B lymphocytes in human subcutaneous adipose crown-like structures

Introduction

Obesity-related insulin resistance (OIR) is a highly prevalent metabolic disorder that contributes to increased mortality through multiple diseases, including type 2 diabetes (T2D), cardiovascular disease, and malignancies.(1) In addition to high body mass and impaired insulin action, OIR is associated with persistently elevated blood levels of inflammatory cytokines, which are thought to be largely derived from adipose tissue.(2)

Many investigators have proposed that visceral adipose tissue (e.g. mesenteric, omental, epididymal) becomes inflamed as a primary event in OIR, a hypothesis initially raised by the presence of adipose-associated lymphoid structures identified in rodent omental fat.(3, 4) Several studies have now confirmed the presence of organized accumulations of immune cells in human adipose, forming “fat-associated lymphoid clusters” (FALC) (5) in visceral depots and macrophage-predominant “crown like structures” (CLS) in both visceral and subcutaneous depots (5, 6). In rodents, the progressive accumulation of macrophages into these CLS is associated with insulin resistance.(7, 8) Our group and others have shown that macrophage infiltration and CLS burden in both subcutaneous and visceral fat predicts the severity of insulin resistance in humans,(7, 9-11) as well as systemic endothelial dysfunction in the peripheral vasculature. (6)

While the CLS has been described as housing macrophages and T lymphocytes to coordinate local inflammation (12-18), the B lymphocyte has not as yet been described as an integral, or “resident,” feature of these lesions despite their described prominence in analogous immunologic structures, such as “milky spots” in rodent mesenteric fat depots.(2) The few human studies investigating the presence of adipose tissue lymphocytes (ATLs) have shown rare B cells in mesenteric fat by flow cytometry using an antibody to CD19, a pan B cell marker, (16) or have not included B cell-specific antibodies in CLS immunohistochemistry. (14) Plasma cells (large B cells that actively produce antibody) have been identified, but are rare and interspersed in visceral fat (19). Recently, Winer, et al proposed that B cells are pathogenic in OIR, showing that a B cell null rodent model lacks pathogenic immunoglobulins that contribute to insulin resistance. (20) Members of our group and others have identified aberrant expression of pathogen-recognition Toll-like receptors (TLR) 4 and 2 on circulating B cells, which produce copious inflammatory cytokines in inflammatory states. (21-23) In contrast to these findings, others have shown a protective role of B lymphocytes in tissue models of artherosclerotic disease.(24) In light of this emerging and conflicting data, we sought to identify B cells in or around the human subcutaneous CLS (sCLS) in an obese population and determine whether their presence is associated with specific clinical parameters.

Methods and Procedures

Results

All subjects (n=32) had class 3 obesity (BMI ≥ 40 kg/m²) with an average BMI of 46 kg/m². Approximately 40% of subjects had T2D, defined by having the diagnosis on the clinic problem list or being treated with a diabetes medication. Thirteen (13) % of subjects were taking metformin; 28% were taking an HMG-coA reductase inhibitor (Statin); and 6 % were taking a thiazolidinedione (TZD). Subjects taking insulin were excluded. Clinical and metabolic parameters of all subjects are displayed in Table 1 according to CD19 (+Bcell) status, described below.

Discussion

Accumulation of macrophages and T cells in adipose tissue present in the form of crown-like structures (CLS) in subcutaneous and visceral depots is a common finding in obesity and predicts the severity of both insulin resistance and endothelial dysfunction in humans. (6) In rodent visceral fat, B lymphocyte migration to adipose tissue heralds immune cell infiltration prior to the onset of insulin resistance, (13, 14, 16) yet recent literature supports either a beneficial (27, 28) or maladaptive (20, 29) nature of adipose B cells. Here we present the novel finding in humans that B cells (CD19+) are present in relatively higher numbers than T cells (CD3+) within sCLS and that B cell predominance (B:T cells) relates to exposure to the drug metformin, a highly effective treatment for OIR.(30) While our data support the hypothesis formed by rodent literature that the B cell in adipose tissue plays an active role in human adipose inflammation, it also poses more questions about whether this role is pathogenic or protective.

Our prior data associating the sCLS with abnormal metabolic and vascular phenotypes may on the surface implicate the sCLS B cell as pathogenic. However, the current literature is conflicted regarding the function of tissue-specific B cells in metabolic disease. As noted above, Winer, et al have found that B cells accumulate in adipose tissue of DIO mice and facilitate the activation of pro-inflammatory macrophages through production of specific, possibly pathogenic, IgG antibodies.(20) This contrasts with in vitro data showing that incubation of adipocytes with the Fc moiety of IgG reduced expression of IL-1β and IL-6 (19), and with the finding that B cells identified in pre-atherosclerotic lesions may protect against atherosclerosis in rodents,(24) possibly explained by the production of antibody reactive to modified low-density lipoproteins (LDLs).(27)

The complexity of the B cell could partially explain this conflict. Based on recent reviews on the interplay between immune cell types in adipose, B cells are thought to contribute to adipose inflammation via the production of antibodies to specific antigens and recent work reflects this hypothesis.(2, 13, 20, 29) However, B cells in humans, particularly in chronic inflammatory diseases, do not necessarily recapitulate murine B cell biology (21, 31, 32). In humans, circulating B cells express functional TLR4 and TLR2 in inflammatory diseases, while B cell TLR expression in healthy controls is nearly absent. (21) By contrast, in mice, B cell TLR2 and TLR4 responsiveness is constitutive in the healthy state. Moreover, we found that TLR ligands produce disease-specific responses from B cells, which can be either pro or anti-inflammatory. (32) Adipose lymphocytes express TLR4 as well (data not shown), suggesting that the B cell may be one of the cell-types found in adipose tissue to affect downstream NF-kB activation and upregulated cytokine release via TLR activation (or suppression).(33) This potential “alternative” function of B cells, via TLRs (either pro or anti-inflammatory), has yet to be explored in adipose tissue. Although B cell function in human adipose remains unknown, based on our finding and that in rodents, the B cell likely contributes to the complex immune environment described in adipose tissue in response to pathologic stimuli and tissue “remodeling.” (8, 34, 35)

The unexpected finding that B cells are present in larger numbers than T cells could reflect timing of adipose infiltration in our patient population, which is difficult, if not impossible, to control in human models of obesity. Duffaut first showed that in mice fed a high fat diet, B cell accumulation occurred as early as 3 weeks, and 9 weeks before T cell infiltration, before the onset of insulin resistance.(16) The authors postulate that perhaps the high fat diet in mice induces an early cell-mediated immune reaction (B cells) in adipose tissue that is followed by maladaptive changes (T cells). Since nearly all of our subjects were severely obese, reflecting long-term excess adiposity and overnutrition, our finding of a common high B:T ratio phenotype may depict a later stage in the natural history of adipose inflammation. It may be that B cell infiltration is earliest, but is persistent over time, making the high B:T sCLS phenotype as a manifestation of chronic overnutrition. Overall, however, it is premature to hypothesize whether chronic B cell presence is beneficial or maladaptive, and if that changes with time.

There were fewer lymphocytes overall in omental tissue compared with subcutaneous adipose, which although of interest, is not surprising given flow cytometry performed by others on mesenteric fat tissue has yielded very small B cell fractions.(16) This is of interest, since the study of adipose inflammation thus far has been largely dedicated to mesenteric depots, which tend to have more CLS and expression of inflammatory cytokines. It is possible that lymphocytes, specifically B lymphocytes, traffick preferentially to subcutaneous depots under certain conditions, and based on our preliminary findings, this may be a clinically important phenomenon and warrants further study.

Knowing at the outset that the adipose B cell population was very small (<4% cells by our FACS), we sought to identify a morphologic correlate with CLS. We acknowledge that the IHC technique does not serve to quantify cells due mostly to the fact that cells are viewed on slides in one dimension, and in one section of an adipose CLS. For this reason, we have reported cell density as others have done (19, 36), and emphasized the relative presence of B cells when compared with T cells. Flow cytometric techniques (FACS) may be more accurate for cell quantification, which overcomes the obvious limitations of IHC and is also able to more specifically identify and characterize cells that express more than one identifying surface marker by using several antibodies simultaneously. However, as illustrated well by B cells in the sCLS, FACS cannot localize cells within a tissue bed, so potentially integral cells in relatively low numbers may be unquantifiable (15). In addition, since lymphocytes are usually isolated from the stromal-vascular fraction of adipose tissue for FACS, the final isolate is likely to include cells originating from the bloodstream due to the degradation of the endothelium. Thus, blood cells can “dilute out” cells originating from adipose, causing FACS to overestimate adipose lymphocyte numbers if care is not taken to isolate only tissue cells.

The independent relationship of sCLS lymphocyte prevalence and the B:T ratio with metformin therapy is preliminary given our small patient sample and cross-sectional study design, yet it is consistent with emerging data on the immunomodulatory effects of this drug. Metformin is the optimal first-line treatment for T2D (30) as its traditional mechanism of action is lowering blood glucose by impairing hepatic glucose secretion and improving insulin sensitivity in peripheral tissues.(37) In overweight and obese cohorts with T2D, metformin therapy is associated with lower mortality.(38) Metformin is also known to have pleotropic effects, including effects on inflammation and cellular apoptosis, thought largely mediated by modulation of AMP-kinase activity.(39) Relevant to our findings, metformin has been found to inhibit the proliferation of T cells, potentially through its effect on fatty acid oxidation and reduction in oxidative stress.(40) Whether metformin is partly responsible for an increased ratio of B cells to T cells in adipose tissue, or whether it reflects unmeasured variable(s) in our cohort, requires further investigation.

Despite its novelty, the design of this study limits the ability to draw conclusions about pathophysiology or B cell function. We did identify a population of IgG+ B cells in the CLS, which may represent the pathologic B cells described similarly in rodents (20), although this is inconclusive given technical limitations and lack of functional data. We additionally did not identify any plasma cells, which actively secrete antibody and are rare in subcutaneous adipose (19). The sample size is likely too small to detect some differences that may be present in a larger population. Moreover, our group was relatively homogenous (all CLS+) without a lean control group, which may have limited our ability to find relationships. For example, the HOMA and BMI did not vary (nearly all patients had BMI >40 and the standard deviation was approximately 2), making it difficult to evaluate these variables as relevant parameters.

In conclusion, we found that B cells are prominent in subcutaneous fat CLS in severely obese humans, and relative predominance appears to vary among individuals. The meaning of this finding is unclear, but a relative predominance of identified B cells over T cells pHPF correlated with metformin therapy, a drug that improves insulin resistance and my have an effect on immune function. It has only recently been accepted that the development of obesity has immunologic consequences, or origins, both systemically and in adipose tissue. What remains poorly understood is how the natural history of the adipose tissue sCLS relates to disease mechanism or severity of obesity. Our findings provide a clue that B cells are not only integral in the sCLS natural history, but also that immunomodulation through B cells may be a target of OIR treatment. While its relationship with metformin needs to be confirmed, the B cell is sufficiently complex that careful study of lymphocytes isolated from blood, CLS and lymphoid tissues will be required to link this cell to metabolic and cardiovascular disease in humans.

Supplementary Material