse fibroblast line transfected with the gene for human CD32. This transfectant, in addition to providing a potential source of stromal cell surface structures and secreted products, can bind the Fc regions of IgG antibodies and thus promotes more extensive cross-linking of CD40 in the presence of CD40 monoclonal antibody (MoAb).7 For mature resting human B cells, culture with this CD32 transfectant and CD40 MoAb (the “CD40 culture system”⁷) maintains their growth for many weeks11 and for GC B cells up to 4 days6 in the presence of IL-4. Thus, the CD40 culture system may provide a means of identifying costimulatory molecules that can maintain relatively long-term growth of GC B cells.

Here we have investigated the ability of cytokines, used individually and in combinations, to stimulate long-term DNA synthesis by and growth of GC B cells in the CD40 culture system. We identify combinations of cytokines which can selectively maintain CD40-dependent DNA synthesis and cell proliferation for at least 10 days and we provide further evidence of a requirement of GC B cells for stromal support. The ability to maintain long-term growth of GC B cells in vitro as described should prove valuable in the study of the molecular and cellular events involved in GC reactions.

Reagents.Cytokines were human recombinant proteins purchased from R&D Systems Ltd (Oxford, UK) with the exception of IL-4, which was a gift from Dr Steven Gillis (Immunex Research and Development Corp, Seattle, WA). Cytokines were initially assessed at three concentrations; for IL-1β, stem cell factor (SCF ), IL-3, and IL-6, these concentrations were the nominal ED50 in reference target cell systems as stated by the manufacturer, together with a 10-fold higher and a 10-fold lower concentration. For IL-10, concentrations were chosen on the basis of effective ranges reported by Rousset et al¹²; for IL-7, as reported by Saeland et al8 and for IL-2 and IL-4, as reported by Holder et al.5 In experiments to investigate synergy, IL-2 was employed at 20 ng/mL and other cytokines at the intermediate concentration from the range shown in Fig 2.

MoAbs G28-5 (CD40; ref 13), BU52 (CD44), and AC2 (CD39; ref 14) were produced from the hybridomas in the Department of Immunology, University of Birmingham (Birmingham, UK) and were purified by ion exchange chromatography on DE52 (Whatman Ltd, Maidstone, UK). The G28-5 clone was obtained from the American Type Culture Collection repository (Rockville, MD) and the BU52 clone from D. Hardie (Department of Immunology, University of Birmingham). Ascitic fluid containing rat IgM CD77 MoAb, 38-13 was a gift from Dr J. Wiels (Institut Gustav-Roussy, Villejuif, France) and was conjugated to FITC by standard procedures. FITC conjugated CD19 and anti-IgD MoAbs and phycoerythrin conjugated CD3 and CD38 MoAbs were purchased from Dako Ltd (High Wycombe, UK). Phorbol myristate acetate (PMA) was obtained from Sigma Chemical Co (St Louis, MO) and ionomycin from Calbiochem-Novabiochem (Nottingham, UK).

Flow cytometry.Flow cytometry was performed on a Becton Dickinson FACScan and data analysis performed using LYSIS software (Becton Dickinson, Mountain View, CA). Gates for viable lymphocytes and CD32-L cells were set on forward and side scatter. Isotype matched conjugates of irrelevant specificity (Dako Ltd) were employed as controls.

Tonsillar B cells.Human tonsils were obtained from patients undergoing routine tonsillectomy. All procedures were performed at room temperature except where stated. GC B cell3 and resting B-cell15 fractions were isolated as described previously. Briefly, cells were extracted by dissection and dispersal of tonsillar tissue in RPMI 1640 (GIBCO Ltd, Paisley, Scotland) and removal of large tissue fragments at 1g. Mononuclear cells were isolated by density gradient centrifugation on Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden) and T cells were depleted by two rounds of rosette formation with amino ethyl isothiouronium bromide treated sheep red blood cells (SRBC) on ice. Rosettes were removed by centrifugation on Ficoll-Paque. B cells were then separated into high- and low-density fractions by centrifugation on isotonic Percoll (Pharmacia Biotech) 60% (vol/vol) in RPMI. Cells that pelleted under these conditions were enriched for resting B cells mainly from the follicular mantles (95% ± 2.7% [mean ± SD] CD19⁺; 72% ± 3.1% IgD⁺; ≤2% CD3⁺). The cell fraction that was buoyant on 60% Percoll was further purified by depletion of contaminating resting cells and follicular mantle cells by treatment with SRBC covalently coupled with CD39 MoAb, AC2 and, independently, sheep polyclonal anti-human IgD antibody (The Binding Site Ltd, Birmingham, UK) followed by rosette depletion as described. This low density fraction was enriched for GC B cells (93% ± 3.4% [mean ± SD] CD19⁺; 88% ± 8.2% CD38⁺/IgD⁻; 71% ± 7.7% CD77⁺; ≤3% CD3⁺). Isolated cells were stored on ice in fetal calf serum (FCS) before use in experiments that were always performed within 24 hours of tonsillectomy. Cell viability as assessed by trypan blue exclusion was ≥96%.

Cell culture.Culture medium (CM) was RPMI-1640 containing penicillin (100 IU/mL), streptomycin (100 μg/mL), 2 mmol/L glutamine (GIBCO, Grand Island, NY), and 10% (vol/vol) FCS (Sera Lab Ltd, Crawley Down, UK). B cells (10⁶/mL) were cultured in flat-bottom 96-well microtiter plates (Becton Dickinson Labware, Oxford, UK) in a total volume of 100 μL at 37°C in a humidified incubator in 5% CO₂/95% air.

Mouse L cells transfected with the gene for human CD32 (CD32-L cells) were obtained from DNAX Research Institute of Cellular and Molecular Biology (Palo Alto, CA). CD32-L cells were cultured in hypoxanthine-aminopterin-thymidine selection medium consisting of CM containing hypoxanthine (0.1 mmol/L), aminopterin (0.4 μmol/L), and thymidine (16 μmol/L) (Sigma). The adherent cells were recovered using 0.02% (wt/vol) disodium EDTA in phosphate-buffered saline (PBS) pH 7.0 and resuspended in CM. They were γ-irradiated with a dose of 20,000 rads before addition to B-cell cultures at a ratio of (B cells:L cells) 10:1.

Measurement of DNA synthesis.[³H]thymidine ([Amersham International, Amersham, UK] 10 μCi/mL in culture medium, 50 μL per well) was added after the specified interval and cells were procured after a further 16 to 18 hours in culture on a Skatron cell harvester. Assays were performed in triplicate.

Measurement of cell proliferation.Cells recovered from triplicate microplate wells were pooled and incubated for 10 minutes with 1/10 volume of 0.2% disodium EDTA in PBS to disperse aggregates then counted in a Neubauer hemocytometer. Viability was assessed by trypan blue exclusion.

Measurement of Ig.Concentrations of human IgA, IgM, IgE, and IgG subclasses in culture supernatants were measured by enzyme-linked immunosorbent assay using commercially available kits (The Binding Site Ltd).

Fixation of CD32-L cells with carbodiimide.CD32-L cells were fixed with 1-ethyl-3-(3′-dimethylaminopropyl) carbodiimide hydrochloride (ECDI; Calbiochem-Novabiochem, Nottingham, UK) as described by Anderson et al.16 Briefly, CD32-L cells were resuspended in ECDI in saline and incubated on ice for 1 hour, resuspending the cells every 15 minutes to prevent agglutination. They were then washed in serum free RPMI. A concentration of 375 mmol/L ECDI was required to inhibit [³H]thymidine uptake by the CD32-L cells by >95%.

Statistical analysis.The criterion used to define synergy between cytokines in stimulation of B-cell DNA synthesis was that the mean of triplicate counts for [³H]thymidine incorporation for cells cultured with two cytokines less the standard error of the mean (SEM) was greater than the sum of the mean counts for cells cultured with each cytokine alone plus the mean of the SEMs for the responses to the individual cytokines. Statistical significance of positive interactions in data from three experiments was estimated by two-way analysis of variance.

Culture with CD40 MoAb and CD32 transfected L cells can stimulate short-term high-rate DNA synthesis by GC B cells in the absence of exogenous cytokines.High rates of spontaneous DNA synthesis in freshly isolated GC B cells fall rapidly in culture such that [³H]thymidine incorporation cannot be detected after 2 days (Fig 1A), reflecting the progress of these cells to apoptosis in the absence of rescue signals. Neither CD40 MoAb (Fig 1A) nor CD32-L cells (Fig 1B) influenced the decline in DNA synthesis when added to cultures alone; when used together they stimulated a large increase in [³H]thymidine uptake that reached a maximum after 2 days (Fig 1B). The magnitude of this response approached that induced by a combination of PMA (1 nmol/L) and ionomycin (0.8 μg/mL) (Fig 1B), which is an optimal stimulus for GC B cells, and was of similar short duration, falling to <50% of its maximum value after 3 days and approaching background values after about 7 days (Fig 1B).

Both the time course and magnitude of the response to the CD40-dependent stimulus distinguished GC B cells from resting B cells isolated from the same tonsils: culture of high-density B cells with CD40 MoAb in the presence of CD32-L cells promoted a slow rise in [³H]thymidine incorporation from basal levels over 4 days to a level that was equivalent to only about 30% of the response induced by PMA and ionomycin but was then sustained at this level for at least 10 days (Fig 1C).

Ability of individual cytokines to maintain CD40-dependent DNA synthesis in GC B cells.As an initial screen for growth activity, GC B cells were cultured with various cytokines in the presence and absence of CD40 MoAb and CD32-L cells and [³H]thymidine incorporation was measured after 5 days, a time at which basal CD40-dependent stimulation approached a minimum (Fig 1B).

We confirmed the reported ability of IL-2 to stimulate GC B-cell DNA synthesis in the absence of other factors (Fig 2A)5 and found that this was the only cytokine to show CD40-independent activity consistently in this regard. IL-7 typically promoted a modest CD40-independent response (Fig 2A) although cells from some tonsils gave responses of a similar magnitude to those stimulated by IL-2 and these were enhanced by both CD40 MoAb and the CD32 transfectant (not shown). IL-4 stimulated dose-related CD40-dependent responses whereas IL-3 and IL-10 each stimulated small CD40-dependent responses that were enhanced in the presence of CD32-L cells (Fig 2A and B). SCF was essentially inactive at the concentrations used.

Cytokines can synergize to stimulate CD40-dependent DNA synthesis in GC B cells.Cytokines from the panel described above (Fig 2) were investigated for potential synergistic interactions in promoting DNA synthesis in 5-day cultures of GC B cells. Four of the 28 possible pairs of cytokines showed synergy in the presence of CD40 MoAb and CD32-L cells. IL-4 synergized with 3 cytokines (Fig 3): IL-1β (P = .019), IL-7 (P = .021), and IL-10 (P = .053) whereas IL-7 synergized with IL-3 (P = .008). We found no evidence of synergy between cytokines in cultures containing either CD40 MoAb or CD32-L cells alone, nor in unsupplemented cultures.

IL-10 is an important costimulatory factor for CD40-dependent DNA synthesis in GC B cells.Having established that synergistic interactions operate between certain pairs of cytokines in stimulating GC B cells we next investigated combinations of three cytokines for cooperative growth stimulation in the presence of both CD32-L cells and CD40 MoAb. Five sets of 3 cytokines showed synergy in 6-day cultures and all of these included IL-10: IL-10+IL-1β+IL-2 (P = .0002), IL-10+IL-1β+IL-3 (P = .0063), IL-10+IL-3+IL-6 (P = .035), IL-10+IL-7+IL-3 (P = .087), and IL-10+IL-7+IL-4 (P = .048). Importantly, three of these combinations exclusively comprised cytokines which stimulated only very small increases in DNA synthesis when used alone at the same concentrations (Fig 2B). Kinetic studies identified a significant CD40-independent component in the activity of the combination of IL-10+IL-1β+IL-2 (see later and Fig 5B). Of the four combinations that were confirmed to show CD40-dependent activity, IL-10 with IL-4 and IL-7 consistently promoted the greatest response (Fig 4A). Resting B cells from the same tonsils did not show enhanced DNA synthesis in 6-day cultures in response to these combinations of factors (Fig 4B) indicating that the responses of GC B cells were not attributable to the outgrowth of contaminating resting B cells.

Cytokine combinations can stimulate a second phase of CD40-dependent DNA synthesis by GC B cells.Having established that certain cytokine combinations could stimulate GC B cells to DNA synthesis in 6-day cultures (see above) we next investigated their activity over longer periods. By measuring [³H]thymidine uptake at intervals more than 10 days we found that the effective cytokine combinations all stimulated a second phase of CD40-dependent DNA synthesis which reached a maximum after 6 to 7 days and was of similar magnitude to the spontaneous levels observed in freshly isolated GC cells (Fig 5A). Only one combination of cytokines (IL-10 with IL-7 and IL-4) enhanced the earlier phase of CD40-dependent DNA synthesis (from day 1 to day 3) (Fig 5A). One cytokine combination, IL-10+IL-1β+IL-2, was found to stimulate DNA synthesis in the absence of CD40 MoAb (Fig 5B).

The time course of [³H]thymidine uptake by GC B cells in the presence of stimulatory combinations of cytokines contrasted with that of resting B cells under the same conditions (Fig 5C). The initial phase of DNA synthesis in high density B cells was enhanced in the presence of all cytokine combinations except for IL-10 with IL-1β and IL-2. Furthermore, no second phase of DNA synthesis corresponding to that seen in GC B-cell cultures was evident for resting B cells (Fig 5C). It is possible that maintenance of DNA synthesis in resting B cells in 10-day cultures in the presence of IL-10+IL-7+IL-4 is mainly attributable to the effect of IL-4 as this is a strong cofactor for CD40-dependent responses in this cell type, whereas that in the presence of IL-10, IL-1β, and IL-2 reflects the IL-2 induced CD40-independent outgrowth of GC B cells that were a significant contaminant in some of these experiments (Fig 5B, C, and D).

Cytokine combinations can stimulate proliferation of GC B cells.Net increases in viable cell numbers could be detected after 3 to 4 days in GC B cells cultured with triple cytokine combinations, which stimulated DNA synthesis, and these increases reached maximum values of 50% to 100% after 7 to 8 days (Fig 6A through E). No net increase in cell numbers was observed in the CD40 culture system in the absence of cytokines (Fig 6F ). A steady accumulation of nonviable cells in all of these cultures leads to increases in total cell numbers of threefold to fourfold; however, the proportion of nonviable cells produced was greater in the later stages of cultures containing IL-10+IL-1β+IL-2 and IL-10+IL-7+IL-4 (Fig 6A and E).

The largest increases in viable cell numbers consistently occurred in cultures containing IL-10+IL-1β+IL-2 (Fig 6A). However, the combination IL-10+IL-7+IL-4 induced the greatest rate of increase, with a doubling time of approximately 24 hours from day 2 to 3 and this mirrored the greater preceding burst of [³H]thymidine incorporation (Fig 6E).

Phenotype of cells cultured with cytokines.Cells recovered after 8 days from GC B-cell cultures containing CD32-L cells, CD40 MoAb, and cytokine combinations were >95% CD19⁺ and <1% CD3⁺ arguing against possible outgrowth of contaminating non-B cells, especially T cells. Most of the viable cells in these cultures formed large aggregates and had the morphological characteristics of large blasts on the basis of Jenner-Giemsa staining (not shown). The great majority of cells recovered (>90%) remained CD38⁺ and a significant proportion of these also retained expression of CD77 but lacked CD44, features typical of centroblasts (Table 1).

Ig secretion by GC B cells cultured with cytokines.GC B cells cultured for 10 days with CD40 MoAb, CD32-L cells, and growth stimulatory cytokine combinations secreted only small amounts of Ig (Fig 7). All such cytokine combinations induced an increase in IgA and IgG3 secretion (Fig 7B and E) and there was a slight enhancement of IgG1 and IgG4 secretion in the presence of certain combinations (Fig 7C and F ). These differences may not be significant, however, when it is considered that cell numbers were appreciably higher in cultures containing cytokines (data not detailed).

In the absence of CD40 MoAb, IL-10 together with IL-1β and IL-2 stimulated large increases in secretion of IgM, IgA, and all IgG subclasses (Fig 7) but not IgE (not shown) by GC B cells. This was the only cytokine combination that stimulated IgG2 synthesis. These effects were inhibited by the presence of CD40 MoAb.

CD32-L cells must be metabolically active to support long-term cytokine-dependent DNA synthesis in GC B cells.Irradiated CD32-L cells may release soluble factors which themselves contribute to the observed CD40-dependent growth of GC B cells. We investigated this possibility by fixing the CD32-L cells with ECDI, a reagent that renders cells metabolically inactive yet allows their surface molecules to interact with receptor ligands.16 The ability of ECDI-fixed and irradiated CD32-L cells to support cultures of GC B cells was then compared.

Although ECDI-fixed CD32-L cells were able to support a similar level of DNA synthesis by GC B cells in the presence of CD40 MoAb to that provided by irradiated transfectants, the enhancement of these responses by IL-10 with IL-4 and IL-7 was not evident when fixed L cells were used (Fig 8A). Thus, the costimulatory effect of the cytokine combination appears to require metabolically active stromal cells, most likely so that they can secrete additional growth factors. ECDI-fixed CD32-L cells showed a much reduced capacity to support resting B-cell cultures in the presence of CD40 MoAb alone than did irradiated transfectants; however, IL-4 and to a lesser extent the combination of IL-10 with IL-4 and IL-7, were able to overcome this (Fig 8B). The similarity of the CD40-dependent IL-4 induced responses in resting B cells supported by irradiated and fixed L cells provides evidence that the ability of the latter to interact via CD32 was not greatly impaired by the fixation process.

As Ig v-gene hypermutation is believed to be initiated in the centroblasts17 the signals that regulate proliferation of GC B cells are likely to be important in the events regulating affinity maturation. Centroblasts in GC dark zones initially arise from antigen-specific primary B-cell blasts, which have colonized a follicle following their activation as resting B cells in T-zones; once established, however, it is possible that this pool may also be supplied by centrocytes which re-enter the dark zone. This possibility was first raised by Holder et al4 by demonstrating that membrane-bound CD40L was capable of maintaining GC cells in active cycle over 3 to 4 days. More recently, Han et al18 have suggested that in order for the population of the GC to remain stable, rescued B cells must divide several times. Although a major role of CD40 is probably to direct selected centrocytes into the memory pathway,2 7 it is not known whether GC B cells that have undergone initial interactions via CD40 are then receptive to additional signals that can drive their re-entry into cell cycle and possibly sustain their cycling status for some time. The present study aimed to identify factors that can maintain CD40-rescued GC B cells in cell cycle and thereby, indirectly, provide a potential first step toward characterizing the external signals underlying affinity maturation.

CD40 MoAb alone has been regarded as a potent rescue signal for GC B cells but a poor stimulator of their proliferation.2 In the presence of CD32-transfected L cells, however, we have now shown that CD40 MoAb can provide a strong, albeit transient, stimulus for GC B cell DNA synthesis. Under these conditions of the “CD40 culture system” not only is the decline in spontaneous DNA synthesis of GC B cells prevented but the initial rate is enhanced to near maximal levels achievable before subsiding after 2 days. A major contribution to this potent stimulation is likely to be the optimal receptor cross-linking facilitated by the CD32 transfectant as no stimulation occurred in its absence. That metabolically inert ECDI-fixed CD32 transfectants supported similar levels of stimulation to their irradiated counterparts makes it unlikely that soluble factors from the L cells played a major role in enhancing basal CD40-induced responses.

As rescue via CD40 is known to extend the viability of GC B cells in culture to about 2 to 3 days2 the kinetics of DNA synthesis in the presence of CD40 MoAb and CD32-L cells could reflect initial rescue of centrocytes which re-enter cell cycle for 2 days but then succumb to apoptosis. Alternatively, continued cycling of centroblasts could also make a contribution to this profile and our current data do not allow us to distinguish between these possibilities. The GC dark zone is reported to be essentially devoid of T cells1 suggesting that opportunities for interactions with CD40L in vivo are unlikely. However, Grammer et al19 have recently reported that a second ligand for CD40 is expressed by activated B cells and that this can costimulate B-cell responses; therefore it is possible that CD40 interactions between neighboring B cells may be important in regulating proliferation of centroblasts. Our observation that GC B cells but not resting B cells can be stimulated to optimal DNA synthesis via CD40 would be, at least, consistent with this notion.

As DNA synthesis by GC B cells in the CD40 culture system had subsided to minimum levels after 5 days we chose this time to assess cytokines for costimulatory activity in maintaining longer term growth. The screen of individual cytokines confirmed the moderate stimulatory activity of IL-2 and IL-4 reported previously5,6 but did not identify others with the exception of occasional, but inconsistent, modest responses to IL-7. The ability of IL-2 to stimulate CD40-independent DNA synthesis of GC B cells served here as a useful control for GC B-cell responses given its previously well-documented actions on this population.5 IL-10 stimulated small levels of CD40-dependent DNA synthesis but these were less evident at higher concentrations suggesting that the relative activities of IL-10 in promoting both growth and differentiation of B cells reported recently12 may be dose-related for GC B cells.

Our finding that IL-10 was a component of all triple combinations of cytokines that acted synergistically in stimulating CD40-dependent DNA synthesis by GC B cells suggests that this T-cell product may be a particularly important growth factor for GC B cells. Levy and Brouet20 have recently reported that IL-10 can rescue splenic GC B cells from apoptosis. Therefore, it is possible that the role of IL-10 that we identified in promoting DNA synthesis was related to its ability to improve survival despite our inability to detect a significant direct effect on the rescue of tonsillar GC B cells from programmed death.

The stimulation of GC B-cell DNA synthesis by cytokine cocktails was reflected in increased cell numbers in cultures (Fig 6). In the experiments documented we did not replenish either the culture medium or any of the stimulatory reagents; therefore, the responses observed in longer term cultures may well have been suboptimal. The greater proportion of nonviable cells found in cultures containing IL-10+IL-1β+IL-2 and IL-10+IL-7+IL-4 (Fig 6A and E) as compared with other stimulatory combinations may be a consequence of nutrient depletion resulting from the preceding higher rates of cell division. An alternative possibility is that this reflects the generation of larger numbers of short-lived blasts under these conditions. The high observed rate of proliferation induced by the combination IL-10+IL-7+IL-4 (doubling time ∼24 hours, Fig 6E) suggests that these culture conditions may be relevant to the cell cycle kinetics of the follicular reaction.

When GC B cells from cultures containing the cytokine combination IL-10+IL-4+IL-7 were recovered and then washed and transferred to cultures with fresh cytokines, CD32-L cells, and CD40 MoAb after 7 or 8 days and then at weekly intervals thereafter, high rates of DNA synthesis could be maintained for up to 4 weeks (data not detailed). Therefore, we believe that long-term maintenance of GC B cells will be feasible using the conditions described.

Five of the seven cytokines which were seen to cooperate in stimulating GC B-cell DNA synthesis have been identified in cells found in the GC suggesting that they might be of some physiological relevance. IL-2, IL-4, and IL-10 have each been identified at the mRNA level in GC T cells and IL-1β²¹ and IL-722 FDC at both the mRNA and protein levels. Although IL-3 is a product of TH cells,23 there are as yet no specific reports of IL-3 production by cells in the GC. IL-6 is an important autocrine B-cell growth and differentiation factor24 whereas a study reporting that IL-10 can be produced by activated human splenic B cells indicates that GC B cells may produce their own IL-10.20 

Three lines of evidence show that the cells being maintained in the CD40 system with triple combinations of cytokines were indeed GC B cells. Firstly, when comparing GC B-cell populations with resting B cells, the cytokine effects were selective: for resting B cells, they failed either to enhance CD40-dependent DNA synthesis at day 5 or to promote a second phase of CD40-dependent [³H]thymidine incorporation, each of which was evident for GC B-cell populations. Secondly, and most importantly, cells recovered from 8-day cultures of GC B cells with cytokine combinations that maintained DNA synthesis retained a GC B-cell phenotype; these cells were not only CD38⁺ (strongly expressed by all GC B cells) but a substantial proportion were also CD77⁺ and/or CD44⁻. Although CD77 is considered as a general but highly specific marker for B cells of GC origin, it is more strongly expressed by centroblasts than centrocytes; the absence of CD44 is a good marker for centroblasts and distinguishes them from centrocytes that are weak expressors of this receptor.25 The finding that these cultures contained significant numbers of CD38⁺/CD44⁻ cells and CD38⁺/CD77⁺ cells (Table 1) is consistent with the notion that proliferating centroblasts are being encouraged under the culture conditions established. Finally, we have seen that withdrawing either the CD40 stimulus or the cytokines results in cessation of DNA synthesis followed by death of the cultured cells within 48 hours (data not detailed); this is in keeping with maintaining a potentially apoptotic GC B-cell population in these cultures. The requirement for continuous stimulation via CD40 may reflect the importance of CD40 in maintaining proliferation of GC B cells in vivo: indeed, studies in mice have shown that antibody to CD40L can abrogate an established GC reaction.18 

GC B cells maintained in cycle by cytokines under the conditions described secreted only small amounts of Ig (Fig 7) indicating that no significant differentiation to plasma cells occurred. The high levels of IgM secretion by GC B cells cultured with IL-10+IL-1β+IL-2, which stimulated long-term CD40-independent DNA synthesis, probably reflects the action of IL-2, which has been reported to promote the outgrowth of a CD5⁺ subset to IgM secreting plasmablasts.26 Our finding that IL-10 and IL-1β cooperated with IL-2 to stimulate DNA synthesis suggests that the former two cytokines may be relevant cofactors in that pathway.

Recently, Arpin et al27 have reported that a combination of IL-2 and IL-10 stimulates CD40-dependent proliferation of GC B cells in 3-day cultures. These cells acquire some of the phenotypic characteristics of memory cells (CD38⁻/CD20⁺) after a further 4 days but if the CD40 stimulus is withdrawn, differentiation toward plasma cells occurs. We did not identify cooperativity between IL-10 and IL-2 in our studies but this may simply reflect kinetic considerations; thus, because our intention was to identify the factors for long-term growth of GC B cells, we specifically screened for interactions among cytokines in stimulation of DNA synthesis in 5-day cultures and later while not investigating earlier time points.

With the one exception (IL-10+IL-1β+IL-2), cooperation among cytokines in stimulating DNA synthesis of GC B cells was identified only in cultures containing both CD40 MoAb and CD32-L cells. This suggests that the majority of cytokine-mediated effects observed may depend on extensive cross-linking of CD40 and/or the presence of stroma. In the absence of exogenous cytokines, ECDI-fixed CD32 transfectants that are metabolically inert supported the basal CD40-dependent DNA synthesis in GC B cells, almost as effectively as did their irradiated counterparts arguing against a major role for feeder cell-derived soluble factors. However, the combination of exogenous IL-4 with IL-7 and IL-10 which enhanced CD40-dependent DNA synthesis in the presence of irradiated CD32-L cells had no effect when ECDI-fixed transfectants were employed. Therefore, soluble factors from the transfectant may act in concert with exogenous cytokines and this would support a role for stromal derived factors in long-term CD40-dependent GC B-cell growth. Thus, this study provides further evidence that growth of GC B cells is likely to be regulated by stroma.

The description of a culture system capable of maintaining B cells of GC origin for at least 10 days, and potentially for several weeks, should aid studies aimed at elucidating mechanisms underlying somatic hypermutation on Ig v-region genes and possibly other events associated with the GC response. The results presented here also provide candidate factors that may be physiologically relevant to the process of sustaining centroblast proliferation.

We are grateful to Michelle Holder for Ig measurements.