Phosphorothioate-modified CpG oligodeoxynucleotides mimic autoantigens and reveal a potential role for Toll-like receptor 9 in receptor revision


  • Anne Doster,

    1. Department of Infectious Diseases, Medical Microbiology and Hygiene, University Hospital Heidelberg, Heidelberg, Germany
    Search for more papers by this author
    • A.D. and S.Z. have contributed equally to this work. This study includes experiments from their doctoral theses.

  • Saskia Ziegler,

    1. Department of Infectious Diseases, Medical Microbiology and Hygiene, University Hospital Heidelberg, Heidelberg, Germany
    Search for more papers by this author
    • A.D. and S.Z. have contributed equally to this work. This study includes experiments from their doctoral theses.

  • Sandra Foermer,

    1. Department of Infectious Diseases, Medical Microbiology and Hygiene, University Hospital Heidelberg, Heidelberg, Germany
    Search for more papers by this author
  • Ralf Joachim Rieker,

    1. Department of Pathology, University Hospital Heidelberg, Heidelberg, Germany
    2. Institute of Pathology, University Hospital Erlangen, Erlangen, Germany
    Search for more papers by this author
  • Klaus Heeg,

    1. Department of Infectious Diseases, Medical Microbiology and Hygiene, University Hospital Heidelberg, Heidelberg, Germany
    Search for more papers by this author
  • Isabelle Bekeredjian-Ding

    Corresponding author
    1. Institute for Medical Microbiology, Immunology and Parasitology (IMMIP), University Hospital Bonn, Bonn, Germany
    • Department of Infectious Diseases, Medical Microbiology and Hygiene, University Hospital Heidelberg, Heidelberg, Germany
    Search for more papers by this author

Correspondence: Isabelle Bekeredjian-Ding, Institute for Medical Microbiology, Immunology and Parasitology (IMMIP), University Hospital Bonn, Sigmund-Freud-Str. 25, D-53105 Bonn, Germany. E-mail:

Senior author: Isabelle Bekeredjian-Ding


Re-expression of recombinase activating genes (RAG) in mature B cells may support autoreactivity by enabling revision of the B-cell receptor (BCR). Recent reports suggest that administration of Toll-like receptor 9 (TLR9) -stimulating CpG oligodeoxynucleotides (ODN) could trigger the manifestation of autoimmune disease and that TLR are involved in the selection processes eliminating autoreactive BCR. The mechanisms involved remain to be elucidated. This prompted us to ask, whether TLR9 could be involved in receptor revision. We found that phosphorothioate-modified CpG ODN (CpGPTO) induced expression of Ku70 and re-expression of RAG-1 in human peripheral blood B lymphocytes and Igλ expression in sorted Igκ+ B cells. Further results revealed unselective binding specificity of CpGPTO-induced immunoglobulin and suggested that CpGPTO engage and/or mimic IgM receptor signalling, an important prerequisite for the initialization of receptor editing or revision. Altogether, our data describe a potential role for TLR9 in receptor revision and suggest that CpGPTO could mimic chromatin-bearing autoantigens by simultaneously engaging the BCR and TLR9 on IgM+ B cells.


The recombinase activating genes (RAG) are essential for editing and revision of the antigen receptors. The overall purpose of these processes lies in diversifying the antigen receptor repertoire and in revising autoreactive receptors to prevent autoimmunity. Consequently, these enzymes become promoters of self-tolerance during lymphocyte differentiation. Once T and B cells mature, RAG expression is turned off and the cells are released to the periphery. However, re-expression of RAG proteins and receptor revision have been reported in mature peripheral blood B cells from patients with autoimmune diseases such as systemic lupus erythematosus (SLE) and juvenile idiopathic arthritis.[1-4] In these studies re-expression of RAG correlated with CD5 expression and was found to be dependent on interleukin-6 (IL-6).[5-7] Albeit RAG re-expression in the autoimmune context may result from abnormal B-cell activation, the molecular mechanisms enabling re-expression and consecutive rearrangement processes remain to be investigated.

Important evidence for a role of Toll-like receptors (TLR) in B-cell-mediated disease comes from studies dealing with autoimmune disorders. In this context, a central role of TLR was demonstrated in promoting the expansion of autoreactive B cells and autoantibody production.[8-10] Moreover, patients with SLE display an elevated frequency of TLR9-expressing B cells[11, 12] and TLR9-reactive CD27 effector memory B cells.[13] Nonetheless, it was also reported that TLR9 exerts tolerogenic effects in murine SLE[14] and that patients with defective TLR signalling display elevated frequencies of autoreactive B-cell receptors (BCR),[15] indicating that TLR might influence receptor editing. However, only recently a clinical trial using TLR9 agonists, e.g. phosphorothioate-modified CpG oligodeoxynucleotides (CpGPTO) as adjuvant was halted because severe autoimmune disease developed in one subject.[16] This unexplained incident encouraged us to investigate the role of TLR9 in B-cell tolerance, i.e. its role in receptor revision.


Cells and cell isolation

The use of human materials was approved by the local ethics committee and written consent was obtained from donors. Total B cells were isolated from peripheral blood mononuclear cells with anti-CD19 microbeads (Miltenyi Biotec, Bergisch-Gladbach, Germany) (purity 98·5 ± 1%). For Igκ+ B-cell purification Igλ+ B cells were depleted with anti-Igλ-phycoerythrin (PE) and anti-PE microbeads before selection of Igκ+ B cells with anti-CD19 microbeads (purity 99 ± 0·5%). IgM+, IgM, CD27+ and CD27 B-cell fractions were isolated as described previously.[17] Plasmacytoid dendritic cells were isolated with anti-BDCA4 beads (Miltenyi Biotec). Culture medium contained 5–10% heat-inactivated autologous serum [or 10% fetal calf serum (FCS; Biochrom, Cambridge, UK) for immunoglobulin assays]. Thymus was homogenized in Trizol (Invitrogen, Karlsuhe, Germany) or RIPA buffer.

B-cell stimulation

Media and reagents were tested for endotoxin as described in ref. [18]; stimuli were used at the following concentrations: CpG ODN 2006 PTO/PO (5′-tcgtcgttttgtcgttttgtcgtt-3′) 1 μm (MWG Biotech, Ebersberg, Germany); UV-irradiated BHK-CD40L and BHK-pTCF (1 : 10); recombinant human (rh) IL-4 (Miltenyi Biotec) 100 U/ml; goat anti-human IgM + IgG + IgA F(ab′)2 fragments (Jackson Immunoresearch, Westgrove, PA) 5 μg/ml; SU6656 (Merck, Darmstadt, Germany) and R406[19] (Rigel Pharmaceuticals, San Franscisco, CA) (in DMSO). One hundred micrograms streptavidin-coated polystyrene beads (Bangs Laboratories, Fishers, IN; 0·13 μm or dragon-green 0·39 μm) were coupled with biotinylated anti-human IgM + IgA + IgG F(ab′)2 or 5' biotinylated, non-PTO ODN (MWG Biotech), i.e. CpG 2006, GpC 2006 and poly-(T)20 (30 min), washed, resuspended in PBS and diluted 1 : 20 for stimulation.

Proliferation assays

B-cell proliferation was assessed after 72 hr with an 8-hr [3H]thymidine pulse (1 μCi/well; Perkin Elmer, Hamburg, Germany). For bromodeoxyuridine (BrdU) assays B cells were stimulated in the presence of 0·5 μm BrdU (Roche, Mannheim, Germany) (4 days) and stained according to the protocol from BD Biosciences.

Antibodies and flow cytometry

Cells were stained following standard procedures. For intracellular staining, cells were fixed with PBS/4% paraformaldehyde and stained in Fix & Perm Medium B (Invitrogen). Measurements were performed on a FACSCanto (BD Biosciences, Heidelberg, Germany). Antibodies were purchased from BD Biosciences: anti-human Igλ-PE (murine IgG1), Igκ-FITC (murine IgG1), IgD-FITC, IgM-PE, CD5-allophycocyanin, CD5-FITC, CD20-Peridinin chlorophyll protein, CD19-PE, CD27-PE, murine IgG1-PE; Santa Cruz: rabbit anti-human RAG-1 [sc-363 (K-20)], goat anti-human RAG-2 [sc-7623 (C-19)], goat anti-rabbit IgG-FITC, donkey anti-goat IgG-FITC; Novus Biologicals, Littleton, CO: mouse anti-human Ku70 mAb; DakoCytomation, Glostrup, Denmark: mouse IgG1; Sigma, Munich, Germany: rabbit anti-mouse IgG-FITC. The mean fluorescence intensity is given as ΔMFI = MFI(primary antibody) − MFI(secondary antibody or isotype control) to account for the differences in antibody binding due to the activation state of the cell.

Immunofluorescence microscopy

Cells were fixed with PBS/4% paraformaldehyde, blocked in PBS/0·1% saponin/5% FCS/2% non-fat dry milk and stained with anti-RAG-1 1 : 50, anti-RAG-2 1 : 50, anti-Ku70 1 : 50, mouse IgG1 1 : 50; goat anti-rabbit IgG-TexasRed 1 : 1000, donkey anti-goat IgG-TexasRed 1 : 1000 (Jackson Immunoresearch), anti-mouse IgG-FITC 1 : 400 and 0·1 μm DAPI (Invitrogen). Specificity of anti-RAG-1 was controlled using the immunization peptide (see Supplementary material, Fig. S1A). B cells incubated with dragon-green microsphere conjugates (3 hr) were stained with Hoechst dye. HEp2G cells were fixed, permeabilized, incubated with B-cell supernatants or intravenous immunoglobulin G (5 μg/ml, Octapharma, Langenfeld, Germany), washed, stained with biotinylated anti-human immunoglobulin, streptavidin-Dy647 (ImmunoTools, Friesoythe, Germany) and Hoechst dye. Images were acquired on a Leica DMI 6000B or a TCS SP5 (Leica Microsystems, Mannheim, Germany).

Western blot analysis

For Western blots 3 × 106 B cells were lysed in RIPA buffer. Nitrocellulose membranes were blocked in Tris-buffered saline/5% dry milk, and incubated with anti-RAG-1 1 : 200, anti-Ku70 1 : 1000, anti-RAG-2 1 : 200, anti-GAPDH (Millipore, Schwalbach, Germany) or anti-β-actin (Cell Signaling Technologies, Danvers, MA).

Real-time RT-PCR

Real-time RT-PCR was performed using a High Pure RNA Isolation Kit (Roche), First Strand cDNA Kit with oligo(dT) primers (Fermentas, St Leon-Rot, Germany), Absolute QPCR SYBR GREEN Low ROX Mix (ABgene House, Epsom, UK), primers (Table 1, MWG Biotech) and a 7900 HT Fast Real Time PCR System (Applied Biosystems, Darmstadt, Germany). Relative expression to β-actin was calculated as rE = 1/(2Ct(target) − Ct(β-actin)).

Table 1. Primer sequences and calculated PCR product size
GeneForward primer [5′–3′]Reverse primer [5′–3′]Product size (bp)
ß-actin agagctacgagctgcctgacagcactgtgttggcgtacag184
rag-1 gagcaaggtacctcagccagagagagggtttcccctcatcaaa197
rag-2 cctctggccttcagacaaaatccagatggaaaactccagtg187
ku70 tgcagcaagagatgatgagggaaaggcagctgcacataca184
tdt aaatgtgctcagtgattctgtcacctgtcatttccaccggtttc170
artemis accgagactcagaggaggaagtcctggggtttgtctcagtt158
polλ gcaggaagggttcctcacaggaagtagagcagggcacagg224
dna ligase-4 ccctggacgacctagaacaatttggctggctatctgttcc167
xrcc4 agaatggactgggacagttttgacttcagctgggttttca245
xlf tccctggtctcccaacattgacaaagggctttccatcac249
polμ accagggcctcatcctgtacgggtgcaactaccaagtc205

Cytokine and immunoglobulin quantification

Interleukin-6 (72 hr) was measured using the OptEIA ELISA kit (BD Biosciences); IgM (13 days) was quantified using the IgM ELISA (Bethyl Laboratories, Montgomery, AL). For polyreactivity ELISA, plates were coated with 10 μg/ml lipopolysaccharide (Sigma), pneumovax (Aventis Pasteur, Lyon, France), tetanus toxoid (Statens Serum Institute, Copenhagen, Denmark) or 100 μg/ml salmon sperm DNA [Sigma; double-stranded DNA (untreated), single-stranded DNA (boiled)], rehydrated, blocked with PBS/3% FCS and incubated with B-cell supernatant and anti-human immunoglobulin-horseradish peroxidase (1 : 5000).


Statistical significance was determined using the paired two-tailed Student's t-test; significant differences are indicated with *P ≤ 0·05 and **P ≤ 0·005.


TLR9 stimulation induces autocrine IL-6 as a prerequisite for RAG re-expression

In the present study we asked whether TLR9 could participate in receptor revision. As IL-6 was previously found to be essential for the expression of RAG proteins in B-cell progenitors[20] and in mature B cells,[5, 6] we first determined the preconditions for induction of B-cell-derived IL-6: CpGPTO represented potent inducers of IL-6 (Fig. 1a), but IL-6 was also stimulated by combination of CD40L and rhIL-4, used as a surrogate for T-cell help (Fig. 1a), and combination of CpGPTO with CD40L synergistically enhanced IL-6 production (Fig. 1a). By comparison, CpGPTO triggered proliferation in all conditions but the combination of CD40L and rhIL-4 (Fig. 1b).

Figure 1.

Comparative analysis of interleukin-6 (IL-6) and proliferation in response to phosphorothioate-modified CpG ODN (CpGPTO). B cells were stimulated with CpGPTO (CpG), BHK-CD40L (40L), BHK-pTCF (pT), the control cell line, recombinant human (rh) IL-4, anti-immunoglobulin (aIg) and combinations of these stimuli. (a) IL-6 secretion. IL-6 levels from n = 3 experiments were normalized to CpG = 100% (149 ± 44 pg/ml). Mean values ± SEM are provided. **P = 0·0008; *P = 0·009. (b) Proliferation was determined by [3H]thymidine incorporation (counts per minute). Data from n = 3 experiments were normalized to CpG = 100% (7376 ± 3236 cpm). An arrow marks the lack of proliferation with CD40L/rhIL4. Mean values ± SEM are given. **P = 0·0002; *P = 0·016.

TLR9 activation triggers RAG-1 re-expression in peripheral blood B cells

Having confirmed this prerequisite for re-expression of RAG, we approached the analysis of RAG expression. RNA and protein lysates from freshly isolated peripheral blood B cells were compared with those from B cells stimulated with CpGPTO, CD40L ± rhIL-4 or a combination of these stimuli. As expected, RAG-1 mRNA was not found in freshly isolated B cells but – paralleling IL-6 induction – became detectable in B cells stimulated for 24 hr or longer with either CD40L/rhIL-4 or CpGPTO, or combinations of CpGPTO with CD40L ± rhIL-4 ± BCR stimulation with anti-human immunoglobulin F(ab′)2 (Fig. 2a). However, RAG-1 mRNA expression levels remained low, and RAG-2 mRNA expression was not detectable, suggesting that RAG expression may be restricted to a B-cell subfraction.

Figure 2.

RAG-1 expression in response to stimulation of CD19+ peripheral blood B cells. B cells were stimulated with phosphorothioate-modified CpG ODN (CpG), BHK-CD40L (40L), BHK-pTCF (pT), the control cell line, recombinant human interleukin-4 (rhIL-4), anti-immunoglobulin (aIg) and combinations of these stimuli. (a) RT-PCR for RAG-1 was carried out in RNA lysates from thymus (upper panel) and from B cells harvested at day 0 or day 2 (lower panel). One representative experiment of n ≥ 3 is shown. (b) Western blot analysis for RAG-1 and GAPDH with protein lysates from B cells harvested on day 0 (d0), 24 hr (d1), 48 hr (d2) and 72 hr (d3) and thymus lysate as positive control. Representative experiments of n ≥ 3 are shown. (c) Flow cytometric analysis of RAG-1 expression on day 4 after stimulation. Mean fluorescence intensities (MFI) are given as ΔMFI = MFI (anti-RAG-1) − MFI (anti-rabbit IgG). The results are given as mean values ± SEM of n = 4 independent experiments. *P (unstimulated : CpG) = 0·047; *P (unstim : CpG + CD40L) = 0·02, *P (CD40L : CpG + CD40L) = 0·033. (d) Western blot analysis. Comparison of RAG1 and β-actin expression on day 2 after stimulation with CpG + CD40L, anti-Ig (aIg), CD40L + IL-4 and CpG + CD40L + IL-4. The graph shows one representative experiment of n = 3 experiments. (e) Immunofluorescence. B cells were harvested after 96 hr and stained for RAG-1 and with DAPI (blue) as nuclear counterstain. Graphs depict images representative for n ≥ 3 experiments. B cells were stimulated with CD40L/rhIL-4 (top), CpGPTO (CpG) (middle) or CpGPTO (CpG) + CD40L/rhIL-4 + anti-Ig (bottom) and stained with rabbit anti-human RAG-1 + anti-rabbit IgG (TexasRed). White arrows: enlarged B cells with pronounced nuclear staining (greater magnification (zoom)).

Western blot analysis of whole cell lysates demonstrated absence of RAG-1 protein in freshly isolated B cells and presence of a 119 000 molecular weight protein band corresponding to RAG-1 in protein lysates from thymus and B cells stimulated with CpGPTO for 24 or 48 hr (Fig. 2b). Paralleling IL-6 production simultaneous engagement of TLR9 and CD40 enhanced RAG-1 protein expression (Fig. 2b), which was corroborated by flow cytometric analysis (Fig. 2c).

Well in line with the results obtained by RT-PCR the flow cytometric analysis further revealed that stimulation with CD40L (Fig. 2c), IL-4 or combined CD40L/IL-4 (data not shown) also induced slight increases in the mean fluorescence intensity corresponding to RAG-1. However, these increases never reached statistical significance when compared with background levels in unstimulated B cells. Notably, RAG-1 protein expression was not detected after BCR stimulation with anti-immunoglobulin, but was observed under combined stimulation with CD40L/IL-4 (Fig. 2d), a stimulatory condition leading to IL-6 induction.

Subcellular localization of TLR9-induced RAG-1

Activity of RAG is bound to its localization within the nucleus so we analysed the subcellular distribution of TLR9-induced RAG-1 in peripheral blood B cells. Immunofluorescence microscopy revealed that RAG-1 expression was nearly absent in CD40L/rhIL-4-stimulated conditions (Fig. 2e, upper panel), but detectable in CpGPTO-stimulated B cells (Fig. 2e, middle panel) and most pronounced in CpGPTO+CD40L (±anti-immunoglobulin) -stimulated B cells (Fig. 2e, lower panel). Remarkably, prominent nuclear staining for RAG-1 was found in B-cell blasts (Fig. 2e, white arrows).

mRNA expression of enzymes downstream of RAG

The RAG heterodimer initiates genomic rearrangement, but a multitude of enzymes are subsequently required to accomplish this process. These executing enzymes were detectable on mRNA level in both unstimulated and stimulated human peripheral blood B cells, indicating their possible involvement in RAG-dependent rearrangement processes (Fig. 3). However, despite the intriguing implications of differential regulation with regard to receptor revision, the changes in mRNA expression levels upon stimulation were not significant. Notably, the overall highest basal mRNA expression levels (≥ 10−2) were measured for Ku70, artemis and polμ, a polymerase recently suggested to selectively catalyse rearrangement processes at the LC (light chain) junction.[21]

Figure 3.

Messenger RNA expression of RAG-associated enzymes was analysed by real-time RT-PCR. B cells were stimulated with BHK-CD40L + recombinant human interleukin-4 (rhIL-4) (40L/IL4), phosphorothioate-modified CpG ODN (CpGPTO) (CpG) or CpGPTO (CpG) + 40L/IL-4 + anti-immunoglobulin (aIg) and harvested on day 0, 2, 3 and 4; mRNA expression was normalized to β-actin. Each graph summarizes the mean values ± SEM from n = 3 experiments.

Proliferation and accumulation of Ku70/80

As these enzymes belong to the non-homologous end joining repair complex (NHEJ) that mediates post-replicative DNA repair, we reasoned that their expression could be stabilized by the proliferative response elicited by CpGPTO and proliferation may, in turn, represent a facilitating factor for receptor revision. Western blot analysis revealed the presence of Ku70/80 protein in B cells stimulated with CpG ODN ± CD40L (Fig. 4a). However, most intriguingly, and similarly to RAG-1, microscopy revealed that CpGPTO-induced B-cell blasts were characterized by strong cytosolic and, most importantly, nuclear Ku70 staining (Fig. 4b, upper panel).

Figure 4.

Ku70 expression. B cells were stimulated with phosphorothioate-modified CpG ODN (CpGPTO) (CpG), BHK-CD40L (40L), recombinant human interleukin-4 (rhIL-4) and combinations of these stimuli. (a) Western blot analysis for Ku70 and GAPDH with protein lysates from B cells harvested on day 0 (d0), 24 hr (d1), 48 hr (d2) and 72 hr (d3) and thymus lysate as positive control. Representative experiments of n ≥ 3 are shown. (b) Immunofluorescence. B cells were harvested after 72 hr and stained for Ku70 and with DAPI (blue) as nuclear counterstain. The graphs are representative for n ≥ 3 experiments. B cells were stimulated with CpGPTO (CpG) (upper and middle panels) or CD40L + rhIL-4 (40L/IL4) (lower panel), stained with anti-Ku70 (upper and lower panels) or murine IgG1 (isotype control; middle panel) and anti-mouse IgG-FITC (yellow; all panels).

By contrast, Ku70 staining was faint and nuclear staining was nearly undetectable in CD40L/IL-4-stimulated B cells (Fig. 4b, lower panel), a finding that coincided with the absence of proliferation (Fig. 1b) and B-cell blast formation under these stimulatory conditions.[17] Full-blown proliferative responses as observed with CpG ODN stimulation might, therefore favour nuclear translocation of Ku70/80, but do not seem to be a prerequisite for RAG re-expression, because RAG-1 was detectable in CD40L/IL-4-stimulated B cells, whereas BCR stimulation failed to trigger RAG-1 expression (Fig. 2d).

Functional evidence for RAG activity

Having confirmed these molecular prerequisites for receptor revision we sought functional evidence for RAG activity. We postulated that re-expression of RAG in peripheral B cells enables Igκ/Igλ rearrangement in response to TLR9 ligation. To prove this hypothesis we purified Igκ+ B cells, and compared Igκ/Igλ expression in B cells stimulated with CpGPTO or CD40L/rhIL-4, two stimuli that result in comparable cellular survival and autocrine IL-6 but that differ in the extent of proliferation. Despite the absence of Igλ+ cells in sorted Igκ+ B cells (Fig. 5a), unstimulated and CD40L/rhIL-4-stimulated B cells, a small population of Igκ-negative Igλ+ B cells became detectable after TLR9 stimulation for 4–6 days (Fig. 5b). Moreover, co-expression of Igκ and Igλ on a subset of B cells (Fig. 5b) was interpreted as indicative for ongoing Igκ/Igλ rearrangement. Staining with the isotype control proved the specificity of the anti-Igλ staining (Fig. 5c). Importantly, the low frequency of the evolving Igλ+ population (Fig. 5b), e.g. for CpGPTO: 0·4 ± 0·2% (n = 6) and for CD40L/IL4: 0·03 ± 0·04% (n = 4) makes Igκ/Igλ rearrangement a rare event, a finding that is compatible with the overall low expression of TLR9-induced RAG-1 and selective accumulation of RAG-1 and Ku70 in a small B-cell subfraction. Taken together, these results provided the notion that stimulation with TLR9-active ODN triggers RAG re-expression and consecutively catalyses LC rearrangements in a subfraction of B cells, so proving functional integrity of TLR9-induced RAG proteins in these cells.

Figure 5.

Flow cytometric analysis of Igκ/Igλ expression. Left: anti-Igκ-FITC/anti-Igλ-PE; right: forward scatter (FSC)/anti-Igλ-PE. (a) Igκ and Igλ expression in unsorted B cells. (b) Intracellular staining for Igκ and Igλ in unstimulated (b, upper panels), phosphorothioate-modified CpG ODN (CpGPTO)- (b, middle panels) and CD40L/interleukin-4 (IL-4) -stimulated (b, lower panels) Igκ+ B cells after 4 days. P2 : % Igλ+ (rearranged) B cells. One representative experiment of n ≥ 4 experiments is shown. (c) Specificity of anti-human Igλ antibody staining. Igκ+ B cells were stimulated with CpGPTO for 4 days. After harvest cells were stained with anti-human Igλ (FITC) + anti-human Igκ (PE) or with mIgG1 (FITC) + anti-human Igκ (PE) and analysed by flow cytometry. FITC-positive cells were only detectable with anti-human Igλ staining but not with mIgG1. The results from two representative donors of n = 4 are shown in the dot plots.

IgM derived from CpGPTO-activated B cells displays unselective binding specificity

The current understanding of receptor editing and revision implies that these processes must be initiated by binding of an autoantigen to the BCR. Of note, earlier reports described binding of CpGPTO to the BCR,[22] which raised the notion that CpGPTO could act as unselective BCR stimuli or might even mimic autoantigens.

In a previous report we further demonstrated that stimulation of TLR9 with PTO-modified ODN selects IgM+ B cells for proliferation and differentiation.[17] As depicted in Fig. 6(a), CpGPTO-induced B-cell blasts originate from IgM+ CD27+ B cells because blast formation in response to CpGPTO is restricted to CD27+ and IgM+ B-cell fractions and is absent in CD27 and IgM (class switched) B-cell fractions. In the present context, we reasoned that binding of PTO ODN to surface IgM could explain selectivity of CpGPTO for IgM+ B cells; ligation of surface IgM could, in turn, explain the initiation of receptor revision.

Figure 6.

Binding specificity of phosphorothioate-modified CpG ODN (CpGPTO) -triggered IgM. (a) CD27, CD27+, IgM+ and IgM (class switched) B cells were stimulated with CpGPTO for 4 days. Upper row: blast formation in forward/side scatter. Bottom row: CD20 and intracellular IgM expression. (b) Cellular supernatants from n = 8 experiments with IgM+ B cells (stimulated with CpGPTO (CpG) or CD40L/recombinant human interleukin-4 (rhIL-4) were analysed for immunoglobulin with binding affinity to dsDNA, ssDNA, lipopolysaccharide (LPS), pneumococcal polysaccharide (PPS) and tetanus toxoid (TT). (c) HEp2G cells were incubated with immunoglobulin-containing cellular supernatants or intravenous immunoglobulin G. Binding to HEp2G cells was visualized with biotinylated anti-human Ig and streptavidin-Dy647. The results shown are representative n = 3 experiments.

In line with this hypothesis, the IgM released from CpGPTO-stimulated B cells (14·6 ± 12 μg/ml) displayed unselective binding specificity, e.g. reactivity to lipopolysaccharide, pneumococcal polysaccharide, double-stranded DNA, single-stranded DNA or tetanus toxoid (Fig. 6b). To investigate whether CpGPTO binds to autoantigens, we incubated HEp2G cells with supernatants from CpGPTO- or CD40L/rhIL-4-treated B cells or intravenous immunoglobulin G. Immunofluorescence microscopy showed binding of CpGPTO-induced immunoglobulin with a faint, mainly cytoplasmic staining pattern suggestive of low-degree autoreactivity (Fig. 6c). Hence, CpGPTO might preferentially target B cells expressing potentially polyreactive IgM, which might belong to the IgM memory pool.[17]

BCR ligation: a prerequisite for TLR9 activity of CpGPTO

In B cells, internalization of antigen is mediated by the BCR. Recent studies suggested that physical linkage of a BCR antigen to a stimulatory nucleic acid represents the most efficient means to induce B-cell activation via TLR9.[9, 23, 24] This prompted us to ask whether CpGPTO trigger receptor revision by simultaneously engaging BCR and TLR9 signalling in a B-cell subfraction. Notably, unmodified (phosphodiester) CpG ODN (CpGPO) lack mitogenicity (Fig. 7a), but the stimulatory activity of CpGPO was coupled to microspheres additionally carrying a BCR stimulus [anti-human immunoglobulin F(ab′)2] (Fig. 7b). However, physical linkage of ODN did not waive the requirement for the TLR9-specific CpG-motif: F(ab′)2-coupled microspheres failed to induce proliferation in the absence of CpGPO or when CpGPO was substituted by a control GpCPO or a poly(T)2o-ODN (Fig. 7c).

Figure 7.

Assessment of B-cell receptor (BCR) involvement. (a,b) Proliferation was assessed by bromodeoxyuridine (BrdU) incorporation in B cells stimulated for 96 hr. Representative experiments of n = 4 are shown. (a) Unstimulated B cells compared with stimulation with phosphorothioate-modified CpG ODN (CpGPTO) or unmodified (CpGPO) CpG ODN. (b) Stimulation with microsphere-coupled CpGPO* and/or anti-human immunoglobulin F(ab′)2*. (c) CD19+ B cells were stimulated with microspheres coupled with F(ab′)2* with or without CpGPO*, GpCPO* or polyTPO. Proliferation was quantified via [3H]-thymidine incorporation (cpm). Due to the donor variability, results were normalized to F(ab′)2* + CpGPO* = 1 (2237 ± 1111 cpm). The diagram depicts the mean values obtained from n = 3 experiments ± SEM. **P [F(ab′)2*: F(ab′)2* + CpGPO*] = 0·00006; **P [F(ab′)2* + CpGPO*]: [F(ab′)2* + GpCPO*] = 0·00005; **P [F(ab′)2* + CpGPO*: F(ab′)2* + polyTPO*) = 0·00007. (d,e) Proliferation in response to CpGPTO was assessed in the presence or absence of syk inhibitor R406 (c), lyn inhibitor SU6656 (SU) (d) or DMSO (DM) solvent control by [3H]thymidine incorporation and normalized to CpGPTO stimulation (mean values ± SEM). (c,d) Syk inhibition. CpG = 10678 ± 1578 cpm. **P = 0·002; *P = 0·007. (e) Lyn inhibition. CpG = 3046 ± 1455 cpm. *P (CpG: CpG + SU 0·1 μm) = 0·048; *P (CpG + DMSO: CpG + SU 10 μm) = 0·025. (f) Human pDC were stimulated with CpGPTO in the presence or absence of syk (R406) or lyn (SU; SU665) inhibitors (1 μm) or the DMSO (DM) solvent control.

CpGPTO mimic BCR signalling by recruiting syk and lyn kinases

Next, we asked whether CpGPTO use BCR-dependent signalling. To answer this question, we stimulated B cells with CpGPTO in the presence or absence of inhibitors selectively targeting tyrosine kinases typically recruited upon BCR activation. In support of our hypothesis we found that CpGPTO-triggered B-cell proliferation was partially inhibited by the syk kinase inhibitor R406 in a concentration-dependent manner (Fig. 7d). By contrast, proliferation was enhanced by 20 ± 0·6% when B cells were pretreated with the lyn inhibitor SU6656 (Fig. 7e), a finding well compatible with hyper-responsiveness of lyn–/– B cells.[25, 26] We concluded that, first, syk and lyn kinases participate in CpGPTO-mediated B-cell activation, and, second, CpGPTO either directly stimulate the BCR or bypass BCR signalling by recruiting molecules associated with proximal BCR signalling.

To further investigate this question we sought to perform CpGPTO stimulation in the absence of the BCR. To this end we used plasmacytoid dendritic cells because they are characterized by TLR9 and a BCR-like signalosome.[27] Despite strong donor variability the trend was clear: similarly to the results obtained in B cells, interferon-α secretion levels were negatively affected by syk inhibition and increased by lyn antagonization (Fig. 7f). These findings were compatible with a role of syk and lyn kinases in TLR-dependent signalling, making discrimination of TLR-dependent and BCR-dependent signalling nearly impossible.


RAG re-expression in mature B cells has been described in a variety of studies.[7, 28-31] Importantly, and in marked contrast to the heavy chain locus, repeated rearrangements are possible at the LC loci. It is therefore not surprising that re-expression of RAG is associated with secondary LC rearrangements.[32, 33] In our study, high mRNA expression levels of polμ in human peripheral blood B cells (Fig. 3) and flow cytometric evidence for Igκ/Igλ rearrangement (Fig. 5) support this concept.

Earlier studies in patient cells correlated re-expression of RAG with CD5 expression and autocrine IL-6 levels.[3, 5, 6, 34, 35] In line with these observations, we previously showed that CpGPTO up-regulate CD5 expression,[17] but we could not confirm a direct association of CD5 and RAG expression (data not shown). Nevertheless, under in vivo circumstances CD5 expression probably reflects strong activation of RAG+ B cells as achieved by stimulation with CpGPTO in vitro.[17] This notion is supported by the finding that a stronger degree of B-cell activation – as it results from combined stimulation with CpGPTO + CD40L ± rhIL-4 – concomitantly increases IL-6 production (Fig. 1a), proliferation (Fig. 1b) and associated expression of RAG-1 (Fig. 2b) and nuclear translocation of Ku70 (Fig. 4a).

Nevertheless, re-induction of RAG expression in the periphery is a controversial issue.[36, 37] It should, however, be noted that Sandel and Monroe[36, 37] proposed that B-cell escape from deletion and induction of RAG expression rely on a pro-survival signal inherent to the bone marrow environment. They further demonstrated that prevention of apoptosis can restore expression of RAG in immature transitional B cells. It can, therefore, not be excluded that a strong survival signal as induced by CpGPTO could enable re-expression of RAG and consecutive receptor revision.

Since RAG-1 and RAG-2 are thought to act as a heterodimer,[38] our data indicate that RAG proteins and associated NHEJ enzymes display functional integrity in a small population of CpGPTO-treated B cells (Figs 2-5). However, despite flow cytometric evidence for Igκ/Igλ rearrangement (Fig. 5b) and detection of RAG-1 (Fig. 2), RAG-2 remained below the detection threshold. Differences in expression levels of RAG-1 and RAG-2 may be explained by a cluster of transcription initiation sites in the RAG-1 promoter that lowers the threshold for transcription.[39] Furthermore, RAG-1 serves as an E3 ubiquitin ligase that adversely regulates RAG-2 expression,[40] a property that may further accentuate differences in expression levels. Consequently, RAG-2 expression determines the limiting threshold for RAG function, which is compatible with the low frequency of Igκ/Igλ rearrangement in our assays (Fig. 5).

Regulatory elements that predispose for TLR-mediated RAG-1 promoter activation include binding sites for interferon regulatory factors, activating protein-1, signal transducer and activator of transcription and myc transcription factors.[39] Interestingly, in murine B cells myc is induced upon TLR9 stimulation via protein kinase B (PKB)/Akt.[41] Accordingly, PKB/Akt-mammalian target of rapamycin signalling is indispensible for CpGPTO-induced human B-cell blast formation, proliferation, IL-6 production and differentiation,[17] and may therefore directly or indirectly contribute to re-expression of RAG as proposed in ref. [42]. Moreover, nuclear factor-κB signalling, another important effector pathway of TLR9, is considered an important regulator of RAG expression[43] and a decisive promoter of secondary LC rearrangement.[44] TLR9-induced RAG re-expression and LC rearrangement may, therefore, result from coordinated PKB/Akt and NF-κB signalling.

In a previous study we demonstrated that stimulation with CpGPTO selectively drives IgM+ B cells into a prolonged proliferative state.[17] As shown in Fig. 7(c) the presence of the TLR9-specific CpG motif is critical for the induction of proliferation. This proliferative burst may, however, also counteract RAG-2 expression, because RAG-2 expression was lately shown to correlate with expression of p27kip, a cell cycle inhibitor.[6] However, ongoing DNA synthesis requires post-replicative DNA repair, and availability of Ku70/80 and other NHEJ enzymes could facilitate RAG-dependent receptor revision. Taking into consideration that PKB/Akt induces proliferation and simultaneously blocks receptor editing via inactivation of FOXO transcription factors in pre-B cells,[45] we suggest that initial PKB/Akt-dependent proliferation triggers RAG-1 expression while gradual ceasing of proliferation and onset of differentiation may evoke FOXO-mediated RAG-2 expression. Finally, receptor revision, i.e. secondary LC rearrangement, may be accomplished in a nuclear factor-κB-dependent manner.[44] Future investigations will have to prove this model correct.

In this study we reasoned that a BCR signal must precede receptor revision and therefore postulated that CpGPTO either activates or mimics BCR signalling. This hypothesis was supported by the finding that inhibition of syk and lyn kinases, molecules essential for proximal BCR signalling, affects the response to CpGPTO (Fig. 7). However, these assays cannot distinguish whether the kinases are recruited as a consequence of BCR engagement by CpGPTO, act downstream of TLR9 (thereby circumventing the requirement for BCR engagement) or synergistically interconnect both pathways.

Previous reports indicate that, indeed, both PTO-modified ODN and multivalent DNA aptamers engage surface IgM.[22, 46] In clear agreement with these findings, either PTO-modification or linkage to a BCR stimulus is required for TLR9 activity of CpG ODN (Fig. 5). In views of the unselective binding specificity of CpGPTO-induced immunoglobulin (Fig. 6b,c), we argued that binding of CpGPTO to the antigen receptor could drive a ‘PTO- or DNA-reactive’ B-cell subset into receptor revision as reported previously.[31] Intriguingly, high expression of RAG-1 and Ku70 marked a subpopulation of CpGPTO-induced B-cell blasts as cells prone for receptor revision that were shown to originate from IgM+ CD27+ B cells (Fig. 6a). Although the concept that IgM memory B cells undergo receptor revision is controversial, the physiological antigen promiscuity of the IgM receptor underscores that receptor revision in these cells could be beneficial. Moreover, it is well-acknowledged that marginal zone B cells (discussed as murine counterparts of human peripheral blood IgM+ CD27+ B cells) are strongly responsive to TLR stimulation.[47-50]

Nevertheless, it was recently suggested that CpGPTO induces proliferation of transitional B cells,[51] a B-cell subset expressing polyreactive IgM and sensitive to treatment with syk inhibitors.[52] Albeit the frequency of these cells in freshly isolated peripheral blood B cells from the donors used in this study was very low (0·1–1%), and blast formation was not observed in the CD27 fraction (Fig. 6a), we cannot exclude transitional B cells as the target subpopulation undergoing TLR9-induced receptor revision. Further studies will be needed to answer this question.

Taken together, our data provide evidence that TLR9 can participate in receptor revision. This was demonstrated for LC rearrangement (Fig. 5) but could also affect VH element replacement.[53, 54] Our study further suggests that CpGPTO can be used to study receptor revision triggered by chromatin-bearing autoantigens. It can, however, only be speculated how TLR9 affects receptor revision in vivo: TLR9 could contribute to exceeding a certain activation threshold necessary to tackle receptor revision or could act as a sensor for chromatin-bearing autoantigens, subsequently licensing receptor revision. Hence, a strong and long-lasting B-cell stimulus such as CpGPTO in vitro or that occurring in vivo, i.e. in autoimmune diseases (or possibly that upon CpGPTO administration) could trigger receptor revision in the periphery in the attempt to correct or eliminate autoreactivity as physiologically seen in the bone marrow. Nonetheless, in the periphery this process might result in increased autoreactivity of the immunoglobulin in predisposed individuals. In earlier studies receptor revision is, therefore, viewed as a pathological event. Our results, describe a mechanism possibly contributing to severe adverse events after CpGPTO treatment. Nevertheless, we can only speculate that the observations made in vitro could be associated with the manifestation of autoimmunity in vivo, e.g. the triggering of Wegener granulomatosis reported in the CpGPTO-adjuvanted hepatitis B vaccination trial.[16]


We thank Prof. Dr Hartmut Engelmann, Munich for provision of the BHK-CD40L cells and Dr Konrad Bode, Heidelberg, Germany for provision the Hep2G cells. The study was funded by the Olympia-Morata programme of the Medical faculty, University of Heidelberg, Germany to I.B.-D. and the DFG collaborative research centre SFB 938 TP C to I.B.-D. and K.H. S.Z. is supported by the LGFG postgraduate programme ‘Differential activation and integration of signaling modules within the immune system'.


The authors declare no financial interests.