HMGB1-neutralizing IgM antibody is a normal component of blood plasma

Abstract

Extracellular High Mobility Group Box 1 (HMGB1) is a prototypic damage-associated molecular pattern (DAMP). While a homeostatic level of extracellular HMGB1 may be beneficial for immune defense, tissue repair and tissue regeneration, excessive HMGB1 is linked to inflammatory diseases. This prompts an intriguing question: How does a healthy body control the level of extracellular HMGB1? Here we have identified, in the plasma of both healthy humans and healthy mice, an anti-HMGB1 IgM autoantibody that neutralizes extracellular HMGB1 via binding specifically to a 100% conserved epitope, namely HMW4 (HMGB1_98–112). In mice, this anti-HMW4 IgM is produced by peritoneal B-1 cells, and concomitant triggering of their B cell receptor and toll-like receptor 4 by extracellular HMGB1 stimulates the production of anti-HMW4 IgM. The ability of extracellular HMGB1 to induce its own neutralizing antibody suggests a feedback loop limiting the level of this DAMP in a healthy body.

Introduction

Damage-associated molecular patterns (DAMPs), also known as alarmins, are endogenous proinflammatory mediators essential for initiating sterile inflammation that occurs independently of exogenous pathogens (1, 2). High Mobility Group Box 1 (HMGB1) is a prototypical and highly conserved DAMP. It is a non-histone chromatin protein expressed constitutively in most eukaryotic cells (3–5). When released from the cells, it activates, in a redox-dependent manner, chemotaxis and release of proinflammatory cytokines via stimulating the chemokine receptor CXCR4 and pattern recognition receptors TLR2/4/9 and RAGE, respectively (3–6). This leads to sterile inflammation.

There is a basal level of extracellular HMGB1 detectable in serum of healthy mammals (3), which in concert with other proinflammatory mediators promotes immune defense, tissue repair and tissue regeneration (3–5). Excessive extracellular HMGB1, on the other hand, is implicated in many human diseases where sterile inflammation plays a pathogenic role, including inflammatory and autoimmune diseases, cancer, neurodegenerative diseases, metabolic disorders, and cardiovascular diseases (5). However, it is unclear how the body keeps extracellular HMGB1 at a homeostatic level.

We previously found in the Apoe^−/− mouse model of atherosclerosis that a high fat/high cholesterol diet (western-type diet or “WTD”) induced concomitantly not only an elevated level of plasma HMGB1, but also an anti-HMGB1 IgM Ab specific for an epitope called HMW4 (“anti-HMW4 IgM”); as a result, most plasma HMGB1 was in the form of an HMGB1:anti-HMGB1 IgM immune complex (7). Anti-HMW4 IgG (mostly IgG2b and IgG2c) was also present in these mice but at a 100-fold lower level, and there was no sign of class switch from anti-HMW4 IgM to anti-HMW4 IgG even after immunization in the presence of Freund’s complete adjuvant (7). Although initially thought to be a product of pathogenesis, anti-HMW4 IgM was also found in commercial immunoglobulin prepared from pooled human plasma (7). This pointed to the possibility that this Ab might rather be a normal component of plasma.

In this report, we show that this Ab indeed exists in healthy humans and healthy mice. Both the human and the mouse anti-HMW4 IgMs can neutralize HMGB1’s chemotaxis and cytokine release activities. Depletion of anti-HMW4 IgM-producing B cells in Apoe^−/− mice increases atherogenesis, which supports the role of anti-HMW4 IgM as an HMGB1-neutralizing Ab in vivo. In mice, anti-HMW4 IgM is produced by peritoneal B-1 cells in response to extracellular HMGB1 in a B cell receptor (BCR)-specific and toll-like receptor 4 (TLR4)-dependent manner. That extracellular HMGB1 acts as a dual ligand for the two receptors explains why it can stimulate its own neutralizing Ab. The co-presence of extracellular HMGB1 and anti-HMW4 IgM may represent a novel mechanism by which a healthy body limits HMGB1 and, potentially, other DAMPs of the BCR/TLR dual-ligand nature.

Materials and methods

Human plasma samples

Single-donor human blood plasma samples were purchased from Innovative Research, Inc. (catalog number: IPLASK2E2ML) or donated by volunteers with informed consent. A total of 27 different samples were collected, all from healthy young donors (age 20−35) that included Africans (4 females and 4 males), Asians (7 females and 4 males), and Caucasians (4 females and 4 males). All experiments with human samples were approved by the Institutional Review Board of the University of Illinois College of Medicine at Rockford.

Mice and diets

BALB/c, C57BL/6, BALB/c Rag1^−/−, and Apoe−/− C57BL/6 (“Apoe^−/−”) mice were originally purchased from the Jackson Laboratory and used with approval by the institutional Biologic Resource Committee. BALB/c, C57BL/6, and BLAB/c Rag1^−/− mice were fed a normal chow diet from Harlan (Harlan Teklad diet 7001). Apoe^−/− mice were fed the chow diet and switched to the western-type diet (WTD) from Research Diets (D12492) when indicated.

Reagents

Streptavidin (SA) and goat anti-human IgM (hIgM) Ab were from Jackson ImmunoResearch Laboratories. Purified human IgM, Oil Red O (ORO), hen ovalbumin (OVA), alkaline phosphatase (AP)-conjugated SA, SA-agarose, p-nitrophenyl phosphate tablets, monophosphoryl lipid A (MPLA), and thioglycollate broth were from Sigma-Aldrich. Biotin, Ficoll-Paque Plus (GE Healthcare), and Opti-MEM I serum-reduced medium were from ThermoFisher. Mouse IgM (mIgM) standard, goat anti-mouse Ig capture Ab, AP-conjugated goat anti-mIgM, protease-free/Ig-free BSA were from SouthernBiotech. Peptides HMW4 (HMGB1_98–112, PSAFFLFCSEYRPKI), raHMW4 (sequence-randomized HMW4, IRYFPFCLKSFEAPS), HMW4b (C-terminus-biotinylated HMW4), HP1 (HSP60_292–308, KVGLQVVAVKAPGFGDN), and HP1b (C-terminus-biotinylated HP1) were synthesized by Biomatik. Flurophore-conjugated Abs and SA, biotinylated mouse anti-hIgM mAb (MHM-88), mouse IgG2b anti-SA mAb (3A20.2), rat anti-mouse CD16/32 mAb (Fc blocker), biologically active recombinant human HMGB1 (rHMGB1), and mouse TNF-α ELISA reagents were from Biolegend. Cell count standard for flow cytometry was from Bangs Laboratories, Inc.

Fully-reduced rHMGB1 was produced from rHMGB1 by mixing rHMGB1 at 200 μg/ml with dithiothreitol (DTT) at 5 mM for 30 min at room temperature.

Human and mouse anti-HMW4 IgMs (epitope-specific) were obtained in two steps. First, total IgMs from pooled human and mouse plasma were purified with the LigaTrap hIgM and mIgM purification kits (LigaTrap Technologies), respectively, as per manufacturer’s protocols. The resultant IgMs were each incubated in PBS/2 mM EDTA/0.5% BSA (“blood buffer”) with HMW4b-charged SA-agarose, generated by combining 110 μg of HMW4b with 2 ml of SA-agarose slurry (50%), for 1 h at 4 °C on a shaker. The resin was washed multiple times with blood buffer to a cumulative dilution factor of ~ 1:5000. HMW4-specific IgM was eluted by incubating the resin in the LigaTrap elution buffer (pH 3.0) for 10 min at 4 °C. The eluate was adjusted to neutral pH with the LigaTrap neutralization buffer, dialyzed in PBS with a 20-kD cutoff dialysis cassette (ThermoFisher), and finally stored at 4 °C.

HMW4-specific antigen tetramer was produced by combining 1 μg of HMW4b with 2.3 μg of PE-labeled SA (SA-PE) in blood buffer/50% glycerin overnight at −20 °C. HP1-specific tetramer was produced and stored the same way except that the HP1b peptide was used in place of HMW4b. Both tetramers were quantified according to their SA moiety.

Depletion of stained cells (DOSC) immune complex was produced by combining 1 μg of HMW4b with 2.3 μg of SA in blood buffer/50% glycerin overnight at −20 °C, followed by combining with 12.5 μg of mouse IgG2b anti-SA mAb overnight at −20 °C. PE-labeled DOSC (DOSC-PE) was produced similarly, with the use of SA-PE in place of SA.

B cell staining mix was formed of 2 μg of Fc blocker, 0.5 μg of anti-mouse (m) CD19-PE/Cy7, 0.25 μg of anti-mCD23-FITC, 0.25 μg of anti-mCD3-APC, 0.25 μg of anti-mF4/80-APC, 0.02 μg of anti-mCD21-APC, 0.4 μg of SA-PE/Cy5, 4 μg of SA, and 0.07 μg of biotin. The final volume was adjusted to 60 μl in blood buffer. Two microliters were used per million total B cells.

B-1a cell staining mix was formed of 2 μg of Fc blocker, 0.5 μg of anti-mCD19-PE/Cy7, 1.25 μg of anti-mCD5-FITC, 0.25 μg of anti-mCD3-APC, 0.25 μg of anti-mF4/80-APC, 0.25 μg of anti-mCD23-APC, 0.4 μg of SA-PE/Cy5, 4 μg of SA, and 0.07 μg of biotin. The final volume was adjusted to 60 μl in blood buffer. Two microliters were used per million total B cells.

Chemotaxis staining mix was formed of 2 μg of Fc blocker, 1.5 μg of anti-mGr-1-FITC (RB6–8C5), 0.6 μg of anti-mCD11b-PE, and 0.3 μg of anti-mF4/80-APC. The final volume was adjusted to 60 μl in blood buffer. One microliter was used for cells from 10% of the peritoneal cavity lavage.

ELISA for human anti-HMW4 IgM in blood plasma

Greiner high binding plates (catalog number: 655081) were coated overnight in coating buffer (0.05 M sodium bicarbonate, pH 9.6) with HMW4 (test) and raHMW4 (specificity control) in separate wells at 0.8 μg/50 μl/well, and blocked for 2 h in coating buffer with 1% Ficoll-Paque Plus. Plasma samples were diluted 1:200 in PBS/0.02% Tween-20 (wash buffer) and plated at 50 μl/well for 30 min. Biotinylated mouse anti-hIgM mAb (MHM-88) was added to the wells at 20 ng/50 μl/well for 30 min, followed by AP-conjugated SA at 10 ng/50 μl/well for 30 min. Both reagents were diluted in wash buffer. p-Nitrophenyl phosphate tablets were dissolved in 1 M diethanolamine/0.5 mM MgCl₂ (pH 9.8) as the substrate. Color was allowed to develop for 1 h and overnight. ELISA standards were constructed in the same plate by coating additional wells with goat anti-hIgM Ab at 0.2 μg/50 μl/well, followed by loading the wells with graded amounts of purified hIgM. Because the standard curve was rarely linear, O.D. readings from both the HMW4 and the raHMW4 wells were first converted into a quantity of IgM, and then the O.D. reading from raHMW4 was subtracted. The difference derived this way from wells loaded with wash buffer instead (i.e., not with any plasma sample) was deemed false positive, and its average value was subtracted from all samples to obtain the truly positive readings. The latter were converted to μg/ml plasma based on how much the plasma samples were diluted when plated.

ELISA for mouse anti-HMW4 IgM in blood plasma and spent medium

The same Greiner plates were coated and blocked as described above. Samples of mouse plasma and spent medium were diluted in wash buffer at 1:200 – 1:800 and at 1:50, respectively, and plated for 30 min. Spent medium samples from cell cultures stimulated with rHMGB1 were plated overnight in PBS/0.01% BSA (to allow the plate-bound HMW4 to compete for anti-HMW4 IgM against residual rHMGB1 in the media). AP-conjugated goat anti-mIgM was diluted in wash buffer and added to the wells at 20 ng/50 μl/well for 30 min. ELISA standards were constructed in the same way as described above, except that goat anti-mIg capture Ab and mIgM standard were used. The rest of the procedures for developing and reading the ELISA were identical to what described for human ELISA.

In vivo chemotaxis assay

C57BL/6 male mice at 6 weeks of age were injected (i.p.) with 0.5 μg of fully-reduced rHMGB1 in 0.5 ml PBS/0.01% BSA. Control mice were injected with 0.5 ml of PBS/0.01% BSA. After 6 h, 3 ml blood buffer were injected (i.p.) and lavage was collected from the peritoneal cavity. Cells in 0.3 ml of the lavage were pelleted and stained with 1 μl of chemotaxis staining mix (described above). Infiltrating neutrophils were identified as SSC^intGr-1^hiCD11b^int cells (8) and counted against 10000 counts of the cell counting standard by flow cytometry.

For testing neutralization of fully-reduced rHMGB1 by anti-HMW4 IgM, fully-reduced rHMGB1 was incubated with human or mouse anti-HMW4 IgM, or with non-specific hIgM or mIgM (as a control), at a 1:9 ratio (w/w) for 1 h on ice. The mix was then tested for chemotaxis, as describe above. For testing dependence of the neutralization on binding to the HMW4 epitope within fully-reduced rHMGB1, anti-HMW4 IgMs each were pre-incubated with the HMW4 peptide or raHMW4 (as a control) at a 4:1 ratio (w/w) for 1 h on ice before being tested for neutralization.

In vitro cytokine secretion assay

RAW 264.7 murine macrophages-like cells were stimulated in a 96-well plate with 1 μg of rHMGB1 in 50 μl of Opti-MEM I medium for 16 h. The spent medium was analyzed for secreted TNF-α with the mouse TNF-α ELISA kit, as per manufacturer’s protocol.

For testing neutralization of rHMGB1 by anti-HMW4 IgM, rHMGB1 was incubated with human or mouse anti-HMW4 IgM, or with non-specific hIgM or mIgM (as a control), at a 1:9 ratio (w/w) for 1 h on ice. The mix was then tested for stimulation of TNF-α secretion, as describe above. For testing dependence of the neutralization on binding to HMW4 within rHMGB1, anti-HMW4 IgMs were pre-incubated with HMW4 or raHMW4, at a 4:1 ratio (w/w) for 1 h on ice before the neutralization test.

Detection of epitope-specific B cells by tetramer staining

Method for HMW4- and HP1-specific tetramer staining was reported (7, 9) but has since been improved. Mouse blood was taken by cardiac puncture, from which leukocytes were enriched by ACK lysis. Bone marrow was flushed out from the femur bone, followed by ACK lysis. Peritoneal cavity cells were obtained from lavage. Splenocytes were obtained by homogenizing the spleen, followed by ACK lysis. Cells from these sources were washed multiple times in blood buffer to a cumulative dilution factor of ≥ 10⁶ to remove endogenous IgM, which may neutralize the tetramers during staining. The resultant cell pellet was stained and blocked with the B cell staining mix or the B-1a cell staining mix (described above) at 2 μl/10⁶ B cells for 10 min on ice. The cells were washed once in 10 volumes of blood buffer, pelleted, and stained with a tetramer at 0.3−1 ng/10⁶ B cells in 30 μl of blood buffer for 10 min on ice; as a negative control, tetramer staining was performed after blocking the BCR with goat anti-mIg capture Ab (2 μg/10⁶ B cells). The cells were then analyzed and sorted (if necessary) with a FACScalibur and a FACS Melody, respectively. Epitope-specific B cells (at various sites) were analyzed as follows: B-1 and B-2 cells (from blood, bone marrow and peritoneal cavity) were identified as FSC^largerCD19⁺CD23^-tetramer⁺ and FSC^smallerCD19⁺CD23⁺tetramer⁺ cells, respectively; B-1 cells from the spleen, as FSC^largerCD19⁺CD21^intCD23^-tetramer⁺ cells; follicular B cells, as FSC^smallerCD19⁺CD21^intCD23⁺tetramer⁺ cells; marginal zone B cells, as FSC^largerCD19⁺CD21^hiCD23^-tetramer⁺ cells. The Bang Laboratories’ cell counting standard was used for counting the tetramer⁺ B cells.

Treating mice with the DOSC immune complex

For dosing DOSC in vivo, graded doses of DOSC were injected (i.p.) in 0.5 ml of PBS/0.01% OVA. For depleting HMW4-specific B cells, mice were injected (i.p.) with 1 ml of 3% thioglycollate broth (10) on day −2, followed by i.p. injection of 23 ng of DOSC in 0.5 ml of PBS/0.01% OVA on days −1 and 0. Mice mock-treated with thioglycollate and tetramer-less “DOSC” were used as a control.

Lesion analyses

Lesions in the aortic root were analyzed by the morphometric method, and lesions in the aortic arch were analyzed by the Oil Red O (ORO) back extraction method, as we described (7, 9).

Production of anti-HMW4 IgM by B cell subsets in culture

B-1 and B-2 cell subsets were flow-sorted in the purity mode with the FACS Melody sorter. The same method was also used to “deplete” B-1 cells of the tetramer⁺ cell fraction by excluding the latter via gating and repeated flow sorting. The collected cells were plated in a 96-well plate at 10⁵ cells/well in 100 μl of RPMI1640/10% FCS supplemented with 0.1 μg/ml MPLA. After 7 days, the spend medium was analyzed for anti-HMW4 IgM by ELISA.

Adoptive transfer of tetramer⁺ B-1 cells into Rag1^−/− mice

Tetramer⁺ and tetramer⁻ B-1 cells from the peritoneal cavity were purified by flow sorting with the FACS Melody and injected (i.p.) separately into two groups of syngeneic Rag1^−/− mice at 2000 cells/mouse. After 6 weeks, blood plasma was obtained from the two groups and compared for anti-HMW4 IgM by ELISA.

Stimulation of anti-HMW4 IgM response in vivo and in vitro

For in vivo stimulation with rHMGB1, C57BL/6 male mice at 6 weeks of age were injected (i.p.) with 4 μg of rHMGB1 in 0.2 ml of PBS/0.01% OVA. Control mice were injected with PBS/0.01% OVA. After 7 days, anti-HMW4 IgM in blood plasma was analyzed by ELISA. For in vitro stimulation with rHMGB1, flow-sorted peritoneal B-1 cells from 6-week old mice were cultured in RPMI1640/10% FCS with or without (control) rHMGB1 at 4 μg/ml. After 7 days, spent medium was analyzed for anti-HMW4 IgM by ELISA. The TLR4 inhibitor TAK242 (dissolved in DMSO) was used at 1 μM. DMSO was used as a control. For in vivo stimulation with receptor-specific agonists, mice at 6 weeks of age were injected (i.p.) with 10 μg of HMW4 and/or 2 μg of MPLA (in 0.2 ml of PBS/0.01% OVA), or with neither. After 7 days, anti-HMW4 IgM in blood plasma was analyzed by ELISA.

Statistics

For analysis of sex difference, data from female and male mice were treated separately; whereas for analysis of a difference common to both sexes, data from both sexes were pooled. The mean probability for a human to be positive for anti-HMW4 IgM was calculated by the equation: 1 − P⁴⁵ < 0.00001, where P is the probability of being positive, 45 is the number of humans tested, and 0.00001 is the probability of all 45 being positive by chance, as calculated by the Wilcoxon Signed-Rank test. Thus, P⁴⁵ > 0.99999; P > 0.99998. The unpaired two-sided Student t-test was used for data involving 2 independent groups. One-way ANOVA with post hoc Turky’s honest significant difference test was used for data involving more than 2 groups. The GraphPad Prism software was used for curve-fitting.

Results

Anti-HMW4 IgM is present in all healthy humans and mice tested

Although anti-HMW4 IgM had been detected in commercial human immunoglobulin from pooled human donors (7), it was possible that the Ab came from a few peculiar donors. To assess this possibility, we collected plasma samples from 45 human donors individually, encompassing women and men of African, Asian, and Caucasian races. All donors were young (20−35 years of age) and healthy. We detected anti-HMW4 IgM in every one of them (Fig. 1), which suggests a probability of 0.99998 for an individual to have the Ab. Consistent with this result, we also found anti-HMW4 IgM in healthy wild-type mice, such as the most widely used BALB/c and C57BL/6 strains, and pre-atherosclerotic C57BL/6 Apoe^−/− (“Apoe−/−”) mice (Fig. 1). Sex difference was noted in both humans and mice; remarkably, humans showed a male bias whereas mice showed a female bias (Fig. 1). In mice there was also strain difference, with BALB/c and Apoe^−/− mice having higher levels of anti-HMW4 IgM than C57BL/7 mice (Fig. 1). Despite these differences, it was obvious that the presence of anti-HMW4 IgM is common in young healthy humans and mice.

Figure 1.

Anti-HMW4 IgM is present in all healthy humans and mice tested. Single-donor plasma samples from healthy humans (24 women and 21 men, at age 20−35) and mice (8−14 per sex for each strain, at age of 6 weeks, and on a normal chow diet) were tested for anti-HMW4 IgM by ELISA. Each symbol represents the result from one subject. A horizontal bar indicates the average of all results within the group. Sex (*p ≤ 0.03 by t-test) and strain (**p ≤ 0.007 by one-way ANOVA) differences are indicated. No comparison between humans and mice is made because the human and mouse plasmas were measured using different reagents.

Anti-HMW4 IgM neutralizes HMGB1 by binding to the HMW4 epitope specifically

As we described before (7), anti-HMW4 IgM binds to the HMW4 epitope of HMGB1 (HMGB1_98–112), which is 100% conserved between mice and humans. At the center of this epitope lies the C₁₀₆ residue, known to be essential for binding of extracellular HMGB1 to TLR4 (11). Because of that, we suspected that anti-HMW4 IgM may block C₁₀₆ and thereby neutralize extracellular HMGB1’s cytokine-stimulating activity. To determine if this is the case, we affinity-purified anti-HMW4 IgMs from pooled human and pooled mouse plasmas, and tested their capacity to neutralize recombinant human HMGB1 (“rHMGB1”), which is 99% identical to murine HMGB1 and is active in mice (12). In a RAW 264.7 cell-based TNF-α release assay (12), both human (Fig. 2A) and mouse (Fig. 2C) anti-HMW4 IgMs neutralized the cytokine-releasing activity of rHMGB1. Critically, the neutralizing activity of the IgM Abs was abolished if they were blocked with HMW4, but not with raHMW4, a randomized peptide of the same a.a. composition, which showed that anti-HMW4 IgMs neutralized the cytokine-releasing activity of rHMGB1 by binding to the HMW4 epitope specifically.

Figure 2.

Anti-HMW4 IgM neutralizes HMGB1 by binding to the HMW4 epitope specifically. (A) Cytokine-secretion assay. RAW 264.7 murine macrophages-like cells were stimulated in culture with rHMGB1 and anti-HMW4 human IgM (αHMW4 hIgM) in indicated combinations. TNF-α secreted by the cells was measured by ELISA of the spent medium. To assess the dependence of neutralization of HMGB1 activity on binding to the HMW4 epitope, anti-HMW4 IgM was preblocked with HMW4 or raHMW4 (control), as indicated. Shown are mean ± SD of 2−4 experiments. *p ≤ 0.02 (one-way ANOVA). (B) Chemotaxis assay. C57BL/6 male mice were injected (i.p.) with fully-reduced rHMGB1 and anti-HMW4 hIgM in indicated combinations. These mice were used because they have the lowest level of endogenous anti-HMW4 IgM (Fig. 1). The resultant influx of neutrophils into the peritoneal cavity was measured by flow cytometry. Neutrophils were counted as % of total cells in the peritoneal lavage (≥ 50,000 total cells were analyzed). Shown are mean ± SD of 3 experiments. *p ≤ 0.01 (one-way ANOVA). (C & D) Same cytokine-secretion and chemotaxis assays with anti-HMW4 mouse IgM (αHMW4 mIgM). Shown in each panel were results from 2−4 experiments. *p ≤ 0.01 (one-way ANOVA). (E) Chemotaxis assay for in vivo sensitivity to rHMGB1. Male mice of the indicated strains with different levels of endogenous anti-HMW4 IgM (as shown in Fig. 1) were injected (i.p.) with fully-reduced rHMGB1 in PBS. Net increase in neutrophil numbers was calculated as: influx of neutrophils to the peritoneal cavity following injection of an indicated amount of fully-reduced rHMGB1 – the same influx following injection of PBS. Shown in each panel were results from 2 experiments. *p ≤ 0.03 (one-way ANOVA).

Moreover, in an in vivo neutrophil chemotaxis assay (8, 13, 14), both human (Fig. 2B) and mouse (Fig. 2D) anti-HMW4 IgMs neutralized the chemotactic activity of fully-reduced rHMGB1. Again, this neutralizing activity was abolished by blocking the IgM Abs with HMW4 but not raHMW4. In summary, these results indicate that the human and mouse anti-HMW4 IgMs, via binding to the HMW4 epitope, can neutralize both the cytokine-releasing and the chemotactic activities of extracellular HMGB1.

Depletion of anti-HMW4 IgM-producing B cells in C57BL/6 Apoe^−/− mice leads to increase in atherosclerosis

To assess whether anti-HMW4 IgM actually functions as a neutralizing Ab endogenously, we injected exogenous HMGB1 into the mouse strains that produce different levels of endogenous anti-HMW4 IgM (Fig. 1), and analyzed their neutrophil chemotaxis response (Fig. 2E). The C57BL/6 strain, which has the lowest level of endogenous anti-HMW4 IgM, showed the strongest response. In comparison, the BALB/c and C57BL/6 Apoe^−/− strains in which physiologically higher titers of the anti-HMW4 IgM are present are relatively protected from the biological action of exogenous HMGB1.

To assess the activity of the endogenous anti-HMW4 IgM further in a disease model, we used the Apoe−/− model of atherosclerosis because the disease is driven partly by chronic low-grade inflammation that is typical of sterile inflammation. We followed the work of Kanellakis et al., which shows that HMGB1 neutralization reduces diet-accelerated atherosclerosis in Apoe^−/− mice (15). As endogenous anti-HMW4 IgM already exists in young, pre-atherosclerotic Apoe^−/− mice (Fig. 1), we reasoned that its removal from these mice would potentiate more severe atherosclerosis if it is indeed a neutralizing Ab. Of note, in our previous study in Apoe−/− mice, we raised the level of this IgM Ab by immunization to show that it correlates with atheroprotection (7). Exploiting here a complementary approach, i.e., to remove this Ab, would allow us to solidify its role in atheroprotection.

Because it was infeasible to remove endogenous Ab directly, we aimed to deplete the anti-HMW4 IgM-producing B cells instead. To target these cells, a HMW4 tetramer of biotinylated HMW4 and streptavidin (“SA”) was first formed (termed “tetramer” hereafter), as we described (7). Then, an immune complex for Depletion Of (tetramer-) Stained Cells was formed (named “DOSC” hereafter) between the tetramer and a mouse anti-SA IgG2b mAb. DOSC showed the same specificity as the tetramer because i.p. injection of DOSC blocked tetramer staining of B cells from the peritoneal cavity in a dose-dependent manner (Fig. 3A). When potentiated with thioglycollate (10), DOSC at the saturating dose (~ 23 ng/injection; Fig. 3A) reduced tetramer⁺ B-1 cells in the peritoneal cavity and blood in 7 days by ~ 70% (Fig. 3B). This was not due to blocking of tetramer staining by DOSC, because DOSC was not found on any B cells 7 days after injection (Fig. 3C). Remarkably, at the dose of ~ 23 ng/injection, DOSC preferentially reduced tetramer⁺ B-1 cells over tetramer⁺ B-2 cells (Fig. 3B), likely because B-2 cells are replenished from bone marrow whereas B-1 cells are not (16). Mechanistically, reduction of tetramer⁺ B-1 cells depended on both of the tetramer and anti-SA IgG2b moieties (Fig. 3D). The reduction was specific to HMW4-tetramer⁺ cells (Fig. 3E), because the injection did not reduce B-1 cells stained by the HSP60-derived HP1-tetramer (9). These results showed that DOSC injection specifically removed the majority (> 70%) of endogenous tetramer⁺ B-1 cells. As a result, DOSC-treated Apoe^−/− mice showed an impaired anti-HMW4 IgM response after being switched to WTD (Fig. 3F), which was associated with increases in atherosclerosis and macrophage (Mac-2⁺) infiltration of the aorta (Fig. 3G). In summary, these results support the role of anti-HMW4 IgM as an endogenous HMGB1-neutralizing Ab.

Figure 3.

Depletion of anti-HMW4 IgM-producing B cells in C57BL/6 Apoe^−/− mice leads to increase in atherosclerosis. Female mice were used because they have a higher level of endogenous anti-HMW4 IgM than male mice (Fig. 1). (A) HMW4-specific tetramer (tetramer) staining of B cells in the peritoneal cavity 1 h after i.p. injection of graded doses of DOSC. Tetramer⁺ B-1 and B-2 cells were calculated as % of total B-1 and B-2 cells, respectively (≥ 60,000 total B-1 cells and ≥ 10,000 total B-2 cells were analyzed). Shown are nonlinearly-fitted dose-response curves from 3 experiments. The maximal depleting effect on B-1 and B-2 cells is at ~ 23 ng. (B) Tetramer staining of B cells in the peritoneal cavity and blood 7 days after i.p. injection with thioglycollate broth and DOSC (DOSC), or with PBS. Data (graphed in a logarithmic scale) are mean ± SD of 3 experiments. *p ≤ 0.01 (t-test); ns, not significant (t-test). (C) B cells recovered from the peritoneal cavity 7 days after i.p. injection with thioglycollate broth and PE-labeled DOSC (DOSC-PE), or with PBS. Few DOSC⁺ cells were found in the B cell population, whereas depletion of tetramer⁺ B-1 cells was detected by tetramer staining as in (B). Data are mean ± SD of 2 experiments. *p = 0.003 (t-test). (D) Tetramer staining of B-1 cells from the peritoneal cavity 7 days after i.p. injection of thioglycollate broth and an indicated agent. Data are mean ± SD of 3 experiments. *p ≤ 0.01 (one-way ANOVA). αSA mAb, anti-streptavidin mAb. (E) HP1-specific tetramer (HP1-tetramer) staining of peritoneal B-1 cells 7 days after i.p. injection with thioglycollate broth and DOSC, or with PBS. Data are mean ± SD of 3 experiments. (F) Level of plasma anti-HMW4 IgM in DOSC-treated mice as a function of time on WTD. Control group was mock-treated with thioglycollate and αSA mAb (mock). Data are mean ± SD of 6 mice/group, from which time-course curves are generated by nonlinear fitting. *p ≤ 0.005 (t-test). (G) Lesion analyses of DOSC- and mock-treated mice after 8 weeks of WTD. Data are mean ± SD of 12 mice/group. *p ≤ 0.04 (t-test).

Peritoneal B-1 cells are the major producer of mouse anti-HMW4 IgM

To locate the B cells that produce anti-HMW4 IgM, we flow sorted B-1 and B-2 cell subsets (to ≥ 99% purity) from the peritoneal cavity, spleen, and bone marrow of healthy young BALB/c and C57BL/6 mice and examined their ability to produce anti-HMW4 IgM in culture in the presence of the TLR4 agonist monophosphoryl lipid A (MPLA) as a general stimulator (17). In both strains, peritoneal cavity B-1 cells produced far more anti-HMW4 IgM than any of the others (Fig. 4A). To assess whether this was due to insufficient endogenous activation of the B cells, Apoe^−/− C57BL/6 mice were fed WTD for 8 weeks, which should activate HMW4-specific B cells as we reported (7). Again, anti-HMW4 IgM was detected mainly from peritoneal cavity B-1 cells (Fig. 4A). Thus, although there have been observations that peritoneal cavity B-1 cells do not secrete IgM (18), this was, clearly, not the case for the secretion of anti-HMW4 IgM. When the peritoneal cavity B-1 cells were further sorted into B-1a and B-1b, both subsets produced anti-HMW4 IgM; however, the B-1a subset produced relatively more (Fig. 4B). Within B cells there are HMW4-tetramer⁺ cells (Fig. 3, A–D) that bind to the tetramer via their BCR (7). To determine whether the tetramer⁺ B-1 cells are the actual producers, we compared the IgM production by total B-1 cells v. B-1 cells depleted of the HMW4-tetramer⁺ population (via flow sorting). The production of anti-HMW4 IgM, indeed, depended on the tetramer⁺ B cells (Fig. 4C). Moreover, tetramer⁺ B cells flow sorted from peritoneal cavity B-1 cells, after being adoptively transferred into syngeneic Rag1^−/− mice, produced the IgM (Fig. 4D). In aggregate, these data show that the tetramer⁺ B-1 cells in the peritoneal cavity are both necessary and sufficient for the production of anti-HMW4 IgM.

Figure 4.

Peritoneal B-1 cells are the major producer of anti-HMW4 IgM. (A) ELISA for anti-HMW4 IgM produced in culture with MPLA. B-1 v. B-2 cells were flow-sorted to ≥ 99% pure from indicated sites in 6-week-old BALB/c and C57BL/6 mice and 14-week-old Apoe^−/− mice that had been either kept on a chow diet or switched from the chow diet to WTD for the last 8 weeks. Each symbol represents the result of 1 experiment. Each horizontal bar indicates the average of 4 results. *p ≤ 0.001 for site differences within the same strain (one-way ANOVA); **p = 0.003 for difference between the C57BL/6 and BALB/c strains (t-test). (B) ELISA for anti-HMW4 IgM produced in culture with MPLA by B-1a v. B-1b cells flow-sorted from the peritoneal cavity of indicated. Each symbol represents the result of 1 experiment. Each horizontal bar indicates the average of 4 results. *p ≤ 0.006 (t-test). (C) ELISA for anti-HMW4 IgM produced in culture with MPLA by peritoneal B-1 cells (Total) v. the same cells depleted of tetramer⁺ cells (Depleted). Both samples were flow-sorted from BALB/c female mice. Data are mean ± SD of 3 experiments. *p = 0.006 (t-test). (D) ELISA for anti-HMW4 IgM in blood plasma of BALB/c Rag^−/− mice after adoptive transfer with tetramer⁺ v. tetramer⁻ peritoneal B-1 cells from BALB/c female mice. Data are mean ± SD of 3 experiments. *p = 0.02 (t-test). No anti-HMW4 IgM was detected in naïve Rag^−/− mice (not shown).

Dual BCR/TLR signaling of peritoneal B-1 cells is necessary and sufficient for their production of anti-HMW4 IgM

To understand what causes these cells to produce the IgM, we assessed the role of extracellular HMGB1 as a stimulant. We injected (i.p.) rHMGB1 into C57BL/6 male mice. These mice were chosen because of their low basal level of endogenous anti-HMW4 IgM (Fig. 1), which might otherwise dampen the effect of rHMGB1 (Fig. 2). The injection increased plasma anti-HMW4 IgM in 7 days (Fig. 5A). rHMGB1 also stimulated the secretion of anti-HMW4 IgM by flow sorted peritoneal cavity B-1 cells in culture (Fig. 5B). Here, the effect of rHMGB1 was TLR4-dependent because it could be blocked by TAK242, a specific inhibitor of TLR4 (19) (Fig. 5B), and it was BCR-specific as well because the production of total IgM was only marginally affected (Fig. 5C). Extracellular HMGB1 has been shown to be both a ligand for TLR4 (4) and an autoantigen for HMW4-specific B cells (7). Our results here show further that its stimulation of the anti-HMW4 IgM response is attributed to both of these properties. To solidify this finding, we injected (i.p.) mice with HMW4 (specific BCR epitope) and MPLA (specific TLR4 agonist), alone or in combination. Despite strain and sex differences, while either HMW4 or MPLA alone had only a modest effect, the two combined stimulated the production of anti-HMW4 IgM strongly (Fig. 5, D&E). Collectively, these results reveal that HMGB1 can stimulate its own neutralizing Ab because it is a dual ligand for BCR and TLR4. This interpretation is consistent with the prior finding that dual engagement of BCR and TLR activates B cells, particularly B-1 cells, independently of T cell help (20).

Figure 5.

Dual BCR/TLR signaling of peritoneal B-1 cells is necessary and sufficient for their production of anti-HMW4 IgM. (A) ELISA for anti-HMW4 IgM in blood plasma of C57BL/6 males 7 days after i.p. injection of 4 μg of rHMGB1. Data are mean ± SD of 3 experiments. *p = 0.04 (t-test). (B) ELISA for anti-HMW4 IgM produced by flow-sorted peritoneal B-1 cells from indicated strains and cultured in the presence of indicated agents. Each symbol represents the result of 1 experiment. Each horizontal bar indicates the average of 4 results. *p ≤ 0.008 for differences between indicated cultures within the same strain (one-way ANOVA); **p = 0.03 for difference between BALB/c and C57BL/6 mice in their B cell response to rHMGB1 (t-test). (C) ELISA for total IgM produced in response to rHMGB1 by flow-sorted peritoneal B-1 cells from BALB/c and C57BL/6 female mice. Data are mean ± SD of 3 experiments. ns, not significant (t-test). (D) ELISA for anti-HMW4 IgM in blood plasma of indicated strains 7 days after i.p. injection of indicated agents. Each symbol represents result from 1 mouse. Each horizontal bar indicates the average from 8 mice. *p ≤ 0.001 v. all other treatments within the same strain (one-way ANOVA); **p = 0.006 for difference between BALB/c and C57BL/6 mice in their response to the injection of the MPLA + HMW4 combination (t-test). (E) ELISA for anti-HMW4 IgM in blood plasma of indicated strains after i.p. injection of the MPLA + HMW4 combination at days 0 and 7. Data are mean ± SD of 4 experiments, from which time-course curves are generated by nonlinear fitting. *p ≤ 0.01 female v. male mice of the same strain (t-test).

Given this mechanism and the fact that there is a physiologically relevant level of extracellular HMGB1 in serum of healthy mammals (3, 5), the presence of anti-HMW4 IgM in the same body is all but inevitable. This explains why this Ab is detected in all the humans and mice tested (Fig. 1). Thus, we conclude that this Ab is a normal component of blood plasma.

Discussion

We have identified, in the plasma of both healthy humans and healthy mice, an anti-HMGB1 IgM autoantibody that neutralizes extracellular HMGB1 via binding specifically to an evolutionarily conserved epitope, namely HMW4 (HMGB1_98–112). The commonness of this antibody in all the healthy subjects that we have analyzed regardless of race, sex, or species indicates that it is a normal and evolutionarily conserved component of blood plasma.

There are other B-1 cell-produced autoantibodies in normal blood, including the IgM subclass of the ABO blood group antibodies made to antigens not present on self red blood cells (21, 22). Most of these antibodies are specific for non-protein epitopes, such as phospholipids or carbohydrates that are evolutionarily conserved between men and bacteria (18). They might have been evolutionarily selected for cross-reactivity against bacteria. Although anti-HMGB1 IgM appears to differ by binding to a protein epitope (HMW4), the epitope is, again, highly conserved (100% identical between men and mice). Thus, these normal autoantibodies appear to target mainly evolutionarily conserved epitopes and might have been co-selected along with their targets.

Why do mice and humans have an antibody to a protein sequence that is conserved and apparently universal? Our study has pointed to a couple of possibilities. Mechanistically, we understand that HMGB1 activates cognate B-1 cells by being a BCR/TLR dual ligand. This is a property that most other endogenous proteins don’t have. As a result, although there is B cell tolerance to self proteins generally, an anti-HMGB1 response is exceptional and occurs when the body must maintain a physiological level of extracellular HMGB1 (3, 5). Our data also show that functionally the anti-HMGB1 IgM neutralizes HMGB1 and might well be evolutionarily selected to allow a humoral feedback loop for negatively regulating the activity of extracellular HMGB1.

Although anti-HMGB1 IgM is detected in all human subjects analyzed, there are substantial interindividual variations in levels (up to a 10-fold difference). This may be due to the small sample size (45 subjects); however, increasing sample size generally tends to reduce standard deviation but not interindividual variations. Thus, a more likely possibility is that there are large variations among individuals naturally. Given that HMGB1 is considered to be one of the major causes of sterile inflammation, such natural variations could predispose individuals to various degrees of vulnerability to sterile inflammation.

Studies have also pointed to the role of HMGB1 in multiple serious disease manifestations, particularly those associated with chronic inflammation such as cancer and autoimmune diseases (5, 23, 24). Future studies, ideally with larger cohorts, are needed to evaluate the differences in the anti-HMGB1 IgM response between healthy individuals and patients with conditions associated with increased plasma HMGB1. Moreover, because HMGB1 has also been shown to play a role in infectious inflammation such as in sepsis (25, 26) and pneumonia (27), future studies are needed to explore a parallel anti-inflammatory role of the anti-HMGB1 IgM in these disease contexts.

Our data show that being a BCR/TLR dual ligand is sufficient for the activation of a humoral feedback loop. This implies that similar loops may exist for other DAMPs that meet this requirement, such as the known endogenous TLR ligands (28, 29) heat shock proteins (30–32) and S100A8/9 (33). Future studies are warranted to determine whether this is the case in order to assess the generality of this loop mechanism.

From the therapeutic standpoint, this study can, potentially, provide a novel target for immune interventions. One way would be to explore the protective, anti-inflammatory efficacy of the anti-HMGB1 IgM directly, as an “injectable drug.” For this purpose, the use of anti-HMW4 IgM mAbs should be advantageous. At the present time, we are working toward the generation of anti-HMW4 IgM mAbs. Alternatively, the DAMP-regulating feedback loops may be targeted in vivo. For instance, endogenous anti-HMW4 IgM and other yet to be identified DAMP-neutralizing antibodies can be raised via strategies such as active immunization, in order to control DAMP-mediated chronic sterile inflammation and its related illnesses.

Key Points.

• Anti-HMGB1 neutralizing IgM autoantibody exists normally in healthy humans and mice.

• In mice, B-1 cells produce the antibody in response to extracellular HMGB1.