View larger version (60K): [in this window] [in a new window] [Download PPT slide]   Figure 2. BND cells express autoreactive antibodies. Recombinant monoclonal antibodies were expressed by transfection of human 293 cells with the variable genes from single BND (totaling 95 clones; n = 17, 21, 30, and 27) or naive B cells (totaling 87 clones; n = 16, 37, 10, 10, and 14 by donor) and tested for binding to dsDNA and to antigens within the human HEp-2 cell line. (A) ELISA analysis to determine the frequency of BND and naive B cells that bind to dsDNA. The variable genes encoding classical anti-DNA antibodies were also added (red lines), including 3H9 (squares) (2), and HN241 (triangles) (29). Both control antibodies were expressed as chimeric mouse/human IgG- recombinant monoclonals identically to the naive and BND antibodies so that all capture and detection reagents used were identical. Absorbencies were measured at OD415 (y axis) and normalized to the 3H9-positive control antibody included on every plate. Binding to DNA is plotted for the various antibodies at 1 µg/ml (6.67 nM) and three additional fourfold serial dilutions (x axis indicates nanomoles). Saturation curves were used to calculate relative affinities to determine if antibodies were positively reactive. (B) ELISA screens were performed to determine the amount of ANA reactivity. The BND and naive monoclonal antibodies from A were loaded at a concentration of 25 µg/ml onto commercially available ANA antigen ELISA plates. Absorbencies were measured at OD415 and compared and normalized to an in house positive clone, C-CS clone B09 (25). The absorbencies of naive versus BND clones were significant by either Student's t test to compare means or Mann-Whitney to compare the medians (P < 0.0001 for either test; median absorbancies are indicated as red lines). (C) Antibody absorbencies from B averaged and compared by donor for naive population and BND population, including three matched comparisons of naive versus BND antibodies indicated with dotted lines. The BND cells were more often ANA reactive on average between donors (P < 0.02 by Student's t test with Welch's correction for unequal variance and P = 0.016 by Mann-Whitney test in comparing the medians). (D) HEp-2 reactivity was verified using immunofluorescence. Of the 95 antibodies derived from BND cells, 72 (75%) were clearly reactive with human HEp-2 cells. 20/87 (23%) of the antibodies from naive cells were HEp-2 reactive. Representative antigen binding patterns on HEp-2 slides for antibodies from BND cells are shown. Significance was established by 2 analysis; P < 0.0001. Commercial HEp-2 slides were stained with recombinant monoclonal antibodies and with antihuman IgG-FITC secondary antibody and visualized using fluorescence microscopy. Bar, 20 µm.

B_ND cells are mature naive-like B cells
As immature B cells before completing primary selection may be autoreactive, we hypothesize that B_ND cells might be immature. However, B_ND cells do not express markers found on immature B cells such as the B cell progenitor marker CD10 and, like mature cells, express only low levels of the development markers CD38 and CD24, which are commonly used to distinguish recent BM immigrant immature/transitional B cells that are CD38^hiCD24^hi (Fig. 3 A) (32). Furthermore, the B_ND and naive populations are both positive for the BCR-associated regulator CD22 that is expressed on mature B cells (33). In addition, recent reports have demonstrated that transitional B cells, as well as CD27⁺ IgG⁺ memory cells, in circulation can be distinguished because they do not express the ATP binding cassette ABCB1 transporter and, subsequently, they retain mitochondrial dyes such as Rhodamine 123 or MitoTracker Green (MTG) (34). Like naive cells, B_ND cells actively extrude MTG, further demonstrating that they are neither transitional B cells nor are they memory cells (Fig. 3 A, middle). Finally, B_ND cells do not produce CD179a (surrogate light chain; VpreB) transcripts, which distinguishes them from an autoreactive population of naive B cells that are in the process of receptor editing (35, 36) (Fig. 3 A, bottom).

View larger version (45K): [in this window] [in a new window] [Download PPT slide]   Figure 3. The BND phenotype is naive mature B cells. Peripheral blood B cells were isolated and stained with anti-CD19, CD27, IgM, and IgD, as well as one of each of the antibodies shown in the figure, to delineate phenotype differences between BND and naive populations for immaturity (A), maturity (B), or activation markers (C). Isotype controls were included and, where appropriate, CD19+ human BM, marginal zone–like peripheral blood B cells (CD19+, CD27+, IgM+, and IgD+), or fractions representing IgG memory B cells (CD27+, IgM–, and IgD–) were included for an internal positive and negative control. Histograms are representative of three to eight donors in at least three independent experiments. RT-PCR analysis was performed on sorted BND and naive cells for the presence of VpreB (CD179a) transcripts. Actin RT-PCR was included as a template control.

B_ND cells are naturally autoreactive, having no evidence of previous immune experience
Activated B cells rapidly proliferate and accumulate somatic mutations in the antibody variable genes to hone the specificity of the BCR. The cells also undergo class switch recombination to alter the antibody isotype. Variable regions from 228 gene sequences were analyzed from B_ND cells of nine blood donors and compared with historical data in the laboratory from various other B cell subpopulations (24, 25, 40–42). B_ND cells were analyzed from the four donors that antibodies were derived plus an additional five blood donors. As indicated in Fig. 4 A, there were no somatic mutations on the B_ND cell variable genes, indicating that the autoreactivity detected is natural and not generated during an immune response. In addition, no two B_ND cells sharing the same VDJ junctions were detected, likely indicating that they have not been clonally expanded. Finally, there is a small population of germinal center, memory, and plasma cells that have class switched to IgD at the genetic level (referred to as C-CS) and so are also IgD⁺IgM^– (43, 44). These cells also express autoreactive IgD receptors (25). However, PCR analysis found no evidence of recombination junctions between the µ and switch regions for B_ND cells (Fig. 4 B), indicating that IgD is expressed as a splice product of VDJ-Cµ-C transcripts and that B_ND cells have not undergone class switch recombination. C-CS B cells also have excessive somatic mutations of the variable genes and use predominantly Ig light chains, further distinguishing them from B_ND cells (Fig. 4 A). Interestingly, further indicating an autoreactive phenotype (1, 24, 35), B_ND cells use predominantly J_H6 gene segments to encode antibodies (Fig. 4 C). In addition, variable genes from B cells of humans with lupus and in mouse models of lupus (mrl/lpr mice) tend to have long CDR segments (45–47). Cells with long CDR3s are also counterselected during normal human B cell development (1). As indicated in Fig. 4, both the frequency of B_ND cells with long CDR3 segments (Fig. 4 D) and the mean CDR3 length (Fig. 4 E) is significantly increased for B_ND cells. In conclusion, B_ND cells are naturally autoreactive with no evidence of ever having been involved in an immune response.

View larger version (41K): [in this window] [in a new window] [Download PPT slide]   Figure 4. The variable region sequences suggest that BND cells are most like naive cells. (A) Sequence analysis from several healthy donors of BND, naive, IgG and IgM memory, and C-CS (C-delta class switch) revealed that BND cells, like naive cells, did not have somatic mutations in the variable region, are not clonally related as determined by common CDR3s, and use predominantly light chains. (B) PCR of genomic DNA for the constant chain switch regions between µ and revealed no DNA recombination at the delta locus and, thus, the BND cells are not class switched to DNA at the genomic level. (C) Comparison of heavy chain sequences for mean (±SD) JH gene segment usage revealed increase use of JH6 (red bars) in BND cells compared with JH6 usage in naive, IgM, and IgG populations (P < 0.001 by Student's t test between the BND JH6 and separately JH6 in naive, IgM, and IgG populations; n values listed under the table in A). (D and E) BND cells had more amino acids on average (±SD) in their CDR3s as determined both by the frequency of cells with long CDR3s (17 aa+; red bars; P < 0.03) and the mean CDR3 length for each donor (points) and averaged for all donors (blue bars) as compared with the other blood B cell populations (E; P < 0.05; scale is amino acids >10).

View larger version (41K): [in this window] [in a new window] [Download PPT slide]   Figure 5. BND cells have attenuated calcium responses compared with naive cells after BCR cross-linking. (A) A representative scheme of isolation and negative enrichment for the BND and naive populations. Negatively enriched B cells from human peripheral blood were split into two equal fractions: one stained with anti-CD27, anti-IgM, and anti-IgG (anti-BCR) to negatively gate BND cells and the other with anti-CD27 and anti-IgG (anti-BCR) to negatively gate naive cells. Cells were loaded with calcium dye Fluo-4 (Invitrogen) and transferred to flow cytometer, warmed to 37°C, and treated with different amounts and combinations of polyclonal Fab'2 BCR cross-linkers. Side stains were performed with these antibodies plus anti-CD19 and anti-IgD to test for population purity. In each case, B cell purities were >98% and BND and naive purities were >96% as indicated. Calcium curves were determined using average mean fluorescence intensity (MFI) over time. (B) Representative kinetic graphs of average Flou-4 MFI over time for naive, BND, and a B cell class-switched memory control (CS B cell) population representing IgA/E+ memory cells (CD27+, IgM–, IgG–, and IgD–; black line). Basal levels were established for 30 s and then the cells were treated with 50 µg/ml of anti-IgM Fab'2/anti-IgD Fab'2, 50 µg/ml of anti-IgD Fab'2 alone (C), or 20 µg/ml ionomycin (D). (E) Maximal peak fluorescence of Fluo-4 was averaged ±SD for B and C (*, P < 0.05; n = 3 donors; Student's t test). (F) Mean fluorescence of Fluo-4 ±SD representing the basal levels (the first 30 s) was calculated showing a slightly higher amount in the BND population.

As reported for anergic B cells in mice (48, 51; for review see reference 15), basal calcium levels for the B_ND population were 10% higher on average than for naive cells, which is consistent with chronic stimulation by self-antigen (Fig. 5 F). We conclude from these experiments that B_ND cells have intrinsically attenuated BCR-induced calcium flux and that this anergized phenotype likely results from chronic exposure to self-antigen in vivo.

B_ND cells accumulate less pTyr when activated by BCR cross-linking
In addition to calcium flux, the signaling cascade seconds and minutes after BCR cross-linkage results in overall increases of cellular pTyr levels. Because relatively few B_ND cells can be isolated at any one time from human peripheral blood, we chose to analyze pTyr levels by using flow cytometry or "PhosFlow." This powerful technique allows the quantification of both the pTyr magnitude for each individual cell and the frequency of cells affected. With this assay, total activated B cells are rapidly fixed seconds after receptor cross-linkage and before any manipulation that might reduce detectable pTyr levels. The phenotypes and pTyr levels of individual cells can then be resolved retrospectively by using immunostaining and detection on a flow cytometer (Fig. 6 A). Each B cell population was fractionated and depleted of other cell types. Naive and B_ND cells were first enriched through sorting with anti-CD27 and anti-IgM antibodies. Naive cells are predominantly in the total CD27^– fraction, B_ND cells are predominantly in the IgM^–CD27^– fraction, and untreated cells that will not be activated by anti-IgM/D treatment are predominantly in the CD27^–IgD^–IgM^– fraction (determined separately to be mostly IgG, IgA, or IgE⁺ B cells). By spiking the anti-IgM/IgD Fab'2 cross-linking mix with a fluorescently labeled (FITC) anti-IgD and then staining with intracellular anti-CD20 and anti-pTyr after stimulation, fixation, and permeabilization, we could gate and resolve each B cell population to high purity (Fig. 6 A, left). As indicated in the middle of Fig. 6 A, poststimulation pTyr levels were much greater in naive B cells than in B_ND cells after treatment with a combination of polyclonal anti-IgD/IgM for an optimal 45 s (determined separately to be the peak of activation; Fig. S2 A). The reduced pTyr induction is not caused by increased cell death or lack of viability in the B_ND population, as both they and naive cells are stimulated to the same level with the phosphatase inhibitor pervanadate (orthovandate H₂O₂; Fig. 6 A, right histogram).

View larger version (43K): [in this window] [in a new window] [Download PPT slide]   Figure 6. BND cells have reduced tyrosine phosphorylation, suggesting that they are natural anergic B cells. (A) Representative histogram of treated BND and naive cells sorted according to a least-touch strategy (see Materials and methods). 105 cells were incubated with anti-IgD Fab'2-FITC + IgM Fab'2 for 45 s at various concentrations, fixed, and permeabilized, followed by staining by anti-pTyr and intracellular anti-CD20. Histograms were built by FACS analysis and treated populations were compared with untreated controls. Cells were also stimulated with pervanadate as a positive control (right histogram). Green lines are used as qualitative references to mark the median fluorescence of the untreated controls, allowing a visual comparison of fluorescence shift with treated samples. (B) Cells were stimulated with anti-IgD Fab'2 FITC + IgM Fab'2 at concentrations of 10, 50, and 100 µg/ml to determine concentration dependency and were plotted as mean fluorescence intensity (MFI) ± SD of pTyr-PE (for each concentration point, n = 3 groups of 1–3 donors; *, P < 0.05 and **, P < 0.01 for 10 and 50 µg/ml, respectively, by Student's t test). Class switched B cells could also be gated (IgD–, IgM–, and CD20+; black line) from the peripheral set and are included as an internal negative control. Isotype controls (gray line) were established for the stimulated fractions. (C) Incubation of BND and naive cells with 50 µg/ml of anti-IgD Fab'2 alone or in combination with anti-IgM. Results of three donors are plotted as MFI of pTyr. Error bars represent the mean ± SD. Color lines represent individual donor variation between naive and BND population. (D) Cells were sorted as in A but stained with antiphosphorylated Src (pTyr 416; pSrc). Representative histograms are shown for naive and BND cells, untreated and treated with anti-IgD Fab'2 FITC + anti-IgM Fab'2. Bar graphs represent the mean absolute number of units/molecules ±SD of phosphorylated Src as determined by Sphero Rainbow Calibration beads (Invitrogen) from three healthy donors and compared between BND (red) and naive (blue) populations for both treated and untreated fractions. Significance for treated fractions (P < 0.05) was established by a Student's t test.

In conclusion, likely because of chronic self-stimulation, B_ND cells are anergic, displaying reduced activation levels after BCR cross-linkage. This analysis suggests that B_ND cells have a decreased ability to respond to antigen binding, thus avoiding the eventual secretion of autoantibodies. Future characterization of the specific signaling pathways disrupted in B_ND cells should provide insight into the intrinsic mechanisms of immune tolerance for B cells and may identify specific pathways that can be therapeutically targeted to treat autoimmune diseases.

	DISCUSSION

Top ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

 
Mature and naive autoreactive (and thus potentially pathogenic) B cells in the periphery have been shown to account for up to 20% of the circulating B cell pool in healthy human adults (1, 25). The presence of these autoreactive cells begs many but ultimately two important questions: why have they survived selection and how are they tolerized? Focusing on the latter, extensive studies of self-tolerant B cells in transgenic mouse models over the last two decades have revealed the complicated systems of B cell selection used to avoid autoimmunity. B cells expressing transgenic surface Ig that bind DNA or protein autoantigens that do not alter the BCR by receptor editing (2, 3) may be eliminated (clonal deletion) (4). Importantly, and pertinent to this work, many autoreactive cells pass these checkpoints for reasons not well understood and survive to the periphery, but with reduced or altered function so they no longer react to the self-antigens (clonal anergy) (7, 8).

A key advance to understanding and treating human autoimmune disease is to link the numerous fundamental discoveries of the past 20 yr concerning autoreactive B cells in transgenic mice to immune tolerance in humans. Through analyses of a mature naive human B cell population, we have shown that, similar to the classical anti-HEL/HEL model of anergy developed by Goodnow (8), IgD⁺ B cells with no surface IgM are predominantly autoreactive and are functionally attenuated. The mechanistic connection between anergy and reduced IgM expression is not known and is not a universal phenotype of all anergic B cells (for review see reference 15) but may represent a consequence of constant self-antigen exposure in the early life of the cell (53).

As described, anergy in the B_ND population is not solely dependent on receptor density (or lack of surface IgM expression), as signaling through IgD alone was also less effective for B_ND than for naive cells (Fig. 5 & 6). However, it is notable that naive cells responded differently to cross-linking by IgM versus IgD, suggesting that down-regulation of surface IgM could have functional consequences for the B_ND cells. Though the initial burst of stimulation was similar with either cross-linking reagent, a subsequent plateau of sustained increased calcium levels was maintained after IgM stimulation but was lost after IgD stimulation (Fig. S2 C). Combined stimulation through both IgM and IgD resulted in an additive response, with an initial burst similar to either reagent alone, followed by a plateau of sustained calcium levels similar to the anti-IgM response. Thus, assuming that the anti-IgD and anti-IgM reagents similarly cross-link their respective target epitopes, the maintenance of anergy in B_ND cells may also be influenced by the lack of surface IgM. The exact mechanism behind the receptor-mediated lack of response in the B_ND population will need to be further elucidated.

B_ND cells exhibit a well-characterized naive phenotype in that they are CD44⁺, CD23⁺, CD38^–, CD95^–, CD80^–, and CD5^–. They were not clonally expanded and had no somatic mutations. Importantly, B_ND cells can also be distinguished from immature new BM immigrant B cells described in recent reports (1, 32) because the immature cells are CD44^– and CD38⁺ as well as being IgD^– and IgM⁺. B_ND cells also do not express surrogate light chains (CD179a), distinguishing them from preB cells, editing B cells, or recent bone emigrants from the BM (35, 36). Thus, B_ND cells appear mature by all measures; however, we cannot rule out the possibility that they have recently transitioned from an immature state. Similarly, although the attenuated signaling of B_ND cells demonstrated that these cells are anergic, we cannot completely rule out some specialized B cell differentiation similar to B1 or MZ cells in mice because these populations have not been well established in humans. However, based on all analyses to date we propose that B_ND cells arise from autoreactive naive cells that have matured in the classical B-2 pathway.

Similarly to previous reports of anergy, B_ND cells show attenuation in Ca²⁺ response and a decrease in accumulation of phosphotyrosines upon treatment with BCR cross-linkers. However, much like anergic B cells in mice, an active mechanism and exposure to autoantigen appears necessary to maintain anergy because rest in tissue culture media overnight restores B_ND cell signaling to levels similar to naive cells (Fig. S3). In addition, the function of B_ND cells can be rescued when treated with the soluble T cell factors Il-4 and CD40L, leading them to proliferate, express CD80 and the proliferation marker CD71, and differentiate into plasmablast-like CD38^hiCD27^hi cells (Fig. S3). Thus, as in previous reports (52, 54, 55), anergy in the B_ND population appears transient and reversible if the cells receive appropriate pro-T/B cell signals or if the cells are removed from the autoreactive environment. Importantly, Chang et al. (56) recently found that IgD⁺IgM^lowCD27^– B cells that we suspect are B_ND cells become increased in SLE patients. Unlike healthy people, in SLE patients these cells also displayed evidence of activation such as up-regulated CD80/86 levels. Thus, B_ND cells may pose a serious danger in the development of acute or chronic autoimmune pathology, especially if T cell tolerance is also breached allowing helper activation of these autoreactive cells. We propose that the B_ND population may be an important source of precursors for pathological autoantibody-secreting cells activated in lupus or other autoimmune diseases.

In conclusion, we have characterized a population of naturally autoreactive B cells in humans that have become anergic. We can now study the molecular and functional characteristics of this manifestation of anergy and its role in avoiding autoimmunity, or if these cells are not properly controlled, in causing autoimmune disease. Further, studies to elucidate the mechanisms inducing anergy may identify molecular pathways that can be exploited pharmacologically to induce an anergic state in reactive B cells to treat autoimmune diseases or avoid organ rejection.

	MATERIALS AND METHODS

Top ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS REFERENCES

 
B cell enrichment, sorting, and FACS analyses.
B lymphocytes were isolated from the buffy coat component from 500 ml of freshly donated peripheral human blood obtained from the Oklahoma Blood Institute (Oklahoma City, OK) according to established National Institutes of Health guidelines and protocols with approval from the Oklahoma Medical research Foundation Institutional Review Board. Lymphocytes were isolated by placement over a Ficoll density gradient to remove RBCs and other leukocytes followed by B cell–specific enrichment by negative magnetic bead selection as described previously (24, 40, 57). Cell sorting was performed by staining the cells with monoclonal antihuman antibodies against CD27-FITC, CD19-PE–Alexa Fluor 610 (Invitrogen), IgM-APC (SouthernBiotech), and IgD-PE (BD). B_ND and naive B cell populations were resolved as in Fig. 1 using a MoFlo flow cytometer (Dako) in the Oklahoma Medical Research Foundation flow core facility. For phenotype profiles, populations were resolved as in Fig. 1 and stained with various antihuman antibodies labeled with either FITC or PE fluorophores (Invitrogen), including monoclonal antibodies against CD38, CD44, CD80, CD71, CD95, CD138, CD5, CD179a, CD23, CD21, and Annexin-V. For analyses of mitochondrial dye retention, MTG FM dye (Invitrogen) was loaded into sorted B_ND and naive B cell populations at a concentration of 100 nM for 30 min and chased with fresh media as previously described (34). For single-cell PCR, bulk cells of either phenotype were first sorted and then resorted into 96-well PCR plates to ensure purity of single cells (98–99% pure). FACS analysis was performed using flow core facility FACSCalibur, FACSAria, or LSR-II flow cytometers (BD) and software analysis was performed with FlowJo Cytometric software (Tree Star, Inc.).

Sequencing and repertoire analysis.
Cells were sorted into the different populations as indicated above and as previously described (24, 40, 57). V region amplification and sequence analysis was performed as described but, in brief, VH4 and VH3 family specific leader sense primers were paired with antisense primers to particular constant region gene (µ or ). These primers included the following: VH4 sense primer, 5'-ATGAAACACCTGTGGTTCTT-3'; Cµ antisense primer, 5'-CACGTTCTTTTCTTTGTTGC-3'; and C antisense primer, 5'-GTGTCTGCACCCTGATATGATGG-3'. PCR products were amplified using TA or TOPO-TA cloning kits (Invitrogen), and plasmid purification was performed using the miniprep kit (QIAGEN). Sequencing was done at the Oklahoma Medical Research Foundation sequencing core. All V regions were analyzed using in house analysis software and the National Center for Biotechnology Information IgBlast server (http://www.ncbi.nlm.nih.gov/igblast/) or the Immunogenetics server (http://imgt.org/IMGT_vquest/share/textes/).

For VpreB (CD179a) analysis, RNA was extracted from 10⁴–10⁵ freshly sorted B_ND and naive fractions using RNAWiz (Ambion), followed by One-Step RT-PCR (QIAGEN) RNA to cDNA amplification in 40 repeated cycles of equal RNA amounts, using the following primer sets: VpreB sense, 5'-GAGTCAGAGCTCTGCATGTCTG-3'; VpreB antisense, 5'-GTGAGGCCGGATTGTGGTTCCAAG-3'; actin sense. 5'-TACCACTGGCATCGTGATGGACT-3'; and actin antisense, 5'TCCTTCTGCATCCTGTCGGCAAT-3'. Direct PCR products were analyzed on 1% agarose gels with TAE buffer.

Production of recombinant monoclonal antibodies from single human B cell.
In order to test the specificity of B_ND cells, recombinant monoclonal antibodies were produced from the variable region genes of isolated single cells. Using a modified strategy similar to previous reports (1, 25, 36), single B cells were sorted by flow cytometry into 96-well plates, and the variable genes amplified by multiplex single-cell RT-PCR to identify and clone the Ig heavy and light chain genes from a random assortment of cells. These variable genes were then cloned into expression vectors and expressed with IgG constant regions in the 293A human cell line. A total of 95 antibodies from B_ND cells (n = 17, 21, 30, and 27 by donor) were compared with 87 antibodies from naive B cells (IgD⁺IgM⁺CD38^–; n = 16, 37, 10, 10, and 14 by donor). Three sets of naive and B_ND antibodies were matched from individual donors. Each antibody was purified from the serum-free tissue culture media.

ELISA and HEp-2 analyses for autoreactivity.
To screen the expressed antibodies for DNA reactivity or polyreactivity, ELISA microtiter plates (Costar; Corning) were coated with 10 µg/ml of calf thymus ssDNA or dsDNA (Invitrogen), LPS (Sigma-Aldrich), or recombinant human insulin (Fitzgerald). Goat anti–human IgG ( chain specific) peroxidase conjugate (Jackson ImmunoResearch Laboratories) was used to detect binding of the recombinant antibodies followed by development with horseradish peroxidase substrate (Bio-Rad Laboratories). Absorbencies were measured at OD₄₁₅ on a microplate reader (MDS Analytical Technologies). Antibody affinities to the various antigens were compared by endpoint regression analyses to predict the absorbencies expected based on the binding curve for four fourfold dilutions of antibody (beginning at 1 µg/ml, 7 nM) against a fixed concentration of antigen. Antibodies with B_MAX absorbencies that were above the 95% confidence interval of the low positive control (HN241) and had measured affinities of <7 nM (half-maximal binding at 1 µg/ml) were scored as DNA binding. All antibodies were screened for ANA reactivity using the QUANTA Lite ANA ELISA kit (INOVA Diagnostics, Inc.) as per the manufacturer's direction. ANA results were verified on a subset of antibodies by immunofluorescence using commercial HEp-2 slides (Bion Enterprises) as previously described (1, 36) and as per the manufacturer's suggested protocol. HEp-2 slides were analyzed using a fluorescent microscope (Axioplan II; Carl Zeiss, Inc.). All HEp-2 and ANA reactivity were compared with positive control sera from a lupus patient and to nonreactive sera from a healthy blood donor.

Calcium flux.
Fresh human PBMCs from buffy coats were enriched for B cells (see B cell enrichment, sorting, and FACS analysis) and split into four equal fractions of 10⁶ cells for each donor (1): naive (2), B_ND (3), ionomycin naive (4), and ionomycin B_ND. Cells were resuspended in preprepared HBSS containing 1 µM Ca²⁺and 1 µM Mg²⁺ ions supplemented with 1% BSA and were warmed to 30°C for 10 min. After temperature equilibration, cells were loaded with Fluo-4 AM calcium indicator (Invitrogen) at a final concentration of 5 µM and incubated for 15 min in the dark. Cells were then washed and resuspended in HBSS/BSA containing 2.5 µM Probenecid (Sigma-Aldrich) for an additional 30 min at 30°C. After calcium dye loading, cells were placed on ice and differentially stained for 30 min. The naive fractions, including ionomycin naive, received anti-CD27 PE (Invitrogen) and anti–IgG-APC (BD) antibodies. The B_ND fractions, including ionomycin B_ND, received anti-CD27 PE, anti–IgM-APC, and anti–IgG-APC. Cells were also stained with anti–CD3-Bio and streptavidin-PE (Invitrogen) to gate out T cells and other non-B cells that escaped column purification. Non–Fluo-4-loaded B cells from each donor were also simultaneously stained, but with the addition of anti-IgD to evaluate the purity of measured fractions. After staining, cells were washed in HBSS/BSA/Probenecid and placed at 37°C for 10 min. Cells were then placed on a LSRII flow cytometer and gates were drawn for the subpopulations naive and B_ND. Base lines were read for 30 s, after which the cells were removed and stimulated with indicated amounts of Fab'2 anti-IgM (Jackson ImmunoResearch Laboratories) and Fab'2 anti-IgD (SouthernBiotech) or Fab'2 anti-IgD alone or Fab'2 anti-/ (SouthernBiotech) antibodies. Calcium curves were read for 5–7 min. In the ionomycin fractions, 5 µl of 1 mg/ml ionomycin/EtOH (Sigma-Aldrich) was added 30 s after baseline reads to control for dye loading and cell activation potential. Kinetic curves for each calcium response were generated using FlowJo software.

PhosFlow analysis.
Human B cells from blood buffy coats were isolated (see B cell enrichment, sorting, and FACS analysis) and split into two pools for sorting, one (pool 1) stained with anti–CD27-FITC and the other (pool 2) with anti–CD27-FITC and anti–IgM-APC. Pool 1 B cells were sorted for CD27-negative cells and represented primarily naive cells. Pool 2 cells, representing the B_ND cells, were sorted for the IgM^–CD27^– population. Each pool was then washed and resuspended in RPMI without FCS and preheated to 37°C. The cells were then split into aliquots (10⁵ cells/tube) and incubated for 45 s with polyclonal anti-IgD Fab 2'-FITC (SouthernBiotech) and polyclonal anti-IgM Fab 2' (Jackson ImmunoResearch Laboratories). Cells were then immediately fixed with PhosFlow Fix Buffer I (BD) for 10 min at 37°C. Cells were then washed two times in PhosFlow Perm/Wash Buffer I (BD), and stained with anti–pTyr-PE (clone PY20; BD) or antiphosphorylated Src (phosphorylated Tyr 416; Cell Signaling Technology). Cells were simultaneously stained with intracellular anti–CD20-PerCP-Cy5.5 (clone H1-FB1; BD) at the amounts suggested by the manufacturer. For the pSrc analysis, Sphero Rainbow Beads (BD) were used at the amounts suggested by the manufacturer to calibrate the FACS machine for direct comparisons from multiple experiments. For a positive control, a separate fraction (10⁵) of each population was incubated for 5 min with 5 µl of the protein tyrosine phosphatase inhibitor pervanadate/H₂O₂ (3% 100 mM orthovanadate and 3% H₂O₂ in dH₂O). Populations were resolved using FACS and compared with an untreated control. All computational analysis was performed with FlowJo software.

Statistical analyses.
Graphing and statistical analysis were performed using Prism software (GraphPad Software, Inc.) or Excel (Microsoft). For comparisons of antibody autoreactivity between total naive and B_ND antibodies the ² statistic was used. To compare the ANA values generated by the ANA ELISA assays, we used both Student's t tests to compare the means and Mann-Whitney tests to compare the medians. The Student's t tests were done with a Welch's correction because of unequal variances between the naive and B_ND datasets (determined by the F test). However, although not significantly so, there was a non-Gaussian distribution notable in these datasets and, therefore, we also used the nonparametric Mann-Whitney test to compare the median distributions. Also, because of the non-Gaussian distribution the medians are noted in Fig. 2 rather than the means. PhosFlow analyses of pTyr levels and variable gene repertoire analyses used two-tailed Student's t tests (paired for PhosFlow analyses and heteroscedastic for the repertoire analyses).

Online supplemental material.
Fig. S1 illustrates the frequency of polyreactive antibodies PCR-cloned from the naive and B_ND cells. Fig. S2 includes graphs of the time course of activation of naive versus B_ND cells as measured by pTyr accumulation (A), calcium flux induced by anti- light chain treatment (B), and an illustration of the qualitative differences in calcium flux comparing cross-linking of naive B cells through the IgM or IgD receptors individually and in combination (C). Fig. S3 illustrates the rescue of a functional state for B_ND cells by stimulation with cofactors such as CD40L, IL-4, and IL-10 (A-D) or after rest in culture ex vivo, presumably without the self-stimulation (E). Online supplemental material is available at http://www.jem.org/cgi/content/full/jem.20080611/DC1.

Related Article

Learning to tolerate ourselves

Amy Maxmen
J. Exp. Med. 2009 206: 2. [Full Text]