Kinin B₁ Receptor Acts in Adipose Tissue to Control Fat Distribution in a Cell-Nonautonomous Manner

The prevalence of obesity and its associated diseases, such as type 2 diabetes, has been increasing worldwide (1). Obesity is a state of positive energy balance where excess energy accumulates as triglycerides (2). While adipose tissue expansion is an adaptative response, the site of fat deposition is key to determining the risk of obesity-related complications (3–5). Fat stored in the subcutaneous white adipose tissue (sWAT) is usually neutral and even protective against obesity-related insulin resistance (3), whereas lipid deposition in the visceral white adipose tissue (WAT) strongly correlates with the metabolic syndrome (2,5). Ectopic fat accumulation and lipotoxicity develop when lipid storage capacity is limited in adipose tissue, such as during lipodystrophy or in certain obese individuals, and it strongly correlates with insulin resistance, hepatic steatosis, and dyslipidemia (2). The pattern of lipid accumulation is genetically determined, and so is the capacity of WAT to expand in an individual (6).

Adipocyte size increases as a consequence of lipid accumulation. But when they become larger than 100 μm, oxygen diffusion in the tissue is compromised, forming hypoxic regions that contribute to adipose tissue dysfunction (7). De novo adipocyte differentiation (hyperplasia) is also an adaptative response to lipid overload (2). A classic example of a mechanism controlling adipocyte hyperplasia is the activation of peroxisome proliferator–activated receptor γ (PPAR-γ), a lipid-sensing transcription factor required during adipogenesis and for lipid accumulation in adipose tissue (8). Insulin signaling at the level of adipocytes also affects adipocyte differentiation and lipid storage (6). Lack of PPAR-γ or insulin and IGF-I signaling in adipocytes creates lipodystrophy, wherein adipose tissue mass is dramatically reduced and lipids accumulate in peripheral organs (9,10).

Adipose tissue is an endocrine organ that participates in whole-body metabolic control (11). The use of adipose tissue transplantation techniques has successfully demonstrated the cell-nonautonomous roles of adipose tissue (3). For example, sWAT transplanted into the visceral cavity decreases adiposity, glycemia, and insulin level in mice (4). Hormones, adipokines, lipids, and miRNAs are some of the many types of mediators secreted by adipose tissue and involved in metabolic regulation (11–14).

In the context of adipose tissue remodeling and signaling, immune cells play an important role (14). Lean adipose tissue is abundant in regulatory cells such as CD4⁺ T regulatory lymphocytes (T_reg) and anti-inflammatory M₂ macrophages (2). Nonetheless, during the establishment of obesity or lipodystrophy, adipose tissue cellularity changes toward a proinflammatory profile (11,15). In brief, the number of CD8⁺ T lymphocytes increases, monocytes migrate from the periphery and differentiate into proinflammatory M₁ macrophages, and local resident macrophages proliferate and are activated (15–17). In such conditions, immune cells produce proinflammatory cytokines such as tumor necrosis factor-α (TNF-α), interferon-γ, and MCP-1, which in turn signal locally to adipocytes/preadipocytes and systemically to other tissues in order to control tissue remodeling and energy portioning (14). When this inflammatory process becomes chronic, it leads to adipose tissue fibrosis and insulin resistance, and contributes to the metabolic syndrome (18).

The kinin B₁ receptor (B₁R) is a G protein–coupled receptor with known roles in inflammation (19,20). The bradykinin metabolite des-Arg⁹-bradykinin is the natural agonist of B₁R (20). We and others have demonstrated that in addition to its role in inflammation, B₁R controls adiposity and glucose homeostasis via regulation of leptin and insulin sensitivity (21–25). B₁R knockout mice (B₁^−/−) are lean, are resistant to diet-induced obesity (DIO), are protected from hepatic steatosis, have improved insulin and leptin sensitivity, and have fewer plasma triglycerides (22–24). We demonstrated that B₁R overexpression in adipose tissue driven by the Fabp4 (aP2) promoter restores the capability of B₁^−/− mice to gain WAT mass upon eating a high-fat diet (HFD) and impairs insulin sensitivity (22). Hence, B₁R expression in adipocytes controls adiposity and in turn contributes to whole-body insulin sensitivity.

To determine whether at least part of these effects were cell nonautonomous, we took advantage of adipose tissue transplantation and inserted visceral WAT from wild-type donors (B₁^+/+) into B₁^−/− mice (B₁^+/+→B₁^−/−) and compared them with autologous controls (B₁^+/+→B₁^+/+ or B₁^−/−→B₁^−/−). These mice then were fed an HFD, and their response was assessed. Wild-type fat reversed the capacity of B₁^−/− mice to resist HFD-induced obesity, mainly by increasing adipocyte hyperplasia, without changing lean mass, glucose tolerance, or insulin sensitivity. Moreover, in B₁^+/+→B₁^−/− mice, expression of adipocyte markers (e.g., Fabp4 and Adipoq [in sWAT only]) was increased and the proportion of lymphocytes and macrophages were somewhat altered in endogenous WAT when compared with that in B₁^−/−→B₁^−/− mice. B₁^+/+→B₁^−/− mice displayed an increased capacity for gluconeogenesis and accumulated lipid in the liver, in contrast to the B₁^−/−→B₁^−/− controls. Finally, circulating levels of insulin, epidermal growth factor (EGF), and leptin were normalized in B₁^+/+→B₁^−/− mice but not in B₁^−/−→B₁^−/− mice. In summary, we conclude that the presence of B₁R in adipose tissue controls WAT expansion and whole-body lipid handling through a cell-nonautonomous mechanism.

Mice were obtained from the Center for the Development of Animal Models in Medicine and Biology of the Universidade Federal de São Paulo, Brazil. Animals were fed a chow diet (Nuvital), and the environment was maintained at 22°C under a 12-h light/12-h dark cycle. Mice were allowed ad libitum access to food and tap water, unless otherwise indicated. Adult males were used in all experiments. We took 0.5 g of epididymal white adipose tissue (eWAT) from B₁^−/− or wild-type (B₁^+/+) donor mice and transplanted it into the subcutaneous region of the host mice. This mass corresponds to ∼50% of the fat content of the wild-type mice at the age of 8–10 weeks. We did not normalize the mass of the transplanted tissue by mouse body weight, although the mean body weights were not statistically different at the time of the transplant. We included three experimental groups in this study: 1) B₁^+/+→B₁^−/− mice, in which B₁^−/− mice received eWAT from B₁^+/+ mice; 2) B₁^−/−→B₁^−/−, in which B₁^−/− mice received eWAT from B₁^−/− mice; and 3) B₁^+/+→B₁^+/+, in which B₁^+/+ mice received eWAT from B₁^+/+ mice. After receiving the eWAT transplant, mice recovered for 10 days before being fed the HFD. Mice were monitored during recovery, and among all the mice used in this study, only two were sacrificed because of graft rejection or infection. At sacrifice, we observed the presence of vascularization in the transplanted tissue of each mouse, confirming the efficacy of the transplantation method. Animals were sacrificed between 11:00 a.m. and 1:00 p.m. at which time we collected the WAT. All experiments were conducted as stated in the National Institutes of Health Guide for the Care and Use of Laboratory Animals (26) and approved by the institutional research committee (no. 2011/1403).

Mice (12 weeks old) receiving the transplanted fat were fed an HFD (45% kcal from fat) or a low-fat diet (10% kcal from fat) (Research Diets). Food intake and body weight were monitored weekly. Mice were deprived of food for 3 h, sacrificed, and then organs and blood were collected. Serum lipid profile was determined by using colorimetric methods (Labtest). Body composition was analyzed with DEXA through the use of a Hologic QDR 4500 scanner (23).

For glucose tolerance tests (GTT) and insulin tolerance tests (ITT), mice were deprived of food for 3 h and injected intraperitoneally (i.p.) with 1 g of glucose or 0.5 units of insulin (Humalog) per kilogram body weight. In this case, the short period without food was used in order to avoid the effect of ketone bodies (27). To measure glycemia under a prolonged period without food, we deprived mice of food for 24 h. We then measured glycemia in mice in the fed state, 1 h after refeeding. The tests were started at 11:00 a.m. Glycemia was measured by using an Accu-Chek Advantage glucometer (Roche).

We quantified insulin (Millipore) and leptin (R&D Systems) in the serum using ELISA. We quantified EGF, TNF-α, MCP-1, and platelet endothelial cell adhesion molecule-1 using Multiplex assays (Millipore), following the manufacturer’s directions.

Tissues were fixed in 10% formalin and embedded in paraffin. Sections were stained with hematoxylin-eosin. We captured images using a digital camera coupled to an optical microscope (Nikon), and we estimated the mean cell diameter using ImageJ software.

Mice were sacrificed and then both epididymal fat pads were removed, minced, and incubated in PBS containing 2% FBS, 5 mmol/L glucose, and 1.25 mg/mL collagenase at 37°C in an orbital shaker (at 150 rpm) for 15 min. After this, cells were filtered through a 100-μm nylon mesh and washed three times with PBS/2% FBS. The solution was centrifuged at 1,000g and the precipitate containing the stromal-vascular fraction (SVF) was collected.

We counted the number of cells in the SVF using a Scepter 2.0 Handheld Automated Cell Counter (Millipore). Cells were incubated for 1 h with specific antibodies: anti-CD4–Pacific Blue, anti-CD8–peridinin chlorophyll protein complex (PerCP), anti-CD25–FITC, anti-CD11b–FITC, and anti-F4/80–PerCP (BioLegend). Cells were washed, fixed, and permeabilized with Cytofix/Cytoperm (BD Biosciences), and then intracellularly stained for Foxp3 (anti-Foxp3–phycoerythrin [BioLegend]). Flow cytometry was performed by using a FACSCanto system (BD Biosciences); 20,000 events were acquired.

Total RNA was extracted by using TRIzol reagent (Life Technologies), following the manufacturer’s instructions, and then was quantified by using a Nanodrop 1000 spectrophotometer (Thermo Scientific). We used 1 μg of RNA for the cDNA reverse transcriptase reaction. We performed real-time quantitative PCR using SYBR Green reagent (Fermentas) and a 7500 Real-Time PCR System (Applied Biosystems). Oligonucleotide sequences (Exxtend Biotechnologia) are available upon request.

Values are presented as the mean ± SEM. We performed statistical analyses using one-way ANOVA followed by the Tukey posttest; we used repeated measures ANOVA followed by the Tukey posttest for analyzing weight gain. The analyses were done in GraphPad Prism 5 software. P < 0.05 was considered to represent significance.

B₁R controls WAT function and impacts whole-body glucose and lipid metabolism (22). We hypothesized that some of the effects of B₁R in adipose tissue could be cell nonautonomous. To assess that, we used an experimental model in which we placed eWAT from wild-type mice into the subcutaneous cavity of B₁^−/− mice (B₁^+/+→B₁^−/−) (Fig. 1A). After receiving the eWAT transplant, mice recovered for 10 days before being fed the HFD (45% kcal from lipids) or the control (low-fat) diet (10% kcal from fat). Mice fed the control diet did not exhibit changes in body weight gain, food intake, or organ weight (Supplementary Fig. 1). In contrast, although the body weight of mice in the control group (B₁^+/+→B₁^+/+) increased by 46.7% after they ate the HFD, homozygous knockout mice (B₁^−/−→B₁^−/−) gained significantly less weight (a 25% increase over the initial body weight) (Fig. 1B). Strikingly, wild-type eWAT transplantation promoted weight gain in B₁^−/− mice (B₁^+/+→B₁^−/− mice), matching the weight gain of the B₁^+/+→B₁^+/+ group and reverting the DIO-resistant phenotype of B₁^−/− mice (23) (Fig. 1B). Despite the body weight changes, food intake was not different among groups (Fig. 1C). Our data demonstrate that B₁R from adipose tissue controls HFD-induced weight gain via a cell-nonautonomous mechanism.

On the other hand, when fat from a knockout mouse was transplanted into a wild-type acceptor mouse (B₁^−/−→B₁^+/+), we observed no changes in HFD-induced body weight gain (Supplementary Fig. 2A). We also confirmed that B₁^−/−→B₁^+/+ mice are as insulin resistant as B₁^+/+→B₁^+/+ mice (Supplementary Fig. 2B). Thus, on the basis of these results, we decided to fully concentrate our efforts on studying the intriguing effects of the B₁^+/+→B₁^−/− transplantation model under the HFD.

To understand the nature of the changes in body mass in HFD-fed B₁^+/+→B₁^−/− mice, we measured body composition and tissue weight. Lean mass was not different among groups (Fig. 1D). The percentage of fat mass was lower in the B₁^−/−→B₁^−/− mice than in the B₁^+/+→B₁^+/+ mice, although this difference did not reach significance; B₁^+/+→B₁^−/− mice had a percentage body fat twofold higher than that in B₁^−/−→B₁^−/− mice (Fig. 1E). Bone mineral content was not different between B₁^−/−→B₁^−/− and B₁^+/+→B₁^+/+ mice, but it was significantly lower in B₁^+/+→B₁^−/− mice than in the other groups when normalized by body weight (Fig. 1F). We found the opposite when bone mineral content was not normalized by body weight (Supplementary Fig. 3). Bone mineral density was also lower in B₁^+/+→B₁^−/− mice than in the other groups (Fig. 1G), whereas organ weights (i.e., heart, kidney, and liver) were not significantly different among groups (Fig. 1H). B₁^−/−→B₁^−/− mice had larger gastrocnemius mass than mice in the B₁^+/+→B₁^+/+ group, but it did not differ from that of the B₁^+/+→B₁^−/− mice (Fig. 1H). sWAT weight was not different among groups. Strikingly, although endogenous eWAT of B₁^−/− mice receiving B₁^−/− eWAT remained smaller than that of B₁^+/+→B₁^+/+ mice, transplantation of wild-type eWAT into the subcutaneous cavity of these animals led to expansion of endogenous eWAT in B₁^−/− mice (B₁^+/+→B₁^−/−)—to an extent that matched the mass found in B₁^+/+→B₁^+/+ mice (Fig. 1H). These data demonstrate that B₁R-expressing eWAT controls its expansion in response to an HFD via a depot-specific, cell-nonautonomous mechanism.

Next, we investigated whether increased adiposity in B₁^+/+→B₁^−/− mice was associated with adipocyte hyperplasia or hypertrophy. Despite the increase in eWAT mass, the mean diameter of adipocytes in endogenous eWAT was smaller in B₁^+/+→B₁^−/− mice than in B₁^−/−→B₁^−/− and B₁^+/+→B₁^+/+ mice, suggesting expansion through adipocyte hyperplasia (Fig. 2A and Supplementary Fig. 4A). Consistent with this phenotype, genes expressed by mature adipocytes (i.e., Adipoq and Fabp4) were expressed more in endogenous eWAT of B₁^+/+→B₁^−/− mice (Fig. 2B). Similarly, expression of PPAR-γ (Pparg), a major regulator of adipogenesis, tended to be high in B₁^+/+→B₁^−/− mice (P = 0.11). The expression of the gene coding for the insulin-responsive GLUT4 was similar among groups, whereas expression of Glut1 was higher in B₁^+/+→B₁^−/− than in B₁^−/−→B₁^−/− mice, but it did not differ from that in the B₁^+/+→B₁^+/+ group. On the other hand, the mRNA expression of fatty acid synthase (Fasn) was 70% and 82% lower in eWAT in B₁^−/−→B₁^−/− and in B₁^+/+→B₁^−/−, respectively, than in B₁^+/+→B₁^+/+ mice (Fig. 2B).

Adipose tissue expansion is linked to changes in the immune cell population (16). B₁^−/−→B₁^−/− and B₁^+/+→B₁^−/− mice exhibited a trend toward a higher percentage of CD4⁺ T lymphocytes in eWAT than what was found in the B₁^+/+→B₁^+/+ group (Fig. 2C), whereas the population of CD8⁺ T lymphocytes (Fig. 2C) and T_reg (CD4⁺CD25⁺Foxp3⁺ cells) was not changed (Fig. 2D). In contrast, the macrophage population (CD11b⁺F4/80⁺ cells) trended to be larger in endogenous eWAT of B₁^+/+→B₁^−/− mice than in that of the other groups (Fig. 2E). Gene expression of the proinflammatory cytokines Ccl2 and Tnf and the anti-inflammatory cytokine Il10 in B₁^+/+→B₁^−/− mice was lower than in B₁^+/+→B₁^+/+ mice but similar to that in B₁^−/−→B₁^−/− mice (Supplementary Fig. 5A); this suggests that inflammation in the endogenous eWAT of B₁^−/− mice remains suppressed despite the transplant.

Although endogenous sWAT mass did not change upon fat transplantation, we investigated whether it exhibited changes in cellularity. Mean diameter of subcutaneous adipocytes was not different in B₁^−/−→B₁^−/− or B₁^+/+→B₁^+/+ mice, but B₁^+/+→B₁^−/− mice had smaller adipocytes than the other groups (Fig. 3A and Supplementary Fig. 4B), consistent with adipocyte hyperplasia. The differences in gene expression were milder in this adipose depot; however, we found upregulation of Fabp4 in endogenous sWAT of B₁^+/+→B₁^−/− mice, but not B₁^−/−→B₁^−/− mice, and higher expression of Pparg in B₁^−/−→B₁^−/− mice; it also tended to be higher in B₁^+/+→B₁^−/− than in B₁^+/+→B₁^+/+ mice (Fig. 3B). Differences in the immune cell population were minor (Fig. 3D and E), but the percentage of T_reg lymphocytes was 1.8-fold higher in B₁^+/+→B₁^−/− than in B₁^−/−→B₁^−/− mice (Fig. 3D). Expression of Ccl2, Il10, and Tnf was not significantly different among the groups, although, surprisingly, B₁^−/− mice had a trend toward increases in Ccl2 and Tnf expression, independent of the transplant (Supplementary Fig. 5B). Our data demonstrate that adipose B₁R alters WAT morphology, gene expression, and cellularity in cell-nonautonomous and depot-specific manners.

Besides being leaner than wild-type mice, B₁^−/− mice have lower plasma insulin levels and are more sensitive to insulin (23). Thus we investigated whether the wild-type eWAT transplant affects glucose metabolism in these mice, as it does adiposity. As expected, B₁^−/−→B₁^−/− mice had better glucose tolerance than B₁^+/+→B₁^+/+ mice (Fig. 4A and B). In contrast, despite being as obese as B₁^+/+→B₁^+/+ mice, B₁^+/+→B₁^−/− mice did not become glucose intolerant, and their blood glucose levels remained similar to those in the B₁^−/−→B₁^−/− mice during the GTT. Insulin sensitivity, as measured by the ITT, was similar between B₁^+/+→B₁^−/− and B₁^−/−→B₁^−/− mice (Fig. 4C and D).

We tested whether postprandial differences in blood glycemia existed in these mice. Glycemia after prolonged food deprivation (i.e., 24 h) was ∼40% lower in B₁^−/−→B₁^−/− mice than in B₁^+/+→B₁^+/+ or B₁^+/+→B₁^−/− mice (Fig. 4E). One hour after refeeding, all the groups exhibited the expected increase in glycemia; however, this increase was less prominent in the B₁^+/+→B₁^−/− mice (Fig. 4E). These data suggest that the presence of B₁R in adipose tissue does not control glucose tolerance or insulin sensitivity in a cell-nonautonomous way, but it contributes to glycemic control after a long period of food deprivation and subsequent refeeding.

Next, we investigated whether wild-type fat transplantation affects lipid handling and accumulation by non–adipose tissues in B₁^−/− mice. Plasma triglyceride and total cholesterol levels were not different among the groups (Supplementary Fig. 6). However, whereas B₁^−/−→B₁^−/− mice were protected from fatty liver disease, B₁^+/+→B₁^−/− mice exhibited the same extent of hepatic lipid accumulation as B₁^+/+→B₁^+/+ mice (Fig. 5A and B). Despite the increase in lipid accumulation, AST, a marker of liver damage, was not significantly higher in B₁^+/+→B₁^−/− mice than in the other groups (Fig. 5C). This phenotype was accompanied by 3.2-fold higher Fasn expression in the B₁^+/+→B₁^−/− mice than in the B₁^−/−→B₁^−/− mice (Fig. 5D). The expression of other lipogenic enzymes such as acetyl-CoA carboxylase (Acaca), glycerol-3-phosphate acyltransferase (Gpam), and stearoyl-CoA desaturase 1 (Scd1), and the gluconeogenic enzyme phosphoenolpyruvate carboxykinase (Pck1), was not different among groups (Fig. 5D). Similarly, macrophage markers (i.e., Adgre1 or F4/80, and Itgax or Cd11c) or lymphocyte markers (i.e., Cd4 and Cd8a) were not differentially expressed (Supplementary Fig. 7). Because B₁^−/− mice have higher hepatic leptin sensitivity (24), we quantified leptin receptor mRNA (Lepr), but again, we found no differences (Fig. 5D). These results demonstrate that introducing exogenous eWAT expressing B₁R induces hepatic lipid accumulation in B₁^−/− mice. These data evidence a cell-nonautonomous mechanism of B₁R-mediated adipose tissue–liver crosstalk.

To explore changes in circulating molecules associated with adipose tissue expansion and hepatic lipid accumulation after wild-type eWAT transplantation in B₁^−/− mice, we measured the levels of plasma proteins (Supplementary Figs. 6 and 8). Inflammatory cytokines (i.e., TNF-α, MCP-1, and platelet endothelial cell adhesion molecule-1) were not changed in the groups, suggesting no changes in systemic inflammation. We also measured proteins related to adipose tissue growth and lipid accumulation in the liver—that is, insulin, EGF, and leptin (9,28–30) (Fig. 6). B₁^−/−→B₁^−/− mice showed lower levels of insulin, EGF, and leptin than the B₁^+/+→B₁^+/+ group; these levels were fully or partially normalized in B₁^+/+→B₁^−/− mice, but not in the B₁^−/−→B₁^−/− group (Fig. 6). Our data demonstrate that B₁R-expressing adipose tissue controls the circulating levels of growth factors such insulin, EGF, and leptin, ultimately contributing to WAT expansion and hepatic lipid accumulation.

In most tissues, inflammatory mediators induce B₁R expression (31). In adipocytes, however, B₁R is expressed under normal physiological conditions (22). To understand how B₁R in adipose tissue contributes to the physiological response to DIO, we transplanted wild-type eWAT into the dorsal subcutaneous region of B₁R knockout (B₁^−/−) mice, generating what we named the B₁^+/+→B₁^−/− model. Wild-type eWAT transplant induces adipose tissue gain in B₁^−/− mice fed an HFD, reversing the DIO phenotype of these mice. These changes are mainly due to adipocyte hyperplasia in the endogenous WAT, supporting a cell-nonautonomous role for B₁R-expressing adipose tissue. Consistent with a cell-nonautonomous function of B₁R in adipose tissue in controlling lipid handling and anabolism in response to an HFD, B₁^+/+→B₁^−/− mice have higher hepatic lipid accumulation and higher levels of circulating growth factors than their knockout counterparts (B₁^−/−→B₁^−/− mice).

We and others have previously demonstrated the importance of B₁R in maintaining weight (23,32). Genetic ablation or pharmacological blockade of B₁R decreases HFD-induced weight gain in rodents, and this was proposed to be due to a reduction in food intake (23,33). However, genetic reconstitution of B₁R in adipose tissue of B₁^−/− mice reverses HFD-induced weight gain without altering food intake, suggesting an alternative mechanism (22). Indeed, our data demonstrate that B₁R-expressing adipose tissue promotes weight gain and adiposity independently of food intake, probably via adipocyte hyperplasia.

Distinct adipose depots have different intrinsic characteristics (3). Adipocyte progenitor cells of eWAT are less prone to differentiate in vitro and more susceptible to hypertrophy in vivo, whereas these cells in sWAT are prone to hyperplasia (3). In our model, wild-type eWAT transplantation into the subcutaneous cavity increases endogenous B₁^−/− eWAT mass; however, the adipocytes in this tissue are smaller and the whole tissue expresses more adipocyte markers such as Fabp4 and Pparg, indicating adipocyte hyperplasia. Moreover, it reduces mean adipocyte diameter in host B₁^−/− sWAT, despite normal tissue mass, again indicating more adipocytes. These data are consistent with those obtained from previously published models of adipocyte hyperplasia, such as mice treated with PPAR-γ agonists, which exhibited high adiposity associated with increased levels of fatty acid binding proteins, induced adipogenesis reflected by the high frequency of small adipocytes, and high insulin sensitivity (34). These data suggest that transplanted eWAT controls adipocyte hyperplasia in a cell-nonautonomous manner, and that the presence of B₁R in this tissue is required for the phenomenon to occur.

Typically, obese subjects with type 2 diabetes have large, insulin-resistant adipocytes (35). A widespread concept in the field is that obesity- and lipodystrophy-associated type 2 diabetes and cardiovascular diseases are the consequence of a limited reservoir for energy storage, that is, when energy availability exceeds the capacity of adipose tissue to store lipids (14). This leads, in turn, to ectopic lipid accumulation and adipose tissue dysfunction. Adipocyte hypertrophy is limited and in some cases aggravates adipose tissue dysfunction, because it creates a hypoxic environment favoring inflammation and fibrosis (7). Conversely, adipocyte hypertrophy may be beneficial and even protect against obesity-associated complications (36). Not surprisingly, antidiabetes drugs such as PPAR-γ agonists are thought to ameliorate insulin sensitivity, at least in part, by inducing adipocyte differentiation (34).

Smaller adipocytes are associated with less adipose tissue inflammation (37). In obesity, immune cells, as well as adipocytes, residing in adipose tissue contribute to installing a proinflammatory niche in the adipose tissue, producing TNF-α, IL-1β, MCP-1, and other inflammatory cytokines (11). In obesity, T_h1 CD4⁺ lymphocytes, CD8⁺ effector lymphocytes, and polarized M₁ macrophages are increased and T_h2 CD4⁺ lymphocytes and T_reg are decreased (2,11,15). The endogenous eWAT of B₁^+/+→B₁^−/− mice shows lower expression of the cytokines Tnf, Il10, and Ccl2, in addition to normal macrophage accumulation, and more CD4⁺ T lymphocytes than in B₁^+/+→B₁^+/+ mice. This increase in the number of CD4⁺ T lymphocytes and reduction in the expression of proinflammatory markers could contribute to improving insulin sensitivity in B₁^+/+→B₁^−/− mice despite increased adiposity, because the transfer of CD4⁺ T lymphocytes has been shown to reverse obesity-associated insulin resistance in lymphocyte-deficient Rag1-null mice (38). The mechanism and dynamics through which CD4⁺ T lymphocytes increase in number in host B₁^−/− mice upon wild-type eWAT transplantation is unclear; however, it might involve the presence of B₁R in transplanted eWAT-resident leukocytes. B₁R is required for proper leukocyte function (39), and activated macrophages in adipose tissue participate in CD4⁺ T lymphocyte antigen presentation (40), which could prime CD4⁺ T lymphocytes and help them to migrate to the host’s expanding adipose tissue. However, more studies are required to define whether changes in immune cell population are the cause or consequence of changes in adiposity and insulin sensitivity in B₁^+/+→B₁^−/− mice.

Despite not being expressed in mouse liver, the B₁R seems to control several aspects of liver function. For example, B₁^−/− mice exhibit less hepatic lipogenesis and are protected from HFD-induced steatosis (24). Here we demonstrate that these processes are directly or indirectly controlled by visceral WAT in a cell-nonautonomous manner. Transplantation of wild-type eWAT into B₁^−/− mice completely reverses their hepatic phenotypes, as revealed by the higher lipid accumulation and elevated fatty acid synthase gene (Fasn) expression, suggesting that hepatic de novo lipogenesis had been induced. Hepatic Fasn expression is high in nonalcoholic fatty liver disease, such as in the livers of ob/ob mice, and this feature is linked to obesity but not necessarily to insulin resistance (41). Tamoxifen treatment also leads to hepatic steatosis through inducing fatty acid synthesis (42); similar to our model, this occurs in spite of unaltered plasma triglyceride and cholesterol levels.

B₁^+/+→B₁^−/− mice have high hepatic lipid accumulation and expansion of endogenous eWAT, but not of sWAT, even though the fat is transplanted into the subcutaneous cavity; this suggests that the fat graft stimulates an endocrine response in the acceptor mice. We previously demonstrated that B₁^−/− mice have low plasma levels of several adipokines such as leptin, resistin (23), intercellular adhesion molecule-1, fetuin A, leukemia inhibitory factor, tissue inhibitor of metalloprotease-1, and oncostatin M (23,24). All of these adipokines have, at some level, been involved in hepatic or adipose tissue function, or both (24); thus they are expected to play a role in the phenotypes observed in the transplanted mice. Here we confirm the reduction in plasma leptin and insulin levels in B₁^−/− mice, and we found EGF to be lower in B₁^−/−→B₁^−/− mice than in B₁^+/+→B₁^+/+ mice. Interestingly, these changes are reversed when eWAT is transplanted to B₁^−/− mice. B₁R stimulation upregulates leptin secretion but inhibits insulin-induced leptin production (23,43); this could explain why leptin levels are higher in B₁^+/+→B₁^−/− than in B₁^−/−→B₁^−/− mice.

Changes in circulating growth factors such as EGF and insulin further support the cell-nonautonomous functions of B₁R-expressing adipose tissue. EGF is highly expressed in muscle, pancreas, and kidney (44), whereas insulin is produced almost exclusively by β-cells (45). High insulin levels in the circulation are usually associated with insulin resistance (46). Despite being obese, B₁^+/+→B₁^−/− mice are not insulin resistant or glucose intolerant, suggesting another cause for hyperinsulinemia. One potential cause is the direct modulation of β-cell function by adipokines. Adipose tissue produces adiponectin, for instance, which acts directly in the pancreas, where it is required for regenerating β-cells and protecting against lipotoxicity (47). Consistent with this notion, adiponectin (Adipoq) expression in endogenous eWAT was increased when B₁^−/− mice were transplanted with control eWAT. As in these mice, Adipoq overexpression in ob/ob mice resulted in improved insulin sensitivity, hyperlipidemia, and WAT expansion (48). Leptin, another adipokine increased in B₁^+/+→B₁^−/− (but not B₁^−/−→B₁^−/−) mice, could help mice to properly store excess calories in adipose tissue and liver, preventing lipotoxicity. Indeed, fat transplantation or administration of adiponectin, leptin, or both reversed metabolic dysfunction in lipodystrophic A-ZIP/F-1 mice (49,50).

Ours is not the only study demonstrating beneficial effects of fat transplantation (3,51). For instance, sWAT transplanted from exercised mice into sedentary mice fed an HFD had cell-nonautonomous effects, improving glucose tolerance, increasing insulin-stimulated glucose uptake in muscle and brown adipose tissue, promoting mitochondrial biogenesis in sWAT, and increasing the circulating FGF21 level (52). Whether B₁R signaling is involved in this process is a matter for future studies.

In conclusion, B₁R-expressing adipose tissue coordinates the metabolic response to DIO in a cell-nonautonomous manner, ensuring that energy is properly stored in fat by promoting adipocyte hyperplasia, and in liver by promoting lipid accumulation. These effects are associated with changes in the plasma levels of growth factors and adipokines; however, the exact molecules that mediate this mechanism remain unidentified, and this is a promising area of investigation for antiobesity therapy.

Acknowledgments. The authors thank Fernanda Malinsky, Elice Carneiro, Carlos Barros, Ana I. Silva, Yasmin C. Nunes (all at the Department of Biophysics, Universidade Federal de São Paulo [UNIFESP], São Paulo, Brazil) and Meire Hyane (Department of Immunology, Universidade de São Paulo [USP], São Paulo, Brazil) for providing great technical support.

Funding. This work was supported by grants from Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP; the São Paulo Research Foundation) (grant nos. 2010/50526-0, 2014/27198-8, 2017/07975-8, and 2017/01184-9) and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brazil.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. V.M.S., T.G.-Z., A.C., M.B., J.B., and C.H.M.C. analyzed the data. V.M.S., T.G.-Z., A.C., M.B., C.B., V.O., E.S., and C.H.M.C. performed the research. V.M.S., N.C., M.A.M., and J.B.P. designed the research. A.C., M.B., C.H.M.C., N.C., and J.B.P. contributed new reagents or analysis tools. V.M.S., M.A.M., and J.B.P. wrote the manuscript. J.B.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.