sign in sign up
<<
Home , Clusterin, LRP2

Diabetes Care 1

Clusterin Impairs Hepatic Insulin David Bradley,1 Alecia Blaszczak,1 Zheng Yin,2 Joey Liu,1 Joshua J. Joseph,1 Sensitivity and Adipocyte Valerie Wright,1 Kajol Anandani,1 Bradley Needleman,3 Sabrena Noria,3 Clusterin Associates With David Renton,3 Martha Yearsley,4 Stephen T.C. Wong,2 and Willa A. Hsueh1 Cardiometabolic Risk https://doi.org/10.2337/dc18-0870

OBJECTIVE Components of the adipose tissue (AT) extracellular matrix (ECM) are recently discovered contributors to obesity-related cardiometabolic . We identified increased adipocyte expression of ECM-related clusterin ( J) in obese versus lean women by microarray. Our objective was to determine 1) whether subcutaneous AT adipocyte (SAd) clusterin and serum clusterin are associated with insulin resistance (IR) and known markers of cardiometabolic risk and 2) how clusterin may contribute to increased risk.

RESEARCH DESIGN AND METHODS We validated increased clusterin expression in adipocytes from a separate group 1 of 18 lean and 54 obese individuals. The relationship of clusterin expres- Diabetes and Metabolism Research Center, Di- vision of Endocrinology, Diabetes and Metabo- sion and plasma clusterin with IR, cardiovascular biomarkers, and risk of cardio- lism, Department of Internal Medicine, Wexner vascular disease (CVD) was then determined. Further investigations in human Medical Center, The Ohio State University, Co- cultured cells and in aged LDLR2/2 mice prone to development of obesity- lumbus, OH 2 associated complications were performed. Department of Systems Medicine and Bioengi- RISK METABOLIC AND CARDIOVASCULAR neering, Institute for Academic Medicine, Hous- ton Methodist Hospital, Weill Cornell Medicine, RESULTS Houston, TX SAd clusterin correlated with IR, multiple CVD biomarkers, and CVD risk, in- 3Center for Minimally Invasive Surgery, Depart- dependent of traditional risk factors. Circulating human clusterin exhibited similar ment of General Surgery, The Ohio State Uni- associations. In human adipocytes, palmitate enhanced clusterin , and in versity Wexner Medical Center, The Ohio State University, Columbus, OH human hepatocytes clusterin attenuated insulin signaling and APOA1 expression 4Department of Pathology, Wexner Medical Cen- and stimulated hepatic gluconeogenesis. LRP2 (megalin), a clusterin , ter, The Ohio State University, Columbus, OH highly expressed in liver, mediated these effects, which were inhibited by LRP2 Corresponding author: David Bradley, david siRNA. In response to Western diet feeding, an increase in adipocyte clusterin [email protected], Willa A. Hsueh, willa expression was associated with a progressive increase in liver fat, steatohepatitis, [email protected], or Stephen T.C. Wong, [email protected] and fibrosis in aged LDLR2/2 mice. Received 20 April 2018 and accepted 19 Decem- CONCLUSIONS ber 2018 This article contains Supplementary Data online Adipocyte-derived clusterin is a novel ECM-related linking cardiometabolic at http://care.diabetesjournals.org/lookup/suppl/ disease and obesity through its actions in the liver. doi:10.2337/dc18-0870/-/DC1. D.B. and A.B. are co–first authors. Obesity is associated with a heightened cardiovascular disease (CVD) risk profile that © 2019 by the American Diabetes Association. encompasses insulin resistance (IR), type 2 diabetes, hypertension, dyslipidemia, and Readers may use this article as long as the work is d properly cited, the use is educational and not for hepatic steatosis all components of the well-described metabolic syndrome (1,2). profit, and the work is not altered. More infor- These factors act synergistically to accelerate morbidity and mortality in obesity (3). IR mation is available at http://www.diabetesjournals stems from adipocyte lipolysis and deposition into insulin target tissues, as well .org/content/license. Diabetes Care Publish Ahead of Print, published online January 18, 2019 2 Adipocyte Clusterin and Cardiometabolic Risk Diabetes Care

as a systemic inflammation instigated by (accession number GSE44000). Of Tissue Procurement, RNA Isolation, adipose tissue (AT) (4). However, iden- 428 ECM-related in the Gene Set and Protein Expression tification of specific mediators relating Enrichment Analysis (GSEA) database, Subcutaneous AT was biopsied near the obesity to IR and other CVD risk factors genes of interest were identified based umbilicus at either elective bariatric sur- has been unsuccessful. on significance, rather than fold change, gery (Roux-en-Y gastric bypass, n = 30; The adipocyte is increasingly recog- to reduce bias, since both up- and down- sleeve gastrectomy, n = 24) in obese nized as an important instigator of in- regulated genes could be included in the subjects or elective surgery (cholecystec- flammation in expanding AT (5–7). In analyses without independently assigning tomy, n = 5; hernia repair, n = 11; Heller obesity, adipocyte enlargement and hyp- cutoffs. Also, using P values allowed us to myotomy, n = 1; Nissen fundoplication, oxia enhance adipocyte production of exclude the potential impact of outliers. n = 1) in lean subjects. Patients fasted extracellular matrix (ECM) , lead- A PubMed search for the top 48 genes from midnight prior to surgery; surger- ing to the accumulation of collagens and included the gene/protein name and ies were performed from 7:00 A.M. to other matrix proteins and, ultimately, the search terms “insulin resistance” 2:00 P.M. under general anesthesia. Sam- fibrosis, which limits AT adaptability and/or “cardiovascular disease.” The ples were transferred to ice-cold saline (8–11). In addition to AT remodeling, studies were then reviewed individually and processed within 15 min of tissue these ECM proteins can enhance diverse to assess for published effects on cardio- excision. SAd and the stromal vascular functions such as AT inflammation metabolic disease. CLU was one of the fraction were isolated as previously through the influx of immune cells identified genes, and we performed gene described (7), and the adipocytes were (12,13). We found that expression of expression, correlation analyses, and blood leukocyte depleted to achieve ,0.1% clusterin was higher in adipocytes of measurements in an independent group leukocyte contamination (Supplementary obese versus lean subjects. Previous of subjects undergoing elective surgery Fig. 1). AT macrophages (ATMs) were studies have suggested that plasma clus- at The Ohio State University. then fractionated from the stromal vas- terin levels correlate with BMI, inflam- Baseline characteristics for this valida- cular fraction with biotinylated antibod- mation, IR (14), and CVD risk (15–17), and tion group of male and female subjects ies against CD14 (eBioscience, San Diego, polymorphisms are associated with without diabetes (n = 18 lean and n = CA) and streptavidin-coupled Dynabeads type 2 diabetes (18). Clusterin is a mo- 54 obese) are included in Table 1. Pre- (Thermo Fisher Scientific, Waltham, MA); lecular that assists with clear- surgery blood samples were obtained there was little contamination of the ance of cellular debris from the ECM via prior to induction of anesthesia, after ATM fraction with adipocytes or T cells and lysosomal degradation an 8-h overnight fast, and were analyzed (Supplementary Fig. 1). The relative ab- (19). Although specific clusterin recep- for fasting plasma (FPG) using the sence of adipocyte precursor cells in the tors for many metabolically active tis- Glucose Assay Kit and fasting plasma adipocyte fraction was verified by analyz- sues, including the liver, have yet to be insulin (FPI) (Millipore Sigma, Billerica, ing of the preadipocyte defined, the most well characterized is MA), (TGs), total marker protein delta homolog 1 (preadi- LDL receptor–related protein 2 (LRP2). (Tchol), and HDL and LDL cholesterol by pocyte factor 1 [Pref-1]) (Supplementary Clusterin binding to LRP2 in the CNS ELISA (Millipore Sigma, Billerica, MA). Fig. 2). affects hypothalamic feeding regulation Serum clusterin was measured by ELISA Quantitative RT-PCR (qRT-PCR) was (20) and leptin signaling (21). Our goal (R&D systems, Waltham, MA), with a performed as previously described (6). was, therefore, to determine 1) whether sensitivity of 200 pg/mL, intra-assay co- In brief, RNA was isolated using Direct-zol adipocytes could secrete clusterin, 2) efficient of variation of ,10%, and inter- RNA MiniPrep kit (Zymo Research), re- whether clusterin was associated with assay coefficient of variation of ,12%. verse transcribed with High-Capacity insulin action and components of the metabolic syndrome, and, if so, 3)how Table 1—Characteristics of the study population undergoing qRT-PCR (group 2) clusterin impacted insulin target tissues. analyses Validation study population RESEARCH DESIGN AND METHODS Lean (N = 18) Obese (N = 54) Microarray analysis was performed on n female/n male (% female) 5/13 (27) 31/23 (57) subcutaneous adipocytes (SAd) obtained Age (years) 44.9 6 12.4 44.2 6 10.4 from postmenopausal female subjects BMI (kg/m2) 23.1 6 1.6 49.7 6 8.7a (n = 7 lean and obese) undergoing elec- SBP (mmHg) 126.9 6 19.6 140.2 6 14.8a tive abdominal surgery. Subjects were DBP (mmHg) 75.8 6 10.2 78.6 6 10.6 not on hormone replacement therapy or Plasma parameters medications for hyperlipidemia, hyper- Tchol (mg/dL) 182.8 6 32.6 167.2 6 31.2 tension, or diabetes. Demographics and HDL cholesterol (mg/dL) 58.0 6 14.6 43.4 6 9.4a baseline data for group 1 and microarray LDL cholesterol (mg/dL) 104.0 6 34.5 95.3 6 29.9 6 6 technology have previously been re- TGs (mg/dL) 104.9 89.9 144.7 53.9 FPG (mg/dL) 85.5 6 13.0 114.2 6 34.7a ported (22). Raw microarray data sets FPI (mIU/mL) 7.3 6 6.0 22.6 6 14.1a are deposited in the National Center HOMA-IR 1.9 6 1.7 5.7 6 4.9a for Biotechnology Information Gene Data are means 6 SD. aP , 0.05 in lean vs. obese patients. Expression Omnibus (GEO) database care.diabetesjournals.org Bradley and Associates 3

cDNA Reverse Transcription kits (Thermo for 20 h before incubation with human protein lysates were collected for West- Fisher Scientific), and analyzed by using insulin (50 nmol/L) for 30 min. Human ern blot analysis as described above. gene-specific primer/probe sets with HepG2 cells were maintained in Eagle’s For quantitative PCR, after 48 h clusterin values normalized to PPIA expression. Minimum Essential Medium (ATCC) (26). was added directly to siRNA media and RNA was analyzed for relative expression Human adipocytes were maintained in pretreated for 4 h, and then 50 nmol/L of CLU. Gene expression values are shown adipocyte maintenance medium contain- insulin was added and cells were incu- as the fold change, defined by 22DDCT as ing DMEM. HEPES (pH 7.4), FBS, biotin, bated for 20 h; Zymo RNA Lysis Buffer recommended by the manufacturer and pantothenate, human insulin, dexameth- (Zymo Research) was added, cells were according to standard protocol (6,7,23). asone, penicillin, streptomycin, and am- collected, and qRT-PCR was carried out as Adipocyte protein levels of clusterin photericin B; stored at 220°C; and used by previously described (6). were analyzed by Western blotting of the expiration date. Human skeletal mus- visceral adipocytes (VAd) obtained from cle myoblasts were maintained in skeletal Investigation of Nonalcoholic lean (n = 8) and obese (n = 6) patients as muscle growth medium containing Steatohepatitis in Mice 2 2 previously described (5,6). DMEM, FBS, BSA, fetuin, human epider- Lean LDLR / mice were obtained from mal growth factor, dexamethasone, hu- The Jackson Laboratory (Bar Harbor, ME), Calculations man insulin, penicillin, streptomycin, and group housed under a 12-h light/dark IR was determined by HOMA of IR amphotericin B; stored at 220°C; and cycle, and fed standard chow (8904; (HOMA-IR) (24). Ten-year CVD risk was used within the expiration date. Protein Harlan Teklad). The age of all mice at calculated according to the Framingham extraction and separation, immunoblot- the time of sacrifice was ;72 weeks. Risk Score algorithm (25). ting, and visualization were performed as For induction of weight gain, steatosis, previously described (6). Phosphorylated and nonalcoholic steatohepatitis (NASH), Statistical Analyses AKT (S473 and T308) and total AKT anti- middle-age mice (52 weeks) were Normality was determined by the bodies were purchased from Cell Signal- switched from chow to a Western diet Kolmogorov-Smirnov criteria. Gene expres- ing Technology. (WD) (40% kcal fat; D12079B; Research sion and other variables between lean qRT-PCR was performed in triplicate Diets,NewBrunswick,NJ)for6,8,10,and and obese groups were compared by on samples from cultured HepG2 cells as 11 weeks prior to the study end point Student t test (normally distributed previously described (6). RNA was ana- (27). The control mice were maintained data) or Mann-Whitney U test (nonnor- lyzed for relative expression of APOA1 on chow diet throughout the duration of mally distributed data). Pearson/Spearman (), AGT (angiotensino- the study. All animal procedures were correlations were calculated to assess as- gen), PON1 (paraoxonase-1), key genes conducted in specific-pathogen-free facil- sociations. Multivariable analysis with in- regulating gluconeogenesis (PCK1 and ities at The Ohio State University Wexner dependent variables of age, sex, BMI, -2 [phospoenolpyruvate carboxykinase Medical Center and approved by The smoking status, and individual gene ex- 1 and 2], GCK [glucokinase], and PKLR Ohio State University Institutional Ani- pression values was performed in step- [pyruvate kinase isozyme R]), and SREBP- mal Care and Use Committee. At time of wise fashion with predicted adjusted 1. Relative expression of putative clus- tissue harvest, the epididymal fat pads values calculated by SPSS, version 24. terin receptors including LRP2 (megalin), and livers were removed. Liver was for- LRP8 (ApoER2), TGFBR1 (TbR1), TGFBR2 malin fixed en bloc. Histological analysis In Vitro Treatment of Cultured Human (TbR2), and VLDLR was determined in was performed to determine steatosis, Hepatocytes, Skeletal Muscle Cells, cultured HepG2 cells, skeletal muscle lobular inflammation, ballooning degen- and Adipocytes cells, and human adipocytes. eration, and fibrosis by an unbiased liver Human preadipocytes (ZenBio, Research For LRP2 knockdown experiments, pathologist (M.Y.), and NAFLD Activity Triangle Park, NC) were differentiated HepG2 cells were cultured according Score (NAS) was calculated (28). Adipo- according to the manufacturer’s instruc- to manufacturers’ protocols and pre- cytes were harvested and subjected to tions. Secreted clusterin in conditioned treated in quiescent medium (0.5% gene expression analyses (6). medium of human cultured adipocytes FBS). After reaching 70–80% confluence, was analyzed by ELISA, following treat- cells were transfected using Lipofect- RESULTS ment with palmitate (200 mmol/L), adi- amine RNAiMAX Transfection Reagent Adipocyte Gene and Protein ponectin (20 mg/mL), IFNg (2 ng/mL), (catalogue number 13778030; Invitro- Expression of Clusterin Is Upregulated interleukin (IL)-6 (50 ng/mL), IL-33 gen) with 50 nmol/L LRP2 SMARTpool in Obesity (50 ng/mL), and TGFb1 (10 ng/mL). siRNA (L-012673-00-0005; Dharmacon) Out of the top differential genes on For examination of clusterin’s effects or control nontargeting siRNA (D- microarray, we identified clusterin as on protein phosphorylation of AKT, hu- 001810-10-05; Dharmacon) in growth me- one of the proteins previously reported man liver HepG2 cells (ATCC, Manassas, dia (Eagle’s minimal essential medium, to influence cardiovascular/metabolic VA) and skeletal muscle cells and adipo- 10% FBS, penicillin, and streptomycin). For risk. Clusterin overexpression was vali- cytes (ZenBio) were cultured according phosphorylated AKT experiments, siRNA dated by qRT-PCR in an independent to manufacturers’ protocols and pre- media was removed after 48 h and re- cohort of healthy lean (n = 18) and obese treated in quiescent medium (0.5% placed with Basal Medium Eagle with 0.5% (n = 54) individuals, with obese subjects FBS) with recombinant human clusterin FBS, penicillin, streptomycin, and clusterin displaying many metabolic syndrome protein (concentrations 0.1, 0.5, and (5 mg/mL) and treated for 24 h. Insulin components but without diabetes (Table 5.0 mg/mL) (BioLegend, San Diego, CA) (50 nmol/L) was added for 30 min and 1). The obese subjects had higher BMI 4 Adipocyte Clusterin and Cardiometabolic Risk Diabetes Care

(P , 0.001), FPG (P = 0.037), FPI (P = Adipocyte Gene Expression of of palmitate (Fig. 2A and B) but not 0.004), systolic blood pressure (SBP) (P = Clusterin and Serum Clusterin Are with adiponectin or proinflammatory cy- 0.022), Tchol–to–HDL cholesterol ratio Independently Associated With tokines (IFNg, IL-6, IL-33, TGFb1 [data not (P = 0.045), and HOMA-IR (P = 0.007) Validated Biomarkers of CVD Risk shown]). These data indicate that the compared with lean participants. There SAd CLU was significantly related to TG adipocyte expresses the clusterin gene was a trend for higher TGs (P = 0.122) in (Fig. 1H) and Tchol–to–HDL cholesterol and protein and that adipocyte clusterin the obese subjects. The obese subjects ratio (Fig. 1I) and inversely associated with secretion is regulated by palmitate. had lower HDL cholesterol (P , 0.001) HDL (Fig. 1J), even after multivariable but no difference in LDL cholesterol (P = adjustment for cardiovascular risk factors Recombinant Clusterin Impairs 0.504) or Tchol (P = 0.221). CLU gene (age, sex, BMI, smoking status) (b =0.341, Hepatic Insulin Sensitivity Through expression was upregulated in SAd (fold P = 0.050). In addition, SAd CLU was the LRP2 Receptor change 2.4 6 0.2 vs. 1.2 6 0.2, P = 0.004), significantly related to SBP (Fig. 1N) To determine whether clusterin inhibits but not subcutaneous AT macrophage and diastolic blood pressure (DBP) (Fig. insulin action, we treated cultured (SATM), of obese versus lean patients 1O), even after multivariate adjustment HepG2 cells, human adipocytes, and hu- b (Fig. 1A). Similar to the observed increase ( =0.349,P = 0.039). SAd CLU was also man skeletal muscle cells with insulin in CLU gene expression (Fig. 1A), protein related to 10-year risk of CVD (composite with or without recombinant clusterin. levels of clusterin were substantially and stroke) (Fig. 1P) In HepG2 cells, clusterin treatment de- higher in SAd of obese versus lean hu- and 10-year risk of cardiovascular-related creased insulin-induced Akt phosphory- mans (P = 0.037) (Fig. 1B). mortality (r = 0.435, P = 0.013). lation (Fig. 2C) but had no effect on We next determined whether circu- skeletal muscle cells or adipocytes to lating clusterin had a similar relationship inhibit insulin action (Supplementary Adipocyte Gene Expression and Serum with cardiovascular risk. Serum clusterin Fig. 4). Clusterin also augmented genes Clusterin Are Associated With IR and positively correlated with TGs and Tchol- representing enzymatic steps in he- Impair Hepatic Insulin Sensitivity to-HDL ratio (Fig. 1K and L). When we patic gluconeogenesis, including a dose- To detect whether clusterin impacts me- accounted for age, sex, BMI, and smoking dependent increase in insulin-suppressed tabolism, we analyzed the relationship status, serum clusterin remained the only expression of GCK (converts glucose to of adipocyte and ATM clusterin mRNA significant positive predictor of these risk glucose-6-phosphate) and PKLR (upregu- expression with HOMA-IR. Only SAd CLU, factors (b = 0.440, P = 0.015, and b = lates gluconeogenesis and reduces gly- but not SATM, was significantly related 0.461, P = 0.014, respectively). We also ob- colysis/glycogen synthesis) (31) (Fig. 2D), to HOMA-IR and FPI (Fig. 1E and F), with served a negative association between further suggesting that clusterin antag- an increase in adipocyte CLU expression serum clusterin and HDL (Fig. 1M)that onizes insulin action. To identify poten- in insulin resistant (HOMA-IR $3) com- exhibited a trend after multivariable ad- tial receptor(s) responsible for the effects pared with insulin-sensitive (HOMA- justment (b = 20.318, P = 0.076). In con- of clusterin on insulin signaling, we next , 6 6 IR 3) subjects (2.5 0.3 vs. 1.4 trast, serum clusterin concentrations were measured gene expression of known 0.2, P = 0.012) (Fig. 1D). Multiple re- not associated with cardiovascular risk/ clusterin receptors in HepG2 cells, skel- gression was performed to predict mortality, SBP, or DBP. etal muscle cells, and VAd and SAd. There HOMA-IR based on adipocyte CLU ex- For understanding of how clusterin was markedly higher expression of LRP2 pression,age,sex,BMI,andsmokingsta- may affect cardiovascular risk, HepG2 (megalin) in HepG2 cells compared with tus (F(5, 3.118), P = 0.014, r = +0.654), with cells, a human hepatoma cell line, human skeletal muscle or VAd and SAd fi only CLU signi cantly adding to the pre- were treated with clusterin. In the pres- (Fig. 2E), which demonstrated higher b diction model ( = 0.492, P = 0.009). ence of insulin, clusterin decreased expression levels of LRP8, TGFBR1 and When a similar analysis was performed APOA1 expression (Fig. 2D), a major -2, and VLDLR, which are other putative using FPI(F(5, 5.957),P = 0.015,r= 0.652), component of HDL cholesterol and a clusterin receptors (Supplementary Fig. only CLU was significant (b = 0.460, P = biomarker of reduced myocardial in- 5). To determine whether LRP2 mediated 0.013). In contrast, CLU SATM gene ex- farction risk (29), without altering ex- the effects of clusterin, HepG2 cells were pression was not related to IR. We mea- pression of AGT, a regulator of blood transfected with nontargeted scrambled sured circulating clusterin levels, which pressure, or PON1, an antioxidant factor siRNA (scRNA) or LRP2-targeted siRNA. were significantly higher in obese com- in HDL (data not shown). The siRNA decreased LRP2 by 60% com- pared with lean subjects (Fig. 1C), pos- pared with scRNA in the presence of itively related to BMI (P = 0.045), and Clusterin Secretion Is Increased in clusterin, which is known to stimulate trended to rise in association with HOMA- Response to Palmitate LRP2 expression (32) (Fig. 2F). Clusterin IR (Fig. 1G), suggesting that serum To determine whether clusterin is se- inhibited insulin signaling in cells treated levels rose in parallel with SAd gene creted from adipocytes, we treated cul- with scRNA but not in samples treated expression. There was no significant dif- tured human adipocytes with factors with LRP2 siRNA (Fig. 2G). Similarly, ference based on sex in CLU adipocyte known to influence IR (30). CLU protein expression of APOA1, GCK, PCK1, and expression (females 2.0 6 1.2 vs. males levels in the cell lysates were measured SREBP1 was altered in the presence of 2.1 6 1.4; P = 0.911) or serum clusterin by Western blot, and secretion was de- scRNA but not LRP2 siRNA (Fig. 2H). (females 362,622 6 94,330 pg/mL vs. termined by ELISA of the adipocyte These data suggest that clusterin binds males 326,115 6 72,483 pg/mL; P = cell media. CLU gene expression and to LRP2 to inhibit insulin signaling and 0.112) (Supplementary Fig. 3). protein levels increased with addition altergene expressionin humanlivercells. care.diabetesjournals.org Bradley and Associates 5

Figure 1—A: SAd and SATM gene expression of CLU in lean and obese subjects. B: Clusterin protein expression by Western blotting in human lean and obese. C: Serum clusterin concentrations in lean and obese by Student t test. D: SAd fold change in CLU in patients with HOMA-IR ,3or$3 by Student t test. *P , 0.05. Multiple linear regression analyses between adipocyte gene expression of CLU byqRT-PCRvs.adjustedHOMA-IR(E)andfastinginsulin(F). G: Multiple linear regression analyses between serum clusterin concentrations (pg/mL) vs. adjusted HOMA-IR. Multiple linear regression analyses between SAd gene expression and serum clusterin by qRT-PCR vs. adjusted TGs (H and K), the ratio of Tchol to HDL level (I and L), and HDL cholesterol levels (J and M). Covariates in regression model included adipocyte CLU expression or serum clusterin and age, sex, BMI, and smoking status. Values on the y-axis are adjusted for dependent variables from the regression model. Patients on HMG-CoA reductase inhibitors (i.e., statins) or other antihyperlipidemic medications were excluded from analyses. Correlation analyses (Spearman) between adipocyte gene expression of CLU by qRT-PCR vs. SBP (N), DBP (O), and 10-year risk of cardiovascular event (P). Patients on antihypertensive medications at the time of measurement were excluded from analyses in G and H, and patients with known CVD were excluded from analyses in I. Ten-year risk score calculated according to the Framingham Risk Score algorithm (25). BP, blood pressure; yr, year.

Adipocyte Gene Expression of of SAd CLU expression, percent liver fat, WD. Serum clusterin (Fig. 3D) and adi- Clusterin Is Temporally Related to and degree of steatohepatitis in aged pocyte clusterin expression (Fig. 3E)were Hepatic Steatosis (.65 weeks old) LDLR2/2 mice fed WD, a increased in WD-fed versus chow-fed In order to determine whether SAd CLU is model of NASH (33). Body weight (Fig. mice at 6, 8, 10, and 11 weeks of WD, associated with liver fat accumulation 3A) and percent body fat (Fig. 3B) in- plateauing at 8 weeks, and tempo- and NASH, we examined the time course creased progressively over time with rally followed the same pattern as the 6 Adipocyte Clusterin and Cardiometabolic Risk Diabetes Care

Figure 2—Clusterin protein in human subcutaneous (SQ) SAd treated with palmitate (PA) (200 mmol/L) and adiponectin (APN) (20 mg/mL) assessed by Western blotting of cell lysates (A)andELISA(B) to determine the clusterin concentrations in conditioned media of the samples in A and the supernatant of the preadipocytes and mature (differentiated) adipocyte cultures without any treatment. *P , 0.05, **P , 0.01, and ***P , 0.001. C: Effect of recombinant clusterin (0.5 and 5.0 mg/mL) on insulin (INS)-stimulated (Akt) phosphorylation by Western blotting in HepG2 cells using two different phosphorylated Akt (p-Akt) antibodies. D: Effect of recombinant clusterin (0.1, 1.0, and 5.0 mg/mL) on insulin-stimulated expression care.diabetesjournals.org Bradley and Associates 7

progressive increases in percent liver fat WD-fed aging LDLR2/2 mice, the pro- mRNA decreased after sleeve gastrec- as measured by MRI (Fig. 3C), as well as gressive increase of adipocyte CLU ex- tomy (43). In obese subjects, plasma histologic features of NASH, including pression paralleled the increase in liver clusterin levels were elevated and pos- ballooning degeneration, lobular inflam- fat, hepatic fibrosis, and steatohepatitis, itively associated withBMI, inflammation mation, fibrosis, and NAS (Fig. 3H–J). In suggesting an adipocyte-clusterin-liver (hs-CRP and retinol-binding protein-4) addition, we found positive correlations relationship in both humans and mice. (16), and IR (14). Our results in hu- between adipocyte gene expression of Our findings are consistent with previ- man SAT considerably enhance these clusterin and percent body fat and steatosis ously observed elevations of serum clus- observations by demonstrating that (Fig. 3F and G). terininobesemiceandhumansand,yet, adipocyte-specific CLU expression and considerably expand upon these findings protein levels are upregulated in human to delineate a unique role for adipocyte- obesity, both CLU expression and serum CONCLUSIONS derived clusterin in propagating the clusterin positively correlate with IR, A strength of our investigation is the metabolic syndrome, IR, and CVD risk, likely and the saturated fatty acid palmitate, isolation and purification of human through its hepatic effects. known to induce IR and increased by SAd, as multiple previous genomic anal- HFD (44), is a potent stimulus for in- yses have used whole fat, which contains Adipocyte CLU (Clusterin) Expression creased adipocyte CLU expression and various mixtures of cells, making gene May Integrate Obesity, the Metabolic secretion. expression difficult to interpret. Our tran- Syndrome, and Cardiovascular Risk Most reports have confirmed an ad- scriptomics analysis of human SAd iden- Recent studies indicate that adaptive and verse effect of serum clusterin on CVD tified multiple ECM-related genes that innate immune functions of the adipo- risk in humans. Plasma clusterin levels differed between lean and obese sub- cyte regulate the overall immune cell followed a graded increase with the jects (6), but only SAd CLU (clusterin) was composition of AT, which is progressively number of metabolic syndrome compo- associated with systemic IR, multiple proinflammatory during the transition nents, associated positively with proin- components of the metabolic syndrome from a lean to obese state (35). In addition, flammatory hs-CRP (16) but negatively (HDL, Tchol–to–HDL cholesterol ratio, and as adipocytes expand, hypoxia stimulates with leptin in obesity-related CVD (45). In TGs, SBP, and DBP), and overall CVD risk. excessive and pathologic ECM production addition, clusterin had a deleterious effect Similarly, serum clusterin associated with (12). Within AT, a complex coordination on the antioxidant activity of paraoxonase- several key lipid biomarkers and trended between adipocytes and stromal cells, 1, whose deficiency reportedly enhances to associate with systemic IR. Supporting particularly ATMs (36), influences the AT (46). the gene expression findings, Western ECM (37). Although both SAd and SATM We describe herein that adipocyte blotting demonstrated increased clusterin express clusterin, SAd production appears expression of the gene CLU is directly protein levels in obese compared with to primarily influence systemic insulin ac- related to multiple high risk biomarkers lean human adipocytes. Moreover, tion and the metabolic syndrome. of CVD (TGs, Tchol–to–HDL cholesterol palmitate, a major circulating saturated Clusterin (encoded by CLU) is a mo- ratio, SBP, and DBP), as well as height- fatty acid and component of a high-fat diet lecular chaperone that assists folding of ened overall cardiovascular risk, and in- (34), enhanced clusterin expression and secreted proteins (19,38) and facilitates versely correlated with HDL. All of these secretion in cultured human adipocytes. the clearing of cellular debris and mis- associations were independent of age, Recombinant clusterin attenuated insu- folded proteins from the ECM (39). In sex, smoking status, and BMI, indicating lin signaling in liver cells, dramatically humans, there are two proteins encoded that adipocyte CLU expression may be a increased genes involved in gluconeogen- by the CLU gene: secretory CLU (sCLU) novel independent IR and CVD risk factor. esis, and decreased expression of SREBP- and nuclear CLU (nCLU) (40). Overexpres- Moreover, serum clusterin levels were an 1 and APOA1, providing several potential sion of sCLU protects cells from apopto- additional biomarker of higher TGs and mechanisms by which clusterin could sis, and a role for clusterin has previously Tchol–to–HDL cholesterol ratio and re- adversely affect systemic insulin sensi- been reported in –related duced HDL. APOA1 is a major protein tivity, promote hyperglycemia, and in- including various (41) associated with HDL cholesterol particles crease CVD risk. Thus, the liver is likely a and Alzheimer disease (42). Its role in in plasma that facilitates efflux of cho- major clusterin target, since clusterin did cardiometabolicdiseasehas alsorecently lesterol from cells, notably from macro- not affect insulin action in cultured hu- emerged. phages within atherosclerotic plaques, man adipocytes or skeletal muscle cells. In the only study of AT CLU, a 2-week to the liver for excretion. Low plasma These hepatic effects appear to be me- very-low-calorie diet reduced plasma APOA1 levels are also a strong predictor diated by LRP2 (megalin). In NASH-prone clusterin levels, and whole fat clusterin of CVD (29). Our finding of reduced hepatic

by qRT-PCR of genes regulating gluconeogenesis (GCK [glucokinase], PKLR [pyruvate kinase isozyme R], and SREBP-1) and APOA1 (apolipoprotein A1) in HepG2 cells. *P , 0.05 vs. insulin samples by one-sample t test. All experiments were completed in triplicate. E: Relative gene expression of clusterin receptor LDL receptor–related protein 2 (LRP2) (megalin) in HepG2, cultured skeletal muscle cells (SKM), and VAd and SAd. F:LRP2expressioninHepG2 cells treated with LRP2 scRNA vs. siRNA and clusterin (2 ng/mL). G: Effect of recombinant clusterin on insulin-stimulated Akt phosphorylation (pAkt) by Western blotting in HepG2 cells transfected with control nontargeting scrambled siRNA (scrambled siRNA) and LRP2-targeted siRNA. Experiments were performed in duplicate; therefore, statistical analysis was not performed. tAkt, total Akt. H: Effect of recombinant clusterin on insulin- stimulated adipocyte gene expression of LRP2, APOA1, GCK, PCK1, and SREBP1 by qRT-PCR in HepG2 cells transfected with nontargeted scRNA and LRP2-targeted siRNA. Veh, vehicle. 8 Adipocyte Clusterin and Cardiometabolic Risk Diabetes Care

Figure 3—Weight (kg) (A), percent body fat (B), percent liver fat by MRI (C), serum clusterin (D), VAd CLU gene expression (E), percent liver fat by histology (H), and ballooning degeneration, lobular inflammation, fibrosis, and NAS (I) in NASH-prone LDLR2/2 mice fed high-fat diet WD vs. normal chow (n =4)for6(n = 7), 8 (n = 5), 10 (n = 5), and 11 (n = 6) weeks (WK). Correlation analyses (Spearman) between adipocyte gene expression of CLU by qRT-PCR vs. percent body fat (F) and percent liver fat (G) by MRI. J: Representative pathologic liver specimens in LDLR2/2 mice fed WD vs. normal chow for 6, 8, 10, and 11 weeks. *P , 0.05 vs. chow. (A high-quality color representation of this figure is available in the online issue.) APOA1 expression induced by clusterin inhibiting effects were not seen in hypertriglyceridemia, and NASH (27,33). provides one potential mechanism for skeletal muscle cells or adipocytes. We now show further resemblance to increased cardiovascular risk that neces- The liver actions are mediated by binding humans, since these mice also have sitates further investigation of its impact to LRP2, which we found to be highly and increased adipocyte CLU expression, on HDL cholesterol levels. preferentially expressed in hepatocytes which is significantly associated with compared with skeletal muscle cells hepatic steatosis and parallels the time Clusterin Promotes Hepatic IR and adipocytes. The LRP2 protein is course of fibrosis and pathologic NASH Through the LRP2 Receptor critical for the reuptake of numerous scoring. We did not find a change in hepatic ligands (47) and acts to facilitate clus- Clusterin’s effects on IR appear to be LRP2 expression with progression of dis- terin-potentiated lysosomal protein deg- predominantly mediated in the liver. ease (data not shown), and human liver radation, most prominently in the Addition of recombinant clusterin to samples were not available from the study human liver cells suppressed insulin sig- hypothalamus (21) and the neurovascu- participants. Nevertheless, these observa- naling and promoted hepatic glucose lature (48). Yet, the role of hepatic tions suggest that adipocyte CLU may be production through upregulation of clusterin binding to LRP2 is poorly defined. GCK and PKLR and downregulation of related to NASH development in mice; further studies in this model, particularly SREBP-1, despite the presence of insulin. Adipocyte CLU Expression Is Related Knockdown of SREBP-1 perpetuates hy- to the Development of Hepatic focused on liver effects of clusterin, will be perglycemia via enhanced gluconeogen- Steatosis in Mice useful. These results also have implications esis and reduced glycolysis/glycogen We used an aged (65–74 weeks old) for CVD risk, since in humans there are synthesis, even though it also limits LDLR2/2 mouse model, which develops multiple epidemiologic studies linking hepatic lipogenesis (31). These insulin- obesity-related hyperinsulinemia, IR, NASH to CVD (rev. in [49]). care.diabetesjournals.org Bradley and Associates 9

This study has several limitations. De- the surgical biopsies and reviewed/edited the 15. Baralla A, Sotgiu E, Deiana M, et al. Plasma spite no significant difference in age or manuscript. M.Y. performed the histologic clusterin and lipid profile: a link with aging and BMI, there was a higher proportion of analysis and reviewed/edited the manuscript. cardiovascular diseases in a population with a S.T.C.W. additionally contributed to the design consistent number of centenarians. PLoS One males than females among lean partic- of the study. W.A.H. designed the study, ob- 2015;10:e0128029 ipants in our validation group, while tained funding, reviewed the data, contributed 16. Won JC, Park CY, Oh SW, Lee ES, Youn BS, Kim sex among obese subjects was evenly to the discussion, and reviewed/edited the MS. Plasma clusterin (ApoJ) levels are associated distributed. Even after multivariate ad- manuscript. D.B. and W.A.H. are the guarantors with adiposity and systemic inflammation. PLoS of this work and, as such, had full access to all the One 2014;9:e103351 justment including sex, however, our data in the study and take responsibility for the fi 17. Aronis KN, Vamvini MT, Chamberland JP, ndings still consistently demonstrated integrity of the data and the accuracy of the data Mantzoros CS. Circulating clusterin (apolipoprotein potentially important relationships be- analysis. J) levels do not have any day/night variability and tween clusterin, cardiometabolic risk fac- are positively associated with total and LDL cho- tors, and IR. In addition, there were no References lesterol levels in young healthy individuals. J Clin – significant differences in serum clusterin 1. Isomaa B, Almgren P, Tuomi T, et al. Cardio- Endocrinol Metab 2011;96:E1871 E1875 18. Liu XF, Yu JQ, Dalan R, Liu AQ, Luo KQ. vascular morbidity and mortality associated with or CLU expression based on sex. As a Biological factors in plasma from diabetes melli- the metabolic syndrome. Diabetes Care 2001;24: cross-sectional study, many of our find- tus patients enhance hyperglycaemia and pul- 683–689 satile shear stress-induced endothelial cell ings are correlative and do not directly 2. Kirk EP, Klein S. Pathogenesis and pathophys- . Integr Biol (Camb) 2014;6:511–522 imply cause and effect. However, our iology of the cardiometabolic syndrome. J Clin 19. Wyatt A, Yerbury J, Poon S, Dabbs R, Wilson data in human cultured cells treated with Hypertens (Greenwich) 2009;11:761–765 M. Chapter 6: the chaperone action of Clusterin 3. DeFronzo RA, Ferrannini E. Insulin resistance. recombinant clusterin support a possi- and its putative role in quality control of extra- A multifaceted syndrome responsible for NIDDM, ble mechanistic role for clusterin in lipid cellular . Adv Res 2009; obesity, hypertension, dyslipidemia, and athero- 104:89–114 and glucose homeostasis. sclerotic cardiovascular disease. Diabetes Care 20. Gil SY, Youn BS, Byun K, et al. Clusterin and In conclusion, our study highlights that 1991;14:173–194 LRP2 are critical components of the hypotha- adipocyte-derived clusterin may be an 4. Glass CK, Olefsky JM. Inflammation and lipid lamic feeding regulatory pathway [published signaling in the etiology of insulin resistance. Cell important protein that integrates obesity correction appears in Nat Commun 2013;4: Metab 2012;15:635–645 with the metabolic syndrome and car- 2912]. Nat Commun 2013;4:1862 5. Graßmann S, Wirsching J, Eichelmann F, 21. Byun K, Gil SY, Namkoong C, et al. Clusterin/ diovascular risk/mortality. HFD in both Aleksandrova K. Association between peripheral ApoJ enhances central leptin signaling through mice and humans increases palmitate, fl adipokines and in ammation markers: a system- Lrp2-mediated endocytosis. EMBO Rep 2014;15: which stimulates adipocyte production atic review and meta-analysis. Obesity (Silver 801–808 – of clusterin, leading to enhanced circu- Spring) 2017;25:1776 1785 22. Yin Z, Deng T, Peterson LE, et al. Transcrip- lating levels. Clusterin binding to the 6. Deng T, Liu J, Deng Y, et al. Adipocyte adaptive tome analysis of human adipocytes implicates fl hepatic LRP2 receptor then could inhibit immunity mediates diet-induced adipose in am- the NOD-like receptor pathway in obesity- mation and insulin resistance by decreasing insulin signaling to promote IR and induced adipose inflammation. Mol Cell Endo- adipose Treg cells. Nat Commun 2017;8:15725 crinol 2014;394:80–87 gluconeogenesis; decrease APOA1 ex- 7. Deng T, Lyon CJ, Minze LJ, et al. Class II major 23. Livak KJ, Schmittgen TD. Analysis of relative pression, leading to decreased HDL cho- histocompatibility complex plays an essential gene expression data using real-time quantita- fl lesterol; and enhance fatty liver/NASH. role in obesity-induced adipose in ammation. tive PCR and the 2(-Delta Delta C(T)) Method. – Further investigation is needed to de- Cell Metab 2013;17:411 422 Methods 2001;25:402–408 8. Baker NA, Muir LA, Washabaugh AR, et al. 24. Matthews DR, Hosker JP, Rudenski AS, termine whether adipocytes communi- fi Diabetes-speci c regulation of adipocyte metab- Naylor BA, Treacher DF, Turner RC. Homeostasis cate with the liver through clusterin to olism by the adipose tissue extracellular matrix. model assessment: insulin resistance and beta- instigate the metabolic syndrome; clus- J Clin Endocrinol Metab 2017;102:1032–1043 cell function from fasting plasma glucose and terin is one of the first ECM factors 9. Sun K, Tordjman J, Clement´ K, Scherer PE. insulin concentrations in man. Diabetologia linking the enlarging adipocyte in hu- Fibrosis and adipose tissue dysfunction. Cell 1985;28:412–419 Metab 2013;18:470–477 man obesity to cardiometabolic disease. 25. D’Agostino RB Sr., Vasan RS, Pencina MJ, 10. Lin, Chun TH, Kang L. Adipose extracellular et al. General cardiovascular risk profile for use in matrix remodelling in obesity and insulin re- primary care: the Framingham Heart Study. sistance. Biochem Pharmacol 2016;119:8–16 Circulation 2008;117:743–753 Acknowledgments. The authors thank the 11. Lackey DE, Burk DH, Ali MR, et al. Contri- 26. Lopez-Terrada´ D, Cheung SW, Finegold MJ, study subjects for their participation. butions of adipose tissue architectural and Knowles BB. Hep G2 is a hepatoblastoma-derived fi Funding. This study was supported by American tensile properties toward de ning healthy and cell line. Hum Pathol 2009;40:1512–1515 Diabetes Association grant 1-16-ICTS-049, Na- unhealthy obesity. Am J Physiol Endocrinol 27. Collins AR, Lyon CJ, Xia X, et al. Age-accelerated tional Institutes of Health KL2 Scholar Program Metab 2014;306:E233–E246 atherosclerosis correlates with failure to up- award KL2TR001068 and grant HL135622, 12. Crewe C, An YA, Scherer PE. The ominous regulate antioxidant genes. Circ Res 2009; John S. Dunn Research Foundation, and Ting triad of adipose tissue dysfunction: inflamma- 104:e42–e54 Tsung and Wei Fong Chao Foundation. tion, fibrosis, and impaired . J Clin 28. Brunt EM, Kleiner DE, Wilson LA, Belt P, Duality of Interest. No potential conflicts of Invest 2017;127:74–82 Neuschwander-Tetri BA; NASH Clinical Research interest relevant to this article were reported. 13. Iyengar P, Combs TP, Shah SJ, et al. Adipo- Network (CRN). Nonalcoholic fatty liver disease Author Contributions. D.B. and A.B. assisted in cyte-secreted factors synergistically promote (NAFLD) activity score and the histopathologic designing the study, collated and analyzed data, mammary tumorigenesis through induction of diagnosis in NAFLD: distinct clinicopathologic and wrote the manuscript. Z.Y., V.W., K.A., and anti-apoptotic transcriptional programs and meanings. Hepatology 2011;53:810–820 S.T.C.W. assisted with the experiments, per- proto-oncogene stabilization. Oncogene 2003; 29. McQueen MJ, Hawken S, Wang X, et al.; formed data analysis, contributed to the discus- 22:6408–6423 INTERHEART study investigators. , lipopro- sion, and reviewed/edited the manuscript. J.L. 14. Flehmig G, Scholz M, Kloting¨ N, et al. Iden- teins, and as risk markers assisted with the experiments and reviewed/ tification of adipokine clusters related to para- of myocardial infarction in 52 countries (the edited the manuscript. J.J.J. reviewed/edited meters of fat mass, insulin sensitivity and INTERHEART study): a case-control study. Lancet the manuscript. B.N., S.N., and D.R. performed inflammation. PLoS One 2014;9:e99785 2008;372:224–233 10 Adipocyte Clusterin and Cardiometabolic Risk Diabetes Care

30. Petersen MC, Shulman GI. Roles of diacyl- (MMPs) as key genes whose expression is hyperinsulinemia, very-low-calorie diet and bari- glycerols and ceramides in hepatic insulin resis- up-regulated in human adipocytes by macro- atric surgery. Physiol Res 2016;65:481–492 tance. Trends Pharmacol Sci 2017;38:649–665 phage-conditioned medium. Pflugers Arch 44. Lee JY, Cho HK, Kwon YH. Palmitate induces 31. Ruiz R, Jideonwo V, Ahn M, et al. Sterol 2009;458:1103–1114 insulin resistance without significant intracellular regulatory element-binding protein-1 (SREBP-1) 38. HumphreysDT, Carver JA, Easterbrook-Smith accumulation in HepG2 cells. Me- is required to regulate glycogen synthesis and SB, Wilson MR. Clusterin has chaperone-like ac- tabolism 2010;59:927–934 gluconeogenic gene expression in mouse liver. J tivity similar to that of small heat shock proteins. J 45. Poulakou MV, Paraskevas KI, Wilson MR, Biol Chem 2014;289:5510–5517 Biol Chem 1999;274:6875–6881 et al. Apolipoprotein J and leptin levels in pa- 32. Ammar H, Closset JL. Clusterin activates 39. Wyatt AR, Yerbury JJ, Berghofer P, et al. survival through the phosphatidylinositol Clusterin facilitates in vivo clearance of extra- tients with coronary heart disease. In Vivo 2008; 3-kinase/Akt pathway. J Biol Chem 2008;283: cellular misfolded proteins. Cell Mol Life Sci 22:537–542 12851–12861 2011;68:3919–3931 46. Navab M, Hama-Levy S, Van Lenten BJ, et al. 33. Gupte AA, Liu JZ, Ren Y, et al. Rosiglitazone 40. Leskov KS, Araki S, Lavik JP, et al. CRM1 Mildly oxidized LDL induces an increased apo- attenuatesage-anddiet-associatednonalcoholic protein-mediated regulation of nuclear clusterin J/paraoxonase ratio. J Clin Invest steatohepatitis in male low-density lipoprotein (nCLU), an ionizing radiation-stimulated, Bax- 1997;99:2005–2019 receptor knockout mice. Hepatology 2010;52: dependent pro-death factor. J Biol Chem 2011; 47. Marzolo MP, Farfan´ P. New insights into the 2001–2011 286:40083–40090 roles of megalin/LRP2 and the regulation of its 34. Sears B, Perry M. The role of fatty acids in 41. Koltai T. Clusterin: a key player in cancer functional expression. Biol Res 2011;44:89–105 insulin resistance. Lipids Health Dis 2015;14:121 chemoresistance and its inhibition. Onco Targets 48. Nelson AR, Sagare AP, Zlokovic BV. Role – 35. Ip BC, Hogan AE, Nikolajczyk BS. Lymphocyte Ther 2014;7:447 456 of clusterin in the brain vascular clearance of roles in metabolic dysfunction: of men and mice. 42. LiX,Ma Y, Wei X, etal.Clusterinin Alzheimer’s amyloid-b. Proc Natl Acad Sci U S A 2017;114: Trends Endocrinol Metab 2015;26:91–100 disease: a player in the biological behavior of – 36. Martinez-Santibanez~ G, Lumeng CN. Macro- amyloid-beta. Neurosci Bull 2014;30:162–168 8681 8682 phages and the regulation of adipose tissue 43. Klouckovˇ a´ J, Lacinova´ Z, Kavalkov´ a´ P, et al. 49. Lonardo A, Nascimbeni F, Mantovani A, remodeling. Annu Rev Nutr 2014;34:57–76 Plasma concentrations and subcutaneous adipose Targher G. Hypertension, diabetes, atheroscle- 37. O’Hara A, Lim FL, Mazzatti DJ, Trayhurn P. Micro- tissue mRNA expression of clusterin in obesity and rosis and NASH: cause or consequence? J Hepatol array analysis identifies matrix metalloproteinases type 2 diabetes mellitus: the effect of short-term 2018;68:335–352