Myelin protein zero is naturally processed in the B cells of monoclonal gammopathy of undetermined significance of immunoglobulin M isotype: aberrant triggering of a patient’s T cells

Abstract

Background Monoclonal gammopathy of undetermined significance of immunoglobulin M isotype is a condition with clonally expanded B cells, recently suggested to have an infectious origin. This monoclonal gammopathy is frequently associated with polyneuropathy and antibodies against myelin protein zero, whereas the role of the T cells remains largely unknown. We analyzed protein zero-specific B cells, as antigen-presenting cells, and their capacity to activate T helper cells.Design and Methods We used a well-characterized monoclonal gammopathy of undetermined significance-derived B-cell line, TJ2, expressing anti-protein zero immunoglobulin M. The ability of TJ2 cells to bind, endocytose, process, and present protein zero was investigated by receptor-clustering and immunofluorescence. The activation of protein zero-specific autologous T cells was studied by measuring interleukin-2 and interferon-γ with flow cytometry, immunobeads, and enzyme-linked immunospot assays.Results Surface-receptor clustering and endocytosis of receptor-ligand (immunoglobulin M/protein zero) complexes were pronounced after exposure to protein zero. Naturally processed or synthetic protein zero peptide (194–208)-pulsed TJ2 cells significantly induced interleukin-2 secretion from autologous T cells compared to control antigen-pulsed cells (P<0.001). The numbers of interferon-γ-producing T helper cells, including CD4⁺/CD8⁺ cells, were also significantly increased (P=0.0152). Affinity-isolated naturally processed myelin peptides were potent interferon-γ stimulators for autologous peripheral blood mononuclear cells, but not for control peripheral blood mononuclear cells.Conclusions We show for the first time that myelin protein zero is naturally processed in B cells from monoclonal gammopathy of undetermined significance of immunoglobulin M isotype, acting as aberrant antigen-presenting cells in activation of a patient’s T helper cells. Our findings cast new light on the important role of autoreactive protein zero-specific B cells in the induction of the pathogenic T-cell responses found in nerve lesions of patients with monoclonal gammopathy of undetermined significance with peripheral neuropathy.

Introduction

Monoclonal gammopathy of undetermined significance (MGUS) is a premalignant B-cell/plasma cell disorder found in 3.2% of people over 50 years of age, with the prevalence increasing to 5.3% among people over 70 years old.1 The disease course is different for immunoglobulin (Ig) G or IgA producing MGUS, as compared to MGUS of IgM class (IgM MGUS).2,3 IgG or IgA MGUS progresses to multiple myeloma at a rate of 1% per year,4 whereas IgM MGUS progresses, if it does, to Waldenström’s macroglobulinemia or chronic lymphocytic leukemia (CLL) and rarely to other neoplasms.4,5 Recent data reveal a significantly increased risk of MGUS after respiratory infections,6 and an association with certain bacterial infections, 7 which has raised the question of an initial microbial trigger followed by cross-reactivity to self-antigens. Peripheral neuropathy is found in 8% to 36% of MGUS patients8,9 and in 50% of patients with IgM-MGUS.10–12 These patients show a slowly progressive, sensory/sensory motor demyelinating neuropathy10 with antibodies13 and T-cell infiltrates in the nerve lesions.14 The etiology and detailed mechanisms of peripheral neuropathy in MGUS (PN-MGUS) are, however, still elusive.

Detailed structural analysis of IgM binding specificity would be valuable in understanding the pathogenesis of IgM MGUS. The antibodies previously described in PN-MGUS frequently target a sulfated trisaccharide epitope, termed HNK-1, present on surface membrane molecules of peripheral nerve Schwann cells, including myelin protein zero (P0), a 28 kDa glycoprotein and member of the Ig super gene family with adhesion molecule function mediating compaction of peripheral nerve myelin.15–17 The HNK-1 oligosaccharide epitope is also found on myelin associated glycoprotein (MAG),9,18–21 gangliosides22,23 and sulfate-3-glucoronyl paragloboside.24,25 Biochemical structural data have shown that mycobacterium bind to P0,26,27 which is of special interest in view of the recently described association between MGUS and mycobacterial infections.7

PN-MGUS nerve lesion biopsies show infiltrating T cells,14 besides the presence of IgM antibodies. Circulating CD4 and CD8 T cells in these patients have an activated phenotype,28 and increased systemic levels of soluble interleukin (IL)-2-receptors have been observed.29 There is also an association with HLA-DR haplotypes carrying a non-polar tryptophan residue at position 9 in the DRβ chain.30 B cells secreting anti-MAG antibody are subject to T-cell regulation in vitro31 and a T-helper 1 (Th1)-like response, with interferon (IFN)-γ secretion in response to peptides from myelin proteins, has been observed in PN-MGUS patients.16

In this study, we investigated whether IgM MGUS B cells are efficient antigen-presenting cells (APC) for activation of memory helper T cells. Normal B cells are well known for their APC function, but there has been some controversy regarding the ability of neoplastic B cells to stimulate T-cell responses. Recently, APC function was shown at least in some B-cell lymphomas i.e. CLL, representing a monoclonal CD5 B-cell expansion.32,33 The CLL cells, however, present antigens aberrantly to T helper (Th) cells, which could explain an autoimmune trigger.32 Although the expanded B-cell clone in MGUS, which it is worth noting is CD5 most of the time, could have a similar role in the development of autoimmune polyneuropathy, 15,16 its role in antigen presentation and T-cell activation has not previously been investigated. We followed the processing of biotin-labeled P0 or native myelin in an established anti-P0-specific IgM-MGUS B-cell line, TJ2, and found that naturally processed myelin P0 peptides aberrantly triggered autologous T cells to release IL2 and IFNγ.

Design and Methods

Antibody reagents and antigens

Antibodies and antibody-conjugates to HLA-DP, HLA-DQ, HLA-DR, IgM, LAMP-2, mouse IgG, IFNγ, CD3, CD4, CD8, CD19, CD23, CD46, CD56, and CD69 are detailed in the Online Supplementary Design and Methods.

Native myelin and P0 full length protein were purified from bovine peripheral nerves as described elsewhere34,35 with some modifications.36 Synthetic P0 peptides representing amino acids 194–208 (P0 194–208) were synthesized based on the human P0 sequence as described previously.16 This peptide was shown to be immunogenic in PN-MGUS patients including the patient in the current study.16 Keyhole limpet hemocyanin (KLH) was purchased from Calbiochem, (Lab Kemi, Stockholm, Sweden) and phytohemagglutinin (PHA) from Sigma-Aldrich, (Stockholm, Sweden).

Patient’s data and patient-derived cells

Blood samples were collected from a 70-year old female MGUS patient with chronic progressive sensory-motor polyneuropathy. At the time of blood sample collection, the patient was not receiving any therapy and her clinical status was stable. She gave informed consent, under the guidelines from the Linköping University Hospital ethics committee, in compliance with the Declaration of Helsinki. The patient’s peripheral blood mononuclear cells (PBMC) were isolated from whole blood using Lymphoprep (Nycomed Pharma A/S, Oslo, Norway). Freshly isolated PBMC were used in the IFNγ enzyme-linked immunosorbent spot (ELISPOT) assay and thawed PBMC were used in the intracellular IFNγ assay.

The Epstein-Barr virus (EBV)-transformed cell line, TJ2, was established previously in our laboratory.36 The patient’s cells expressed HLA-DR4, -DQ8, -DR7, -DQ2.30 Serum antibodies from this patient were of IgM isotype and showed reactivity against myelin P0, MAG, and the LK-1 gycolipid, all expressing the HNK-epitope. It was concluded that TJ2 IgM had HNK-1 oligosaccharide epitope specificity. The TJ2 cell line produces IgM, λ monoclonal antibodies of mutated Ig heavy chain variable 3–15*07 (IGHV3-15*07) genotype. Phenotypic characterization of the cell line revealed a high expression of B-cell markers CD19, CD20, CD22 and CD23. CD80 and HLA-DR, two receptors necessary for antigen-presentation, were also expressed at high levels. CD5 could be found on the TJ2 cells at the initial stage of culture, but expression decreased over time. Neither the T-cell marker CD3 nor the macrophage marker CD14 was expressed by the TJ2 cells. TJ2 cells were cultured in Iscove’s modified Dulbecco’s medium (IMDM) mixed with Opti-MEM medium (50:50), supplemented with 10% fetal calf serum (FCS), 1% glutamine and 0.5% penicillin-streptomycin (Invitrogen Ltd., Paisley, UK).

Major histocompatibility complex class I and class II typing and immunoglobulin heavy chain gene sequencing

MHC polymerase chain reaction (PCR)-based typing using commercial kits was performed according to the manufacturer’s instructions. IGHV gene sequencing was carried out as previously described36 and detailed in the Online Supplementary Design and Methods.

Receptor clustering and co-localization

Receptor clustering was used to investigate whether surface IgM or MHC class II (MHCII), on TJ2 cells, would bind myelin and P0. The clustering/co-capping method has been described previously37 and is detailed in the Online Supplementary Design and Methods.

B-cell receptor mediated endocytosis of protein zero

The B-cell receptor (BCR) mediated uptake of biotinylated P0 in the endosomal compartment of TJ2 cells was investigated. TJ2 cells, at a concentration of 1×10 cells/mL, were pulsed with 10 μg/mL biotinylated P0 for 30 min on ice, after which any unbound antigen was washed away and the cells incubated at 37°C for 0 min, 20 min, 40 min, 60 min, 4 h or 24 h at a cell concentration of 10×10 cells/mL to allow endocytosis of the antigen-receptor complex. The cells were fixed on microscope slides using 4% paraformaldehyde and permeabilized in a balanced salt solution, 1% HEPES, 0.1% saponin buffer and stained for intracellular antigen and LAMP-2, a late endosomal marker, as described previously.38 The primary antibodies used were Alexa488-conjugated streptavidin (2 μg/mL), mouse anti-human LAMP-2 (1:100 dilution) and Alexa594-conjugated goat anti-mouse IgG antibody (4 μg/mL). Three hundred cells were analyzed for co-localization of P0 and LAMP-2 on a Nikon-C1 laser scanning confocal unit attached to a Nikon Eclipse E600 fluorescence microscope (Nikon Instruments Inc., Melville, NY, USA).

Lysosomal staining

TJ2 cells were incubated at 37ºC for 24 h in the presence of 40 μg/mL biotinylated myelin and 40 μg/mL Texas Red-conjugated dextran (Molecular Probes). The cells were washed three times with cold PBS, fixed on microscope slides using 4% paraformaldehyde and permeabilized in a balanced salt solution, 1% HEPES, and 0.1% saponin buffer. Intracellular myelin was stained with Alexa488-conjugated streptavidin and co-localization of dextran and myelin was investigated on a Carl Zeiss Axiovert 200M fluorescence microscope (Carl Zeiss).

Purification of HLA-DR, HLA-DP, and HLA-DQ peptides

We isolated naturally processed myelin peptides presented in TJ2 cell MHCII receptors. TJ2 cells were grown and expanded to 1×10 cells and incubated with crude myelin (550 μg per 10 cells) for 24 h. The cells were then solubilized for 30 min on ice in RIPA-buffer containing protease inhibitors, and the lysate was affinity-purified on a mouse anti-human HLA-DP,-DQ,-DR protein G Sepharose column (GE Healthcare, Uppsala, Sweden) using a cross-linker, disuccinimidyl suberate (DSS), for covalent immobilization of the monoclonal antibody. MHCII-peptide complexes bound to the affinity column were eluted by 10 mM Tris-HCl buffer, pH 2.0. The peak OD fractions were desalted on a Sepharose G-25 column (GE Healthcare), followed by separation on a 3 mL reversed-phase chromatography resource column (GE Healthcare) by fast protein liquid chromatography (FPLC) using an acetonitril gradient (0–100%) in 0.1% triflouroacetic acid. Eluted peptide fractions were freeze-dried.

Interferon-γ enzyme-linked immunosorbent spot assay

In order to investigate whether the purified myelin peptide fractions could activate autologous T cells, an IFNγ ELISPOT assay was performed. Autologous PBMC from a PN-MGUS patient and control PBMC from a healthy donor were prepared and the IFNγ ELISPOT assay was performed as previously described in detail.39

T-cell activation assay

The capacity of the P0-specific PN-MGUS derived B-cell line TJ2 to act as APC and activate autologous T cells was investigated by measuring intracellular IFNγ in T cells after 6 h in culture with antigen-primed TJ2 cells. In addition, secreted IL2 was measured in culture supernatants after 7 days. We chose to analyze IL2 in the supernatants, since it was previously shown by us and others that EBV-transformed B-cell lines can produce IFNγ.40 In addition it was recently shown that potent T-cell and NK-cell reactivity with IFNγ production exist in EBV-seropositive patients against autologous EBV transformed lymphoblastoid cell lines.41 Considering that the TJ2 cell line was EBV-transformed, we first carefully eliminated the anti-EBV reactive T cells and NK cells by co-culturing autologous PBMC with TJ2 cells for 15 h. B cells (CD19), NK cells (CD56) and activated (presumably by EBV-related antigens), IFNγ-secreting cells were then depleted according to the manufacturer’s description using CD19/CD56 immunomagnetic MACS microbeads and an IFNγ secretion assay (Miletenyi Biotec). The purity of negatively selected PBMC was determined by flow cytometry on a FACSCanto flow cytometer (BD Biosciences). The depleted PBMC contained less than 4.8% NK cells, 9.9% B cells and 0.21% IFNγ-secreting T cells.

Freshly propagated TJ2 cell cultures were then antigen-primed by incubation with bovine P0, KLH (control antigen) or P0 peptide 194–208. Full-length protein antigens were used at a final concentration of 10 μg/mL for 16 h at 37°C. Sixteen hours was established as the optimal time for antigen presentation in the receptor co-capping studies. The P0 synthetic peptides were incubated with the TJ2 cells for 1 h at a final concentration of 1 μg/mL. The antigen-primed TJ2 cells were subsequently co-cultured in duplicate with autologous PBMC depleted of NK cells, B cells and IFNγ-secreting cells. The cells were either used for flow cytometric intracellular IFNγ staining, or measurement of secreted IL2. PHA stimulation (10 μg/mL) of PBMC was used as a positive control.

Intracellular interferon-γ staining

Antigen-primed TJ2 cells and autologous PBMC were co-cultured for 6 h at 37°C in the presence of 20 μg/mL brefeldin A. The cells were washed twice in cold PBS, 0.1% BSA and stained with anti-CD3 PerCP, anti-CD4-FITC, anti-CD8-APC Cy7, anti-CD19-PE-Cy7, and anti-CD56 APC monoclonal antibodies for 30 min on ice. The cells were washed twice, fixed in cold 4% paraformaldehyde, washed twice and permeabilized in PBS, 0.1% BSA, and 0.1% saponin. Intracellular IFNγ was stained by incubating the cells with anti-IFNγ PE monoclonal antibodies for 30 min on ice. The cells were then washed twice, resuspended at 1×10 cells/mL and analyzed on a FACSCanto flow cytometer (BD Biosciences). At least 30,000 cells were collected and IFNγ expression was analyzed both for CD3 T lymphocytes and CD3 T lymphoblasts. The percentages of IFNγ-positive cells after antigen stimulation (P0, KLH or P0 peptides 194–208) were calculated and the various T-cell populations were compared using Fisher’s exact test.

Interleukin-2 immunoassay

Supernatants from the 7 day co-cultures of antigen-incubated TJ2 cells (P0, KLH or P0 peptides 194–208) with autologous PBMC were collected and analyzed in six replicate experiments for the presence of the T-cell specific cytokine IL2 using a Milliplex human cytokine/chemokine immunoassay (Millipore, Billerica, MA, USA). The assay was performed according to the manufacturer’s description except that one additional standard reference point of 0.64 pg/mL was added to the standard curve. The samples were analyzed on a Luminex 100 instrument (Luminex Corporation, Austin, USA) with STarStation software (V1.1, Applied Cytometry Systems, Sheffield, UK). Mean values were compared using one way ANOVA and Tukey’s post hoc test.

Bioinformatic prediction of major histocompatibility complex-binding protein zero peptides

The prediction tools used were PROPRED,42 SVMHC43 IEDB-AR44– 47 and HLA-pred. Details of the procedures are given in the Online Supplementary Design and Methods.

Results

The aim of this study was to assess the potential role of the expanded CD5 B-cell clone in PN-MGUS as APC. For this purpose, we used the P0-specific B cell line, TJ2, previously established in our laboratory.36 We typed and sequenced the MHC class I (MHCI), MHCII and IGHV genes respectively. High resolution MHCI typing of HLA-A and HLA-B was performed for the TJ2 cell line and revealed genotype HLA-A*0201, *3101; HLA-B*1302, *4001. High resolution MHCII typing of HLA-DRB1, HLA-DQA1 and HLA-DQB1 for the TJ2 cell line showed genotype HLA-DRB1*0403, *0701; DQA1*0201, *0301; DQB1*0202, *0302. PCR amplification and sequencing of the IGHV gene from the TJ2 cell line was performed. TJ2 cells express IGHV3-15*07 gene with 90.39% homology to germline gene and a very short HCDR3 with only eight amino acids (ATGGLVGA). Codon 118 is mutated, resulting in the lack of a W-anchor (W is replaced by V), however, the W(F)-G-X-G motif is conserved and in frame. Figure 1 gives an overview of the experimental design and strategy of this study. In general, experiments involving nerve antigens were performed both with purified native myelin and pure P0.

Docking of protein zero ligand to the B-cell recpetor

First, we showed that myelin and P0 bound specifically to surface IgM of TJ2 cells, using two-color immunofluorescence and receptor clustering. The cells were pulsed with biotinylated myelin for 30 min followed by incubation with unbiotinylated antigen and microscopy reading at several time points. P0 binding to IgM was investigated at two time points. Receptor clustering of surface IgM was induced and the frequency of myelin to IgM co-localization on myelin-positive cells was analyzed. Surface IgM clustering (red) induced co-capping of myelin (green) in the majority of TJ2 cells at all time points investigated (Figure 2A–B). At 16 h, the ratio of cells displaying co-localization of surface IgM and myelin decreased from 89% to 50%, followed by an increase to 75% at 24 h. Parallel observations on P0 showed 98–100% co-localization with surface IgM (Figure 2C). We concluded that myelin and P0 specifically bound to the surface IgM BCR of TJ2 cells.

Figure 1.Proposed model and experimental design of the current study. Myelin protein zero (P0) is a glycoprotein present in the membrane of Schwann cells in the peripheral nervous system. The crystal structure of the extracellular domain shown is from rat61 and is reprinted with kind permission from Dr Petri Kursula (Oulu, Finland).48 We hypothesize that P0 is exposed to the CD5⁺ P0 specific MGUS B cells, perhaps due to a nerve injury. (1) P0 binds to and induces clustering of BCR on the myelin-reactive MGUS B cells. (2) The BCR-antigen complex is endocytosed, processed and P0 peptides are loaded onto MHCII molecules. (3) P0 peptides are presented to and activate T cells. (4) The activated T cell starts producing IFNγ and IL2. The numbers shown in the Figure also correspond to the different experiments presented in the Results section. (1) Receptor clustering of surface IgM and co-localization of P0. (2) Co-localization of P0 and late endosomal/lysosomal marker LAMP-2. (3) Receptor clustering of MHCII and co-localization of P0. (4) Activation of autologous T cells by P0 incubated MGUS B cells as measured by intracellular IFNγ and secreted IL2. The ability of naturally processed P0 peptides to activate autologous T cells with the secretion of IFNγ was also investigated.

Protein zero uptake and processing

BCR-mediated receptor internalization and intracellular trafficking of P0 was studied by investigating the co-localization of antigen with late endosomal/lysosomal marker LAMP-2 or lysosomal marker dextran conjugated with Texas Red. To follow the P0 trafficking, TJ2 cells were pulsed with biotinylated P0 for 30 min, at 0°C, washed to remove unbound antigen, and then analyzed at various time points (0 min, 20 min, 40 min, 60 min, 4 h, and 24 h at 37°C). At the onset (time 0) 87% of TJ2 cells showed surface-location of P0, not co-localized with LAMP-2, as expected (Figure 2D). Already at 20 min, P0 co-localized with the endosomal LAMP-2 marker in 75% of P0-positive cells. At 4 h and 24 h an endosomal accumulation of P0 was found in 32% and 45%, respectively, of P0-positive TJ2 cell (Figure 2E). Confocal images of P0 trafficking in TJ2 cells at the various time points investigated revealed an accumulation in late endosomes (Figure 2D). We confirmed the processing and trafficking of P0 by using native myelin and tracked its location to the late endosomal/lysosomal compartment with the Texas Reddextran conjugate. At 24 h, distinct 0.5 μm large endosomal vesicles containing both myelin and dextran were observed (Figure 2F).

Figure 2.Myelin/P0 binding to surface IgM, presence of P0 in the late endosome/lysosome compartment and presentation of myelin/P0 peptides in MHCII on the P0 specific MGUS B cell line TJ2. (A) Immunofluorescence images of myelin (green) co-clustered with surface IgM (red) on TJ2 cells after 5 h and 24 h. Yellow regions in the overlay images indicate co-localization. (B) Myelin binding to surface IgM was seen on a majority of TJ2 cells at all investigated time points (2.5 h, 5 h, 8 h, 16 h and 24 h). (C) Similar results were seen for P0 and surface IgM. (D) Images show uptake of P0 (green) in TJ2 cells after 0 h, 20 min and 24 h. Endosomal marker, LAMP-2, is shown in red. Yellow regions in the overlay images indicate co-localization, also shown by arrows. (E) Kinetic analysis of BCR-mediated P0 uptake in TJ2 cells. Percent P0-positive and LAMP-2- positive cells are shown after 0 h, 20 min, 40 min, 60 min, 4 h, and 24 h. Results from two independent experiments are shown (white and gray bars). For each time point, at least 300 LAMP-2 positive TJ2 cells were analyzed for the presence of P0. (F) The simultaneous uptake of P0 (green) and dextran (red) in the lysosomal compartment of TJ2 cells was investigated after 24 h. Yellow regions in the overlay images indicate co-localization. (G) Immunofluorescence images of myelin (green) co-clustered with HLAII (red) after 5 h and 24 h respectively. Yellow regions in the overlay images indicate co-localization. (H) Percentage of TJ2 cells exhibiting co-clustering of myelin with HLAII is shown for 2.5 h, 5 h, 8 h, 16 h, and 24 h. (I) Receptor clustering of MHCII and P0. (J) Negative control showing clustering of myelin, but no co-localization with control protein CD46.

Myelin peptide presentation in major histocompatibility complex class II

We then followed the trafficking of biotinylated myelin after endocytosis and entry into the endosomal/lysosomal (LAMP-2) compartment by analyzing its association with MHCII. Approximately 50% of cells showed co-localization of P0 and LAMP-2 to a large extent already at 1 h, as detailed above. Non-endocytosed myelin P0, however, remaining at the surface was always associated with IgM (Figure 2A–C). We, therefore, assumed that the biotinylated myelin that was bound to MHCII was present as processed peptides and not as native protein. The physical association of myelin peptides with MHCII and the kinetics of processed myelin peptides appearing at the surface membrane was investigated by tracking biotinylated myelin or biotinylated P0 followed by dual color staining after receptor-clustering. Biotinylated myelin was associated with MHCII in 13–16% of cells after 2.5 h to 5 h. At 16–24 h 52% of TJ2 cells displayed myelin peptide-MHCII complexes (Figure 2G–H). The kinetics of the biotinylated P0 peptide-MHCII physical association revealed a similar profile to that of native myelin processing and peptide presentation (Figure 2I). Figure 2J shows a negative control experiment, in which an irrelevant surface antigen, CD46, was shown not to co-cluster with myelin. The fluorescence patterns of the two proteins do not overlap at all.

T helper cell activation by naturally processed myelin peptides

The myelin peptides that were generated upon processing in TJ2 cells were isolated by affinity chromatography using an anti-HLA-DR, -DP, -DQ column, followed by purification by reverse phase chromatography on FPLC. All FPLC fractions were screened for biological activity by analyzing IFNγ release from T cells. Freshly isolated autologous and control PBMC were pulsed with FPLC peptide fractions and surveyed in an IFNγ ELISPOT assay. Triplicate samples were analyzed and the mean value was used for each test. Peptides were eluted in fraction 24–28, as shown by the OD₂₈₀ nm peak (Figure 3). These very fractions also induced IFNγ secretion in autologous PBMC (Figure 3A), but not in control cells (Figure 3B). Fraction 28 generated the maximal IFNγ response with a mean value of 72 spots.

Figure 3.Naturally processed myelin peptides induce IFNγ secretion from PN-MGUS patient’s PBMC, but not from control PBMC. Affinity purified and FPLC separated myelin peptide fractions # 1 to 50 were analyzed by ELISPOT for the ability to stimulate IFNγ secretion in (A) autologous PBMC from a PN-MGUS patient and (B) control PBMC from a healthy donor. The spontaneous background was subtracted and mean values from triplicate samples are shown (left axis). Peptide fraction # 24–28 induced an increased secretion of IFNγ in autologous PBMC, but not in control PBMC, and fraction # 28 generated the highest IFNγ response. Peptide fraction # 24–28 coincides with the OD₂₈₀ absorbance peak from the FPLC separation (line) shown on the right axis.

Figure 4.The PN-MGUS derived B-cell line TJ2 is capable of P0-specific T-cell activation. (A) Experimental design of the T-cell activation assay. CD19⁺ B cells, CD56⁺ NK cells and IFNγ-secreting, presumably EBV reactive cells, were eliminated from the PN-MGUS patient’s PBMC by magnetic separation. PBMC were subsequently co-cultured with antigen-pulsed TJ2 cells for 16 h. Intracellular IFNγ and secreted IL2 were measured after 6 h and 7 days, respectively. (B) Significantly increased levels of secreted IL2 in the 7-day cell cultures were seen after stimulation with P0 pulsed TJ2 cells (P<0.001), compared to control antigen, KLH-pulsed TJ2 cells. The results are mean values (±SD) of six separate experimental cultures. More than 30,000 cells from each culture were analyzed in FACS. (C, left panel) An increased level of intracellular IFNγ was seen in T lymphocytes after incubation with P0-primed TJ2 cells as compared to KLH (control antigen)-primed TJ2 cells. This increase was mainly seen in the CD4⁺ T lymphocyte population although a slight increase was also noted in the CD8⁺ T lymphocyte population. (C, right panel) T lymphoblasts also exhibited an increased accumulation of intracellular IFNγ after P0-primed TJ2 cell stimulation as compared to KLH (control antigen)-primed TJ2 cells. The IFNγ increase was seen in the CD4⁺ T blasts and in CD4⁺/CD8⁺ T blasts, but could not be detected in CD8⁺ T blasts. The P0 versus. KLH populations in C (both panels) were statistically analyzed using Fisher’s exact test and showed a significantly different response (P=0.0152).

We investigated whether the P0-specific B-cell clone pulsed with P0 or synthetic P0 peptides 194–208 could act as APC for the activation of autologous T cells. For this purpose TJ2 cells were pulsed with P0 or control antigen KLH for 16 h, or with P0 peptides 194–208 for 1 h. The antigen-pulsed TJ2 cells were then co-cultured for 6 h or 7 days with autologous PBMC from a PN-MGUS patient; the PBMC were depleted of CD56 NK cells, CD19 B cells and potential EBV-reactive IFNγ-secreting T cells (Figure 4A). T-cell-secreted IL2 was measured in 7 day co-cultures of the TJ2 B-cell clone plus autologous T cells. Significantly increased levels of IL2 were seen after stimulation with P0-pulsed TJ2 cells as compared to control KLH antigen-pulsed TJ2 cells (P<0.001) (Figure 4B). B cells pulsed with P0 peptides 194–208 were also potent inducers of T-cell IL2 release (P<0.001, results not shown).

The induction of IFNγ in a patient’s T cells after exposure to peptide-pulsed TJ2 cells was analyzed by flow cytometry after 6 h co-cultures. In the control T-cell population (without any prior antigen pulse), a distinct lymphoblast population was seen in the forward scatter/side scatter dot plot. IFNγ production was therefore analyzed in two separate FACS experiments both in the T lymphocyte population and in the T blast population. Taken together, P0-pulsed TJ2 cells induced a significantly increased IFNγ production above control values of non-stimulated PBMC in five of the six populations (CD4, CD8 and CD4/CD8 in the lymphocyte and blast populations). In comparison, KLH-pulsed TJ2 cells did not show an increase in any of six populations (P=0.0152, Fisher’s exact test). Increase above control was defined as more than 9% above control, which was the total coefficient of variation in this assay (Figure 4C). Interestingly, a CD4/CD8 T lymphoblast population was also present in the PN-MGUS patient. This population also exhibited increased IFNγ levels after stimulation with P0-pulsed TJ2 cells (Figure 4C). TJ2 cells pulsed with P0 peptides 194-208, however, did not stimulate the patient’s CD3 T lymphocytes or T lymphoblasts to increased IFNγ release (results not shown). Five percent of T lymphocytes and 9% of T lymphoblasts were IFNγ-positive after control antigen- pulsed TJ2 cell stimulation. This might reflect EBV-reactive T-cell populations not eliminated by the magnetic separation prior to the activation assay.

Bioinformatics: prediction of major histocompatibility complex-binding protein zero peptides

The location of the highest ranking MHCII-binding peptide, amino acids 107–115, in the extracellular domain of P0 can be seen marked in black in the protein crystal structure (Figure 5). Further MHCII and MHCI candidates are detailed in the Online Supplementary Results and Online Supplementary Figure S1.

Discussion

In this study, we explored the interaction between autoreactive, P0-specific monoclonal CD5 B cells and T cells in an IgM MGUS patient with chronic progressive sensory-motor polyneuropathy. The main finding is that the patient’s authentic MGUS B-cell line, TJ2, had a potent antigen-processing capacity, able to trigger IL2 and IFNγ cytokine production in myelin-specific autologous T cells. The TJ2 cell line expresses a surface IgM BCR with a mutated IGHV3-15*07 gene. The TJ2 IgM binds to an epitope that includes the HNK-1 oligosaccharide linked to Asn93 in the P0 protein sequence.48 The PN-MGUS patient’s B-cell clone expanded in vitro was authentic to the patient’s monoclonal anti-P0 autoreactive cells, and we found that it retained surface BCR with affinity for P0, efficiently mediated BCR-ligand uptake, processed the antigen and presented myelin peptides in MHCII molecules. We show here for the first time that these naturally processed P0 peptides presented by the B-cell clone induce production of the T-cell-specific cytokine IL2 from autologous T cells and production of the Th1 cytokine IFNγ from both CD4 and CD8 autologous T cells.

The involvement of T cells in the pathogenesis of PNMGUS has been previously suggested14,16,28–30 and our results presented in this study add further evidence on how these MGUS-derived T helper cells can be triggered by the patient’s monoclonal B-cell population that aberrantly presents myelin peptide antigens. The results presented in Figure 3 show that the FPLC peptide fractions induced IFNγ in autologous PBMC, probably due to the presence of anti-P0 antigen memory T cells, which may have arisen after broken tolerance to self-antigen after an infection, as we have interpreted the data. A transforming event in a self-reactive cell may then lead to expansion of an autonomous MGUS-clone. These memory T-cell clones that we observe might be remnants of a disease process that occurred some time previously.

Figure 5.Crystal structure of the P0 extracellular domain from rat according to Shapiro et al.61 and Kursula P.48 Glycosylation site, Asn93, is shown as spheres and the HLA-DRB1*0701 predicted binding peptide amino acids 107–115 in black.

A recent study identified P0 as the key CD4 T-cell antigen in the NOD-B7-2KO autoimmune peripheral neuropathy mouse model.49 NOD-B7-2KO mice deficient of IFNγ did not develop peripheral neuropathy, suggesting an inflammatory Th1 response to P0 in these mice. The authors also generated a P0-specific TCR transgenic mouse (NOD-POT) and CD4 T cells from this mouse proliferated in vitro when exposed to peripheral nerve lysate and P0. T cells from the NOD-POT mouse exposed to peripheral nerve lysate or P0 in vitro also produced IFNγ and IL17.

A contrasting hypothesis was suggested by Horna et al.,33 who showed that malignant B cells efficiently present tumor antigens to antigen-specific CD4 T cells, resulting in a strong antitumor effect. This intrinsic antigen-presenting ability of malignant B cells was, however, overridden by tolerogenic bone marrow-derived APC, leading instead to T-cell unresponsiveness and lack of antitumor effect. In the case of MGUS, however, it is difficult to overlook that in PN-MGUS nerve lesion biopsies there are infiltrating T cells, with both CD4 and CD8 cells.14,28 In addition, we previously compared IL4 and IFNγ responses in eight PN-MGUS patients with those in eight normal donors, and in four patients with polyneuropathy other etiology and found significantly more IFNγ than IL4 release in PN-MGUS patients.16

The malignant CLL B-cell clone can act as potent APC and aberrantly activate Rh autoreactive T helper cells, thus driving the autoimmune hemolytic anemia seen in many CLL patients.32 Furthermore, CLL cells have been found to be efficient APC, presenting, in CLL cells, the highly upregulated protein fibromodulin in HLA-A2.50,51 Pre-activation in vitro also enabled the expanded, fibromodulin-specific T cells to secrete IFNγ upon recognition of the antigen.51 This CD5 B-cell-specific antigen-presentation thus enabled the expansion of autologous tumor-specific T cells. Another example illustrating the important role of B cells is experimental autoimmune encephalitis, the experimental animal model for multiple sclerosis. Experimental autoimmune encephalitis can be induced by exposing wild-type mice to myelin oligodendrocyte protein, but in mice deficient of B cells, there is no induction, suggesting that B cells have a clear antigen-presenting role in this condition.52

In this study (Figure 2 D–F), we show that myelin peptides are processed in the endosome/lysosome compartment (LAMP-2) and physically associated with MHC class II (Figure 2 G–I). Non-endocytosed myelin could remain on the surface membrane, which would dim the interpretation. However, based on the data illustrated in Figure 2A, we found that all native surface-bound myelin was associated with surface IgM (none was found in other sites). Secondly, the FPLC fractions analyzed (Figure 3) contained peptides only, no native myelin. Specific anti-peptide reagents/monoclonal antibodies would, however, be helpful in distinguishing between processed and unprocessed myelin.

Previous studies have shown Th1 activation and IFNγ production by PBMC from a PN-MGUS patient after synthetic P0 peptide 194–208 stimulation.16 We confirmed in this study that P0 peptide (194–208)-pulsed TJ2 cells could stimulate a patient’s T cells to increase IL2 secretion significantly compared to control antigen. P0 peptide (194–208) did not appear as one of the candidates in the bioinformatic search for HLA-DRB1*0701, HLA-A or HLA-B binding P0 peptides. One possible explanation is that P0 peptide (194–208) is presented in a HLA-DR, DP, DQ receptor of a different genotype. The MHCII peptide binding prediction tools used did not allow prediction of peptides binding to the PN-MGUS patient’s HLA-DQ or HLA-DRB1* 0403 allele. Interestingly, the bioinformatics data did show that one of the top ranking HLA-DRB1*0701 binding peptides VGKTSQVTL (P0 amino acids 107–115) is localized in the extracellular domain of P0 (Figure 5), thus making it an appealing target for further T-cell activation studies.

During detailed analysis of the IFNγ-producing cell compartment, we noted the presence of an activated CD4/CD8 double-positive T blast population (6.2% of T cells) producing IFNγ in the investigated PN-MGUS patient after stimulation with P0. This double-positive population was also found after control KLH antigen stimulation, but at a lower percentage (4.2%). The appearance of double-positive CD4/CD8 T lymphocytes had previously been observed after in vitro stimulation of human PBMC with mitogenic PHA.53 It is noteworthy that human herpes viruses are able to induce CD4-expression on human CD8 T cells.54 The PBMC from the MGUS patient investigated in this study contained 3.5% CD4/CD8 lymphocytes prior to antigen exposure, which is a level similar to that found in normal individuals (2–3%).55,56 An elevated amount of CD4/CD8 lymphocytes had previously been found in MGUS patients,55 as well as in other autoimmune conditions, such as myasthenia gravis57 and multiple sclerosis.58 The appearance of double-positive CD4/CD8 T blasts after antigen stimulation reflects an immune abnormality in these autoimmune conditions, as suggested also in our study.

One potential limitation of the present study is that it is based primarily on detailed findings from one patient. However, several different experiments were carried out on this patient and consistently pointed in the same direction. We also expanded the P0 peptide analysis to three additional PN-MGUS patients, which all showed increased P0-peptide-induced IFNγ release as compared to control peptides. This release was abrogated by elimination of CD19 cells. Hence, there is no reason to believe that the findings presented in this study would not be of general relevance.

The induction of autoantibody production by bacterial molecular mimicry has previously been suggested in PNMGUS, based on cross-reactivity of the patients’ IgM anti-MAG antibodies with bacterial polypeptides.59 Novel findings also show that the CD5 surface receptor binds certain microbial glycan structures with high affinity.60 Furthermore, recent data show that the risk of developing multiple myeloma and MGUS is significantly increased among men who have had prior infectious disorders, in particular poliomyelitis for MM (RR = 3.69) and pneumonia for MGUS (RR =2.48).6 In addition, an association between MGUS and mycobacterial infections was recently revealed in a population of 17,398 patients.7 Based on these observations and our own results it is tempting to speculate about an infectious background to PN-MGUS. A microbial-triggered B-cell expansion may generate (by molecular mimicry) antigen spreading with B-cell clones reacting with neo-epitopes of myelin molecular motifs in P0 thus aberrantly presenting antigens that induce autoreactive T helper cells. P0 molecular structure studies showing homology to mycobacterial proteins provide support to this idea.27 Activation of autoreactive T helper cells and production of nerve-specific autoantibodies would, in the end, result in the neuropathy seen in PN-MGUS patients.

In conclusion, the data presented here strongly suggest that the expanded P0-specific IgM MGUS clone TJ2 is an efficient professional APC able to generate deleterious T-cell activation. These effects are paralleled in vivo where CD4 and CD8 cells, together with myelin-bound autoantibodies, have been observed in nerve lesions in situ, and are most likely involved in the pathological destruction of peripheral nerves seen in PN-MGUS patients.

Acknowledgments

the authors wish to express their sincere thanks to Dr Marie Larsson for valuable advice on T-cell activation assays. Many thanks also to Dr Petri Kursula for the crystal structure image of P0, and Karin Backteman, Mari-Anne Åkeson, Jeanette Svartz, Maria Petersson, and Petra Cassel for advice and laboratory assistance.

Footnotes

• Funding: this work was supported by funding provided by the Swedish Research Council, Swedish East Gothia Cancer Foundation, Linköping University Hospital Funds, and Swedish Cancer Association N. 3171- B04-16XBB. GSD.
• Authorship and Disclosures EH and MK planned and performed research, analyzed data, and wrote the paper; AS performed research, analyzed data, and wrote the paper; MV collected patients’ samples, planned research, analyzed data and wrote the paper; JE and AR designed and supervised the research, analyzed data and wrote the paper.
• The authors declare no conflicts of interest.
• The Online version of this paper has a Supplementary Appendix.

• Received August 1, 2009.
• Revision received September 15, 2009.
• Accepted October 5, 2009.