Maintenance of Long-Lived Plasma Cells and Serological Memory Despite Mature and Memory B Cell Depletion during CD20 Immunotherapy in Mice¹

The B1, marginal zone, and germinal center B cell subsets contribute to the circulating natural Ab pool, T cell-independent (TI)⁴ IgM Ab responses, and adaptive immunity (1, 2, 3, 4) by terminal differentiation into plasma cells, the effector cells of humoral memory and immunity (5, 6). Ag activation of mature B cells leads initially to the development of germinal centers, the transient generation of plasmablasts that secrete Ab while still dividing, and short-lived plasma cells that secrete Ag-specific germline-encoded Abs (7). Extrafollicular plasma cells with half-lives of 3–5 days appear first in the T cell zones (8) and then migrate to the red pulp of the spleen (9). In germinal centers, Ag-specific B cells proliferate, diversify their receptors, and give rise to pools of long-lived memory B cells (10). Memory B cells generated during the second week of primary Ab responses express mutated Ag receptors with enhanced affinities for Ag (11, 12, 13). The cellular precursors contributing to memory Ab responses are classically defined as surface IgM⁺IgD⁻B220⁺ or IgM⁻IgD⁻B220⁺ Ag-binding B cells that do not express syndecan-1 (CD138) (12, 14, 15, 16). Memory B cells persist in substantial numbers for at least 8 wk after Ag challenge, rapidly expand during secondary responses, and terminally differentiate into plasma cells (17). Thereby, postgerminal center memory B cells give rise to Ab isotype-switched plasmablasts and then long-lived plasma cells once outside of germinal centers.

Persistent Ag-specific Ab titers are thought to derive from long-lived plasma cells (5, 17). Ab-secreting plasma cells generated in the spleen migrate to the bone marrow where they may persist for the life of the animal without the need for self-replenishment or turnover (6, 7, 18, 19, 20, 21). In fact, bone marrow cell transfers confer long-lived Ab production, presumably due to the transfer of long-lived plasma cells (19). Nonetheless, Radbruch et al. (5) state in a recent review, “The biology of plasma cells… remains enigmatic and is of considerable debate. Of particular interest is whether plasma cells are long-lived and form a pool of memory plasma cells that is separate from the pool of memory B cells.” In one model for the maintenance of serum Ab levels, primary and secondary immune responses generate a pool of long-lived plasma cells that occupy essential survival niches within the bone marrow where they survive indefinitely, resulting in long-lived serum Ab levels (6, 8, 20, 22). In a second model, persisting Ag drives the memory B cell pool to chronically generate plasmablasts and short-lived plasma cells that result in long-lived Ab production (23). In a third model, memory cells themselves (24, 25) or pre-plasma cell precursors (26) continuously give rise to plasma cells in an Ag-independent manner as a means of maintaining serum Ab levels for extended time periods. In this third model, Ag-independent cytokines or TLR signals continuously induce a subset of memory B cells to differentiate into plasma cells with defined half-lives. These three models are not mutually exclusive, but all hinge on understanding whether long-lived plasma cells are distinct from the memory B cell pool or other precursor populations.

Mature CD20⁺ B cell depletion is proving to be effective in treating B cell malignancies and some autoimmune diseases (27, 28). CD20 is a B cell-specific molecule that is first expressed on the cell surface during the pre-B to immature B cell transition, but is presumed to be lost upon plasma cell differentiation (29, 30). However, it has been difficult to determine the precise effects of B cell depletion on human B cell subsets and humoral immunity since most CD20 mAb therapy patients are also given immunosuppressive drugs (27). To assess the effects of CD20⁺ B cell depletion on germinal center B cells, memory B cells, plasmablasts, Ab-forming plasma cells, existing Ab levels, and primary and memory Ab responses in mice, B cells were depleted using CD20 mAb (30, 31, 32, 33, 34, 35, 36). Although mature and memory B cells were CD20⁺ and were effectively depleted by CD20 mAb treatment in vivo, CD20⁺ B cell depletion did not deplete long-lived plasma cell numbers or negatively affect long-lived serum Ab levels. Thus, the bone marrow plasma cell pool does not require ongoing contributions from the memory B cell pool for its maintenance.

Wild-type C57BL/6 and Rag1^−/− (B6.129S7-Rag1^tm1Mom/J) mice were obtained from The Jackson Laboratory. Hemizygous transgenic mice expressing human CD19 (TG-1 line) on a C57BL/6 genetic background were as described previously (37, 38). In some experiments, mice received drinking water containing BrdU (1 mg/ml; Sigma-Aldrich). Mice were housed in a pathogen-free barrier facility and were used at 6–8 wk of age. These studies were approved by the Duke University Animal Care and Use Committee.

Mouse CD20-specific mouse mAbs were as described elsewhere (30). Other mAbs included B220 mAb RA3-6B2 (provided by Dr. R. Coffman, DNAX Corporation, Palo Alto, CA) and CD1d (1B1), CD21 (7G6), CD24 (M1/69), CD138 (281-2), and GL-7 mAbs from BD Pharmingen. TCRβ chain mAb (H57-597) was purchased from eBioscience. Alexa Fluor 350-conjugated streptavidin was obtained from Invitrogen Life Technologies. A BrdU Flow kit (BD Pharmingen) was used to detect intracellular BrdU labeling. VLA-4 (clone PS/2) and LFA-1 (clone M17/4) blocking mAbs were purified from hybridoma (American Type Culture Collection) tissue culture supernatant fluid. Isotype-specific and anti-mouse Ig or IgM secondary Abs were purchased from Southern Biotechnology Associates.

Sterile and endotoxin-free CD20 (MB20-11) and isotype-matched control mAbs were injected i.v. through lateral tail veins. Mice were given 250 μg of mAb in 200 μl of PBS, unless otherwise indicated. For thioglycolate-induced peritonitis, 1 ml of thioglycolate solution (3% w/v; Sigma-Aldrich) was injected i.p. 1 day before mAb treatment. In some cases, function-blocking LFA-1 and VLA-4 mAbs (200 μg each) were injected i.p. as indicated.

Two-month old mice were immunized i.p. with trinitrophenyl (TNP)-LPS (50 μg; Sigma-Aldrich) or DNP-Ficoll (25 μg; Biosearch Technologies, San Rafael, CA) in 200 μl PBS. Mice were also immunized i.p. with DNP conjugated to keyhole limpet hemocyanin (KLH) (100 μg; Calbiochem-Novabiochem) in 200 μl of CFA (50% v/v) and were boosted i.p. 28 days later with DNP-KLH in 200 μl of IFA (50% v/v). In other experiments, mice were immunized i.p. with 100 μg of 4-hydroxy-3-nitrophenyl acetyl conjugated to chicken γ-globulin (NP-CGG) precipitated in alum as described previously (10). Mice were bled before and after immunizations as indicated.

Single-cell leukocyte suspensions were stained on ice using predetermined optimal concentrations of each primary and secondary Ab for 20–60 min and fixed as described elsewhere (37). Cells with the forward and side light scatter properties of lymphocytes were analyzed on FACScan or FACSCalibur flow cytometers (BD Biosciences). Background staining was determined using unreactive control mAbs (Invitrogen Life Technologies) with gates positioned to exclude ≥98% of the cells. Bone marrow and spleen B cell subsets were gated as previously described (32). Frozen sections of spleen were fixed in acetone and stained with FITC-conjugated GL-7, PE-conjugated TCRβ chain, and biotin-conjugated B220 mAbs followed by Alexa Fluor 350-labeled streptavidin at predetermined optimal dilutions in saline containing 1% BSA. Ab reactivity was visualized by fluorescence microscopy.

DNP- and TNP-specific Ab levels in individual serum samples were determined in duplicate using Ab isotype-specific ELISA as previously described (38). Sera from TNP-LPS-immunized mice were diluted 1/400, whereas sera from DNP-Ficoll- and DNP-KLH-immunized mice were diluted 1/1000 for analysis using ELISA plates coated with DNP-BSA (Calbiochem-Novabiochem) or TNP-BSA (Biosearch Technologies). Relative Ag-specific IgM and IgG titers were determined for each group of mice using pooled serum samples with results plotted as OD vs dilution (log scale). NP-specific serum Ab was detected by ELISA with IgG or IgG1 anti-NP Ab concentrations estimated by comparisons to standard curves generated using the high-affinity NP-specific H33Lγ1 mAb (39). The relative affinities of serum NP-specific Ab responses were assessed by comparing the relative amounts of Ab bound to NP₅-BSA vs NP₂₅-BSA as described elsewhere (10). Briefly, half of a 96-well ELISA plate was coated with NP₅-BSA or NP₂₅-BSA, and a set of sera was applied to the plate. OD values of NP₅-specific and NP₂₅-specific Abs of each sample were used to calculate the relative ratio of NP-specific low- to high-affinity Abs. Relative NP-specific IgM and IgG titers were determined for each group of mice using pooled serum samples with results plotted as OD vs dilution (log scale).

Spleen B cells from mice immunized i.p. with 100 μg of NP₁₈-CGG in alum 42 days before harvest were enriched by T cell depletion using Thy1.2 mAb-coated magnetic beads (Invitrogen Life Technologies; >94% B220⁺ cells and <0.7% Thy1.2⁺ cells). Spleen T cells from mice immunized i.p. with 100 μg of CGG in alum 30 days before harvest were enriched by B cell depletion using B220 mAb-coated magnetic beads (Invitrogen Life Technologies and Dynal; <0.8% B220⁺ cells). Equal numbers of B and T cells (3 × 10⁷) were injected i.v. into Rag1^−/− mice immediately before i.v. injection of MB20-11 or isotype-matched control mAbs. One day after mAb treatment, all mice were given soluble NP₁₈-CGG (100 μg) i.v., with sera collected at the times indicated. For ELISPOT assays, bone marrow and spleen cells were harvested from individual mice 10 days after boost immunizations.

The frequency of NP-specific Ab-secreting cells (ASCs) from single-cell splenocyte and bone marrow suspensions were estimated by ELISPOT assays using NP₆-, NP₂₅-, or NP₃₃-BSA conjugates as previously described (10). Bone marrow and spleen cells were added to Immobilon-P Multiscreen 96-well plates (Millipore) that were precoated with NP-BSA (5 μg/ml) at either 10⁴, 10⁵, or 10⁶ cells/well in culture medium (100 μl; RPMI 1640 containing 10% FCS, 10 mM glutamine, 100 U/ml penicillin/streptomycin, and 55 μM 2-ME). After incubating the plates for 3 h at 37°C in a humidified CO₂ incubator, the plates were washed three times and incubated with alkaline phosphatase-conjugated polyclonal goat anti-mouse IgG or IgG1 Abs (Southern Biotechnology Associates) for 1 h at room temperature. After washing, the plates were developed using BCIP/NBT substrate (Sigma-Aldrich). Similarly, the total number of splenic and bone marrow IgG and IgM ASCs was determined using plates coated with polyclonal anti-mouse Ig and developed with alkaline phosphatase-conjugated goat anti-mouse IgM or IgG Abs (Southern Biotechnology Associates). For sorting of cell populations before use in ELISPOT assays, the MACS system (Miltenyi Biotec) was used to purify CD20⁺ and CD20⁻ cells. Cells labeled with biotinylated MB20-11 mAb and anti-biotin microbeads were used to positively select CD20⁺ cells. Flow-through cells were harvested and run a second time over the magnetic column to remove contaminating CD20⁺ cells. The purity of the CD20⁻ population was >98%, while CD20⁺ cell purity was >95%.

Purified B cells were differentiated in culture by stimulation with LPS or CD40 mAb plus IL-4 as previously described (40). Briefly, T cell-depleted splenocytes (>93% B220⁺) were labeled with 1 μM CFSE (CFSE Vybrant; Invitrogen Life Technologies and Molecular Probes) according to the manufacturers’ instructions. B cells were cultured in 24-well flat-bottom tissue culture plates (3 × 10⁶ cells in 1.5 ml of culture medium/well) at 37°C with 5% CO₂ in the presence of LPS (10 μg/ml) or CD40 mAb (1 μg/ml. HM40-3, BD Biosciences) in combination with IL-4 (100 U/ml, Sigma-Aldrich) for 72 h. CD20 expression levels were assessed by immunofluorescence staining of harvested cells with flow cytometry analysis.

All data are shown as means ± SEM. The Student t test was used to determine the significance of differences between sample means.

The influence of B cell depletion on serum Ig levels was assessed in littermates after CD20 (MB20-11) or control mAb treatments (250 μg). At this mAb dose, circulating, spleen and lymph node B cell numbers were reduced by >95% by day 2 and began to recover on day 57 (data not shown and Refs. 31 , 32 , and 34). Despite effective B cell depletion, serum IgM, IgG, and IgA levels were similar for both CD20 and control mAb-treated littermates for up to 58 days following mAb treatment (Fig. 1 A). Thus, CD20⁺ B cell depletion did not significantly reduce basal serum Ig levels.

Since peritoneal B cells are not efficiently depleted by CD20 mAb treatment (31, 32), their contribution to basal serum Ig levels was assessed. Thioglycolate given i.p. 1 day before CD20 mAb treatment induces significant peritoneal B cell depletion within 7 days (32). B cell depletion prevented the increase in serum Ig levels that follows thioglycolate-induced inflammation. In addition, thioglycolate plus CD20 mAb treatment reduced serum IgM ∼40% from initial levels (p < 0.05), while serum IgG1, IgG2b, IgG3, or IgA levels did not fall below the normal range (Fig. 1,B). To assess whether prolonged CD20 mAb treatment might also affect serum Ig levels, 3-mo-old CD19-transgenic mice were treated biweekly with a CD20 mAb dose (5 μg) that effectively depletes circulating and peripheral B cells (34). Serum Ig levels increase at a faster rate than normal in autoimmune-prone CD19-transgenic mice, as described (38). Continuous CD20⁺ B cell depletion did not significantly affect basal IgM or IgG levels, but did prevent the development of hypergammaglobulinemia (Fig. 1 C). Thus, CD20⁺ B cell depletion significantly attenuated increases in serum Ig, but had minimal effects on basal serum Ig levels.

Ab responses to the T cell-dependent (TD) Ag DNP-KLH were assessed in mice treated with CD20 mAb 7 days before immunization. CD20 and control mAb-treated littermates generated similar primary IgM Ab responses 7 days after DNP-KLH immunization (Fig. 2,A). However, IgM Ab responses in CD20 mAb-treated mice decayed rapidly between 2 and 4 wk following immunization relative to control mAb-treated littermates. By contrast, DNP-specific IgG responses were 20-fold lower in CD20 mAb-treated mice compared with control mAb-treated littermates on day 28. All mice were retreated with CD20 or control mAb on day 21 and were rechallenged with DNP-KLH on day 28. Secondary IgM and IgG Ab responses were impaired (15- and 40-fold lower, respectively) in CD20 mAb-treated mice by day 35, while significant secondary Ab responses were observed in control mAb-treated littermates (Fig. 2,A). When mice were treated with CD20 mAb plus thioglycolate, primary IgM responses to DNP were significantly decreased, with 6-fold lower IgM Ab titers on day 7 and 77-fold lower IgG titers on day 28 (Fig. 2 B). Secondary IgM and IgG titers were 30- and 36-fold lower on day 35, respectively. Thus, initial TD IgM Ab responses progressed normally after B cell depletion due predominantly to peritoneal B cells, while class-switched and secondary Ab responses were severely inhibited.

The influence of B cell depletion on TI humoral immune responses was assessed in littermates treated with CD20 or control mAb 7 days before TNP-LPS (TI-1 Ag) or DNP-Ficoll (TI-2 Ag) immunizations. Following TNP-LPS immunizations, hapten-specific IgM levels were similar in both CD20 and control mAb-treated littermates, while TNP-specific IgG titers were significantly lower in CD20 mAb-treated mice (Fig. 2,C). With thioglycolate and CD20 mAb treatment before TNP-LPS immunization, both TNP-specific IgM and IgG levels were significantly decreased (Fig. 2,D). Hapten-specific IgM and IgG responses to DNP-Ficoll were also significantly reduced in CD20 mAb-treated mice compared with control mAb-treated littermates (Fig. 2 E). Thus, CD20⁺ B cell depletion reduced TI-1 IgG and TI-2 IgM and IgG responses, while peritoneal B cell depletion was necessary to abrogate TI-1 IgM responses.

The effect of CD20 mAb treatment on germinal center B cells was assessed by immunofluorescence staining. Most B220⁺GL-7^high germinal center B cells expressed CD20 at levels 7-fold higher than those found on B220⁺GL-7⁻ follicular B cells (Fig. 3,A). However, a small population (1–2%) of germinal center B cells expressed little to no CD20. Since germinal center cells are maximally established by day 8 following immunizations with NP-CGG (9), mice were treated with CD20 or isotype control mAb on day 7 after immunization. On day 14, B220⁺GL-7^high germinal center cell numbers were decreased by 99% in naive mice following CD20 mAb treatment and by 75% in immunized mice (Fig. 3, B and C). The majority of B220⁺ B cells remaining within the spleens of immunized mice following 7 days of CD20 mAb treatment represented recent pre-B and immature B cell emigrants from the bone marrow that had not begun to express CD20 at significant densities, as described previously (32). By contrast, mature bone marrow (B220^highIgM⁺), blood (B220⁺), mature spleen (B220⁺CD24⁺CD21⁺), spleen T1 (B220⁺CD24^highCD21⁻), spleen T2 (B220⁺CD24^highCD21⁺), and marginal zone (B220⁺CD1d^highCD21^high) B cells were depleted by 92–99% following CD20 mAb treatment in naive mice and mice immunized with NP-CGG or DNP-KLH (Fig. 3,C and Table I). Histological examination of spleen sections revealed dramatically reduced follicular and germinal center B cell staining 7 days after CD20 mAb treatment (Fig. 3 D). Therefore, immunization does not affect the overall efficiency of B cell depletion and most germinal center B cells expressed CD20 at high levels, thus making them sensitive to CD20 mAb-mediated depletion.

The effect of CD20⁺ B cell depletion on memory Ab responses was assessed using DNP-KLH-immunized mice that were then treated with CD20 or control mAb 21 days later. Despite effective B cell depletion for 57 days as described (31, 32), significant DNP-specific IgM and IgG Ab levels were maintained in CD20 and control mAb-treated mice for at least 77 days following primary immunizations (Fig. 4,A). The mice were then retreated with CD20 or control mAb on day 82 and rechallenged with DNP-KLH 7 days later. Secondary Ab responses were elicited in control mAb-treated mice, but IgM levels were 26-fold lower (day 96) and IgG titers were 6-fold lower (day 104) in CD20 mAb-treated mice (Fig. 4,B). To compare primary and secondary Ab responses, naive mice were immunized with Ag in parallel with the rechallenge of their previously immunized littermates. These primary IgM responses were greater than secondary responses in previously immunized mice after CD20 mAb treatment, but IgG responses were relatively modest (Fig. 4, A and B). Thus, CD20 mAb treatment inhibited memory IgM and IgG B cell responses, but did not deplete preformed serum Ab.

Adoptive transfer experiments were conducted to verify the effect of CD20 mAb treatment on memory Ab responses and to exclude the induction of plasma cells from primary immune responses. NP-CGG-primed B220⁺ B cells or naive B cells were mixed with equal numbers of CGG-primed Thy1.2⁺ T cells and transferred into Rag1^−/− mice before treatment with CD20 or control mAbs and challenge with NP-CGG. NP-specific IgG1 Ab titers and avidities were determined by ELISA using low-valency NP₅-BSA or high-valency NP₂₅-BSA. NP-specific IgG1 Abs (400 ± 50 μg/ml) were easily observed in control mAb-treated Rag1^−/− mice receiving NP-primed B cells, while Ab levels were not measurable in CD20 mAb-treated mice (Fig. 4,C). Control mAb-treated Rag1^−/− mice given naive B cells had low NP-specific IgG1 Ab levels. The NP-specific Abs produced in control mAb-treated Rag1^−/− mice receiving primed B cells were high avidity (NP₅:NP₂₅ = 0.91), while NP₅:NP₂₅ ratios were lower (NP₅:NP₂₅ = 0.13) in Rag1^−/− mice given naive B cells (Fig. 4,D). The numbers of plasma cells and plasmablasts measured functionally as NP-specific IgG1 ASCs in the bone marrow and spleens of control mAb-treated Rag1^−/− mice that received primed B cells were 13 ± 4 and 643 ± 196 (per 10⁵ cells), respectively. However, NP-specific ASCs were undetectable in CD20 mAb-treated mice (Fig. 4 E). Rag1^−/− mice given naive B cells generated spleen ASCs at very low numbers, but none were detected in the bone marrow. Thus, CD20 mAb treatment effectively depleted memory B cells and thereby eliminated secondary Ab responses.

Plasmablasts and plasma cells express CD138, a commonly used marker for ASCs (41, 42). Plasmablasts may be induced in vitro by stimulating B cells with LPS to induce TLR-dependent plasma cell differentiation or with CD40 mAb in combination with IL-4 to mimic TD plasma cell differentiation (40). Therefore, B cells were labeled with CFSE, cultured with LPS or CD40 mAb plus IL-4, and divided into three subpopulations based on their relative CFSE intensities and CD138 expression (43). After LPS stimulation, proliferating plasmablasts (CFSE^lowCD138⁺) expressed CD20 at significant levels (Fig. 5 A). CFSE^lowCD138⁻ proliferating cells and CFSE^highCD138⁻ nonproliferating cells also expressed CD20, but at levels 60% lower than plasmablasts. Similar results were obtained following B cell activation with CD40 mAb plus IL-4 stimulation (data not shown). Thus, plasmablasts generated in vitro expressed cell surface CD20.

Although CD20 may be expressed on in vitro-derived ASCs, CD20 is down-regulated in human B cells differentiating to plasma cells (29). To determine whether CD20 is down-regulated in murine ASCs in vivo, CD20 expression was assessed by immunofluorescence staining of spleen and bone marrow from naive mice. B220^lowCD138^high cells in the bone marrow expressed CD20, but at densities 80 and 75% lower than splenic and bone marrow B220⁺CD138⁻ B cells, respectively (Fig. 5,B). In fact, when spleen and bone marrow single-cell suspensions from naive mice were sorted with magnetic beads into CD20⁺ and CD20⁻ populations, the CD20⁺ populations of cells contained the majority of ASCs (Fig. 5 C). Thus, ASCs expressed CD20 at low, but detectable levels in vivo.

Because CD20 expression is drastically reduced on ASCs in vivo, the effect of CD20 mAb on ASC numbers in naive mice was determined. Using CD138 staining with in vivo BrdU-labeling techniques, in vivo-generated short-lived plasmablasts (CD138^highBrdU⁺) and long-lived plasma cells (CD138^highBrdU⁻) have been phenotypically recognized (44). Therefore, mice were fed BrdU for 2 wk before treatment with CD20 or control mAb. Seven days later, spleen and bone marrow cells were harvested and assessed for CD138 and BrdU staining. Despite effective B cell depletion after CD20 mAb treatment, short-lived and long-lived plasma cell numbers were unchanged (Fig. 5 D).

To directly assess the effect of CD20 mAb treatment on ASC numbers in vivo, naive mice were given CD20 or control mAb 7 days before spleen and bone marrow cells were harvested, with ASC numbers determined by ELISPOT. IgM and IgG ASC numbers in the bone marrow of mice receiving CD20 and control mAb were comparable (Fig. 5,E). However, CD20 mAb treatment significantly decreased the number of splenic IgG and IgM ASCs. Thus, bone marrow ASCs expressed insufficient CD20 levels for their depletion in vivo, while some splenic ASCs were depleted. Administration of thioglycolate induces systemic inflammation that enhances B cell depletion by CD20 mAb in vivo (32). However, cotreatment of mice with thioglycolate and CD20 mAb did not enhance the degree of ASC depletion (Fig. 5 E).

Since CD20 mAb did not reduce the bone marrow ASC compartment, an alternative approach was tested to determine whether long-lived plasma cells could be depleted. Because bone marrow plasma cells express the LFA-1 and VLA-4 adhesion molecules (26, 45), whether function-blocking LFA-1 and VLA-4 mAbs could purge bone marrow ASCs from their survival niche (22) was assessed. Coinjection of LFA-1 and VLA-4 mAbs reduced NP-specific IgG ASC numbers by ∼75% in the bone marrow of mice immunized with NP-CGG 6 wk earlier (Fig. 6 A). NP-specific ASC numbers in the bone marrow were reduced for up to 28 days, after which NP-specific ASCs returned. Bone marrow ASCs were only depleted when LFA-1 and VLA-4 mAbs were used in combination and similar results were obtained for NP-specific spleen ASCs (data not shown). Thereby, coblockade of LFA-1 and VLA-4 caused temporary bone marrow ASC depletion.

Repopulation of the bone marrow with NP-specific ASCs 4 wk after LFA-1/VLA-4 mAb injection may have been due to mobilized bone marrow ASCs surviving by circulating within other mouse tissues, followed by ASC relocalization to the bone marrow once the mAbs were cleared; alternatively, mobilization of ASCs from the bone marrow survival niche may have led to ASC death due to loss of survival signals (22), with repopulation of the bone marrow due to new ASC generation from memory B cells. To determine how the bone marrow came to be repopulated with Ag-specific ASCs, immunized mice were treated with either CD20 or control mAbs in combination with LFA-1/VLA-4-blocking mAbs. Bone marrow cells were harvested 1, 4, and 10 wk later. After 1 wk of mAb treatment, NP-specific IgG ASC numbers were reduced by ∼70% (p < 0.02) in mice receiving LFA-1/VLA-4 mAbs or both LFA-1/VLA-4 and CD20 mAbs (Fig. 6, B and C). Both total (NP₃₃-specific) and high-affinity (NP₆-specific) IgG ASCs were reduced similarly by LFA-1/VLA-4 mAb treatment. By weeks 4 and 10, CD20⁺ B cell depletion prevented the recovery of NP-specific ASCs in the bone marrow of mice receiving LFA-1/VLA-4 mAbs, while ASCs repopulated the bone marrow of mice receiving only LFA-1/VLA-4 mAbs. Thus, repopulation of the bone marrow after ASC purging with LFA-1/VLA-4 mAbs was due to new ASC production because CD20⁺ B cell depletion prevented bone marrow ASC repopulation when given in combination with LFA-1/VLA-4 mAbs.

To verify that repopulation of the bone marrow ASC pool in LFA-1/VLA-4 mAb-treated mice was not due to failed depletion of the CD20⁺ mature/memory B cell pool, the effects of LFA-1/VLA-4 plus CD20 mAb treatments on B cell depletion in tissues was assessed. Mature bone marrow B cells, and splenic marginal zone and T2 B cells were cleared 1 and 4 wk after treating mice with CD20 mAb (Fig. 6, D–G), but returned by week 10. By contrast, LFA-1/VLA-4 mAb treatment only depleted mature bone marrow B cells, splenic marginal zone B cells, and ∼50% of spleen T2 B cells after 1 wk, with full repopulation of these subsets by 4 and 10 wk. One week after LFA-1/VLA-4 mAb treatment, the number of mature spleen B cells increased, but these numbers normalized by 4 wk after mAb administration. Serum from mice receiving LFA-1/VLA-4 mAbs did not contain detectable levels of rat IgG beyond 1 wk after mAb injection, suggesting that the effect of LFA-1/VLA-4 mAb treatment does not persist after mAb clearance (data not shown). Finally, consistent with reduced bone marrow plasma cell numbers after combined LFA-1/VLA-4 and CD20 mAb treatment, serum NP-specific IgG levels were also significantly reduced (Fig. 6 H). Thus, B cells had the ability to repopulate the bone marrow with ASCs when the plasma cell niche was depleted. However, bone marrow repopulation was abrogated by the concurrent depletion of mature and memory B cells with CD20 mAb.

These studies are the first to demonstrate in mice that most nascent humoral immune responses were blocked by CD20⁺ B cell depletion before immunization, while predeveloped serum Ab levels were not affected by either short- or long-term CD20⁺ B cell depletion (Figs. 1, 2, and 4). Abrogation of primary and secondary Ab responses is explained by the effective depletion of follicular, marginal zone, B1, germinal center, and memory B cell subsets (Figs. 3 and 4 and Table I). Remarkably, long-lived Ab titers persisted in the absence of memory B cell function and recall responses to Ag challenge (Figs. 1, 2, and 4). Thus, long-lived Ab production does not require persisting Ag to drive the memory B cell pool to chronically generate plasmablasts and short-lived plasma cells as previously suggested (23). Likewise, maintenance of serum Ab levels does not require Ag-independent memory B cell differentiation into plasma cells, as once suggested (24). In fact, inflammation-induced increases in serum Ig levels and hypergammaglobulinemia were inhibited by CD20⁺ B cell depletion without an effect on basal serum Ig levels (Fig. 1, B and C). Therefore, Ag-independent cytokine or TLR signaling may induce memory or other B cell subsets to differentiate into plasma cells (24, 25), but this is not necessary for the maintenance of long-lived serum Ab levels. Thus, long-lived plasma cells or their immediate precursors are indeed intrinsically long-lived and are distinct from the memory B cell pool.

Germinal center B cells, in vitro-generated plasmablasts, and some ASCs expressed cell surface CD20 at significant levels (Figs. 3,A and 5, A and B), which allowed their removal from B cell preparations (Fig. 5,C). By contrast, in vivo CD20 mAb treatment depleted only ∼75% of splenic ASCs and had no effect on bone marrow ASC numbers (Fig. 5,E) or established Ag-specific Ab responses (Figs. 4 and 6). Likewise, CD20 mAb treatment did not significantly deplete either short-lived CD138^highBrdU⁺ plasmablasts or long-lived CD138^highBrdU⁻ plasma cells (44) in vivo (Fig. 5 D). Spleen ASC depletion despite normal CD138^high B cell numbers suggests that the use of CD138 as a marker for plasma cells may only be appropriate for a mature CD20⁻ subset of ASCs (46). Nonetheless, the in vivo decrease in splenic ASC numbers after CD20 mAb treatment is likely due to residual CD20 expression. Alternatively, the depletion of marginal zone, germinal center, and memory B cells may prevent the formation of new plasmablasts, while short-lived plasma cells died off naturally during the 7 days of CD20 mAb depletion. Even though some spleen ASCs were depleted in vivo, they did not appear to contribute significantly to serum Ig levels in comparison to fully mature bone marrow plasma cells. The lack of long-lived bone marrow ASC depletion was most likely attributable to either their low-density CD20 expression or the absence of effector monocytes within their microenvironmental niches. Thus, although CD20⁺ B cell depletion was effective at inhibiting primary and secondary humoral immune responses, serologic “memory” resulting from long-lived plasma cells was unaffected.

Function-blocking mAbs against LFA-1 and VLA-4 purged Ag-specific ASCs from the bone marrow (Fig. 6,A), potentially inducing their death due to loss of survival niche signals (22). The bone marrow was repopulated with plasma cells as soon as the LFA-1 and VLA-4 mAbs were cleared from the circulation. However, coinjection of mice with LFA-1/VLA-4 and CD20 mAbs led to prolonged ASC depletion from the bone marrow (Fig. 6, B and C). Therefore, CD20⁺ B cells, and most likely memory B cells, were required for repopulating empty plasma cell niches in the bone marrow. The prolonged ASC depletion in LFA-1/VLA-4- and CD20 mAb- cotreated mice was also reflected in serum Ig levels; Ag-specific serum IgG was significantly reduced in mice receiving both CD20 and LFA-1/VLA-4 mAbs (weeks 7–12) compared with mice receiving either mAb alone. Others have shown that the marginal zone B cell subset can be specifically mobilized and depleted by LFA-1/VLA-4-blocking mAbs (47); however, the current studies now demonstrate that LFA-1 and VLA-4 adhesion molecules also regulate mature B cell and plasma cell localization within the bone marrow. Moreover, this study is the first to show that Ag-specific long-lived plasma cells can be purged from the bone marrow by blocking adhesion receptor function, with memory B cells required for repopulation of a plasma cell-deficient bone marrow. Therefore, long-lived plasma cells and memory B cells contribute in interdependent ways to long-lived serological memory; mature and memory B cells are not required for maintaining bone marrow plasma cell numbers, but are required for repopulation of plasma cell-deficient bone marrow through Ag-dependent (10) or -independent (24) mechanisms. Thus, CD20 mAb treatment in combination with LFA-1/VLA-4-blocking mAbs could provide a new therapeutic avenue for bone marrow plasma cell depletion that may also lead to the identification of additional similar strategies for depleting plasma cells from other tissues.

CD20⁺ B cell depletion did not inhibit IgM responses to TNP-LPS or DNP-KLH immunizations, while IgM responses to DNP-Ficoll were significantly inhibited by CD20 mAb treatment (Fig. 2). That CD20 mAb treatment rapidly and effectively depletes marginal zone B cells but not peritoneal B cells (32) indicates that peritoneal B cells may preferentially generate IgM responses to TI-1 and TD Ags. IgM Ab generation by peritoneal B cells is further supported by the observation that primary IgM Ab responses were not generated in CD20 mAb-treated mice after rechallenge with DNP-KLH (Fig. 2) since CD20 mAb eventually reduces peritoneal B cell numbers by the time that these mice were boosted with Ag (32). CD20 mAb plus thioglycolate treatments also reduced basal serum IgM Ab levels by up to half (Fig. 1,B), most likely due to the depletion of peritoneal B cells and their natural Ab products (48). Otherwise, CD20 mAb plus thioglycolate treatments did not significantly alter bone marrow and spleen ASC numbers or basal serum Ig levels (Fig. 5), arguing that peritoneal B cells only produce a portion of serum IgM and relatively little IgG. Thereby augmenting the innate immune response using agents like thioglycolate to mediate more effective B cell depletion may not augment the reduction of serum Ig levels. Thus, the majority of serum Ab appears to be continually produced by long-lived plasma cells that do not require replenishment once generated.

The absence of Ag-specific Ab isotype switching and IgG-subclass Ab responses to TI-1, TI-2, and TD Ags in CD20 mAb-treated mice is likely due to the depletion of germinal center B cells and their precursors (Figs. 2 and 3). Most B cells with a germinal center phenotype and histologically recognizable germinal centers were depleted following 7 days of CD20 mAb treatment, with at least 70% of germinal center B cells depleted in immunized mice (Fig. 3, B–D and Ref. 32). Thereby, germinal center-dependent clonal expansion, somatic hypermutation, and isotype switching are unlikely to occur to a significant extent following CD20 mAb treatment. CD20 mAb treatment also prevented high-affinity Ag-specific Ab formation and the expansion of ASC in bone marrow and spleens of recipient mice during adoptive transfer experiments. Thus, memory B cells expressed sufficient levels of CD20 to be effectively targeted in vivo, with memory Ab responses virtually eliminated by CD20 mAb treatments in vivo (Fig. 4). Thereby, the elimination of germinal center and memory B cells explains the absence of isotype-switched Ab responses in CD20 mAb-treated mice.

That CD20 mAb treatment impairs primary and recall Ab responses while maintaining serum Ig levels in mice is consistent with studies in patients receiving rituximab, a chimeric anti-human CD20 mAb. This mAb can cause decreased Ab responses to some recall Ags (49), while total Ig levels in lymphoma and rheumatoid arthritis patients generally remain within the normal range (49, 50, 51, 52). Rituximab treatment also decreases primary Ab responses and class switching in baboons (53). Although antibacterial Ab levels are preserved in autoimmune disease patients following rituximab-mediated B cell depletion (52), a positive clinical response is associated with a significant fall in autoantibody levels in some patients (54, 55, 56, 57, 58), but not all (59). However, these conclusions are complicated by the fact that most patients also receive supplementary immunosuppressive therapies and the full extent of B cell clearance in humans given rituximab is unknown. In autoimmune mice, continuous CD20 mAb treatment does not affect circulating Ig or autoantibody levels (34, 36). Thus, CD20 mAb therapy may not be beneficial if pathogenic autoantibody-secreting cells are long-lived or do not express CD20. However, CD20 mAb treatment may deplete memory B cells or the precursors of pathogenic autoantibody-secreting plasma cells, thereby reducing the sources of new Ab-secreting plasma cells in autoimmune disease where autoreactive memory B cells may be continually “tickled” with Ag, leading to perpetual plasma cell differentiation. It is important to note that the effects of rituximab in human patients may not completely recapitulate the results detailed in this study. Rituximab and the anti-mouse CD20 mAb used in this study are vastly distinct mAbs, with potentially different binding capabilities and effects in vivo (31, 32, 33).

The current studies demonstrate that depleting the majority of peripheral B cells by CD20 mAb treatment does not have a dramatic negative effect on preexisting Ab levels because plasma cells are long-lived and survive independent of repopulation by the B cell compartment. These results contrast with significant reductions in serum Ab levels within 2 wk of CD19⁺ B cell depletion in mouse models (60), consistent with the measured half-lives of serum Abs (61). Thus, serum Ab deficiency may not be a significant consequence of long-term B cell depletion using CD20 mAbs, except when new Ags and pathogens are encountered. However, alternative therapeutic strategies for depleting long-lived plasma cells will have to be devised to combat the negative clinical implications of humoral memory once initiated. Thus, a balance between effective B cell depletion, therapeutic benefit, and immunodeficiency will also need to be established as more potent B cell depletion strategies move into clinical practice.

We thank Drs. Jonathan Poe and Jean-David Bouaziz for assistance with these studies.

Thomas F. Tedder is a paid consultant for MedImmune and a consultant and a shareholder for Angelica Therapeutics.