ORIGINAL ARTICLE Adipocyte Death, Adipose Tissue Remodeling, and Obesity Complications Katherine J. Strissel, Zlatina Stancheva, Hideaki Miyoshi, James W. Perfield, II, Jason DeFuria, Zoe Jick, Andrew S. Greenberg, and Martin S. Obin OBJECTIVE—We sought to determine the role of adipocyte death in obesity-induced adipose tissue (AT) inflammation and obesity complications. dipose tissue (AT) macrophages (ATM⌽s) pro- mote obesity-associated AT inflammation and RESEARCH DESIGN AND METHODS—Male C57BL/6 mice insulin resistance (1–5). Infiltrating ATM⌽s se- were fed a high-fat diet for 20 weeks to induce obesity. Every 4 crete proinflammatory mediators that are ele- weeks, insulin resistance was assessed by intraperitoneal insulin A tolerance tests, and epididymal (eAT) and inguinal subcutaneous vated in AT of obese mice and humans and are implicated AT (iAT) and livers were harvested for histological, immunohis- in the development of insulin resistance, including tumor tochemical, and gene expression analyses. necrosis factor (TNF)-␣, interleukin (IL)-6, and monocyte chemotactic protein (MCP)-1 (5–9). In obese mice and RESULTS—Frequency of adipocyte death in eAT increased humans, M⌽s infiltrate AT as circulating monocytes in from Ͻ0.1% at baseline to 16% at week 12, coincident with response to AT secretion of MCP-1, which recruits mono- increases in 1) depot weight; 2) AT macrophages (ATM⌽s) cytes expressing the C-C chemokine receptor (CCR)2 expressing F4/80 and CD11c; 3) mRNA for tumor necrosis factor (1,2,10–12,13). CCR2ϩ M⌽s expressing the M⌽ differen- (TNF)-␣, monocyte chemotactic protein (MCP)-1, and interleu- ⌽ tiation marker F4/80 and CD11c, a dendritic cell (DC) kin (IL)-10; and 4) insulin resistance. ATM s in crown-like marker (13) upregulated in M⌽ foam cells (14), were structures surrounding dead adipocytes expressed TNF-␣ and IL-6 proteins. Adipocyte number began to decline at week 12. At recently reported to infiltrate AT of obese mice and were ϳ distinguished from noninflammatory resident ATM⌽s (F4/ week 16, adipocyte death reached 80%, coincident with maxi- ϩ Ϫ Ϫ mal expression of CD11c and inflammatory genes, loss (40%) of 80 /CCR2 /CD11c ) isolated from nonobese mice eAT mass, widespread collagen deposition, and accelerated (12,15). However, the relationship between this obesity- hepatic macrosteatosis. By week 20, adipocyte number was associated ATM⌽ “phenotypic switch” (12) and the devel- restored with small adipocytes, coincident with reduced adipo- opment and progression of obesity complications is cyte death (fourfold), CD11c and MCP-1 gene expression (two- unclear (16,17). fold), and insulin resistance (35%). eAT weight did not increase at In addition to their roles in innate immunity, M⌽s week 20 and was inversely correlated with liver weight after perform vital functions in developmental and homeostatic week 12 (r ϭϪ0. 85, P Ͻ 0.001). In iAT, adipocyte death was first tissue remodeling (18,19). In AT, M⌽s promote angiogen- detected at week 12 and remained Յ3%. esis and vascular remodeling required for postnatal growth of mouse epididymal AT (eAT) (20). ATM⌽ pro- CONCLUSIONS—These results implicate depot-selective adi- pocyte death and M⌽-mediated AT remodeling in inflammatory duction of matrix-degrading proteinases is implicated in and metabolic complications of murine obesity. Diabetes 56: the matrix remodeling associated with adipocyte enlarge- 2910–2918, 2007 ment and AT expansion (21). We previously hypothesized that infiltrating ATM⌽s play an important role(s) in obe- sity-associated AT remodeling, based on our observation that ATM⌽s in obese mice and humans localize to dead adipocytes, which increase in frequency in obesity (22). At sites of adipocyte death, ATM⌽s aggregate to form a crown-like structure (CLS) that envelopes and ingests the From the Obesity and Metabolism Laboratory, Jean Mayer-U.S. Department of moribund adipocyte and its potentially cytotoxic remnant Agriculture Human Nutrition Research Center on Aging (JM-USDA HNRCA) at lipid droplet (22). As a consequence of lipid scavenging, Tufts University, Boston, Massachusetts. ATM⌽s within CLS become lipid-laden “foam cells” (22). Address correspondence and reprint requests to Martin S. Obin or Andrew ⌽ S. Greenberg, USDA, HNRCA at Tufts University, Boston, MA 02111. E-mail: M fusion within CLS is common, with multinucleate [email protected] or [email protected]. giant cells (MGCs) frequently observed (22). We propose Received for publication 5 June 2007 and accepted in revised form 27 that clearance of dead adipocytes by ATM⌽s is an initial August 2007. Published ahead of print at http://diabetes.diabetesjournals.org on 11 Sep- remodeling event required for AT repair and differentia- tember 2007. DOI: 10.2337/db07-0767. tion of new adipocytes at sites of adipocyte loss. M⌽- Additional information for this article can be found in an online appendix at mediated cell killing is a feature of various forms of tissue http://dx.doi.org/10.2337/db07-0767. ⌽ AT, adipose tissue; ATM⌽, AT macrophage; AUC, area under the curve; remodeling (18), rendering it plausible that ATM s ac- CCR, C-C chemokine receptor; CLS, crown-like structure; DC, dendritic cell; tively participate in adipocyte execution (22). DIO, diet-induced obesity; eAT, epididymal AT; HF, high fat; iAT, inguinal The clearance of dead adipocytes is likely to promote subcutaneous AT; IL, interleukin; ITT, intraperitoneal insulin tolerance test; proinflammatory ATM⌽ activation, reflecting both the LF, low fat; MCP, monocyte chemotactic protein; MGC, multinucleate giant cell; NEFA, nonesterified fatty acid; TNF, tumor necrosis factor. necrotic-like morphology of adipocyte death and ATM⌽ © 2007 by the American Diabetes Association. fusion (22). M⌽ fusion, which synergistically increases The costs of publication of this article were defrayed in part by the payment of page ⌽ ␣ charges. This article must therefore be hereby marked “advertisement” in accordance M absorptive capacity, requires TNF- autocrine/para- with 18 U.S.C. Section 1734 solely to indicate this fact. crine signaling (23), suggesting that CLS and MGCs may be 2910 DIABETES, VOL. 56, DECEMBER 2007 K.J. STRISSEL AND ASSOCIATES TABLE 1 were rat anti-mouse F4/80 (Serotec), Mac-2 (Cedarlane Labs), goat anti-mouse Primer sequences used for gene expression analysis by real-time TNF-␣, goat anti-mouse IL-6 (Santa Cruz), and rabbit anti-mouse perilipin (27). PCR Negative controls were nonimmune IgG and peptide-neutralized primary antibody. Gene target Sequences (5Ј–3Ј) Adipocyte death. Dead/dying adipocytes were identified by light microscopy as perilipin-negative lipid droplets surrounded by M⌽ crowns (22). The F4/80 F: CTTTGGCTATGGGCTTCCAGTC frequency of adipocyte death was calculated from micrographs as (number R: GCAAGGAGGACAGAGTTTATCGTG dead adipocytes/number total adipocytes) ϫ 100. CD11b F: CGGAAAGTAGTGAGAGAACTGTTTC Adipocyte volume and adipocyte number. Adipocyte volume was calcu- R: TTATAATCCAAGGGATCACCGAATTT lated from cross-sectional area obtained from perimeter tracings using Image CD11c F: CTGGATAGCCTTTCTTCTGCTG J software (28) (Sun Microsystems). Adipocyte number was calculated from fat pad weight and adipocyte volume (29) and corrected for percentage of R: GCACACTGTGTCCGAACTC dead adipocytes. CD68 F: CTTCCCACAGGCAGCACAG; Quantitative PCR. Adipose tissues (lymph nodes removed) were dissected, R: AATGATGAGAGGCAGCAAGAGG frozen in liquid nitrogen, and stored at Ϫ70°C. Total RNA was extracted, TNF-␣ F: ATGGCGTTTCCGAATTCAC quantified, and analyzed by SYBR Green real-time PCR on an Applied R: GAGGCAACCTGACCACTCTC Biosystems 7300 Real Time PCR system (30). Fold difference in gene MCP-1 F: ACTGAAGCCAGCTCTCTCTTCCTC expression was calculated as 2Ϫ⌬⌬Ct using cyclophilin B as the endogenous R: TTCCTTCTTGGGGTCAGCACAGAC control gene with mice fed the LF diet for 1 week as the “comparer” (31). IL-6 F: TCCAGTTGCCTTCTTGGGAC Genes and primer sequences are listed in Table 1. R: GTGTAATTAAGCGCCGACTTG Measures of insulin resistance. Fasted (overnight) serum insulin was measured by enzyme-linked immunosorbent assay using mouse insulin as a Adiponectin F: GAATCATTATGACGGCAGCA standard (Crystal Chem). Intraperitoneal insulin tolerance tests (ITTs) were R: TCATGTACACCGTGATGTGGTA performed on nonanesthetized mice fasted for 4–6 h in the morning. Glucose Cyclophilin B F: ATGTGGTTTTCGGCAAAGTT measures were obtained from whole–tail vein blood using an automated R: TGACATCCTTCAGTGGCTTG glucometer at baseline and 15, 30, 45, 60, and 90 min following intraperitoneal injection of human insulin (0.75 mU/kg). Glucose area under the curve F, forward; R, reverse. (AUC0–90) was determined for HF-fed and LF-fed cohorts and the difference (⌬AUC) reported. chronic sources of TNF-␣. Moreover, because each dead Biohumoral measures. Blood was obtained from fasted (overnight) mice by adipocyte “recruits” dozens of ATM⌽s, a low frequency of cardiac puncture. Serum leptin, resistin, adiponectin, and MCP-1 were mea- sured by enzyme-linked immunosorbent assay (Lincoplex mouse adipokine; adipocyte death may be sufficient to cause AT inflamma- Millipore Bioscience), and nonesterified fatty acids (NEFAs) were measured tion and promote insulin resistance. However, the proin- using the NEFA-C kit (Wako Chemicals). flammatory milieu at sites of tissue remodeling is typically Statistics. ANOVA or general linear model procedures were used in conjunc- transient, giving way to a “repair” program that promotes tion with Tukey’s honestly significant difference test (SYSTAT v10). Frequency