Several types of CpG-oligodeoxynucleotides (ODN) have been recently characterized. In mice, type A(D) CpG-ODNs primarily stimulate macrophages and dendritic cells, but fail to stimulate B cells. On the contrary, type B(K) CpG-ODNs are excellent B cell activators. Type C CpG-ODNs combine features of both types A(D) and B(K) CpG-ODNs. Despite cell type preferences, all CpG-ODNs require the presence of TLR9 for activation. In this study, we show that a subset of B cells from lupus mice responds to type A(D) CpG-ODN stimulation vigorously and directly with increased CD25 and CD86 expression and IL-10 secretion. Furthermore, these CpG-ODNs induce high surface IgM expression and promote 50- to 100-fold higher IgM and IgG3 secretion in lupus B cells than in controls. This response is similar to that seen with bacterial DNA stimulation of B cells. Type A(D)-responsive cells are enriched within lupus B cells with the marginal zone (MZ) phenotype. These cells are at least twice more numerous in lupus mice than in controls. The ability of lupus B cells to respond to type A(D) CpG-ODN stimulation is not due to differential TLR9 expression. Therefore, type A(D) CpG-ODNs may contribute to the lupus pathogenesis by inducing MZ-B cell activation, costimulatory molecule expression, and polyclonal Ig secretion. Through increased IL-10 secretion, MZ-B cells may also modify the activity of other cell types, particularly dendritic cells and macrophages.
The engagement of TLR receptors, particularly TLR9 on B cells and dendritic cells, seems to play an important role in the pathogenesis of systemic autoimmunity (1, 2, 3, 4). B cells respond to TLR9 activation by proliferating, ceasing a spontaneous cell death program, up-regulating MHC class II and B7 family members, and secreting IgM and cytokines (e.g., IL-6, IL-12p40, and IL-10) (5, 6, 7). Chromatin/IgG complexes trigger the proliferation of autoimmune B cell clones with rheumatoid factor specificity, and this response requires TLR9 (3). Moreover, when autoimmune B cells are activated through TLR9 by bacterial DNA or CpG-oligodeoxynucleotides (ODNs),3 but not through the BCR or CD40 receptor, they can potently down-regulate the activity of other immune cells apparently in an IL-10-dependent manner (8). Additionally, the treatment of prediseased lupus mice with bacterial DNA or CpG-ODNs results in increased production of anti-dsDNA Abs (9), and may cause progression of lupus nephritis (10, 11).
Three major types of TLR9 ligands have been identified (types A(D), B(K), and C (7, 12, 13, 14)). Type A(D) CpG-ODNs primarily stimulate dendritic cells (in both humans and mice) and macrophages (only in mice). These ODNs can also promote NK cell cytotoxic activity and IFN-γ secretion through high IFN-α secretion (15). Interestingly, optimal concentrations of type A(D) CpG-ODNs fail to stimulate B cells directly and may actually compete with type B(K) CpG-ODN-induced activation (16, 17). Although TLR9 is required for all types of CpG-ODNs (14, 18), HEK293 cells stably transfected with mouse TLR9 respond poorly to type A(D) CpG-ODNs (13, 14). This clearly suggests that additional coreceptors or adaptor molecules are needed to facilitate type A(D) CpG-ODN-mediated effects.
Due to the possible role of TLR9 in lupus pathogenesis, we studied the activation of lupus B cells with various TLR9 ligands. Surprisingly, lupus B cells, but not control B cells, responded vigorously to type A(D) CpG-ODN stimulation as well as to type B(K) CpG-ODN stimulation and bacterial DNA stimulation. Lupus B cells activated by type A(D) CpG-ODN up-regulated CD86 and CD25, and secreted IL-10 and Igs, particularly of IgM and IgG3 isotypes. We identified marginal zone (MZ) B cells as the primary B cell subset responsive to this type of CpG-ODN. These findings suggest that lupus B cells show a very low threshold for stimulation with different CpG-ODNs, and that the B cell response to TLR9 stimulation may have a role in lupus pathogenesis.
Materials and Methods
We used female prediseased lupus-prone Palmerston North (PN) mice (kindly provided by B. Handwerger, University of Maryland, Bethesda, MD), lupus-prone New Zealand Black/New Zealand White (NZB/NZW) (F1), and control mice DBA/1, BALB/c, C57BL/6, and 129/Sv (The Jackson Laboratory). Lupus-prone MRL-Faslpr/lpr mice were used in some experiments and showed results comparable to PN and NZB/NZW (F1) strains. All mice were between 5 and 8 wk of age when sacrificed for in vitro experiments. Animal protocols were approved by the University of Iowa Animal Care and Use Committee.
Purification of mouse B cell subsets
Mice were killed by CO2 asphyxiation. Spleens were removed aseptically, and RBCs were lysed with ammonium-chloride. All cultures were performed in 10% FCS-RPMI 1640 medium with supplements. Conventional B cells (follicular and MZ-B cells) were obtained by negative selection using anti-CD43-coated magnetic beads and passage over the MIDI-MACS magnet (Miltenyi Biotec). The pass-through fraction reproducibly contained between 97 and 99% of B220+ CD19+ cells. Such B cells were used for purification of either MZ-B cells or IgD+ B cell subsets by sorting on FACSDiva cell sorter. The CD21highCD23− subset (MZ-B cells) was obtained by staining purified total B cells from lupus mice with FITC anti-CD21 (clone 7E9) and PE anti-CD23 (clone B3B4), and by gating on CD23−CD21bright (MZ-B cells IgD−IgM+) and CD23+CD21int (follicular B cells IgD+IgM+) B cells. In some experiments, total B cells were labeled with PE anti-IgD Ab (clone 11.26) and FACS sorted into IgDbright (nonswitched B cells) and IgD− populations. Purified B cell subsets were cultured at 0.5–0.7 × 106/ml in 10% FCS-RPMI 1640 at 37°C, 5% C02 for 18 h or 6 days.
Type B(K) CpG-ODN, stimulatory ODN-1826, 5′-T*C*C*A*T*G*A*C*G*T*T*C*C*T*G*A*C*G*T*T-3′, and control ODN-1982, 5-T*C*C*A*G*G*A*C*T*T*C*T*C*T*C*A*G*G*T*T–3′ (*, denotes phosphorothioate linkages); type A(D) CpG-ODN, stimulatory ODN-1585, 5′-G*G*GGTCAACGTTGAG*G*G*G*G*G-3′; stimulatory ODN-2216, G*G*GGGACGATCGTCG*G*G*G*G*G-3′, and control ODN-2118, 5′-G*G*GGTCAAGCTTGAG*G*G*G*G*G-3′. All ODNs used in this study were endotoxin free and were provided by the Coley Pharmaceutical Group. Highly purified low-endotoxin bacterial DNA (from Escherichia coli, Ultrapure DNA) was from Sigma-Aldrich.
PE-, HRP-, or biotin-labeled polyclonal Abs against IgG2a, IgG2b, IgG3, and IgA were from Southern Biotechnology Associates. Purified and biotin-labeled mAbs against IgE (R35-72 and R35-118), IgG1 (A85-3 and A85-1), and IgG2a (R11-89 and R19-95) were from BD Pharmingen. Polyclonal Abs (purified and HRP labeled) against IgM were from Bethyl Laboratories; PE-labeled anti-IgD Ab (clone 11.26) was from Southern Biotechnology Associates. FITC-labeled Ab against CD21 (7E9) was a gift from L. Tygrett, University of Iowa. PE-labeled Abs against CD45R (B220; clone RA3-6B2), CD23 (clone B3B4), CD40 (clone 1C10), and FITC-labeled anti-CD86 (clone GL1) and anti-CD25 (clone PC61.5) were all from eBioscience. Isotype-matched controls were from different sources.
RT-PCR analysis of TLR9 expression
RNA was purified from freshly isolated total B cells or B cell subsets (2 × 106 cells each) obtained from DBA/1 and PN mice. Total RNA was isolated using RNeasy mini kit (Qiagen), following the manufacturer’s instructions. cDNA was synthesized from 0.4 μg of total RNA using Omniscript and Sensiscript reverse transcriptases for 30 min at 50°C (Qiagen One-Step RT-PCR kit). For PCR, the following TLR9-specific primers were used: upstream primer, GCA CAG GAG CGG TGA AGG T, and downstream primer, GCA GGG GTG CTC AGT GGA G (19). PCR conditions were the following: 94°C for 30 s, 56°C for 30 s, and 72°C for 45 s (PTC-200; MJ Research). TLR9 cDNA fragment was amplified for 30 cycles in a final volume of 50 μl containing 2.5 mM magnesium dichloride, dNTP mix (400 μM of each dNTP), HotStarTaq polymerase (2 μl per PCR condition, initially activated by incubation at 95°C for 15 min), and 0.6 μM each primer. PCR products were resolved by 1.2% agarose gel electrophoresis and visualized by ethidium bromide staining. A single band of 838 bp was identified. Identical results were obtained with mRNA diluted 5×, or 25×, and with PCR performed for 25 and 28 cycles (data not shown). Saturation of TLR9 occurred when PCR was performed for >35 cycles. As a control for equal RNA loading, RT-PCR was performed for 30 cycles with specific primers for GADPH (0.6 μM each).
All results are expressed as means ± SEM. Student’s t test (two tailed) was used to compare differences between the PN and age- and sex-matched DBA/1 mice. Values of p < 0.05 were considered significant.
Type A(D) CpG-ODNs up-regulate CD86 and CD25, but not CD40 in lupus B cells
It is well established that optimal concentrations of type A(D) CpG-ODNs (∼33–330 nM) fail to directly stimulate highly purified human or mouse B cells (12, 13, 18, 20). Higher concentrations may induce backbone-dependent CpG-independent stimulation of B cells.
As predicted, in contrast to the type B(K) CpG-ODN-1826 (Fig. 1⇓, left two columns), we found that type A(D) ODN-1585 was unable to induce CD86 (B7.2), CD25, or CD40 expression in control DBA/1 B cells. However, ODN-1585 induced significant CD86 and CD25, but not CD40 expression, in a subset of autoimmune B cells obtained from young prediseased PN lupus mice (Fig. 1⇓, PN, far right column; ODN-1585 treated, thick line; control ODN treated, thin line; medium treated, shaded area). Type A(D) CpG-ODNs failed to induce CD80 expression on B cells.
Type A(D) CpG-ODNs induce high IL-10 secretion in lupus B cells
High serum IL-10 levels are typically found in patients with systemic lupus erythematosus, as well as in several animal strains that spontaneously develop lupus-like disease (21). We recently showed that mixed spleen cells from lupus-prone PN mice produce high levels of IL-10 in response to both type A(D) and B(K) CpG-ODNs (8). In this study, we show that lupus B cells, but not control B cells, respond to type A(D) CpG-ODN-1585 stimulation with high IL-10 production (Fig. 2⇓).
Type A(D) CpG-ODNs induce polyclonal Ig secretion from lupus B cells
Polyclonal hypergammaglobulinemia and anti-nuclear Ab formation are immunologic hallmarks of systemic lupus. When control DBA/1 B cells were stimulated with optimal concentrations of type A(D) CpG-ODNs, only trace amounts of IgM and IgG3 could be detected even when B cells were cultured over extended periods of time (up to 10 days) (Fig. 3⇓, A and B). A similar lack of induction of IgM and IgG3 with type A(D) CpG-ODNs, but excellent response to type B(K) CpG-ODN stimulation and LPS stimulation, was observed in several other control strains, e.g., C57BL/6 and 129/Sv (Fig. 3⇓, C and D). Moreover, neither IgG1 nor IgE was detectable when type A(D) CpG-ODN stimulation was combined with IL-4 (data not shown). Similarly, no IgG2a secretion could be detected in response to type A(D) CpG-ODNs when IgG-depleted B cells were used (data not shown). In contrast, lupus B cells from PN mice responded to type A(D) stimulation with significantly higher IgM (Fig. 3⇓A, p < 0.001) and IgG3 secretion (B, p < 0.001) closely resembling their response to a natural TLR9 ligand, bacterial DNA (Fig. 4⇓). We also observed induction of a high IgM surface phenotype in lupus B cells, although not in control B cells, when stimulated with the ODN-1585 (Fig. 5⇓, right panels). In contrast, type B(K) CpG-ODNs enhanced surface IgM expression in both control and lupus B cells (Fig. 5⇓, left panels).
MZ-B cells primarily account for type A(D) CpG-ODN stimulatory response
We further tested whether differential responsiveness to type A(D) CpG-ODNs in lupus B cells could be attributed to a particular B cell subset by investigating the response of MZ-B cells vs follicular B cells. We first tested the response of highly purified nonswitched IgD+ B cells. As shown in Fig. 6⇓, IgD+ B cells from lupus mice failed to produce IgM (A) and IgG3 (C) when stimulated with ODN-1585. In contrast, IgD− B cells (Fig. 6⇓, A and C) as well as B cells with the MZ phenotype (CD23− CD21+) (B and D) primarily responded to type A(D) CpG-ODN stimulation. In addition, MZ-B cells were at least twice as numerous in lupus mice as in controls (Fig. 6⇓, E and F).
Figs. 7⇓ and 8⇓ demonstrate that the abnormal response of lupus B cells to type A(D) CpG-ODN stimulation is not limited to PN strain. Similarly to PN mice, NZB/NZW (F1) mice up-regulated CD86 expression in a subset of B cells, and secreted more IgM and IgG3. CD40 expression was minimally changed in both control and lupus B cells. The MZ-B cell compartment was 2- to 3-fold enlarged in NZB/NZW (F1) mice compared with control BALB/c mice, and MZ-B cells were the primary source of IgM following stimulation with bacterial DNA (Fig. 8⇓).
Finally, we compared the TLR9 expression at the mRNA level between total B cells obtained from control and lupus strains, as well as between the different B cell subsets from lupus mice (Fig. 9⇓). Similar levels of TLR9 expression were detected independent of the B cell type or strain of mice. Results were confirmed with real-time PCR analysis.
Therefore, lupus B cells may have a very low threshold of activation when stimulated with bacterial DNA or unusual CpG-ligands; this abnormality may be specific to MZ-B cells and does not depend on TLR9 expression.
Two different classes of CpG-ODN have been identified based on their distinct activities on plasmocytoid dendritic cells and B cells: type A(D) and type B(K). The prototype mouse type A(D) CpG-ODN sequence is exemplified by ODN-1585 (22). ODN-2216 is the best type A(D) CpG-ODN in the human system, but is also active in mice (12, 20). Poly(G) tails and central phosphodiester palindromic sequences with unmethylated CpG motifs structurally characterize this type of CpG-ODNs. Human plasmocytoid dendritic cells produce large amounts of IFN-α in response to ODN-2216, which in turn causes strong activation of γδ T cells and NK cells (12). In mice, type A(D) CpG-ODNs primarily activate macrophages and dendritic cells and induce potent NK cell cytotoxic activity and IFN-γ secretion through IL-12/IFN-α production (15, 22).
The major distinction between the two types of CpG-ODNs may lie in the inability of type A(D) CpG-ODNs to induce direct B cell activation. When poly(G) tails are removed, as in the case of recently discovered type C-CpG ODNs that have the palindromic sequence at the 3′ end, CpG-ODNs can regain the ability to stimulate both dendritic cells and B cells (14, 23). Therefore, poly(G) tails may actively suppress B cell activation in cis. Interestingly, ODNs with potent and specific inhibitory properties on TLR9 activation have recently been discovered (24, 25, 26). These inhibitory ODNs all contain a stretch of three to four guanosines at or near the 3′ end and a pyrimidine-rich (preferably CCT) sequence at the 5′ end. They potently inhibit the activation of various cell types induced with either type A(D) or type B(K) CpG-ODNs, acting proximal to the earliest events in the NF-κB (25) and AP-1 (27) activation, probably at the level of TLR9/MyD88 interaction. One can imagine a scenario in dendritic cells/macrophages in which poly(G) tails in type A(D) CpG-ODNs may fail to interact with the putative inhibitory (poly(G)) receptor, while in B cells this interaction results in complete inhibition of CpG-mediated downstream signaling events.
We now show that lupus B cells, in contrast to control B cells, respond to type A(D) CpG-ODN stimulation by up-regulating CD86 and CD25 and by secreting IL-10 and Igs. Apart from identifying type A(D) CpG-ODNs as a new group of lupus B cell stimulators, these results further stress the role of TLR9 activation in the pathogenesis of lupus (2, 3). As previously envisioned (1, 2), CpG stimulation of lupus B cells may be responsible for increased B cell survival, hypergammaglobulinemia, and IL-6 secretion. In addition, CpG stimulation of lupus B cells also induces high IL-10 secretion, which may negatively regulate certain macrophage and dendritic cell functions (8, 41). IL-10 may be particularly responsible for decreased IL-12 secretion, and therefore may cause skewing of the immune response from a Th1 pattern to probably more favorable Th2 pattern. However, in view of the detrimental role of IFN-γ in animal models of lupus (28), and the increased pathology of lupus mice with the deleted IL-10 gene, B cell-derived IL-10 may also be of benefit by preventing or down-regulating ongoing T cell-mediated tissue destruction (29).
We have identified MZ-B cells as major responders to bacterial DNA and type A(D) CpG-ODNs in lupus mice. These nonrecirculating long-lived B cells are highly enriched in the MZ of the spleen, where they perform important function in the initiation of the immune response in particular to T-independent Ags (30). These cells have a lower threshold for activation and differentiation into Ab-secreting cells, which allows them to rapidly respond to low-affinity Ags, and to limiting amounts of Ag (30). MZ-B cells have immediate access to blood-borne particulate Ags, and are responsible for an early IgM production (31). In newborns, the MZ is severely underdeveloped, which explains the inability of infants to mount a strong response to T-independent Ags, e.g., pneumococcal polysaccharide (32, 33).
Additionally, MZ-B cells may function as excellent APCs as they can process and present Ags, and deliver costimulatory signals to T cells much more efficiently than other B cells (30). Indeed, within a few hours of exposure to an Ag, MZ-B cells with anti-hen egg lysozyme specificity could induce costimulatory molecules and present Ags to T cells (30). This directly relates to a possible role of MZ-B cells in autoimmunity. Self-reactive clones are enriched within the MZ compartment (34, 35, 36). Estrogens may rescue autoreactive B cell precursors from deletion, and most of these cells will differentiate along the MZ pathway, e.g., like B cells from transgenic mice spontaneously secreting anti-DNA (37). Enlargement of MZ-B cell population in NZB/NZW (F1) mice (and in PN mice; this work) occurs before signs of disease (38) and pathogenic MZ-like B cell lines have been found in salivary glands in mice with Sjogren’s syndrome (39).
The primary defect responsible for the low threshold of MZ lupus B cells to CpG stimulation has yet to be discovered. This may either be an intrinsic B cell abnormality, or may be secondary to abnormal B cell activating factor belonging to the TNF family (BAFF)/BAFF-receptor stimulation, programmed death-1 dysfunction, CD40 engagement, IFN priming, or continuous TLR9 signaling in vivo. There is also the possibility of an imbalance between the TLR9 expression and low expression of the putative inhibitory DNA receptor (yet to be characterized) in MZ lupus B cells, preventing G-rich tails from exerting their negative influence, and allowing type A(D) CpG-ODNs to stimulate B cells. However, our finding that TLR9 expression in lupus B cells is not different from controls speaks against this possibility. One can envision that in mice or humans with adequate genetic background, strong hormonal (estrogenic) influence may initially lead to an expanded MZ-B cell pool. Such B cells may respond to very low concentrations of TLR9 ligands frequently present in the bloodstream, including less favorable CpG sequences, like those found in hypomethylated mammalian DNA (4, 40). The dual engagement of BCR and TLR9 by these TLR9 ligands may preferentially expand low-affinity B cell clones specific for DNA, or clones with the rheumatoid factor specificity that recognize chromatin/IgG immune complexes. Through T cell help and somatic hypermutation, these clones may give rise to higher avidity autoreactive B cells, resulting in autoantibody production, complement activation, and immune complex-mediated pathology. Therefore, attempts to pharmacologically block TLR9 signaling at an early step may still allow MZ-B cells to mount productive T-independent responses, but yet prevent unwanted activation and expansion of autoreactive B cells by unmethylated CpG-containing DNA.
In conclusion, lupus B cells, but not control B cells, respond to stimulation with type A(D) CpG-ODNs by up-regulating CD86 and CD25. In addition, lupus B cells produce 50- to 100-fold more IgM and IgG3 and high amounts of the regulatory cytokine IL-10. Interestingly, requirements for CD40 up-regulation may be more stringent than those for CD86 or CD25, as type A(D) CpG-ODN fail to up-regulate CD40 expression in B cells. Therefore, increased ligand/receptor interactions may be required for A(D) CpG-ODNs, as compared with type B(K) CpG-ODNs, to induce expression of some B cell surface molecules. Type A(D) CpG-ODNs may contribute to the lupus pathogenesis primarily through increased Ig secretion.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This study was supported by National Institutes of Health Grant AI047374-01A2.
↵2 Address correspondence and reprint requests to Dr. Petar Lenert, Department of Internal Medicine, C312 GH, University of Iowa, Iowa City, IA 52242. E-mail address:
↵3 Abbreviations used in this paper: ODN, oligodeoxynucleotide; MZ, marginal zone; NZB, New Zealand Black; NZW, New Zealand White; PN, Palmerston North.
- Received September 10, 2004.
- Accepted December 10, 2004.
- Copyright © 2005 by The American Association of Immunologists