Aging induces an Nlrp3 inflammasome-dependent expansion of adipose B cells that impairs metabolic homeostasis

Aged white VAT shows a unique increase in memory-like B lymphocytes in FALCs

As there is an age-related decrease in VAT macrophages (Camell et al., 2017) along with an expansion of VAT CD3⁺ lymphocytes and Tregs (Cipolletta et al., 2015; Lumeng et al., 2011), we sought to investigate the impact of aging on AT B lymphocytes. To discriminate tissue resident cells from leukocytes in the AT vasculature, circulating cells were labelled with intravenous injection of a CD45.2 antibody (Figure S1A). Mice were sacrificed at 3 min post injection, which permitted labeling of all circulating cells, without CD45.2 antibody leaking into tissues or labelling cells extravasating into various tissues (Anderson et al., 2014). Quantification of tissue resident CD3⁺ and B220⁺ cells in mice revealed an increase in CD3⁺ lymphocytes and a 10-fold increase in the percentage of B220⁺ lymphocytes in the VAT from 24-month-old wild-type (WT) mice (Figure 1A & B). While only 20% of B220⁺ cells in AT from 3-month-old mice were tissue resident, nearly 100% of B220⁺ cells from 24-month-old mice were tissue resident in the AT (Figure 1C). Furthermore, age-related expansion in the frequency and cellularity of AT resident B cells was observed by 15-months of age and was markedly pronounced in female mice and was less profound in male animals (Figure 1D&E). There was no expansion of B cells in the spleen or VAT of ovariectomized female mice (Figure S1B) suggesting that ovarian-derived steroid and endocrine hormones do not regulate AT B cell expansion. A second cohort of male mice showed ~2-fold increase (Figure S1C), indicating that there is an age-related expansion in both genders, but it is of higher magnitude in female animals (~10 fold). Moreover, tissues from 2- or 24-month-old WT female mice housed at a separate animal facility (Pennington Biomedical, Baton Rouge) also showed a similar increase in VAT B cells (Figure S1D). These results demonstrate that the expansion of resident AT B cells initiates by 15-months-of age and is housing facility-independent, with greater cell expansion in female mice.

Excess white VAT is associated with increased risk for metabolic impairments, whereas AT depots such as white subcutaneous or brown AT have been linked to insulin sensitivity and thermogenic capacity (Despres and Lemieux, 2006; Zamboni et al., 1997). Our quantification of resident B cells from various depots (Figure S1E), show that the age-related expansion in resident B cells was specific to white VAT (Figure 1F). Moreover, B220⁺ cells in the AT of 29-month-old WT mice express CD19 and do not express SiglecH (Figure S1F), a marker for B220⁺ plasmacytoid dendritic cells (pDCs). Histopathological scoring analysis revealed significant increases in the lymphocyte infiltration in the visceral, but not subcutaneous AT of 20-month-old mice (Figure 1G). These results suggest an association of AT B cell expansion with visceral adiposity seen in aging.

Previous reports describe immune cells within white VAT niches influence lipolysis, energy metabolism and tissue functionality (Camell et al., 2017; Xu et al., 2013). To understand the distribution of lymphocytes, whole mount imaging using an inverted microscope on fixed mesenteric and gonadal AT, revealed T and B cells in mesenteric lymph nodes and not throughout the AT of 7-month-old WT mice (Figure S1G). However, in 24-month-old WT mice, mesenteric and gonadal AT revealed FALCs throughout the tissue (identified previously: (Benezech et al., 2015; Lumeng et al., 2011; Moro et al., 2010). Lymphocytes accumulated within FALCs (Figure 1H&I) that were distinct from lymphocytes seen in a capsule-lined lymph node in subcutaneous inguinal AT (Figure 1J). B220⁺ B and CD3⁺ CD4⁺ T cells were found in the FALCs of the 22-month-old WT mice (Figure S1H). These data demonstrate that aging induces an increase in resident AT B cells localized to FALCs.

Adipose B cell express Ki67 and have access to the vasculature

Resident AT immune cells, such as macrophages in response to a high-fat diet, (Amano et al., 2014; Zamarron et al., 2017) and CD4⁺ Foxp3⁺ T regulatory cells during aging (Kolodin et al., 2015), are self-maintained through proliferation. Approximately 10% of B cells from spleen and AT of aged mice are positive for Ki67 (Figure 1K), a nuclear antigen expressed in actively dividing cells. These data indicate that a portion of age-expanded AT B cells are actively proliferating.

Lymphoid clusters in omentum contain high endothelial venules (HEVs), specialized post-capillary venules essential for naïve lymphocyte trafficking (Buscher et al., 2016; Rangel-Moreno et al., 2009). Whether HEVs and lymphatic vessels are present in the FALCs of aged VAT to transport B cells is unclear, therefore we imaged FALCs from aged reporter mice specific for HEV or lymphatic vessel markers (Ruddle, 2014; Truman et al., 2012). Hec6ST reporter mice at 22-months of age showed HEVs (GFP⁺) within FALCs and in close proximity to B220⁺ clusters (Figure 1L). Likewise, 22-month-old reporter mice (ProxTom mice), which express tdTomato under the control of a Prox1 promoter, a transcription factor required for lymphatic vessel formation, revealed the presence of lymphatic vessels (tomato⁺⁾ within FALCs (Figure 1M). Lymphatic vessels in AT that were not within FALCs also contained B cells (Figure S1I). To confirm the presence of Prox1⁺ lymphatic vessels in WT aged mice, whole mount nuclear staining for Prox1 was performed on mesenteric lymph nodes (Figure S1J). Prox1 staining surrounded FALCs even at low magnification (Figure S1K) and confocal microscopy revealed clear Prox1⁺ nuclear staining in vessels near and surrounding FALCs (Figure S1L). Taken together these experiments identified a striking increase in the resident population of B lymphocytes in FALCs of white VAT and that B cell expansion may be supported by both active proliferation and local trafficking via lymphatic vessels and HEVs.

Age-expanded adipose B cells are a unique subset of tissue resident ABCs

During acute infection B1 cells, an innate-like B cell expressing CD11b and secreting natural IgM antibodies, in FALCs are activated and contribute to pathogen clearance (Jackson-Jones et al., 2016; Moro et al., 2010). CD11b⁺ AT B cells express higher levels of CD5, indicative of the B1-subset (Figure S2A). CD11b mean fluorescence intensity (MFI) and the proportions of CD11b⁺ (B1) or CD11b⁻ (B2) resident B220⁺ cells were comparable in the resident VAT B cells from 3- or 24-month-old mice (Figure S2B, Figure 2A), suggesting that B cell subsets are equally expanded with age. To better understand the phenotype of age-expanded VAT B cells, we sorted B220⁺ cells from AT of 3- or 24-month-old WT mice for whole transcriptome analysis. Linear support vector regression analysis of those sorted cells revealed they express a memory-like B cell transcriptional profile (Figure 2B) (Newman et al., 2015).

During aging, the accumulation of a unique subset of B cells, termed age-associated B cells (ABCs), in the spleen, bone marrow and blood has been well described (Hao et al., 2011). ABCs are CD93⁻ CD21⁻ CD23⁻ and evidence suggests that they are antigen-experienced, arising from their encounter with nucleic acid-containing antigens in the context of promoting cytokine production (Naradikian et al., 2016). Consistent with these reports, in 24-month-old WT mice 20–40% of splenic B cells were CD93⁻CD21⁻CD23⁻, but nearly 100% of resident AT B cells lacked CD93, CD21 or CD23 (Figure 2C). Notably, both young and aged AT-conditioned media was able to drive the phenotype of CD21⁻ CD23⁻ B cells from B cells from 3-month-old WT mice (Figure 2D). We next addressed how similar age-expanded AT B cells are to ABCs in spleen, which are reported to express the transcription factor Tbet and CD11c. AT B cells did not show Tbet expression (Figure 2E, S2D and E) or CD11c expression (Figure 2F and S2F). Furthermore, RNA sequencing analysis on sorted splenic ABCs and AT B cells (Figure S2G) revealed distinct clustering by tissue in principal component analysis (Figure 2F) and identified 2307 genes that were differentially regulated between the two populations (Figure 2G and S2H). Of interest is the differential regulation of pro-inflammatory IL-1β and chemokines in AT B cells which are not traditionally considered to be produced by splenic B cells (Figure 2G). These data are indicative of the inflammatory nature of AT B cells. Consistent with flow cytometry, Tbet (gene: Tbx21) was downregulated in the AT B cells (Figure S2I). This indicates that although AT B cells, now referred to as aged adipose B cells (AABs), share some similarities with splenic ABCs, they express a transcriptome that is based on their unique tissue-specific microenvironment.

Depletion of adipose B cells restores insulin sensitivity of aged mice

As the age-related expansion in AABs is localized to VAT and not to other AT depots (Figure 1F), we next determined whether a reduction in AABs would improve age-related insulin resistance. Intra-adipose injection of murine specific CD20 depleting antibody (CD20mAB) specifically depleted AABs (Figure S3A) and avoided systemic depletion of B cells. Body-weight and VAT weight in isotype (ISO)-treated and CD20mAB-treated mice were comparable (Figure S3B & C), but mice given CD20mAB showed a reduction in B cells in visceral, but not in subcutaneous AT, spleen, bone marrow or peritoneal cavity (Figure 3A, B, and S3D). When challenged with insulin, 20-month-old WT mice given the intra-adipose injection with CD20mAb showed improved insulin tolerance (Figure 3C). These results suggested that lipolytic capacity of aged AT may also be improved. We fasted the intra-adipose injected mice to induce lipolysis, a process that can be measured by release of glycerol from AT (Schweiger et al., 2014); however, there were no differences in glycerol release from the fed (Figure 3D) or fasted (Figure 3E) VAT in ISO-treated or CD20 mAb-treated mice. Collectively, these data indicate that specific depletion of the AABs was capable of improving age-related impaired insulin sensitivity, but not impaired fasting-induced lipolysis.

Inhibition of systemic B cell expansion restores metabolic homeostasis in aged mice

To examine the consequence of both age-expanded ABCs in lymphoid tissues and AABs in metabolic dysregulation, we performed intraperitoneal (i.p.) injection of the ISO or CD20 mAb into 21-month-old mice to systemically deplete all B cells (Figure S3E). CD20 mAb reduced the percentage of B cells in the spleen and VAT in the 21-month-old mice (Figure 3F). FALCs tended to appear smaller and displayed reduced staining for B220 (Figure 3G) and percentages of ABCs in the spleen were significantly reduced (Figure 3H). To test whether B cell depletion can reverse age-related alterations in AT immune cell composition, we quantified the PD1⁺ effector T cells and FoxP3⁺ T regulatory (Treg) cells. The frequency of total CD4⁺ cells were increased (Figure 3I) in the B cell depleted mice, but there was a statistically significant VAT-specific reduction in the Treg and CD4⁺PD1⁺ sub-populations that typically increase in aged VAT (Figure 3I, S3H and S3I). These data indicate a single CD20 mAb injection to deplete B confers in part, a more young-like resident immune cell distribution in VAT from old mice.

Consistent with previous findings showing restored insulin tolerance in obese mice depleted of B cells and in aged mice depleted of Tregs (Bapat et al., 2015; Winer et al., 2011), we found that 20-month-old B cell-depleted mice had improved insulin sensitivity (Figure S3J and S3K). To address whether systemic depletion of B cells could restore lipolytic signaling and metabolic substrate availability in aged mice, 22-month-old control or B cell-depleted mice were fasted for 24 hours. The adipocyte lipases, adipose triglyceride lipase (ATGL) and hormone sensitive lipase (HSL), are required for triglyceride hydrolysis during fasting and are reduced in fasted aged VAT (Camell et al., 2017). B cell depletion resulted in increased levels of phosphorylated HSL and total ATGL during fasting, indicating that the age-related reduction in lipolytic signaling was restored with B cell depletion (Figure 3J). CD20mAB-treated mice also had increased isoproterenol-stimulated lipolysis (Figure S3L). Furthermore, i.p. transfer of ABCs cells from AT and spleen of 24-month-old mice impaired fasting induced-glycerol release in the 3-month-old hosts (Figure 3K).

Given that white AT lipolysis induced fatty acids are required for the generation of heat to induce thermogenesis and protect against cold stress (Schreiber et al., 2017; Shin et al., 2017) and that aged mice have impaired maintenance of proper core body temperature during cold stress (Figure S3M) (Berry et al., 2017), we tested whether B cell depletion improves defense against cold stress during aging. Consistent with the restored lipolysis providing additional metabolic substrates, 20-month-old B cell-depleted mice were able to maintain core body temperature when placed at four degrees Celsius (Figure 3L).

Age-induced adipose B cell expansion and FALC formation in adipose is regulated by the Nlrp3 inflammasome

In the context of acute inflammation, macrophage-derived TNFα promotes stromal cell activation and the accumulation of cells in FALCs (Benezech et al., 2015). We next studied the spatial association of macrophages with FALC resident B cells using aged mT/mG;LysM-cre reporter mice. FALCs contained dense capillary networks indicative of the vascular network (Figure 4A; S4A) (Benezech et al., 2015), and numerous B220⁺ cells, some of which could be found in close contact with macrophages (Figure 4A; S4A). Given this close association of cells within FALCs and our previous findings showing macrophage and Nlrp3 inflammasome regulation of AT homeostasis (Camell et al., 2017; Vandanmagsar et al., 2011), we asked whether the age-related expansion in AABs requires the Nlrp3 inflammasome. Compared to aged WT female mice, the aged Nlrp3-deficient mice showed a significant reduction in the percentage and the total numbers of B cells in VAT (two individual aging cohorts in Figure 4B and S4B). Although CD3⁺ T cell subsets, CD4⁺ and CD8⁺, are increased with age, the WT and Nlrp3-deficient mice showed comparable frequencies in aged AT (Figure S4C, D and E), indicating that the Nlrp3 inflammasome specifically promotes the age-related B cell expansion in VAT at 19-months, without affecting age-related T cell expansion.

To determine whether the Nlrp3 inflammasome is required for VAT T cell expansion at an earlier age, we quantified cell subsets in young WT mice and middle-aged WT or Nlrp3-deficient mice at a time (10-months) at which B cell expansion has not yet occurred. Consistent with previous findings (Camell et al., 2017), the 10-month old WT mice showed reduced CD11b⁺ myeloid cells, which was prevented by Nlrp3-deficiency (Figure S4F). Furthermore, AT CD4⁺ T cells were significantly elevated in the 10-month-old WT mice, and prevented by Nlrp3-deficiency (Figure S4G), with no age-related expansion in AABs at 10-month-of age (Figure S4H). These data support a model in which the Nlrp3 inflammasome initially controls age-related increases in total CD4⁺ T cells, and by 18-months of age, the Nlrp3 inflammasome is specifically required for resident AAB expansion.

We next asked whether the Nlrp3 inflammasome is also required for FALC development. Compared to 24-month-old WT mice, there was a significant reduction in the number of FALCs per AT in the aged Nlrp3-deficient mice (Figure 4C); however, we found that FALCs in the aged WT and Nlrp3/- contained Prox1⁺ lymphatic vessels (Figure 4D). These data suggested that although the Nlrp3 inflammasome is required for age-related B cell expansion, reduction in AABs is not because of reduced lymphatic vessel formation due to Nlrp3 inflammasome deficiency.

To address how Nlrp3-deficiency in aging protects against AAB expansion, we evaluated phenotype using flow cytometry. AABs from 19-month-old WT or Nlrp3-deficient mice had no difference in Ki67⁺ expression (Figure 4E), similar B1/B2 subset proportions (Figure 4F), ABC-phenotype (Figure 4G) and CD11c expression (Figure 4H). We then investigated the inflammatory potential of these age-expanded FALC resident B cells using unbiased RNA sequencing analyses. Surprisingly, this analysis revealed that compared to 19–22-month-old WT mice, the CD19⁺ B220⁺ CD93⁻ CD21⁻ CD23⁻ AABs from age-matched Nlrp3-deficient mice displayed almost no differentially expressed genes (Figure 4I, J and S4I), with a total of 19 genes being upregulated and 17 genes being downregulated. Furthermore, the pro- or anti-inflammatory genes were not differentially regulated by Nlrp3-deficiency. Taken together with our previous findings (Camell et al., 2017), these data describe an Nlrp3 inflammasome-dependent role for age-related changes in AT resident macrophages and B cells. In contrast to the Nlrp3-dependent transcriptional regulation identified in macrophages (Camell et al., 2017), the transcriptional profile in B cells is primarily regulated in an Nlrp3-independent manner. Furthermore, Nlrp3 inflammasome-derived inflammation from macrophages acts upstream of VAAB expansion without altering their transcriptional phenotype. Thus, the Nlrp3 inflammasome-mediated reduction in AT inflammation during aging may result in part from numerical reduction in B cells.

Multiple receptor-ligand interactions among macrophages and B cells contribute to Nlrp3-mediated FALC development in aging

Given the role of the Nlrp3 inflammasome in resident B cell and FALC accumulation in aged AT, we asked whether aged AT macrophages directly control B cell recruitment. Using our previous AT macrophage whole genome sequencing analysis, (GSE93202) (Camell et al., 2017), we examined the mean chemokine gene expression in WT or Nlrp3-deficient AT macrophages from 3- or 24-month old mice. Hierarchical clustering analysis of chemokine expression did not reveal the group structure within the dataset suggesting that the Nlrp3 inflammasome is not a predominant regulator of B cell trafficking in aging AT (Figure S4J). To address the possibility that macrophage-B cell interaction contributes to the Nlrp3 inflammasome-regulated resident B cell expansion, we generated a macrophage-B cell interaction model by first curating a list of all possible receptors and ligands. To focus on the Nlrp3-dependent contribution, we identified macrophage ligands and receptors which are regulated in an Nlrp3 inflammasome-dependent manner and which have corresponding receptors expressed in the AABs (Figure S4K). All macrophage ligands and receptors identified showed potential to interact with B cells in one or more ways. Some interactions implied inflammatory signaling regulation (eg. Tlr4-S100a8/9 and Tnfrsf18-Tnfsf18 interaction) between macrophages and B cells, but no single interaction appears sufficient to explain the development of FALCs, suggesting that in context of low-grade chronic age-induced inflammation multiple interactions contribute to AAB expansion during aging.

Nlrp3-mediated inflammation and enhanced healthspan is associated with lower splenic ABC accumulation

During aging, the decrease in Nlrp3 inflammasome activation lowers systemic inflammation and protects against functional decline (Youm et al 2013). The increase in an antigen-experienced subset of ABCs in the spleen has been attributed to chronic levels of toll-like receptor activation and B cell receptor signaling (Hao et al., 2011; Rubtsova et al., 2015; Russell Knode et al., 2017). We wondered whether the age-related increase in splenic ABCs requires Nlrp3 inflammasome. There was significant elevation in the frequency of ABCs in the 24-month-old WT mice that was less pronounced in the 24-month-old Nlrp3-deficient mice (Figure 5A). We identified a corresponding age-regulated decrease in marginal zone and follicular B cells that was not restored with Nlrp3-deficiency (Figure 5B), suggesting that the Nlrp3 inflammasome is required for age-related ABC expansion, but not marginal or follicular B cell alterations. There were no differences in B cell subsets in bone marrow of 19-month-old WT and Nlrp3-deficient mice (Figure S5A), suggestive of tissue-specific protective mechanisms. Given that we had previously seen a lack of transcriptional control of aged AABs by Nlrp3-deficiency, we wondered whether splenic ABCs would reveal transcriptional profile unique to Nlrp3-deficiency. Consistent with AABs, the RNA sequencing analysis revealed limited transcriptional differences when comparing the WT and Nlrp3-deficient splenic ABCs (Figure 5C and D). Taken together, these data indicate a role for Nlrp3 inflammasome in driving age-related changes in B cell numbers in VAT and spleen, without altering the phenotype of those cells.

Reduction in adipose and splenic B cells is associated with increased longevity in mice

Reduction in GH-IGF1 somatotropic activity confers longevity in multiple animal models (Masternak and Bartke 2012). The long-lived mice with reduced growth hormone (GH) signaling have improved insulin sensitivity and reduced Nlrp3 inflammasome activation (Spadaro et al., 2016). Along with their extended lifespan and increased insulin sensitivity (Coschigano et al., 2003; Masternak and Bartke, 2012), male and female GHRKO mice have lower body weight, reduced VAT weight and decreased cellularity (Figure S5B and C). Quantification of ABCs in WT or GHRKO spleens revealed significant decreases in the percentage and total number of ABCs (Figure 5E, S5D and E). Furthermore, the VAT of GHRKO mice had significantly reduced percentage of B220⁺ cells (Figure 5F, S5F and G), but not T cell subsets (Figure S5H and I). These data highlight that reduction of aged B cells is associated with extended lifespan and may be related to improved metabolic outcomes associated with enhanced healthspan.

IL-1 signaling drives adipose B cell proliferation, inflammation and impaired lipolysis

Nlrp3 inflammasome activation results in release of the pro-inflammatory cytokines IL-1β and IL-18, which are implicated in age-related inflammation and pathologies (Youm et al 2013; Camell et al 2017). Our RNA sequencing analysis revealed significantly higher levels of IL1R on adipose B cells and this was the only interleukin receptor identified to be more highly expressed on the adipose B cells (Figure 6A). To address whether IL-1β signaling through the IL-1R drives B cell expansion and impaired lipolysis, 20-month-old female mice were depleted of B cells and co-treated with a vehicle control or IL-1R antagonist three times per week (Figure S6A). There were no differences in the body-weight in either group (Figure S6A and B); however, there was a reduction in the frequency of Ki67⁺ AABs upon IL-1R blockade (Figure 6B). Additionally, there was an inverse correlation between the frequency of Ki67⁺ and the frequency of B cells for the IL-1 receptor antagonist-treated group only (Figure 6C). Splenic ABCs showed comparable frequency of Ki67+ cells and no correlation for either treated group (Figure 6D& E). These results indicate that IL-1R signaling is required for B cell proliferation in the VAT, but not the spleen. There was a parallel reduction in pro-caspase-1 and active (p20) caspase-1 in the VAT of the mice treated with the IL-1R antagonist (Figure 6F and S6D). To address whether the reduced B cell proliferation and IL-1R inhibition is sufficient to increase metabolic capacity of aged AT, treated mice were fasted to evaluate lipolysis in the tissue. VAT of the mice treated with the IL-1R antagonist had elevated glycerol release (Figure 6G). These results show that IL-1 signaling drives B cell proliferation in the AT contributing to age-related inflammation and impaired lipolysis.

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dr. Vishwa Deep Dixit (vishwa.dixit@yale.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

METHOD DETAILS

QUANTIFICATION AND STATISTICAL ANALYSIS

DATA AND SOFTWARE AVAILABILITY

The RNA sequencing data has been uploaded to Gene Expression Omnibus (GSE137144). Previously published datasets that were used in the analysis are indicated by GEO number in the text.