B-1b Cells Secrete Atheroprotective IgM and Attenuate Atherosclerosis

Abstract

Rationale

B cells contribute to atherosclerosis through subset specific mechanisms. Whereas some controversy exists about the role of B-2 cells, B-1a cells are atheroprotective due to secretion of atheroprotective IgM antibodies independent of antigen. B-1b cells, a unique subset of B-1 cells that respond specifically to T cell-independent antigens, have not been studied within the context of atherosclerosis.

Objective

To determine whether B-1b cells produce atheroprotective IgM antibodies and function to protect against diet induced atherosclerosis.

Methods and Results

We demonstrate that B-1b cells are sufficient to produce IgM antibodies against oxidation specific epitopes (OSE) on LDL both in vitro and in vivo. Additionally, we demonstrate that B-1b cells provide atheroprotection after adoptive transfer into B and T cell deficient (Rag1^−/−Apoe^−/−) hosts. We implicate Id3 in the regulation of B-1b cells as B cell-specific Id3 knockout mice (Id3^BKOApoe^−/−) have increased numbers of B-1b cells systemically, increased titers of OSE-reactive IgM antibodies, and significantly reduced diet-induced atherosclerosis compared to Id3^WTApoe^−/− controls. Finally, we report that the presence of a homozygous SNP in ID3 in humans that attenuates Id3 function is associated with an increased percentage of circulating B-1 cells and anti-MDA-LDL IgM suggesting clinical relevance.

Conclusions

These results provide novel evidence that B-1b cells produce atheroprotective OSE-reactive IgM antibodies and protect against atherosclerosis in mice, and suggest that similar mechanisms may occur in humans.

INTRODUCTION

Murine studies provide clear evidence that B cells regulate atherosclerosis^{1, 2} and that this effect is subset dependent. While much evidence supports a pro-atherogenic role of B-2 cells, there are experimental data supporting a protective role as well^{1, 3–7}. Alternatively, there is near uniform agreement of an atheroprotective role for B-1a cells ^8–11.

B-1 and B-2 cells are developmentally and functionally distinct subsets in mice¹². B-1 cells produce the majority of total circulating IgM¹³, which they secrete in a T cell independent manner¹⁴. B-1 cells are further subclassified by the surface expression of CD5; B-1a cells being CD5⁺ and B-1b cells being CD5⁻. B-1a cells produce so-called germ line encoded IgM that harbor few non-templated insertions, termed natural antibodies (NAb), in response to non-antigenic activation. A substantial proportion of these antibodies bind oxidation-specific epitopes (OSE), as found on oxidized LDL (OxLDL), as well as apoptotic cells, and reduce the development of atherosclerosis^15–17. In contrast, B-1b cells are activated by both non-antigenic and antigen-dependent stimuli, producing antigen specific IgM, and to a smaller extent IgG3, and subsequently become TI memory B cells^{18, 19}. Whether B-1b cells can secrete atheroprotective OSE reactive IgM antibodies or provide protection against atherosclerosis remains undetermined.

Previous work in our lab has demonstrated that global deletion of inhibitor of differentiation 3 (Id3) resulted in early and significantly increased atherosclerosis⁷. Id3 is a member of the helix-loop-helix transcription factor family known to be important in lymphocyte function²⁰. These global Id3 knockout mice developed equivalent numbers of B-2 cells, but reduced numbers of atheroprotective B-1a cells compared to wild-type²¹. However, this was not phenocopied by B cell-specific deletion of Id3, providing evidence that Id3 can regulate B cell subsets and atherosclerosis through effects in non-B cells. Yet the effect of B cell-specific deletion of Id3 on B-1b cells has not been reported.

A putative human equivalent to murine B-1 cells was described by Rothstein and colleagues as CD20⁺CD3⁻ B cells that are CD27⁺CD43^{+ 22}. They demonstrated that these cells functionally resemble murine B-1 cells in that they spontaneously secrete IgM, are enriched in umbilical cord blood, stimulate T cells and have tonic intracellular signaling. Our lab previously published that a non-synonymous single nucleotide polymorphism (SNP) in the coding region of the ID3 gene (rs11574 G->A) causes an amino acid change (A->T) and this results in decreased Id3 binding to E proteins ²³. Whether individuals harboring the homozygous SNP have a modified proportion of B-1 cells or OSE reactive IgM is unknown.

In the present study, we addressed the role of B-1b cells in atherosclerosis by demonstrating that B-1b cells are sufficient to produce atheroprotective antibodies reactive to OSE and that transferring B-1b cells into B and T cell deficient hosts attenuates atherosclerosis. Furthermore, we report that B cell-specific Id3 knockout (Id3^BKOApoe^−/−) mice develop a systemic increase in B-1b cells, increased titers of atheroprotective, OSE reactive IgM antibodies, and attenuated atherosclerosis. Finally, we report that a cohort of patients bearing the rs11574 SNP within ID3 have an increased proportion of circulating B-1 cells and OSE reactive IgM, suggesting possible clinical relevance for our murine findings in humans.

METHODS

Mice

All animal protocols were approved by the Animal Care and Use Committee at the University of Virginia. Id3^fl/fl mice were a generous gift of Dr. Yuan Zhang (Duke University). CD19^Cre/+ mice and Rag1^−/− mice were provided by Dr. Timothy Bender (University of Virginia). Apoe^−/− mice were purchased from Jackson Laboratory. All mice, purchased or generated, were backcrossed at least 10 generations to C57BL/6J mice.

Serum cholesterol determination

Cholesterol levels were determined as previously described by the University of Virginia Medical laboratories ⁷.

Analysis of atherosclerotic lesions

Hearts and aortas were removed and prepared as previously described ^{7, 9}. Briefly, hearts were embedded in OCT compound (Tissue-Tek) and snap frozen. Serial 5μm sections were cut by Cryostat (Leica biosystems) from the beginning of the three aortic leaflets to the aortic arch. Aortas were fixed in 4% paraformaldehyde then opened longitudinally, pinned, and stained using Sudan IV (Sigma). Plaque areas were assessed using Image-Pro Plus software (Media Cybernetics). For aortic sinus measurements, maximum plaque area was used for comparison. For aortic plaque measurements, the percentage of positive staining was used for comparison.

Immunofluorescence and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) analysis of aortic sinus sections

Slides of the aortic sinus, described above, were stained for macrophage content as previously described ⁷ using biotinylated Mac-2 as the primary antibody (Cedarlene CL8942B) and Streptavidin Alexa Fluor 488 as the secondary antibody (Invitrogen Molecular Probes S11223) then counterstained with DAPI and mounted (Vectashield H-1500). For staining of apoptotic cell bodies the TUNEL method was used following the protocol from ApopTag Peroxidase In Situ Apoptosis Detection Kit (emdMillipore S7100). Imaging for both was done using an Olympus BX51 high magnification light microscope. Images were analyzed using ImageJ (http://imagej.nih.gov/ij/).

Adoptive transfer of B-1b and B-1a cells into Rag1^−/−Apoe^−/− hosts

Following fluorescence-activated cell sorting (FACS) from the PerC of Apoe^−/− mice, 1x10⁵ B-1b or B-1a cells were transferred interperitoneally (IP) into 8-week-old, male, Rag1^−/−Apoe^−/− mice. Mice were maintained on chow diet for one to four weeks following transfer then switched to Western diet for 16 weeks at the end of which time the animals were euthanized and hearts were collected for histological analysis of atherosclerotic lesions as described above.

Immunization

Mice were immunized with DNP-KLH as described previously²⁴. Briefly, 100μg DNP-KLH in complete Freund’s adjuvant were injected IP. 21 days later mice were boosted with DNP-KLH in PBS. Blood was collected at days 0, 7, 21, and 28. Bone marrow was collected at time of euthanization.

Preparation of tissues for flow cytometry and cell sorting

PerC cells, splenocytes, PBMCs and bone marrow (BM) cells were harvested and processed for flow cytometry as previously described^{7, 25}.

In vitro stimulation assays

Post electronic cell sorting, Apoe^−/− B-1b and B-1a cells were plated at 1x10⁴ cells per well in a 96 well plate in 200 μl of B cell culture media with 50μg/mL lipopolysaccharide (LPS) (Sigma, L4391)or PBS for 72 hours. The supernatant was collected for measurement of immunoglobulins by ELISA.

Enzyme-linked Immunosorbent Assay (ELISA)

Specific Ab levels to given antigens in plasma from mice or humans were determined by chemiluminescent ELISA as previously described ^{10, 24, 26, 27}. Data were expressed either as μg/ml, based on standard curves of isotype standards (Online Figure I), or relative light units (RLU/100ms) where quantitative standards were unavailable. For these epitopes, dilution curves are shown in the supplement (Online Figure I).

ELISPOT

BM cells were added to sterile MultiScreen IP-Plates (Millipore, MSIPS4510) according to manufacturer’s protocol after coating wells with unlabelled anti-mouse IgM antibody (Southern Biotech). IgM producing cells were visualized using streptavidin alkaline phosphatase (Abcam) followed with BCIP/NBT. Each spot represented an IgM secreting cell.

Human genotyping

Id3 SNP (rs11574, Assay ID# C_2462609_10) genotyping was performed using the ABI Taqman SNP Genotyping assay from LifeTechnologies. All genotypes were analyzed and assigned automatically using the ABI SDS 2.3 software.

Statistical methods

Data was analyzed using Prism 6.0b (GraphPad Software, Inc.). Results are displayed containing all replicated experiments and values shown are mean ± SEM.

RESULTS

B-1b cells secrete MDA-LDL and CuOx-LDL IgM in response to TLR stimulation

B-1 cell-mediated atheroprotection has been strongly linked to the production of atheroprotective IgM⁹. In particular, IgM that bind OSE on OxLDL, such as anti-MDA-LDL and anti-CuOx-LDL, are considered to be atheroprotective ²⁸. To test whether B-1b cells have the capacity to secrete OSE targeted IgM and how that compares to B-1a cells, B-1b cells (CD19⁺B220^+/−IgM^hiCD23⁻CD5⁻) and B-1a cells (CD19⁺B220^+/−IgM^hiCD23⁻CD5⁺) were sorted at greater than 99% purity from PerC of apolipoprotein E deficient (Apoe^−/−) mice and placed in cell culture with the TLR4 ligand LPS, which has been shown to stimulate antibody secretion from B-1 cells²⁹ (Figure 1A). Absolute concentration of IgM and relative concentration of anti-MDA-LDL and anti-CuOx-LDL IgM were measured using established chemiluminescent ELISA^{10, 26} and calculated against IgM standard and pooled dilution series respectively (Online Figure IA). The data demonstrate that stimulation induced a robust IgM response from both B-1b and B-1a cells (Figure 1B). Additionally, IgM reactive to MDA-LDL and CuOx-LDL was greatly increased for both B-1b and B-1a cells compared to unstimulated controls, though cultured B-1b cells secreted lower titers compared to B-1a cells (Figure 1C&D). Titers of IgG3 were found to be unmeasurable in the culture supernatant (data not included). These data reveal that B-1b cells produce OSE-reactive IgM, though at lower titers than B-1a cells in culture.

Figure 1. B-1b cells produce IgM reactive to MDA-LDL and CuOx-LDL in vitro and in vivo.

A) B-1b (CD19⁺B220^+/−IgM^hiCD23⁻CD5⁻) and B-1a (CD19⁺B220^+/−IgM^hiCD23⁻CD5⁺) cells were sorted from PerC of Apoe^−/− mice and placed in culture then stimulated with LPS to induce antibody secretion. Culture media was collected after 72 hours and total IgM, anti-MDA-LDL IgM and anti-CuOx-LDL IgM concentration was determined by ELISA. B–D) Quantification of total IgM (μg/ml) (B), anti-MDA-LDL IgM (RLU/100ms) (C) and anti-CuOx-LDL IgM (RLU/100ms) (D). E) B-1b and B-1a cells were sorted from PerC of Apoe^−/− mice and transferred IP into Rag1^−/−Apoe^−/− mice. Plasma was collected 4 weeks later and measured for total IgM and OSE reactive IgMs by ELISA. F–H) Quantification of total IgM (F), anti-MDA-LDL IgM (G), and CuOx-LDL IgM (H). Data are mean ± SEM. ND = not detectable

B-1b cells secrete measurable titers of OSE reactive IgM in vivo

To determine if B-1b cells secrete OSE reactive antibodies in vivo, PerC B-1b and B-1a cells were sorted from Apoe^−/− mice in the same manner as for in vitro testing. Recombination activation gene 1 deficient (Rag1^−/−) mice, which are devoid of T and B cells, crossed with Apoe^−/− mice (Rag1^−/−Apoe^−/−) were injected with equal numbers (1x10^5) of purified B-1b or B-1a cells, or PBS control (Figure 1E). Blood was collected four weeks after transfer and absolute titers of IgM and relative titers of IgM reactive to MDA-LDL and CuOx-LDL were calculated against IgM standard and pooled dilution series respectively (Online Figure IB). As expected, PBS-injected Rag1^−/−Apoe^−/− mice had undetectable titers of IgM while both B-1b and B-1a injected mice produced measurable titers of IgM and OSE-reactive IgM. Notably, in contrast to in vitro findings, B-1b recipient mice had higher titers of total IgM and MDA-LDL IgM than B-1a recipient mice (Figures 1F–G). Titers of CuOx-LDL IgM were equivalent compared to B-1a recipients (Figure 1H).

B-1b cells confer atheroprotection to Rag1^−/−Apoe^−/− recipients

To determine whether B-1b cells function in an atheroprotective manner, B-1b cells were sorted as described above and transferred into Rag1^−/−Apoe^−/− hosts. The recipient mice were then fed a Western diet for 16 weeks. Weight and lipid analysis demonstrated that B-1b cell adoptive transfer did not significantly modify weight gain, or plasma total cholesterol, trigylcerides, or HDL cholesterol levels compared to PBS controls (Online Table I). Flow cytometry was used to demonstrate that the transferred B-1b cells populated the recipient mice and that their immunophenotype did not change over the course of the 16 week diet as the transferred cells did not upregulate CD23 or CD5 on their cell surface (Figure 2B&C, Online Figure IIA&B). An average of 3.6x10⁴±0.7 CD19⁺ B cells were recovered from the PerC of the recipient mice (n=9) of which <1% were CD23⁺ and 3.7% were CD5⁺, though with lower CD5 expression than on B-1a cells from Apoe^−/− mice. B-1b cells were also detected in the spleen, although in fewer numbers, while very few transferred cells were found in the blood (Online Figure II). Plasma was collected from recipient mice and absolute titers of IgM, IgG1, IgG2c, and IgG3 were measured by ELISA and calculated in comparison to immunoglobulin standards (Online Figure IC). High titers of IgM were measured in host mice that received B-1b cells with very low levels of IgG2c and IgG3, whereas IgG1 was not detectable (Figure 2D). Additionally, relative titers of IgM reactive to native LDL (nLDL), MDA-LDL, CuOx-LDL, α-1,3-dextran, and E06/T15, (the NAb to the phosphorylcholine (PC) moiety of oxidized phospholipid (OxPL)¹⁵) were measured based on pooled dilution series (Online Figure ID). The data demonstrate that the B-1b recipient mice maintained high titers of IgM antibodies reactive to MDA-LDL and CuOx-LDL as well as low titers of IgM reactive to E06/T15 (Figure 2D). Very low levels of IgM to nLDL were noted, which may reflect binding to spontaneously generated oxidative epitopes on LDL plated on the microtiter wells (Figure 2D) Importantly, levels of IgM reactive to α-1,3-dextran, which is a classic TI bacterial surface antigen that was used as a non-OSE type antigen control, were not measurable (Figure 2D).

Figure 2. B-1b cells produce atheroprotective IgM antibodies and confer atheroprotection to Rag1^−/−Apoe^−/− recipients.

A) PerC B-1b cells were sorted, as described in Figure 1A, from Apoe^−/− mice and 1x10^5 cells (n=9) or PBS (n=6) were injected IP into 8-week-old chow fed, male, Rag1^−/−Apoe^−/− recipients which were then fed a Western diet for 16 weeks. PerC B cells were recovered for flow cytometry, plasma was collected for antibody measurements and aortic sinuses collected for lesion analysis. B–C) Representative flow cytometry plots and quantification of recovered B-1b cells from PerC of Rag1^−/−Apoe^−/− hosts after 16 weeks of Western diet feeding. D) Absolute titers of IgM, IgG1, IgG2c and IgG3, quantified in μg/ml, and relative titers of IgM antibodies reactive against native LDL (nLDL), AB1-2 (T15/E06 anti-idiotype), MDA-LDL, CuOx-LDL, and α-1,3-Dextran reported as RLU/100msE) Representative images of maximum lesion area within the aortic sinus as determined by Oil Red O staining from 10x magnification images. F) Quantification of maximum lesion area from aortic sinus sections. Data are mean±SEM, ND = Not detectable, unpaired, two-tailed, Students T-test was used to compare B-1b cell numbers, two-tailed Mann-Whitney test was used to compare lesion areas due to unequal distribution of variance. *p<0.05 **p<0.01

To compare atherosclerotic plaque size, aortic sinuses were collected for histochemical measurement of cross sectional plaque area demonstrating that B-1b recipient mice exhibited significantly reduced atherosclerosis compared to PBS recipient controls (2.05x10⁵ ± 0.26x10⁵ μm² vs 3.16x10⁵± 0.51x10⁵ μm², p<0.05, Figure 2E&F). Comparison of plaque size to total cholesterol demonstrated that there was no correlation for these values (Online Figure IIC). These data, taken together, provide evidence that B-1b cells produce atheroprotective IgM and attenuate atherosclerosis independent of T cells or other B cells.

B cell-specific Id3 deficiency increases B-1b cell numbers in peritoneum, spleen, and blood

We had previously implicated Id3 in B cell-mediated atheroprotection utilizing adoptive transfer of CD43⁻ splenocytes into uMTApoe^−/− mice⁷ as no mouse with B cell-specific deletion of Id3 was available at that time. To directly assess the loss of Id3 specifically in B cells in an atherogenic Apoe^−/− mouse, we generated B cell-specific Id3 knockout mice (Id3^BKO) by crossing Id3^fl/fl mice with CD19^cre/+ mice and then bred these with Apoe^−/− mice (Id3^BKOApoe^−/−) and confirmed the deletion with Western blot analysis as previously reported²¹. PerC, spleen, and blood were collected from 8-week-old, male, Id3^BKOApoe^−/− and littermate control (Id3^WTApoe^−/−) mice and B-2, B-1a, and B-1b cell subsets quantified by flow cytometry. Representative flow cytometry with our gating strategy to differentiate these subsets is depicted in Figure 3A. Id3^BKOApoe^−/− mice contained significantly greater numbers of B-1b cells compared to Id3^WTApoe^−/− controls. No differences in the number of B-1a or B-2 cells were detected (Figure 3B–D). Importantly, the comparable numbers of B-1a cells suggest that deletion of Id3 did not result in down regulation of CD5 expression. Additionally, CD4 and CD8 T cell numbers were not modified (Online Figure III).

Figure 3. B cell-specific deletion of Id3 results in increased B-1b cell numbers in PerC, spleen, and blood.

A) Representative flow cytometry plots subsetting B-2 (CD19⁺B220^high), B-1a (CD19^highB220^mid/lowIgM^highCD5⁺), and B-1b cells (CD19^highB220^mid/lowIgM^highCD5⁻) from PerC of Id3^WTApoe^−/− and Id3^BKOApoe^−/− mice.. B–D) Quantification of B cell subsets from chow fed, 8-week-old male Id3^WTApoe^−/− (n=12) and Id3^BKOApoe^−/− mice (n=18) from PerC (B), spleen (C), and blood (D). Data are presented as the mean±SEM. Unpaired two-tailed Students T-test was used to compare B cell subsets from PerC, two-tailed Mann-whitney test was used to compare spleen and blood samples after D’Agostino & Pearson normality test revealed non-parametric distribution. **p<0.01, ***p<0.001, ****p<0.0001

Id3^BKOApoe^−/− mice have increased BM IgM producing cells and B-1b cells

It was recently reported that a large proportion of IgM producing B-1 cells are found in the BM of mice ²⁵. To determine whether Id3^BKOApoe^−/− mice had increased numbers of IgM producing cells in their BM, total BM cells were extracted from leg bones of 8-week-old mice and placed into ELISPOT wells (Figure 4A). Id3^BKOApoe^−/− mice had significantly more IgM⁺ immunoglobulin secreting cells (ISC) in their BM than Id3^WTApoe^−/− mice (Figure 4B&C). A portion of the BM tissue was also analyzed by flow cytometry using the gating strategy published by Baumgarth and colleagues (Online Figure IV, ²⁵). BM from Id3^BKOApoe^−/− mice contained significantly more total B-1 cells compared to Id3^WTApoe^−/− mice and the difference was entirely attributable to increased B-1b cells (Figure 4D). Finally, it was determined that Id3^BKOApoe^−/− mice of the same age had significantly higher titers of IgM compared to Id3^WTApoe^−/− mice (Figure 4E). These findings demonstrate that the increased B-1b population of Id3^BKOApoe^−/− mice is also detected within the BM, which is responsible for a large proportion of circulating IgM ²⁵.

Figure 4. B cell-specific loss of Id3 increases bone marrow B-1b cell numbers and total bone marrow IgM producing cells.

A) BM was collected from Id3^WTApoe^−/− (n=8) and Id3^BKOApoe^−/− (n=9) mice and total BM cells were analyzed for IgM production by ELISPOT while a small portion was immunophenotyped by flow cytometry. B) Representative images of ELISPOT wells showing IgM positive spots from Id3^WTApoe^−/− and Id3^BKOApoe^−/− BM samples compared to negative control. C) Quantification of IgM⁺ BM cells from ELISPOT calculated from total BM cells. D) Quantification of total B-1, B-1a, and B-1b cells from BM. Data are mean ± SEM. E) Comparison of total IgM in 8-week-old, litter matched Id3^BKOApoe^−/− and Id3^WTApoe^−/− mice as determined by ELISA. Data are mean ± SEM. Unpaired, two-tailed Students T-test was used to compare differences, *p<0.05, **p<0.01

Id3^BKOApoe^−/− mice exhibit reduced Western diet induced atherosclerosis

To determine if B cell-specific deletion of Id3 resulted in reduced development of atherosclerosis, Id3^BKOApoe^−/− and Id3^WTApoe^−/− mice were fed a Western diet for 16 weeks, and then aortic sinuses and aortas were harvested for histochemical analysis of cross sectional lesion area and aortic lipid deposition respectively (Figure 5A). Id3^BKOApoe^−/− mice exhibited significantly reduced atherosclerosis within the aortic root compared to Id3^WTApoe^−/− mice (3.99x10^5 ±.19x10^5 μm² vs. 4.96x10^5 ±.19x10^5 μm², P<0.01, Figure 5B&C). Additionally, immunofluorescent macrophage staining using an anti-Mac2 antibody identified significantly decreased macrophage content within lesions of Id3^BKOApoe^−/− mice compared to Id3^WTApoe^−/− (7.18 ± 2.23% vs. 18.48 ± 3.53%, P<.05, Figure 5D&E) Finally, Tunel staining demonstrated that lesions in Id3^BKOApoe^−/− mice contained a significantly decreased percentage of apoptotic cell bodies than did Id3^WTApoe^−/− (0.37*10⁻³ ± 0.12*10⁻³ % vs. 0.84*10⁻³ ± 0.17*10⁻³ %, P<0.05, Figure 5F&G). En face aortic preparation with Sudan IV staining demonstrated that the aortas of Id3^BKOApoe^−/− mice had reduced lipid deposition, though not statistically significant, compared to Id3^WTApoe^−/− (10.62 ± 1.17% vs 12.58 ± 0.95%, P=0.10, Figure 5H&I) Lipid analysis demonstrated that there were no significant differences in body weight, weight gain, or plasma lipids at time of euthanasia (Online Table II). Additionally, B-1b cell number remained greater in Id3^BKOApoe^−/− mice after the 16 weeks of Western diet feeding (Online Figure V). Interestingly, although not different at baseline (Figure 3), the number of PerC B-2 cells was greater in Id3^BKOApoe^−/− mice compared to WT controls after Western diet feeding (Online Figure V). This effect was not seen in the blood or spleen and is of unclear consequence.

Figure 5. B cell specific deletion of Id3 reduces diet induced atherosclerosis.

A) 8-week old, male, chow fed Id3^WTApoe^−/− (n=10) and Id3^BKOApoe^−/− (n=7) were fed a Western diet for 16 weeks then plasma, aorta, and aortic sinus were collected for analysis. B–C) Representative images and quantification of maximum lesion areafrom 4x magnification images. D–E) Representative images and quantification of Mac2⁺ lesion area normalized to total plaque area from 10x magnification images. A = Adventitia, P = plaque, L = lumen. F–G) Representative images and quantification of Tunel⁺ area normalized to plaque area from 10x images. Enhanced image is zoomed in from boxed area to show positive staining. H–I) Representative images and quantification of Sudan IV staining of aortas prepared en face. Two-tailed Mann-Whitney test was used to compare measured values. *p≤0.05, **p<0.01

B cell-specific deletion of Id3 increases circulating anti-OSE IgM antibodies

OSE IgM reduce OxLDL uptake by lesion macrophages, slowing the development of foam cells and the expansion of intimal plaques ^{10, 30} while also binding apoptotic cell bodies, enhancing their clearance and decreasing sterile inflammation ³¹. The finding that Id3^BKOApoe^−/− mice had decreased intimal macrophage content and fewer apoptotic cell bodies suggests that IgM antibodies could be involved in the atheroprotection exhibited in those mice. To determine if total IgM and OSE reactive IgM were increased in Id3^BKOApoe^−/− mice compared to Id3^WTApoe^−/−, we measured total IgM and IgG, as well as IgM reactive to the OSE described in Figure 2D in the respective mice. Absolute titers of immunoglobulins were determined using standard curves (Online Figure IC). Id3^BKOApoe^−/− mice had significantly higher titers of total IgM compared to Id3^WTApoe^−/− controls with no differences detected in titers of IgG isotypes (Figure 6A). Id3^BKOApoe^−/− mice also contained significantly greater relative IgM titers reactive to MDA-LDL, and CuOx-LDL than Id3^WTApoe^−/− mice (Figure 6B). An approximately 30% increase in the amount of the E06/T15 NAb was also detected though it was not statistically significant (Figure 6B). No significant difference was detected in IgM reactive to α-1,3-Dextran (Figure 6B, ³²). Again, low levels of IgM to nLDL were noted (Figure 6B). Pooled dilution series were used for all relative titers (Online Figure IE). Importantly, real-time PCR analysis of secreted IgM (sIgM) mRNA, expressed per cell, from sorted B cell subsets from Id3^BKOApoe^−/− and Id3^WTApoe^−/− mice demonstrated no differences in expression suggesting that Id3 does not directly regulate IgM produced per cell (Online Figure VII). Rather, taken together, data suggest that B cell-specific Id3 deletion results in an increased number of IgM producing B-1b cells that generate OSE reactive IgM.

Figure 6. B cell-specific loss of Id3 results in increased titers of atheroprotective IgM antibodies after Western diet feeding.

Plasma was collected from Id3^WTApoe^−/− and Id3^BKOApoe^−/− mice fed a Western diet for 16 weeks and absolute titers of immunoglobulin isotypes and relative titers to indicated antigens were determined by chemiluminescent ELISA. A) Comparison of absolute titers of IgM and IgG isotypes reported as μg/ml. B) Comparison of relative IgM titers to indicated antigens as reported in legend of Figure 2 reported as RLU/100ms. Values are mean ± SEM, unpaired, two-tailed Students T-test was used to compare differences. *p<0.05, **p<0.01, ***p<0.001

To determine whether loss of Id3 in Apoe^−/− mice might also modify the T cell dependent (TD) immune response, Id3^BKOApoe^−/− mice were immunized with DNP-KLH in complete Freund’s adjuvant and bone marrow plasma cells and plasma were analyzed (Online Figure VIA). Greater numbers of IgM⁺ plasma cells (PC) were measured by flow cytometry in Id3^BKOApoe^−/− mice compared to Id3^WTApoe^−/− mice, both those immunized with DNP-KLH and adjuvant controls (Online Figure VIB) supporting the findings of Figure 4 that Id3 is important for BM IgM production. In contrast, there were no differences in other isotype PCs providing evidence that the increase in BM IgM producing cells is not due to inhibition of isotype switching. Analysis of serum anti-DNP antibodies by ELISA demonstrated a T cell-independent increase in IgM in Id3^BKOApoe^−/− mice (Online Figure VIC). Both Id3^BKOApoe^−/− and Id3^WTApoe^−/− mice had robust antigen specific IgG1, IgG2a, and IgG2c responses, although the IgG1 and IgG2c responses were blunted in Id3^BKOApoe^−/− mice. IgG3 was significantly increased in Id3^BKOApoe^−/− mice likely due to B-1b isotype switching as has been previously demonstrated³³. These findings provide evidence that the predominant effect of B cell-specific deletion of Id3 is enhanced T cell-independent IgM production possibly due to B-1b derived PCs as have been described previously²⁹.

Humans harboring SNP rs11574 in ID3 have increased proportion of circulating B1 cells

We have previously shown that presence of the SNP rs11574 encodes an Id3 protein with attenuated function ²³. To determine whether patients bearing the homozygous allele of rs11574 that alters Id3 function have increased B-1 cells in circulation, human peripheral blood mononuclear cells (PBMCs) were analyzed by flow cytometry and genotyped for the presence of the SNP. The flow cytometry strategy utilized was a modification of the original Rothstein strategy²², incorporating additional stains and gates to eliminate initial concerns about possible contamination by T cells and possible overestimation of numbers. Representative flow cytometry of human B cells is presented in Figure 7A&B. Consistent with results from Rothstein and colleagues^34,22 we detected a small but clear population of circulating CD20⁺ cells that are live, singlet, CD3⁻CD27⁺CD43⁺. Quantification of the percentage of CD20⁺ cells from human that are CD27⁺CD43⁺ demonstrated that patients homozygous for the minor allele of the rs11574 SNP had a significantly greater percentage of B cells that are CD27⁺CD43⁺cells in circulation (Figure 7B). These patients did not have differences in the proportion of total, naïve, or memory B cells (Figure 7A&B). Additionally, antibody titers against MDA-LDL, which has been previously shown to inversely associate with Framingham risk score, metabolic syndrome criteria²⁷, and CVD³⁵, were tested from plasma of these patients. This analysis demonstrated that the patients with the minor allele had significantly higher titers of IgM against MDA-LDL compared to patients with the common allele (Figure 7C). We did not observe any differences in titers of IgG reactive to the same epitopes (Figure 7C). These findings suggest that Id3 may also regulate the relative percentage of B-1 cells in humans, suggesting human relevance of our murine findings.

Figure 7. Patients homozygous for the minor allele at rs11574 in the ID3 gene (encoding an Id3 protein with attenuated function²³) have increased percentage of B-1 cells and IgM reactive to MDA-LDL.

DNA was isolated from PBMCs for genotyping of the rs11574 SNP (ancestral allele GG N=50, minor allele AA N=8). Representative flow cytometry and quantitation of B cells by ID3 genotype at rs11574 for (A) total B cells and (B) memory B cells (CD27⁺CD43⁻), naïve B cells (CD27⁻CD43⁻) and, B-1 cells (CD27⁺CD43⁺). C) Comparison of IgM and IgG antibody titers reactive to MDA-LDL by ID3 genotype at rs11574 measured in RLU/100ms by ELISA. Data from experiments comparing genotypes are presented in Box-and-Whiskers format with 95% CI and outliers as dots. Unpaired two-tailed Students T-test was used to compare differences. *p<0.05.

DISCUSSION

Results of the present study are the first to demonstrate that B-1b cells produce IgM to OSE and attenuate diet-induced atherosclerosis. B-1b cells are developmentally and functionally unique compared to B-1a and B-2 cells ^{36, 37}. B-1a cells produce IgM NAbs in an antigen-independent manner, whereas B-1b cells respond to T cell-independent antigens providing them the capacity to form memory ^{12, 38}. The work of Haas et al ³³ and Alugupalli et al ³⁹ established B-1b cells as the source of T cell-independent memory and support B-1b cells as targets of T cell-independent antigen directed vaccination, which has been discussed as a novel therapeutic against atherosclerosis ^{18, 19, 30, 40}. As such, understanding the impact of B-1b cells on atherosclerosis and elucidating factors that regulate their activity is of clear importance.

Studies presented here provide novel evidence that B-1b cells produce atheroprotective IgM antibodies. The finding that Rag1^−/−Apoe^−/− B-1b recipient mice have higher relative titers of OSE reactive IgM compared to B-1a recipients, despite transfer of equal numbers of cells, suggests that either B-1b cells produce more OSE reactive IgM or have better cell viability and persistence in vivo after transfer. IgM antibodies reactive to OSEs on OxLDL are highly conserved and their levels associate with reduced coronary artery disease and cardiovascular events ^{17, 41, 42}. Their protective function is thought to occur by reducing the uptake of OxLDL by tissue resident macrophages¹⁰ and blocking the proinflammatory properties of oxidized lipid moieties ³¹. Additional protective mechanisms are attributed to IgM binding to apoptotic bodies within atherosclerotic plaques which increases their clearance reducing sterile necrosis ^{31, 43}. Taken together these data suggest that B-1b mediated production of OSE reactive IgM antibodies could be important for their atheroprotective function. Our data underscore the importance of future studies to determine if IgM production is essential for B-1b-mediated atheroprotection and to elucidate other potential atheroprotective pathways.

Of note, the magnitude to which a single injection of B-1b cells reduced atherosclerosis in Rag1^−/−Apoe^−/− mice was modest. B-1a cells transferred into splenectomized hosts (3 injections of 1x10^5 cells over 8 weeks) resulted in a greater reduction in diet-induced atherosclerosis compared with our one injection of 1x10^5 B-1b cells prior to 16 weeks of Western diet feeding⁹. Studies to directly compare the relative atheroprotective contributions of equal numbers of injected B-1b cells to B-1a cells are underway in our lab. Nevertheless, while B-1b cells are not the only cell type mediating atheroprotection they clearly contribute significantly to protection from diet-induced atherosclerosis.

Id3 has been implicated in atherosclerosis in mice and humans^{7, 23}. Yet, the specific mechanisms whereby Id3 regulates plaque development are just beginning to be elucidated. Id3 is a broadly expressed transcription factor known to be important throughout development.²⁰ Previous studies by our lab have reported that global loss of Id3 in Apoe^−/− mice leads to enhanced atherogenesis⁷. Follow-on studies demonstrated that Id3^−/−Apoe^−/− mice had reduced IL-33-stimulated IL-5 production by natural helper cells²¹. This defect, and not loss of Id3 in B cells, led to reduced B-1a cell numbers in Id3^−/−Apoe^−/− mice. Id3 has also been implicated in the regulation of VCAM-1 expression by vessel wall cells with associated increased accumulation of macrophages in lesions⁴⁴. In addition, Id3 regulates aortic homing of CD43⁻ splenocytes (primarily B-2 cells) through regulation of the expression of chemokine receptors such as CCR6⁷. These findings from Id3^−/−Apoe^−/− mice underscore the importance of defining cell type-specific effects of Id3 and support the use of B cell-specific Id3 knockout mice to identify B cell-specific mechanisms that may impact on atherosclerosis.

An early study by Pan et al suggested that Id3 may be important for the B-2 mediated T cell dependent immune response²⁴. In the present study, chow fed Id3^BKOApoe^−/− mice responded to T cell-dependent immunization though in a blunted manner (Online Figure V) raising the possibility that loss of Id3 in B-2 cells could also contribute to an atheroprotective phenotype. However, IgG isotype titers were not different between Id3^BKOApoe^−/− and Id3^WTApoe^−/− mice after Western diet feeding suggesting that the TD antigen response was likely not important for the atheroprotection we saw in our Id3^BKOApoe^−/− mice. It is possible that Id3 may regulate activation state or other B-2 functions given that the Id3^BKOApoe^−/− is knocked out for all B cell subsets. Thus, it will be important in future studies to establish whether Id3-mediated regulation of B-2 cells is important for their function in atherosclerosis.

Results of the present study confirm prior findings that B cell-specific deletion of Id3 did not alter the B-2 or B-1a cell population²¹ and provide novel evidence that the numbers of B-1b cells in Id3^BKOApoe^−/− mice are significantly greater than in WT control. Interestingly, in contrast to global knockout Id3^−/−Apoe^−/− mice, Id3^BKOApoe^−/− mice exhibited significantly reduced atherosclerosis in the aortic sinus though not in the aorta, possibly due to a physiological difference in the development of atherosclerosis between the sinus and aorta⁴⁵. Taken together, results provide evidence that Id3 regulates atherosclerosis through unique cell-type dependent mechanisms. Notably, patients both heterozygous and homozygous for the ID3 SNP at rs11574 have increased clinical measures of cardiovascular disease²³, yet this same SNP is associated with an apparently specific increase in the percentage of B-1 cells.

The role of B-1 cells in human immunity is poorly understood owing to the lack of a clearly defined human B-1 cell subset. Recently, Rothstein and colleagues have identified a population of CD20⁺ B cells that are CD27⁺CD43⁺ and possess key features of murine B-1 cells including spontaneous production of IgM ^{22, 34}. More recently this subset has been linked to secretion of atheroprotective IgM after T cell-independent immunization ⁴⁶. It is unknown whether the subclassification of B-1 cells into B-1a and B-1b is appropriate in humans as it is in mice. Interestingly, we have shown that a human cohort homozygous for the SNP rs11574, which expresses a modified Id3 protein with reduced function, have increased B-1 cells in circulation as a percentage of total B cells without a difference in total, naïve, or memory B cells (Figure 7). Additionally, the same patients had increased titers of IgM against MDA-LDL raising the possibility of a functional association although additional studies of larger cohorts will be needed to confirm these findings. Taken together, these findings suggest attenuated Id3 function associates with increased B-1 cells and could serve as a target for enhancing B-1 directed therapies in the future.

Novelty and Significance.

What Is Known?

• Murine B cells regulate atherosclerosis based on subset specific functions; innate B-1a cells are atheroprotective and adaptive B-2 cells have multiple functions. B-1b cells have not been studied in the context of atherosclerosis.

• Inhibitor of Differentiation 3 (Id3) regulates B cell development and atherosclerosis through multiple regulatory pathways.

• A putative human B-1 equivalent was described as CD20⁺CD3⁻CD27⁺CD43⁺CD70⁻ and functionally resembles murine B-1 cells.

What New Information Does This Article Contribute?

• B-1b cells secrete atheroprotective IgM antibodies.

• B cell-specific loss of Id3 causes a significant increase in B-1b cell numbers suggesting it is an important regulator of B-1b cells.

• Patients with a single nucleotide polymorphism (SNP) in the ID3 gene at rs11574, which attenuates Id3 function, have increased B-1 cells as a percentage of total B cells and increased titers of IgM to MDA-LDL in the circulation.

B cells have been demonstrated to have both atheroprotective and atherogenic functions based on subset and context. To date the contributions to atherosclerosis of B-1b cells, an important subset for T cell-independent immunity, have not been studied. Here we show that B-1b cells secrete IgM antibodies reactive to oxidative epitopes on LDL and that these cells are atheroprotective when introduced into B and T cell deficient Rag1^−/−Apoe^−/− mice. We implicate the helix-loop-helix transcription factor Id3 in the regulation of B-1b cells as B cell-specific Id3 knockout mice (Id3^BKOApoe^−/−) developed significantly increased numbers of B-1b cells systemically without modifying the number of other B cell subsets. Additionally, Id3^BKOApoe^−/− mice had significantly increased titers of atheroprotective IgM and developed attenuated atherosclerosis. Finally, patients homozygous for the Id3 SNP at rs11574, previously shown to attenuate Id3 function, have an associated increase in putative B-1 cells as a percentage of total B cells in the circulation and increased titers of IgM to MDA-LDL compared to patients with the common allele. Taken together, our findings suggest targeting Id3 in B cells may attenuate atherosclerosis by augmenting atheroprotective B cell numbers and that this could be relevant in humans.

Nonstandard Abbreviations and Acronyms

OSE

Oxidation-specific epitopes

BAFFR

B cell activating factor receptor

Id3

Inhibitor of differentiation 3

NAb

Natural antibodies

NHC

Natural helper cells

Id3^BKO

B cell specific knockout of Id3

Id3^WT

wildtype littermates to Id3^BKO mice

TUNEL

Terminal deoxynucleotidyl transferase dUTP nick end labeling

LPS

Lipopolysaccharide

MDA

Malondialdehyde

CuOx

Copper oxidized

PC

Phosphorylcholine

OxPL

Oxidized phospholipid

PPS-3

Pneumococcal polysaccharide