Protective role for B1b B cells and IgM in obesity-associated inflammation, glucose intolerance, and insulin resistance

Abstract

Objective

Little is known regarding the role(s) B cells play in obesity-induced metabolic dysfunction. The present study utilized a mouse with B cell-specific deletion of Id3 (Id3^{Bcell KO}) to identify B cell functions involved in the metabolic consequences of obesity.

Approach and Results

Diet-induced obese (DIO) Id3^{Bcell KO} mice demonstrated attenuated inflammation and insulin resistance in visceral adipose tissue (VAT), and improved systemic glucose tolerance. VAT in Id3^{Bcell KO} mice had increased B-1b B cells and elevated IgM natural antibodies (Nabs) to oxidation-specific epitopes (OSE). B-1b B cells reduced cytokine production in VAT M1 macrophages, and adoptively transferred B-1b B cells trafficked to VAT and produced NAbs for the duration of thirteen week studies. B-1b B cells null for Id3 demonstrated increased proliferation, established larger populations in Rag1^−/− VAT, and attenuated diet-induced glucose intolerance and VAT insulin resistance in Rag1^−/− hosts. However, transfer of B-1b B cells unable to secrete IgM had no effect on glucose tolerance. In an obese human population, results provided the first evidence that B-1 cells are enriched in human VAT and IgM antibodies to OSEs inversely correlated with inflammation and insulin resistance.

Conclusions

Nab-producing B-1b B cells are increased in Id3^{Bcell KO} mice and attenuate adipose tissue inflammation and glucose intolerance in DIO mice. Additional findings are the first to identify VAT as a reservoir for human B-1 cells and to link anti-inflammatory IgM antibodies with reduced inflammation and improved metabolic phenotype in obese humans.

Introduction

Obesity is an epidemic in the Western world, and dramatically increasing rates in developing countries have made it one of the prominent global health concerns of the 21^st century. Obesity-induced visceral adipose tissue (VAT) inflammation leads to systemic insulin resistance and glucose intolerance¹. While macrophages and T cells have been implicated in this process^{2, 3}, emerging evidence suggests that B cells also modulate obesity-induced adipose tissue inflammation and insulin resistance.

B cells have been identified in murine and human adipose tissue^4-8, and localize to sites of inflammation⁹. Recently, IgG from diet-induced obese (DIO) mice was shown to drive adipose tissue inflammation and promote systemic insulin resistance⁴. These pathogenic IgG antibodies were localized in VAT and came from adaptive B-2 B cells – the major B cell subset that differentiate into memory B and plasma cells capable of producing class-switched, high affinity antibodies¹⁰. Importantly, IgM antibodies did not impair metabolic function in the same mice, suggesting an isotype-dependent effect.

B-1 B cells, a self-renewing innate B cell population that differ from B-2 cells in response to stimuli and antibody repertoire¹¹, produce a major portion of plasma IgM in uninfected mice¹². B-1 B cells are further divided into B-1a and B-1b subsets that have overlapping functions, but exhibit differences in activation and response to infection^13-16. Both subsets produce evolutionarily-conserved IgM antibodies, which are termed natural antibodies (NAbs) as they contain few non-templated additions (N-additions), few mutations, and often recognize epitopes expressed on both invading pathogens and dying or damaged self-cells^{17, 18}. In contrast to IgG, B-1-derived IgM NAbs have direct and indirect anti-inflammatory functions^{19, 20}, and are thought to protect in specific instances of diet-induced chronic inflammation^{20, 21}. As B-1 B cells are enriched in omental adipose tissue⁵, they may be important regulators of VAT function. Despite this, the roles for B-1 B cells and their IgM NAbs in mediating obesity-induced adipose tissue inflammation and insulin resistance are poorly understood.

While IgM NAbs similar to those derived from murine B-1 B cells are well documented in humans^{22, 23}, identifying a human B-1 B cell has been elusive. Recently however, a subset of circulating B cells in humans was shown to share several functional properties unique to murine B-1 B cells²⁴. However, these cells have yet to be identified outside of the circulation and it is unknown whether, like murine B-1 cells, these B cells are enriched in adipose tissue.

Id3 is a ubiquitously expressed dominant-negative transcription regulator that, along with its binding partners, mediates various stages of B cell development and function^{25, 26}. Mice globally null for Id3 have impaired antigen-specific antibody responses²⁵ and altered levels of circulating IgM^{27, 28}. Recent work from our group has shown a role for Id3 in B cell regulation of diet-induced chronic inflammation^{28, 29}. Additional studies using a mouse model of obesity showed that mice with global deletion of Id3 are protected against diet-induced VAT expansion³⁰. Together, these findings suggest Id3 may be a key factor that links B cell function and obesity. Previous studies of the role of Id3 in atherosclerosis have identified cell type-specific mechanisms whereby Id3 regulates disease pathology^{31, 32}, underscoring the importance of utilizing B cell-specific deletion of Id3 to define mechanisms of B cell regulation in DIO.

In this report, we use mice null for Id3 specifically in B cells (Id3^{Bcell KO}) to test whether B-1 B cells and IgM NAbs mediate the inflammatory and metabolic effects of DIO. We expand upon our murine results with analysis of IgM NAbs and adipose tissue B cells in patients undergoing bariatric surgery. Together, results demonstrate that B-1b B cells attenuate the metabolic effects of DIO in an IgM-dependent manner. Furthermore, we identify B-1 B cells in human VAT and provide evidence that specific IgM NAbs negatively associate with insulin resistance in an obese human population.

Materials and methods

Expanded materials and methods can be found in the online supplement.

Results

B cell specific deletion of Id3 attenuates glucose intolerance and VAT insulin resistance in DIO mice

To evaluate whether Id3 expression is important for B cell-mediated effects on DIO, Id3^{Bcell KO 31, 32} and WT littermates were fed either chow or high-fat diet (HFD) for 12 weeks. As expected, there was a marked increase in body and visceral depot weights in the HFD fed group compared to chow, yet there were no genotype-dependent differences in epididymal adipose tissue mass or body weight (Figure 1A). While B cell-specific loss of Id3 did not completely correct the glucose intolerance due to DIO, Id3^{Bcell KO} mice had significantly improved glucose clearance compared to littermate controls (Figure 1B). There were no genotype-dependent differences in systemic insulin resistance or serum free fatty acid (FFA) levels (Supplemental Figure IA). However, insulin sensitivity as measured by insulin-stimulated AKT phosphorylation was elevated in omental adipose tissue (Figure 1C). No differences were observed in skeletal muscle or liver (Supplemental Figure IB). Taken together, results suggest that B cells may contribute to tissue-specific effects that may improve metabolic function associated with DIO in an Id3-dependent manner.

Figure 1. Loss of Id3 in B cells attenuates glucose intolerance and adipose tissue insulin resistance in DIO mice.

(A) Body and epididymal adipose tissue weights, and (B) GTT analysis in Id3^{Bcell KO} and WT littermates fed standard chow (WT n=6; Id3^{Bcell KO} n=9) or a HFD (WT n=10; Id3^{Bcell KO} n=14). (C) Insulin-induced AKT phosphorylation in omental fat of WT and Id3^{Bcell KO} littermates fed a HFD (n=3). (D) DIO μMT mice received either an i.p. vehicle (V, n=6) saline injection or adoptive transfer of 10⁷ B-2 cells from DIO WT (n=7) or Id3^−/− (n=7) donors and were continued on HFD for six additional weeks. (E) Body and epididymal adipose tissue weights. (F) GTT at two (left panel, representative of two independent experiments) and six (right panel, composite of two independent experiments) weeks post-transfer. Error bars represent ± SEM. *p<0.05, #p<0.05 WT vs. V.

To test whether the attenuated glucose intolerance in the DIO mice stemmed from loss of Id3 in a B-2 cell, either 10⁷ splenic B-2 cells from HFD-primed⁴ WT or Id3^−/− donors, or vehicle control, were injected i.p. into DIO μMT hosts. As μMT mice contain a deletion in the mu heavy chain required for surface BCR expression and essential for both B-1 and B-2 development, these mice lack mature B cells³³. Recipient mice were continued on a HFD and tested for glucose tolerance at two and six weeks post-transfer (Figure 1D). No differences in body weight, fat mass (Figure 1E), or B cell engraftment (Supplemental Figure IC) were observed. While all three groups demonstrated similar glucose tolerance at week two (Figure 1F), mice receiving WT B-2 cells had significantly impaired glucose clearance compared to vehicle controls six weeks post-transfer (Figure 1F). This corroborates previous findings⁴ that B-2 cells impair glucose homeostasis. However, hosts receiving WT and Id3^−/− B-2 cells had nearly identical glucose clearance, providing evidence that improved glucose tolerance in Id3^{Bcell KO} mice is not due to loss of Id3 function in a B-2 B cell, and suggesting that other B cell subsets may also modulate HFD-induced glucose intolerance.

DIO Id3^{Bcell KO} mice have increased B-1b B cells, total IgM, and T15-IgM antibodies in adipose tissue

Immune cells within adipose tissue can impact glucose homeostasis in a subset-dependent manner^{2, 3}. Flow cytometry studies in epididymal fat from DIO Id3^{Bcell KO} mice revealed no differences in F4/80⁺CD206⁻CD11c⁺ M1 or F4/80⁺CD206⁺CD11c⁻ M2 macrophages³⁴ or total CD3ε⁺ T cells (Figure 2A). There was a trend toward an increase in B-2 cells, although this change did not reach statistical significance. In contrast, Id3^{Bcell KO} mice had significantly elevated numbers of B-1 B cells within epididymal fat compared to WT littermates (Figure 2A). Further analysis revealed a specific increase in epididymal B-1b, but not B-1a, B cells in DIO Id3^{B cell KO} mice (Figure 2B).

Figure 2. Adipose tissue-specific increases in B-1 cells, total IgM, and T15 natural IgM antibodies in Id3^{Bcell KO} mice with DIO.

(A) Epididymal adipose tissue F4/80⁺CD206⁻CD11c⁺ M1 and F4/80⁺CD206⁺CD11c⁻ M2 macrophages (left panel, WT n=6; Id3^{Bcell KO} n=6), CD3ε⁺ T cells (middle panel, WT n=8; Id3^{Bcell KO} n=11), and B220^mid/loCD19^hi B-1 and B220^hiCD19^mid/lo B-2 B cells (right panel, WT n=6; Id3^{Bcell KO} n=8). (B) Epididymal adipose tissue CD5⁺ B-1a and CD5⁻ B-1b B cells (n=4-6). (C) Total IgM by ELISA in epididymal adipose tissue and serum (n=6). (D) T15-specific IgM by ELISA in epididymal adipose tissue and serum (n=6). (E) Serum MDA-LDL specific IgM and LDL IgM IC (n=13-15). Error bars represent ± SEM. *p<0.05, **p<0.01, ***p<0.001.

Among the many known IgM antibodies produced by B-1 B cells, the best studied are members of the T15 NAb family that bind phosphocholine (PC)³⁵, an epitope present on pneumococcal cell membranes as well as on phospholipids of oxidized LDL and apoptotic cells³⁶. These antibodies (T15-IgM or PC-IgM) have anti-inflammatory capacities^{19, 20}, suggesting they may have important regulatory functions within VAT. Epididymal adipose tissue IgM (Figure 2C) and T15-IgM (Figure 2D) levels were elevated in DIO Id3^{Bcell KO} mice compared to WT. Additionally, circulating MDA-LDL IgM and LDL-IgM immune complexes (IC), both associated with reduced inflammation and reduced risk of myocardial infarction in a human population^{37, 38}, were also elevated or showed a trending increase in Id3^{B cell KO} mice (Figure 2E). We observed no variations in circulating or adipose tissue IgG antibodies, including B-2-derived pathogenic IgG2b and IgG2c isotypes (Supplemental Table I). Our findings demonstrate that loss of Id3 in B cells leads to increased B-1 B cell numbers and anti-inflammatory IgM antibodies.

Loss of Id3 in B cells leads to increased T15-IgM production in omental adipose tissue

B-1 B cells populate coelomic cavities and can be divided into CD5⁺ B-1a and CD5⁻ B-1b B cells. While both produce IgM antibodies¹¹, B-1a and B-1b B cells exhibit differences in activation and response to infection^{13-16, 18}, suggesting they may have distinct functions in adipose tissue. Baseline analysis of Id3^{B cell KO} mice revealed that, similar to DIO mice, loss of Id3 in B cells led to a specific increase in B-1b B cell numbers within epididymal (Figure 3A) and omental (Figure 3B) adipose tissue, as well as in peritoneal fluid (Supplemental Figure IIA). No B-1b B cell differences were observed in Id3^+/+CD19^Cre/+ mice, indicating that population changes were independent of CD19 haplo-insufficiency (Supplemental Figure IIA). Consistent with our initial findings, no differences in B-1a, B-1b, FO, or MZ B cell subsets were identified in the spleens of Id3^{Bcell KO} mice (Supplemental Figure IIB, data not shown).

Figure 3. Loss of Id3 in B cells leads to increased adipose tissue B-1b B cells and IgM secretion in chow-fed mice.

(A) Epididymal and (B) omental adipose tissue B-1a and B-1b B cells in Id3^{Bcell KO} (n=7-8) and WT (n=5) littermates. (C) Representative flow cytometry and 10× confocal microscopy images of murine epididymal (top panels) and omental (bottom panels) adipose tissue. Scale bar=200μm. (D) T15-specific IgM in supernatant of Id3^{Bcell KO} (n=3) and WT (n=4) omental adipose tissue cultures. Error bars represent ± SEM. *p<0.05, **p<0.01.

Omental fat refers to the fatty band attached to the greater curvature of the stomach³⁹. Murine and human omental adipose tissue is similar in structure⁴⁰ and contains B cell-rich clusters of immune cells termed “milky spots.” Enriched B cell numbers help distinguish milky spots from fat-associated lymphoid clusters (FALCS) found in the mesentery that contain high numbers of natural helper cells, but few B cells⁴¹. Consistent with previous reports⁵, substantially more B-1 B cells were found in omental fat than epididymal fat relative to tissue mass (Figure 3A & B), and only omental adipose tissue contained milky spot clusters that stained heavily for the B cell marker B220 (Figure 3C). In addition, omental fat cultured ex vivo from Id3^{Bcell KO} mice produced three-fold more T15-IgM than controls (Figure 3D), providing evidence that PC-specific IgM is produced in omental adipose tissue in proportion to the number of B-1b cells. Together, our findings indicate that Id3 is an important regulator of adipose tissue B-1b B cell population size, and that loss of Id3 leads to significantly more B-1b B cells, and subsequently elevated local T15-IgM NAb production.

Loss of Id3 promotes omental B-1b B cell survival

B-1 B cells survive better than B-2 B cells and have the ability to self-renew, allowing for self-governing population maintenance. To test whether Id3 regulates population maintenance in mature B-1b B cells, we adoptively transferred equivalent numbers of sorted B-1b B cells from WT or Id3^{Bcell KO} donors into B and T cell-deficient Rag1^−/− hosts. Rag1^−/− mice were used instead of μMT mice because B-1 B cells transferred into μMT mice experience T cell-mediated attack and do not survive⁴². Both WT and Id3^{B cell KO} B-1b B cells were identified in omental fat (Figure 4A) and peritoneal fluid (Supplemental Figure III) three weeks post-transfer. Id3^{B cell KO} B-1b B cells showed significantly greater engraftment compared to WT in both. No B-1b B cell engraftment was detected in the spleen (Supplemental Figure III), indicating compartment-specific repopulation. Together, our findings suggest that enhanced B-1b B cell numbers in Id3^{Bcell KO} mice is intrinsic to loss of Id3 in mature B-1b B cells.

Figure 4. Id3 regulates B-1b B cell survival.

(A) B-1b B cells recovered in omental fat of Rag1^−/− hosts three weeks after adoptive transfer of 80k WT or Id3^{Bcell KO} B-1b B cells (V (vehicle), n=6; WT, n=6 ; Id3^{Bcell KO}, n=5). (B) BrdU incorporation (left) and Annexin V staining (right) of B-1b B cells from WT and Id3^{Bcell KO} mice treated with vehicle (n=3-4) or 40μg LPS (n=5-7). Error bars represent ± SEM. #<0.05 vs V, $<0.05 WT vs Id3^{B cell KO} *p<0.05.

Population maintenance depends on a balance between cell proliferation and cell death, and increased cell numbers can result from defective regulation of either process. To test whether Id3 mediates proliferation or apoptosis in B-1b B cells independent of adoptive transfer, WT and Id3^{Bcell KO} mice were treated with LPS – a rapid B-1b activator⁴³. Mice were then injected with BrdU to label proliferating cells, and B-1b B cells in omental fat were analyzed for proliferation and apoptosis. LPS-activated B-1b B cells in omental adipose tissue displayed no Id3-dependent differences in BrdU incorporation (Figure 4B). However, loss of Id3 in B cells led to lower Annexin V staining (Figure 4B), indicating reduced apoptosis, and suggesting Id3 as an important mediator of B-1b B cell survival.

Loss of Id3 in B cells attenuates HFD-induced inflammation in omental adipose tissue

Omental fat of Id3^{Bcell KO} mice contains higher numbers of B-1b B cells and produces more anti-inflammatory T15-IgM NAbs. To determine if loss of Id3 in B cells attenuates HFD-induced adipose tissue inflammation, inflammatory cytokine secretion by omental adipose tissue was evaluated in DIO WT and Id3^{Bcell KO} mice. Indeed, while no differences were observed between chow-fed animals (data not shown), omental fat from DIO Id3^{Bcell KO} mice produced significantly less TNFα and IFNγ compared to diet-matched WT controls (Figure 5A). While recent studies have identified IL-10 as a mechanism by which B-1 B cells limit inflammation, we did not observe differential IL-10 secretion in adipose tissue of Id3^{B cell KO} DIO mice (Supplemental Figure IV).

Figure 5. Reduced adipose tissue inflammation in Id3^{B cell KO} DIO mice and decreased adipose tissue M1 macrophage inflammation after treatment with B-1b-conditioned media.

(A) TNFα (left) and IFNγ (right) ELISA analysis of supernatant of omental adipose tissue cultures from WT and Id3^{Bcell KO} DIO littermates (n=5-7). (B) LPS-induced TNFα (left) and MCP-1 (right) cytokine levels in adipose tissue M1 macrophages from WT and Id3^{Bcell KO} mice fed chow or HFD detected by flow cytometry (n=5-8). (C) LPS-induced intracellular TNFα (left) and MCP-1 (right) cytokine levels in adipose tissue M1 and M2 macrophages detected by flow cytometry. Cells from DIO mice were treated with control media or media conditioned with supernatant from activated WT B-1b B cells (B-1b media) (n=5). Error bars represent ± SEM. *p<0.05, #p<0.05 vs all groups.

DIO is characterized by infiltrating adipose tissue M1 macrophages that produce inflammatory cytokines such as TNFα and MCP-1. Therefore, we hypothesized that despite no difference in macrophage number (Figure 2A), pro-inflammatory cytokine production may be reduced in omental M1 macrophages of Id3^{B cell KO} DIO mice. Indeed, M1 macrophages from omental adipose tissue of Id3^{B cell KO} DIO mice lacked the DIO-induced TNFα and MCP-1 expression present in WT M1 macrophages (Figure 5B). We next sought to determine whether secreted B-1b B cell products can attenuate adipose tissue macrophage production of these cytokines. Omental depots from WT DIO mice were treated with conditioned media from control and activated B-1b B cells, and flow cytometry was used to measure LPS-induced TNFα and MCP1 production in adipose tissue macrophages. Results demonstrated that secreted factors from activated B-1b B cells blunted TNFα and MCP-1 production in M1, but not M2, macrophages (Figure 5C). Together, these findings provide evidence that B-1b B cells can play a regulatory role in reducing adipose tissue inflammation.

B-1b B cells lacking Id3, but not the ability to secrete IgM, attenuate diet-induced glucose intolerance

Findings in the Id3^{Bcell KO} mouse suggest that improved B-1b B cell survival leads to increased B-1b B cell number and elevated local IgM production that may protect against downstream metabolic dysfunction. To test this, FACS-sorted B-1b B cells from WT, Id3^{Bcell KO}, or sIgM^−/− donors were adoptively transferred into Rag1^−/− hosts that were then placed on a HFD for 12 weeks (Figure 6A). While B cells from sIgM^−/− mice express surface IgM and secrete IgG, they cannot secrete IgM⁴⁴. Total plasma IgM, LDL IgM IC, and MDA-LDL IgM levels were tested at 2 and 12 weeks after diet initiation (Supplemental Figure V). Antibody levels increased with time, indicating continued B-1b survival and active IgM production through the duration of the experiment. Mice receiving Id3^{B cell KO} B-1b B cells showed elevated total IgM and LDL IgM-IC, but similar levels of MDA-LDL IgM compared to WT. Interestingly, B-2 adoptive transfer hosts showed very low antibody levels, providing evidence that B-1b B cells are major IgM producers. While no body weight differences were observed (Figure 6B), hosts that received Id3^{Bcell KO} B-1b B cells had improved glucose tolerance compared to vehicle controls, and WT hosts showed a trend toward improved glucose clearance (Figure 6C). No differences were seen between hosts that received B-1b B cells from sIgM^−/− and vehicle controls (Figure 6C). Additional analysis indicated significantly improved insulin signaling in omental adipose tissue from hosts that received B-1b B cells from Id3^{B cell KO} donors compared to vehicle controls (Figure 6D). These results suggest a role for B-1b B cells in attenuating the metabolic consequences of obesity that is enhanced in the absence of Id3 and is dependent on the ability of the B-1b B cell to secrete IgM NAbs.

Figure 6. B-1b B cells lacking Id3 attenuate diet-induced glucose intolerance, while B-1b B cells unable to secrete IgM have no effect.

(A) Rag1^−/− mice received either an i.p. vehicle (V, n=5) saline injection or adoptive transfer of 8.0×10⁴ B-1b B cells from WT (n=4), Id3^{B cell KO} (n=5), or sIgM^−/− (n=3) donors. After a recovery week, hosts were placed on a HFD for 12 weeks. (B) Body weight, (C) GTT, and (D) omental insulin signaling of DIO Rag1^−/− hosts. #p<0.05 Id3^{B cell KO} vs V, *p<0.05.

PC-IgM NAbs in circulation and within omental adipose tissue of humans negatively correlate with MCP-1

PC-recognizing IgM NAbs derived from B-1 B cells have anti-inflammatory characteristics^{19, 20}. To determine if PC-IgM was associated with a serum marker of inflammation, we measured circulating levels of MCP-1 in a cohort of bariatric surgery patients (cohort 1; 122 patients enrolled between May, 2009 and August, 2010). The baseline clinical characteristics of this cohort are shown in Supplemental Table II. Circulating MCP-1 – a macrophage chemoattractant protein known to be highly predictive of insulin resistance⁴⁵ – inversely associated with PC-IgM in both the circulation and within omental adipose tissue (Supplemental Table III). Our studies provide evidence that PC-IgM NAbs are present in human omental adipose tissue and are associated with reduced MCP-1.

IgM NAbs and apoB-immune complexes correlate with HDL levels and improved LP-IR scores

Several studies have revealed that in addition to PC-IgM, IgM NAbs directed at other oxidation-specific epitopes associate with reduced inflammation and decreased risk of death, myocardial infarction, and stroke in prospectively followed subjects from the general community^{37, 38} To determine if other IgM NAbs were associated with metabolic dysfunction, we scored each patient in cohort 1 from 0 (most insulin sensitive) to 100 (most insulin resistant) using an NMR-based Lipoprotein Insulin Resistance (LP-IR) panel. Because NMR has the capacity to measure both lipoprotein size and particle concentration, LP-IR accounts for parameters missed by normal lipid panels and is highly predictive of progressive insulin resistance after multivariate analysis^{46, 47}. In contrast to methods used for measuring beta cell function such as HOMA-IR, LP-IR measurements are not affected by acute changes in glucose or insulin. Thus, LP-IR is reflective of long-term metabolic function and better reflects overall metabolic health. While there was no correlation with LP-IR scores and circulating PC-IgM in this population (data not shown), LP-IR had a significant negative association with IgM apoB immune-complexes (IgM-IC) and a trending inverse correlation with IgM antibodies to MDA-LDL (Table 1). Both IgM-IC and IgM MDA-LDL were positively associated with HDL. In contrast to IgM, no associations were observed with IgG-IC or IgG MDA-LDL. Together, our results provide evidence that IgM NAbs directed at oxidation-specific epitopes may be associated with protective phenotypes in an obese human population, suggesting a role for B-1 B cells in regulating metabolic syndrome in humans.

Table 1.

IgM autoantibodies and apoB-immune complexes correlate with HDL levels and improved LP-IR scores in obese humans (cohort 1).

Measurement	LP-IR		HDL
	Spearman coefficient	p-value	Spearman coefficient	p-value
IgM-IC	−0.24	0.01	0.32	0.0005
IgG-IC	−0.002	0.98	0.08	0.35
IgM MDA-LDL	−0.15	0.12	0.19	0.04
IgG MDA-LDL	−0.08	0.38	0.02	0.85

Human omental adipose tissue contains B-1-like cells and PC-IgM

Recently, Griffin et al. identified a subset of circulating B cells (CD20⁺CD27⁺CD43⁺) in humans that share several unique properties of murine B-1 B cells²⁴. To determine if this subset of B-1 B cells are found in human omentum, we performed flow cytometry on sections of omental fat harvested from a second smaller group of patients undergoing bariatric surgery (cohort 2; 16 patients enrolled between October, 2012 and October, 2013). Interestingly, a marked enrichment of B-1 cells was found in omental fat compared to subcutaneous fat or blood in four of 16 patients analyzed. Figureure 7A is a representative flow cytometry plot of the patients whose samples displayed marked enrichment. B-1 cells in omental adipose tissue directly correlated with plasma PC-binding IgM levels (Figure 7B). Together, these findings show for the first time that B-1 B cells are enriched in human omental adipose tissue and the percentage of B-1 cells directly correlates with circulating IgM to PC.

Figure 7. B-1 B cells in human omental adipose tissue correlate with serum PC-IgM levels.

(A) Flow cytometry on omental fat, subcutaneous (SC) fat, and blood from a single patient. (B) Fraction of omental B-1 (CD20⁺CD27⁺CD43⁺) B cells plotted against serum PC-IgM levels in patients from cohort 2 (n=16). Solid line represents Spearman correlation and dashed lines represent 95% confidence intervals.

Discussion

Recent studies have demonstrated subset-specific roles for B cells in both aggravating and regulating the perturbations associated with obesity. Winer and colleagues demonstrated that B-2 B cells promote insulin resistance and glucose intolerance through production of pathogenic antibodies⁴. Additional studies have shown that sorted splenic B-2 B cells from obese mice secrete more IL-6 and MIP-2 and less IL-10 compared to lean controls⁴⁸, and circulating B cells (of which the vast majority are B-2) from T2D patients are also skewed toward a pro-inflammatory phenotype after TLR stimulation⁴⁹. Our adoptive transfers of B-2 B cells provide further evidence that B-2 B cells promote metabolic dysfunction during DIO. In contrast, B-1a B cells were recently shown to protect against the inflammatory and metabolic consequences of DIO through a combination of IgM and IL-10 production⁵⁰, and B cell-derived IL-10 reduces adipose tissue inflammation and insulin resistance caused by DIO^{51, 52}. Our findings add to these studies and indicate for the first time a role for B-1b-derived IgM NAbs in attenuating adipose tissue inflammation and metabolic dysfunction in obesity.

Although CD5⁻ B-1b and CD5⁺ B-1a B cells are sister populations, they have distinct developmental origins⁵³ and functional properties¹⁶. While it has long been recognized that B-1a B cells spontaneously produce IgM NAbs against oxidation specific epitopes generated during inflammation and cell death, such as E06/T15 against PC and NA17/E014 against MDA²², the role of B-1b B cells in producing these important antibodies and their impact in DIO is poorly understood. We found Id3^{B cell KO} mice to display an elevated B-1b population at baseline that persisted during obesity progression. Interestingly, we only observed inflammatory and metabolic differences following DIO. It is possible that B-1b B cells function to limit inflammatory processes that become exaggerated in obesity. This explanation fits with our findings that B-1b B cells specifically attenuated M1 macrophage inflammation, as M1 macrophages are rare in lean adipose tissue.

In addition to attenuated obesity-induced inflammation, Id3^{B cell KO} mice showed improved insulin-stimulated pAKT signaling specifically in omental fat, but not in skeletal muscle or liver. As skeletal muscle is the main tissue that accounts for insulin-stimulated glucose disposal⁵⁴, Id3-mediated B cell effects on insulin signaling in adipose alone may not be sufficient to result in systemic effects. Indeed, we did not observe differences in systemic insulin resistance as measured by ITT in Id3^{B cell KO} mice.

B-1-derived IgM NAbs can reduce inflammation and promote apoptotic cell clearance^{17, 22, 55}; processes that become dysregulated during the progression of obesity^{39, 56}. However, it is currently unclear whether specific IgM antibodies play important roles in mediating the effects of obesity. The well-studied T15 family of PC-IgM NAbs have known anti-inflammatory properties¹⁹ and are thought to prevent macrophage lipid uptake and apoptotic cell accumulation in models of atherosclerosis^{21, 57}. However, Shen et al. reported no metabolic effect when PC-IgM was infused into DIO μMT mice⁵⁰. Interestingly, a partial loss of B-1a B cell attenuation of glucose intolerance was observed when the B-1a B cells could not secrete IgM⁵⁰. While these seemingly disparate findings could be explained by insufficient amount or localization of the infused antibody, it is likely that other IgM antibodies are needed for the full effect on glucose handling. Notably, our results in mice and humans support this hypothesis as we demonstrated complete loss of the protective effect provided by B-1b B cells when they were unable to secrete IgM. No associations with PC-IgM and insulin sensitivity were found in obese humans, but trending and significant associations between insulin sensitivity and IgM antibodies and immune complexes to other oxidation-specific epitopes were identified. It is of interest that it has recently been shown that IgM to both MDA-LDL and OxLDL epitopes display a remarkably high heritability, in the range of h(2) – 0.6-0.8, suggesting that there may be a strong genetic component to these “anti-inflammatory” innate responses, which may have confounded these association studies⁵⁸. Together these results suggest that certain IgM clones may play differing roles in mediating inflammation and metabolic dysfunction in obesity. Identifying these associations is an important first step toward understanding the role IgM antibodies play in obesity-associated metabolic disease.

Much of the knowledge about immune regulation of disease has come from murine studies, and in many cases, the relevance to humans remains to be deciphered. While immunohistochemical analysis has demonstrated B cells within human omental fat^{7, 8}, the lack of a known human B-1 cell equivalent, along with the multiple markers needed to identify subsets and the challenges of performing flow cytometry on human omental adipose tissue had made subset analysis difficult. Recently, a circulating human CD20⁺CD27⁺CD43⁺ B cell with several characteristics similar to murine B-1 B cells was identified²⁴. Here, we demonstrate for the first time using multicolor flow cytometry of human omental fat, that like murine B-1 B cells, CD20⁺CD27⁺CD43⁺ B cells are not only present in omental fat, but in some patients were markedly enriched compared to subcutaneous adipose tissue and blood. Not all omental adipose tissue samples demonstrated this marked enrichment, possibly due to uneven clustering and distribution of milky spots within individual depots, variable sampling at the time of surgery, or because of differences in individual patients, such as the genetic regulation of B-1 cell Nab production implied by the heritability of IgM to oxidation specific epitopes noted above. Nonetheless, the percentage of omental B cells that were CD20⁺CD27⁺CD43⁺ B cells correlated with circulating PC-IgM levels. Our findings add to the known similarities between murine B-1 B cells and human CD20⁺CD27⁺CD43⁺ B cells, and indicate that further analysis of omental adipose tissue may improve our understanding of the role B-1 B cells play in metabolic regulation.

In summary, the present study provides the first evidence that B-1b B cells and the anti-inflammatory IgM NAbs they produce attenuate diet-induced glucose intolerance in mice. Moreover, our findings are the first to identify CD20⁺CD27⁺CD43⁺ B cells and PC-IgM within human omental adipose tissue and show an inverse correlation of specific IgM antibodies with MCP-1 levels and insulin resistance. While much work remains to be done, these results provide evidence that B-1 B cells and their IgM NAbs may be important modulators of inflammation and insulin resistance in human obesity.

Significance.

While immuno-regulation has emerged as a critical component of the metabolic dysfunction associated with obesity, functional roles for specific immune subsets in this context remain unclear. In this study, we use a mouse null for the transcription factor Id3 specifically in in B cells to demonstrate a role for innate B-1b B cells and natural IgM antibodies in regulating adipose tissue inflammation, insulin resistance, and systemic glucose intolerance in a mouse model of diet-induced obesity. Furthermore, we identified a recently-discovered human B-1 B cell in visceral adipose tissue of patients undergoing bariatric surgery, and found an inverse association between IgM antibodies and insulin resistance. Results from this study are the first to link B-1b B cells and their IgM antibodies with improved metabolic function and further expand our understanding of the complex interactions between the immune system and metabolic function.

Non-standard Abbreviations and Acronyms

VAT

visceral adipose tissue

DIO

diet-induced obesity

NAbs

natural antibodies

PC

phosphocholine

Id3^{Bcell KO}

B cell-specific Id3 knockout mouse

HFD

high-fat diet

MDA-LDL

malondialdehyde-LDL

LP-IR

lipoprotein insulin resistance