Effect of 8 Weeks of Overfeedingon Ectopic Fat Deposition andInsulin Sensitivity: Testing the“Adipose Tissue Expandability”HypothesisDiabetes Care 2014;37:2789–2797 | DOI: 10.2337/dc14-0761

OBJECTIVE

The presence of large subcutaneous adipocytes in obesity has been proposed tobe linked with insulin resistance and type 2 diabetes through the “adipose tissueexpandability” hypothesis, which holds that large adipocytes have a limited ca-pacity for expansion, forcing lipids to be stored in nonadipose ectopic depots(skeletal muscle, liver), where they interfere with insulin signaling. This hypoth-esis has, however, been largely formulated by cross-sectional findings and to datehas not been prospectively demonstrated in the development of insulin resistancein humans.

RESEARCH DESIGN AND METHODS

Twenty-ninemen (26.86 5.4 years old; BMI 25.56 2.3 kg/m2) were fed 40%morethan their baseline requirement for 8 weeks. Before and after overfeeding, insulinsensitivity was determined using a two-step hyperinsulinemic-euglycemic clamp.Intrahepatic lipid (IHL) and intramyocellular lipid (IMCL) were measured by1H-MRS and abdominal fat by MRI. Subcutaneous abdominal adipose and skeletalmuscle tissues were collected to measure adipocyte size and markers of tissueinflammation.

RESULTS

Subjects gained 7.6 6 2.1 kg (55% fat) and insulin sensitivity decreased 18% (P <

0.001) after overfeeding. IHL increased 46% from 1.5% to 2.2% (P = 0.002); how-ever, IMCL did not change. There was no association between adipocyte size andectopic lipid accumulation. Despite similar weight gain, subjects with smaller fatcells at baseline had a greater decrease in insulin sensitivity, whichwas linkedwithupregulated skeletal muscle tissue inflammation.

CONCLUSIONS

In experimental substantial weight gain, the presence of larger adipocytes did notpromote ectopic lipid accumulation. In contrast, smaller fat cells were associatedwith a worsened metabolic response to overfeeding.

1Pennington Biomedical Research Center, BatonRouge, LA2Charles Perkins Centre and School of BiologicalSciences, University of Sydney, Sydney, NewSouth Wales, Australia3College of Osteopathic Medicine, Touro Univer-sity California, Vallejo, CA4Department of Kinesiology, University of Texasat El Paso, El Paso, TX

Corresponding author: Eric Ravussin, ravusse@pbrc.edu.

Received 25 March 2014 and accepted 25 May2014.

Clinical trial reg. no. NCT01672632, clinicaltrials.gov.

A slide set summarizing this article is availableonline.

© 2014 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered.

Darcy L. Johannsen,1 Yourka Tchoukalova,1

Charmaine S. Tam,1,2 Jeffrey D. Covington,1

Wenting Xie,1 Jean-Marc Schwarz,3

Sudip Bajpeyi,1,4 and Eric Ravussin1

Diabetes Care Volume 37, October 2014 2789

CARDIOVASCULA

RANDMETA

BOLIC

RISK

Obesity prevalence (BMI $30 kg/m2)has increased dramatically from the1980s to 2010 (1) and is being mirroredby an increase in type 2 diabetes (T2DM)(2). Based on the “experiment of na-ture” that has happened in the UnitedStates, with the average adult gaining.10 kg over the past 30 years, Ungerand Scherer (3) proposed that the met-abolic syndrome (insulin resistance, hy-perlipidemia, elevated abdominal fat,and hypertension [4]) preceding T2DMonset occurs specifically through lipo-toxicity, whereas the state of obesityper se is of lesser importance. However,only a well-controlled study of experi-mental weight gain mimicking the routeto obesity can provide insight into thedeterminants of insulin resistance andmetabolic syndrome.The underlying mechanism linking

lipotoxic obesity with insulin resistance iswidely believed to be impaired adipogen-esis, which is manifested as the presenceof enlarged subcutaneous adipocytes (5).Larger adipocytes are thought to be closeto a hypothesized critical volume wherefurther expansion is no longer possible;therefore, excess lipid is shunted insteadto nonadipose tissues (skeletal muscle,liver, heart, and pancreas), where it inter-feres with insulin signaling and causestissue insulin resistance (reviewed bySamuel and Shulman [6]). Enlarged adi-pocytes may also secrete chemoattrac-tants or have localized hypoxia, whichtrigger macrophage infiltration and acti-vate the inflammatory process in adiposetissue, thus worsening insulin resistance(7,8). Adipose tissue containing primarilylarge adipocytes is more insulin resistant,as illustrated by reduced suppression offree fatty acid production, resulting inelevated free fatty acids (FFAs), whichcan directly activate inflammation viaToll-like receptors (9).While evidence in support of the adi-

pose expansion theory of insulin resis-tance is well documented in the animalliterature (reviewed by Virtue and Vidal-Puig [5]), less is known about the pathwayin humans. Obese people who are “met-abolically healthy” have greater adipogen-esis (smaller subcutaneous fat cells) alongwith less visceral adiposity and hepaticlipid accumulation, decreased inflamma-tion, and preserved insulin sensitivitycompared with metabolically unhealthyobese (10–12). Clinically, one of themechanisms by which thiazolidinedione

treatment works to improve insulin sensi-tivity in T2DM is by stimulating adipogen-esis and subcutaneous fat accumulation,reducing circulating and hepatic lipids,and improving insulin sensitivity despiteweight gain (13).

The question remains: Does the pres-ence of relatively smaller subcutaneousfat cells protect against the accumulationof lipid in ectopic depots and develop-ment of insulin resistance during the pe-riod of weight gain en route to obesity?To address this question, we overfedmenby 40% of their energy requirement for8 weeks to determine the influence ofbaseline adipocyte size on insulin sensitiv-ity, lipid accumulation in liver andmuscle,tissue inflammation, and other facets ofthe metabolic syndrome in response toexcess energy. According to the prevailingtheory, we hypothesized that individualswith larger fat cells would have less ca-pacity to expand their subcutaneous adi-pose tissue (SAT) andwould depositmorelipid into ectopic depots, worsening themetabolic response.

RESEARCH DESIGN AND METHODS

ParticipantsVolunteers aged 20–40 years with a BMIbetween 22.0 and 32.0 kg/m2 were eligi-ble to participate. Subjects underwentscreening tests before enrollment,including a physical exam, blood andurine analyses, and detailed medical his-tory. Those reporting a history of chronicdisease (diabetes, heart or liver disease,high blood pressure, gastrointestinal dis-order), eating disorder, recent weightloss or gain (.2.5 kg over the past 6months), or abnormal blood or urine val-ues were excluded. Subjects must nothave had a BMI .32 kg/m2 at any pointin their lifetime. The study was approvedby the Pennington Biomedical ResearchCenter (PBRC) institutional review board,and all volunteers provided written in-formed consent before participation.

DietsBefore overfeeding, participants com-pleted a 2-week measure of free-livingenergy expenditure using doubly la-beled water (DLW) (14). During the sec-ond week of DLW, participants were fedto energy balance using a publishedequation (15). The baseline energy re-quirement was calculated as the aver-age of the measured 2-week energyexpenditure by DLW and the 1-week

calorie level provided during feeding atweight maintenance, which was multi-plied by 1.4 to determine the overfeed-ing prescription. This calorie level wasinitiated on the first day of overfeedingand was maintained until the last day(day 56). All meals were prepared bythe PBRC Metabolic Kitchen using a val-idated 5-day rotating menu (16), andthe diet was composed of 41% carbohy-drate, 44% fat, and 15% protein. The fatcontent comprised a mixture of satu-rated (40%), monounsaturated (37%),and polyunsaturated (23%) fatty acids.Participants consumed all meals (3 perday, 7 days per week) in the PBRC Inpa-tient Unit under the direct supervision ofnursing staff but were free-living the re-mainder of the time. On day 56, thenumber of calories needed to maintainenergy balance at the new body weightwas calculated using the run-in equa-tion, and subjects were fed a weight-maintaining diet for 3 days beforemetabolic testing.

Measurements at Baseline and AfterOverfeeding

Overview

The study protocol is illustrated in Fig.1A. Before and after overfeeding, sub-jects were admitted to the InpatientUnit for 3 days to complete metabolictesting. On day 2, they underwent adi-pose and skeletal muscle (vastus latera-lis) biopsies under fasting conditions,followed by measurement of intrahe-patic lipid (IHL) and intramyocellularlipid (IMCL) by 1H-MRS, abdominal fatby MRI, and whole-body compositionby DXA. Insulin sensitivity was measuredon day 3 using a 2-step hyperinsulinemic-euglycemic clamp.

Fat Depots

Abdominal SAT and visceral adipose tis-sue (VAT) volumes were measuredusing a 3.0T scanner (Excite HD System;General Electric, Milwaukee, WI). Im-ages (240–340) were obtained fromthe highest point of the liver throughthe pubic symphysis and were analyzedby a single trained technician using An-alyze software (AnalyzeDirect, OverlandPark, KS). The mean coefficient of varia-tion for three readings of the same scanwas 9.9% for VAT and 1.8% for SAT. Es-timates of MRI volumes were convertedto mass using an assumed density of0.92 kg/L (17). Because of an inabilityof patients to lie in the scanner or

2790 Adipocyte Size and Overfeeding Diabetes Care Volume 37, October 2014

technical issues, data are reported for26 of the 29 subjects.Soleus and anterior tibialis IMCL and

IHL were measured by 1H-MRS using thepoint resolved spectroscopy box tech-nique (18). Three water-suppressedpoint-resolved spectroscopy voxels werecollected and peak areas of interest weredetermined by time domain fitting us-ing jMRUI (The MRUI Project) (19) anda set of information from Rico-Sanz et al.(20). Water-suppressed signals were de-convoluted with the unsuppressed wa-ter signal acquired from the same voxellocations and the resulting metabolitesignals were analyzed with AMARES, anonlinear least square fitting algorithmoperating in the time domain (21). Datawere normalized to an external oil phan-tom (peanut oil), as described by Perseghin

et al. (22), and in our hands the coeffi-cient of variation is 8.3% for IMCL and9.9% for IHL.

Percentage fat of the whole bodywas measured using DXA (QDR 4500A;Hologics, Bedford, MA), and total fatmass (FM) and fat-free mass (FFM)were calculated from the percentagebody fat and measured metabolicbody weight.

Insulin Sensitivity

Insulin sensitivity was measured using atwo-step hyperinsulinemic-euglycemicclamp. Insulin was infused for 180 minat 10 mU/min zm2 and for 150 min at 50mU/min zm2, and glucose (20%) was in-fused at a variable rate to maintainplasma glucose concentration at 90mg/dL. The average amount of glucose

required during the final 30 min of eachinsulin infusion step (glucose infusionrate [GIR]) was considered the measureof insulin sensitivity (23) and was ex-pressed per kilogram of estimated met-abolic body size (FFM + 17.7 kg) (24). In asubset of subjects (n = 19), endogenousglucose production (EGP) was deter-mined in the basal state (0.22 mmol6,6-2H2 z kg

21 z min21 for 4 h) and dur-ing insulin-stimulated conditions (1%6,6-2H2 enrichment of the 20% glucosesolution).

Circulating Biochemicals

Insulin and hs-CRP were measured byimmunoassay with chemiluminescentdetection (Siemens), and FFA concen-trations were measured by enzymaticreaction with colorimetric detection

Figure 1—A: Schematic of the study protocol. B: Average weekly weight gain from daily fasting weights (days 1–56) (left panel). Composition of thefinal weight gain as determined by DXA and fasting weight at metabolic testing (right panel).

care.diabetesjournals.org Johannsen and Associates 2791

(Wako). Adiponectin and leptin weremeasured by radioimmunoassay, andhigh–molecular weight adiponectin wasdetermined using ELISA (Millipore). Theenrichment of 6,6-2H2 glucose in plasmawas analyzed using gas chromatography–mass spectrometry (Agilent Technologies).

Adipose and Skeletal Muscle Tissue

Abdominal SAT and skeletal muscle tis-sue were collected using the techniquedescribed by Bergstrom (25), and lipo-suction was used to collect additionaladipose tissue. We chose to focus onabdominal fat because of its closer asso-ciation with metabolic risk than otheradipose depots (26). Adipocyte sizewas measured using an osmium tetrox-ide method adapted from Hirsch andGallian (27) and Pasarica et al. (28). Cellswere counted using the Multisizer 3Coulter Counter (BeckmanCoulter,Miami,FL) using a 400-mm aperture (linearrange, 12–380 mm; only cells largerthan 22 mm were included in this analy-sis), and the average of two runs for ap-proximately 1,000–3,000 cells each wascollected to calculate mean adipocytesize. Total RNA was extracted using themiRNeasy kit (Qiagen), and real-timePCR was performed using the ABI PRISM7900 (Applied Biosystems) and normal-ized to the expression of ribosomal pro-tein P0.

Statistical AnalysisStatistical analyses were performed us-ing SAS software version 9.2 (SAS Insti-tute Inc., Cary, NC). Data are presentedas mean 6 SD, with the a level set at0.05; statistical tests were two-tailed.Changes in continuous variables be-tween baseline and after overfeed-ing were analyzed by paired t test orWilcoxon signed rank test, dependingon the normality of the data. Associa-tions between subcutaneous adipocytesize at baseline, with and without ad-justing for baseline FM, and overfeeding-induced changes in metabolic parameters(fat depots, glucose infusion, circulat-ing and tissue markers) were analyzedusing linear regression models. Sub-jects also were grouped by tertile ac-cording to their baseline FM–adjustedfat cell size (FCS), and differences be-tween those in the lowest tertile(smallest fat cells) and highest tertile(largest fat cells) were analyzed by one-way ANOVA.

RESULTS

Participant CharacteristicsThirty-three men began overfeeding andfour withdrew in the first week (two be-cause of too much food and two becauseof interference with job activities); there-fore, 29 men (19 Caucasian, 10 AfricanAmerican) aged 26.8 6 5.4 years com-pleted the overfeeding protocol. Baselineenergy requirements were 3,054 6 396kcal/day, and the overfeeding prescriptionwas 4,235 6 470 kcal/day. Subjects con-sumed an excess of 66,156 6 9,960 kcalabove baseline energy requirements overthe 8-week period (range 42,476–80,052kcal). Averageweight gainwas 7.662.1 kg

and comprised 4.2 6 1.4 kg FM (55%)and 3.4 6 1.5 kg FFM (45%); BMI rosemore than two units (Table 1 andFig. 1B).

Clinical Response to OverfeedingMean changes with overfeeding areshown in Table 1. From DXA analysis,subjects gained 1.9 6 0.8 kg fat (45% oftotal fat gain) in peripheral depots(arms and legs) and 2.1 6 0.8 kg fat(50% of total) in the trunk. Within theabdomen, SAT and VAT increased by31% and 62%, respectively (seen onMRI). IHL increased by 0.7 6 2.8%(1.5–2.2%; P = 0.002); however, there

Table 1—Clinical changes with overfeeding (n = 29)

ParameterBefore

overfeedingAfter

overfeeding

AnthropometryWeight (kg) 81.9 6 10.3 89.5 6 9.4*BMI (kg/m2) 25.5 6 2.3 27.8 6 2.5*FM (kg) 16.0 6 4.8 20.2 6 5.6*Body fat (%) 19.4 6 4.9 22.3 6 5.2*FFM (kg) 65.9 6 7.3 69.4 6 7.3*Waist (cm) 84.6 6 6.6 92.0 6 7.2*Hip (cm) 98.0 6 5.4 104.1 6 7.0*Waist-to-hip ratio 0.87 6 0.04 0.89 6 0.05†

Fat depotsAbdominal SAT (kg) (n = 26) 4.1 6 1.5 5.4 6 1.8*VAT (kg) (n = 26) 0.58 6 0.49 0.94 6 0.58*IHL (% [mean 6 SEM]) 1.50 6 0.6 2.19 6 1.0†IMCL, soleus (%) 0.45 6 0.24 0.49 6 0.24

Blood chemistryTriglycerides (mg/dL) 87 6 42 96 6 68Total cholesterol (mg/dL) 171 6 25 196 6 31*HDL (mg/dL) 55 6 12 57 6 11LDL (mg/dL) 99 6 23 120 6 28*HDL-to-LDL ratio 0.59 6 0.20 0.50 6 0.16*

Leptin (ng/mL) 6.4 6 4.9 11.1 6 6.4*Adiponectin (total, mg/mL) 4.70 6 2.80 4.60 6 2.48hs-CRP (mg/L) 0.87 6 0.87 1.10 6 1.18‡Alanine transaminase (IU/L) 27.4 6 12.4 38.3 6 18.9*Fasting insulin (mU/mL) 5.4 6 4.0 8.3 6 7.8‡Fasting glucose (mg/dL) 91.0 6 6.7 92.7 6 6.9

Insulin sensitivityGlucose infusion** (mU/min z m2 insulin)10 2.87 6 0.94 2.35 6 0.7 *50 11.51 6 2.54 10.91 6 2.46§

Basal EGP (mg/min/kg) 1.69 6 0.18 1.59 6 0.17†EGP during low-dose insulin infusion (mg/min/kg) 0.46 6 0.34 0.56 6 0.27‡EGP suppression (%) 74 6 18 66 6 15†

EGP during higher-dose insulin infusion(mg/min/kg) 0.06 6 0.16 0.28 6 0.30†

EGP suppression (%) 96 6 10 82 6 20†FFAs (nmol/L)Basal 0.26 6 0.08 0.30 6 0.11§Insulin infusion (mU/[min z m2])

10 0.04 6 0.03 0.05 6 0.04†50 0.03 6 0.04 0.02 6 0.02

Data aremean6SDunless otherwise indicated. *P,0.001, †P,0.01, ‡P, 0.05, §P, 0.10. **GIR(mg/min z [FFM+17.7]).

2792 Adipocyte Size and Overfeeding Diabetes Care Volume 37, October 2014

was no change in IMCL in any musclegroup studied (MRS). Waist and hip cir-cumferences increased, as did thewaist-to-hip ratio, indicating that agreater amount of weight was depos-ited in the abdomen relative to thehips. Blood pressure (systolic, 113 6 8to 1156 11 mmHg [P = 0.26]; diastolic,73 6 7 mmHg before and after over-feeding [P = 0.95]) and heart rate(66 6 10 to 67 6 7 bpm; P = 0.30) re-mained stable.Total and LDL cholesterol (P, 0.001),

alanine transaminase (P , 0.001),hs-CRP (P = 0.04), and FFAs (P = 0.08)increased with overfeeding, demon-strating dysregulated lipid metabolism,hepatic stress, and systemic inflamma-tion, but plasma triglyceride concen-trations did not change. The diet’srelatively high fat content may have con-tributed to the altered lipid profile, par-ticularly the increase in LDL cholesteroland systemic inflammation. Concurrentwith the increase in FM, leptin levelsnearly doubled (P, 0.001) and the ratioof leptin per kilogram FM increased 45%after overfeeding (0.366 0.18 to 0.5260.20; P , 0.001), showing that leptinconcentrations increased significantlymore than could be accounted for by in-creased FM. Ghrelin decreased from675 6 240 to 617 6 200 pg/mL (P .0.001). There was no change in eithertotal or high–molecular weight adipo-nectin. Although fasting glucose re-mained unchanged, fasting insulin levelsincreased by 54% (P = 0.01) (Table 1).African American and Caucasian men re-sponded similarly to overfeeding, with nosignificant effect of race on clinical ormetabolic outcomes (data not shown).The amount of glucose required to

maintain euglycemia at 90 mg/dL de-creased by 18% (P , 0.001) duringlow-dose insulin infusion and by 5%(P = 0.052) during higher-dose insulininfusion (Table 1). Decreases in GIR oc-curred despite similar levels of insulinduring steady-state (SS) periods at base-line and after overfeeding; insulin duringthe low-dose infusion (10 mU/min z m2)SS was 13.4 6 4.5 mU/mL at baselineand 13.06 5.2 mU/mL after overfeeding(P = 0.57) and during the higher-dose in-sulin infusion (50 mU/min z m2) SS was59.06 11.3 and 60.86 14.0 mU/mL (P =0.30), respectively. The decrease in GIRafter overfeeding despite similar SSplasma insulin demonstrates the

presence of insulin resistance. Con-versely, the ability of insulin to suppressFFA release into the circulation (presum-ably coming mostly from adipose tissue)was largely unaffected by overfeeding.At baseline, FFAs during the clampwere reduced from basal levels by 85 611% during low-dose insulin infusionand by 906 15% during higher-dose in-fusion, and the percent suppression waslargely unchanged after overfeeding(low-dose insulin, 83 6 11% [P = 0.19];higher-dose insulin, 936 8% [P = 0.28]).However, basal FFA levels were slightlyhigher after overfeeding (P = 0.08),possibly reflecting mild adipose or hepa-tic resistance during the non-insulin–stimulated state.

To determine whether the insulin re-sistance was hepatic-driven or systemic,we examined results of the 19 partici-pants with complete tracer data. Atbaseline, basal EGP was 1.69 6 0.18mg/min/kg and decreased to 1.59 60.17 mg/min/kg after overfeeding (P ,0.01), likely because of higher fastingplasma insulin concentrations. Duringlow-dose insulin infusion, EGP was sup-pressed by 74 6 18% at baseline butwas reduced to 66 6 15% after

overfeeding (P = 0.01). During higher-dose insulin infusion, EGP was sup-pressed by 96 6 10% at baseline butdecreased to 82 6 20% after overfeed-ing (P = 0.002), demonstrating hepaticinsulin resistance (Table 1). Adjustingthe GIR for EGP increased the absoluteamount of glucose disposed but didnot fully account for the reduction inglucose disposal rate observed afteroverfeeding (3.57 6 0.94 to 3.03 60.64 mg/min 3 [FFM + 17.7] [P =0.001] for low-dose insulin infusion).This implies that, in addition to the liver,other tissues (namely, skeletal muscle)also became more insulin resistant.

Associations Between Adipocyte Sizeand Clinical/Metabolic Response toOverfeedingDespite similar absolute and percentweight gain, subjects with larger adipo-cytes at baseline had a smaller percentincrease in abdominal SAT (r = 20.52;P = 0.007) and VAT (r = 20.43; P =0.03) in response to overfeeding andoverall tended to have a smaller in-crease in percent body fat (r = 20.35;P = 0.06). They did not, however, accu-mulate more lipid in the liver (r = 0.15;

Table 2—Linear associations between adipocyte size (adjusted for FM) andchanges in gene expression after overfeeding (n = 29 unless otherwise indicated)

Correlation coefficient P value

Adipose tissueInflammatory markersInterleukin-6 (n = 20) 0.09 0.73p65/RelA subunit of NF-kB (n = 21) 20.03 0.89CCL2 (n = 17) 20.17 0.54CD68 (n = 24) 0.17 0.36

Extracellular matrix remodelingCollagen I (n = 23) 0.07 0.74Collagen III (n = 24) 0.09 0.66Collagen VI (n = 23) 20.09 0.67SPARC (n = 24) 0.04 0.86

AdipogenesisSREBP-1 (n = 19) 20.13 0.60PPARg (n = 21) 20.18 0.43

Skeletal muscleInflammatory markersCD40 20.56 0.002CD11c 20.37 0.05CD68 20.35 0.07p65/RelA subunit of NF-kB 20.48 0.01

Extracellular matrix remodelingCollagen I 20.43 0.02Collagen III 20.42 0.03Collagen VI 20.44 0.02SPARC 20.44 0.02

CCL2, macrophage chemoattractant protein-1; mTOR, mammalian target of rapamycin; NF,nuclear factor; SPARC, secreted protein acidic and rich in cysteine.

care.diabetesjournals.org Johannsen and Associates 2793

P = 0.43) or skeletal muscle (r = 0.08; P =0.69). Moreover, subjects with smaller,not larger, adipocytes had a greater de-crease in insulin sensitivity (DGIR at low-dose insulin: r = 0.49, P = 0.008) (Fig. 2Aand B).Subjects who had larger fat cells also

had a higher BMI (r = 0.46; P = 0.01),FM (r = 0.56; P = 0.002), and percentfat (r = 0.45; P = 0.01) to begin withcompared with those with smallercells; thus, the lesser relative expan-sion of subcutaneous fat could simplyreflect higher FM at baseline. There-fore, we repeated the analysis afteradjusting FCS for FM. Whereas the as-sociation between adipocyte size andD percent VAT lost significance, allother associations were unchanged.In addition, after adjusting FM, larger

adipocytes tended to be related to lessIHL accumulation after overfeeding(r = 20.35; P = 0.06). Of the weightgain, a larger percent was depositedas FFM (r = 0.39; P = 0.04) in subjectswith larger fat cells compared withmore as FM in those with smaller cells(r = 20.39; P = 0.04).

Associations Between Adipocyte Sizeand Tissue Molecular Markers WithOverfeedingWe recently demonstrated that markersof tissue inflammation were increased inskeletal muscle but not in abdominal adi-pose tissue after overfeeding; however,both tissues showed significant up-regulation of extracellular matrix geneexpression (29). Here, we questionedwhether adipocyte size influenced

the change in these inflammatory or ex-tracellular matrix markers in eithertissue. At baseline, expression of adipo-genic markers peroxisome proliferator–activated receptor (PPARg) (r = 0.47; P =0.03) and SREBP-1 (r = 0.55; P = 0.01) washigher in adipose tissue of subjects withlarger fat cells, but there were no rela-tionships in skeletal muscle tissue. Afteroverfeeding, adipocyte size was not as-sociated with changes in the expressionof markers of adipogenesis, inflamma-tion, or extracellular matrix remodelingin adipose tissue (Table 2). Conversely,in skeletal muscle the expression ofmarkers of inflammation (CD40, CD11c,CD68 [trend], RelA) and extracellular ma-trix remodeling (collagens I, III, V, and VI;SPARC) were upregulated in subjects whohad smaller adipocytes. Tertile analysis

Figure 2—A: Linear association between adipocyte size at baseline and the change in GIR with overfeeding with low-dose insulin infusion (10mU/min/m2), expressed as milligrams per minute z (estimated metabolic body size [EMBS]) (EMBS = FFM + 17.7). The association remainedsignificant after using Pearson correlation (r = 0.42; P = 0.03) but was attenuated after removing the subject with the largest adipocytes (r = 0.36;P = 0.06). B: The three-dimensional relationship between baseline adipocyte size and change in GIR across baseline levels of FM illustrates thatsmaller adipocytes are associated with larger decreases in GIR for any given value of FM.

2794 Adipocyte Size and Overfeeding Diabetes Care Volume 37, October 2014

showed that FCS increased by 0.50 60.38 nL (109% increase) in the lowertertile (i.e., the smallest fat cells) withoverfeeding but did not change in theupper tertile (20.01 6 0.40 nL; 1% de-crease). Compared with the uppertertile, the lower FCS tertile hada greater increase in FM (4.7 6 1.3 vs.3.6 6 1.6 kg; P = 0.09) and decreasedglucose disposal by 32% and 13% (low-and higher-dose insulin, respectively),whereas the upper tertile decreasedglucose disposal by only 9% and ,1%,respectively. Logistic regression anal-ysis revealed that those in the lower(vs. upper) tertile had an 11.6-foldgreater risk of developing a 20% orlarger reduction in glucose disposal(P = 0.03, low-dose insulin). Therewere no differences between thegroups in the change in VAT (P =0.70), IHL (P = 0.14), IMCL (soleus; P =0.35), or total weight gain (P = 0.75) inresponse to overfeeding. The larger in-crease in FM despite similar weightgain indicated that the lower FCS tertilegroup gained a larger percentage of theweight as fat (676 14% vs. 506 15% inthe upper tertile; P = 0.02) (Table 3).

CONCLUSIONS

Eight weeks of 40% overfeeding in a groupof young, healthy men caused a 7.6-kgweight gain and induced visceral and he-patic lipid deposition and mild hepatic andskeletal muscle insulin resistance and trig-gered systemic and skeletal muscle in-flammation. The purpose of the studywas to determine the influence of base-line adipocyte size on depot-specific fat

expansion and the development of in-sulin resistance among subjects. In linewith the adipose expandability theory,we found that subjects who had largeradipocytes expanded their subcutaneousfat depot significantly less; however, de-spite similar total weight gain, they didnot accumulate more lipid in the liver orskeletal muscle. Moreover, in contrast toour hypothesis and contrary to the ex-pandability theory, subjects who hadsmaller adipocytesdeven after adjustingfor baseline FMdhad a larger decline ininsulin sensitivity after overfeeding.

These results from a well-controlledprospectively designed study are in con-trast to previous observational (30) andcross-sectional reports (31). Largermean subcutaneous adipocytes wereshown to change very little in size in re-sponse to weight gain but were associ-ated with the onset of insulin resistanceand, independent of body fatness, pre-dicted future development of T2DM(30). In offspring of patients withT2DM, the size of the large adipocytefraction was associated with lower insu-lin sensitivity independent of BMI (31),suggesting that larger fat cells impair in-sulin sensitivity, even in the absence ofobesity. Conversely, in obese adolescents,a higher ratio of visceral to subcutaneousfat was associated with insulin resistance,impaired adipogenesis/lipogenesis, and asmaller proportion of large adipose cells(32). Gene expression of markers of denovo lipogenesis and lipid uptake wasfound to be lower in subjects with agreater proportion of small adipocytes; inthis sense, small fat cells were considered

detrimental because of the diminishedcapacity for fat storage and subsequentlyhigher circulating plasma FFA and depo-sition of lipid in ectopic (liver) depots.These results are more in line with ourlongitudinal findings and suggest that agreater number of small adipocytes (notlarger) and hypertrophy of adipocytesrepresent impaired adipogenesis andlead to insulin resistance. Notably, suchdiscrepancy between data from cross-sectional versus longitudinal studies isreminiscent of early studies of energymetabolism, where cross-sectional anal-ysis showed obese subjects had higherenergy expenditure than lean adults;however, when examined longitudinally,adults who became obese had lowerenergy expenditure than those who re-mained lean (33). This popular miscon-ception (that obese subjects had higherthan normal energy expenditure) led ini-tially to erroneous conclusions aboutcauses of obesity.

Mechanistically, what might be driv-ing the link between small adipocytesat baseline and adverse outcomes tooverfeeding? First, smaller adipocyteswere associated with a larger percentincrease in circulating FFAs after over-feeding (r = 20.38; P = 0.04). Adiposetissue–derived nonesterified fatty acidshave been shown to induce muscleinsulin resistance (34,35) possibly bypromoting the formation of lipid inter-mediates in skeletal muscle, includingdiacyglycerols, ceramides, and long-chain acyl-CoA, which activate kinases(protein kinase C, c-Jun-NH2-terminal ki-nase) that interfere with insulin signal-ing (36). Although we did not observe anincrease in skeletal muscle triglyceridewith overfeeding, we cannot rule outincreases in other lipid species. Thereis also evidence that FFAs may act as aligand for the Toll-like receptor 4 com-plex, thereby activating the classicalinflammatory response, including mac-rophage accumulation (9,37). Indeed,we observed a greater increase in theexpression of inflammatory factors(p65/RelA subunit of nuclear factor-kB)andmacrophagemarkers (CD40, CD11c)in the skeletal muscle of subjects withsmaller fat cells at baseline.

Second, whereas mean adipocyte sizeincreased from 0.76 6 0.40 to 1.06 60.33 nL after overfeeding (P , 0.001),smaller fat cells increased most in size inresponse to excess energy (r = 20.56;

Table 3—Changes in fat stores and glucose disposal with overfeeding amongparticipants in the lower and upper tertiles of baseline adipocyte size

FCS

P valueUpper tertile

(n = 10)Lower tertile

(n = 10)

Excess kilocalorie intake (8 weeks) 67,944 6 9,491 66,952 6 11,246 0.78

Weight gain (kg) 7.3 6 2.5 7.6 6 2.1 0.75

Weight gained as fat (%) 50 6 51 67 6 14 0.02

DFFM (kg) 3.7 6 1.5 2.9 6 1.6 0.26

DTotal FM (kg) 3.6 6 1.6 4.7 6 1.3 0.09

DAbdominal SAT (kg) 1.1 6 0.5 1.4 6 0.7 0.27

DVAT (kg) 0.3 6 0.2 0.3 6 0.2 0.70

DIHL (%) 20.4 6 2.2 0.9 6 1.3 0.14

DIMCL, soleus (%) 0.12 6 0.22 0.00 6 0.31 0.35

DGlucose disposal (mg/min z [FFM + 17.7])Low insulin 20.26 6 0.53 20.96 6 0.53 ,0.01High insulin 20.04 6 1.59 21.73 6 1.35 0.02

care.diabetesjournals.org Johannsen and Associates 2795

P , 0.001). The large increase in adi-pocyte size suggests that the smalleradipocytes had a greater capacity toexpand during overfeeding, whereasthe larger cells may have been closerto a hypothesized “critical volume” andinstead relied on adipogenesis to ac-commodate excess lipid (38), prevent-ing storage of lipid in liver or muscle.Supporting this hypothesis, the higher“adipogenic potential” at baseline insubjects with larger fat cells (as repre-sented by greater expression of PPARgand SREBP-1) may have primed theadipose tissue for differentiation/proliferation upon exposure to excessenergy, whereas those with smaller exist-ing fat cells were able to accommodatethe excess lipid by cellular hypertrophy.Indeed, an approximation of fat cellnumber using adipocyte size (nanoliters)and total subcutaneous adipose volume(liters) by MRI suggested that partici-pants with larger adipocytes had agreater increase in fat cell number withoverfeeding (r = 0.46; P = 0.02). A recentstudy of obese adolescents showedthat a higher ratio of visceral to subcu-taneous fat was associated with insulinresistance, impaired adipogenesis/lipo-genesis, and a smaller proportion oflarge adipose cells (32). Gene expressionof markers of de novo lipogenesis andlipid uptake was lower in subjects with agreater proportion of small adipocytes,and in this sense, small fat cells wereconsidered detrimental because of a di-minished capacity for fat storage andsubsequently higher circulating plasmaFFAs and deposition of lipid in ectopic(liver) depots. These results are morein line with our longitudinal findingsand suggest that a greater number ofsmall adipocytes (not larger) and hyper-trophy of adipocytes represent im-paired adipogenesis and lead to insulinresistance.Although the expansion of subcuta-

neous adipocytes to accommodate lipidoversupply is considered to be a “safe”way to store fat, the rapid enlargementof fat cells may trigger macrophage in-filtration, hypoxia, and cellular fibrosis(39). Supporting such pathologicalexpansion, we found that smaller adipo-cytes were associated with upregulatedexpression of markers of inflamma-tion and matrix remodeling, possiblyindicating fibrotic changes, but, inter-estingly, this was limited to skeletal

muscle and not detected in adipose tis-sue. We speculate that adipokines re-leased from the rapidly expanding fatcell may have interfered with insulinsensitivity in skeletal muscle throughtissue cross talk.

In conclusion, our data suggest thatamong young healthy men with no pre-existing metabolic disorder, havinglarger subcutaneous adipocytes did notpromote ectopic or visceral lipid accu-mulation during overfeeding-inducedweight gain and did not induce insulinresistance. Rather surprisingly, partici-pants with smaller fat cells had a largerreduction in insulin sensitivity despite agreater expansion of the subcutaneousadipose depot. We propose that duringweight gain toward the development ofobesity, smaller fat cells are not protec-tive and respond to excess energy byrapidly increasing in size. This rapid ex-pansion may induce insulin resistancethrough the release of adipokines thatpromote inflammation/extracellularmatrix remodeling even in nonadiposetissues.

Acknowledgments. The authors thank thededicated staff of the Pennington BiomedicalResearch Center inpatient and outpatient unitsand imaging center for their contributions.Funding. This work was funded by NationalInstitute of Diabetes and Digestive and KidneyDiseases grants K01-DK-089005 (to D.L.J.) andR01-DK-060412 and 2P30-DK-072476 (to E.R.).Duality of Interest. No potential conflicts ofinterest relevant to this article were reported.Author Contributions. D.L.J., Y.T., C.S.T., andJ.D.C. performed the experiments and wrotethe manuscript. W.X. assisted with statisticalanalysis. J.-M.S. performed the experimentsand contributed to intellectual discussion. S.B.contributed to intellectual discussion. E.R. de-signed the study and wrote the manuscript. E.R.is the guarantor of this work and, as such, hadfull access to all the data in the study and takesresponsibility for the integrity of the data andthe accuracy of the data analysis.Prior Presentation. A preliminary form of thiswork was presented as an abstract at the 72ndScientific Sessions of the American DiabetesAssociation, Philadelphia, PA, 8–12 June 2012.

References1. Flegal KM, Carroll MD, Kit BK, Ogden CL.Prevalence of obesity and trends in the distri-bution of body mass index among US adults,1999-2010. JAMA 2012;307:491–4972. Chen L, Magliano DJ, Zimmet PZ. The world-wide epidemiology of type 2 diabetes mellitusdpresent and future perspectives. Nat Rev Endocrinol2012;8:228–2363. Unger RH, Scherer PE. Gluttony, slothand the metabolic syndrome: a roadmap to

lipotoxicity. Trends Endocrinol Metab 2010;21:345–3524. Eckel RH, Grundy SM, Zimmet PZ. The meta-bolic syndrome. Lancet 2005;365:1415–14285. Virtue S, Vidal-Puig A. Adipose tissue ex-pandability, lipotoxicity and the MetabolicSyndromedan allostatic perspective. BiochimBiophys Acta 2010;1801:338–3496. Samuel VT, Shulman GI. Mechanisms for in-sulin resistance: common threads and missinglinks. Cell 2012;148:852–8717. Skurk T, Alberti-Huber C, Herder C, Hauner H.Relationship between adipocyte size and adipo-kine expression and secretion. J Clin EndocrinolMetab 2007;92:1023–10338. Le KA, Mahurkar S, Alderete TL, et al. Sub-cutaneous adipose tissue macrophage infiltra-tion is associated with hepatic and visceral fatdeposition, hyperinsulinemia, and stimulationof NF-kB stress pathway. Diabetes 2011;60:2802–28099. Shi H, KokoevaMV, Inouye K, Tzameli I, Yin H,Flier JS. TLR4 links innate immunity and fattyacid-induced insulin resistance. J Clin Invest2006;116:3015–302510. Kloting N, Fasshauer M, Dietrich A, et al.Insulin-sensitive obesity. Am J Physiol Endocri-nol Metab 2010;299:E506–E51511. Karelis AD, Faraj M, Bastard JP, et al. Themetabolically healthy but obese individualpresents a favorable inflammation profile. JClin Endocrinol Metab 2005;90:4145–415012. O’Connell J, Lynch L, Hogan A, Cawood TJ,O’Shea D. Preadipocyte factor-1 is associatedwith metabolic profile in severe obesity. J ClinEndocrinol Metab 2011;96:E680–E68413. Miyazaki Y, Mahankali A, Matsuda M, et al.Improved glycemic control and enhanced insu-lin sensitivity in type 2 diabetic subjects treatedwith pioglitazone. Diabetes Care 2001;24:710–71914. Redman LM, Heilbronn LK, Martin CK,et al.; Pennington CALERIE Team. Metabolicand behavioral compensations in responseto caloric restriction: implications for themaintenance of weight loss. PLoS ONE 2009;4:e437715. Heilbronn LK, de Jonge L, Frisard MI, et al.;Pennington CALERIE Team. Effect of 6-monthcalorie restriction on biomarkers of longevity,metabolic adaptation, and oxidative stress inoverweight individuals: a randomized con-trolled trial. JAMA 2006;295:1539–154816. Bray GA, Smith SR, de Jonge L, et al. Effect ofdietary protein content on weight gain, energyexpenditure, and body composition duringovereating: a randomized controlled trial.JAMA 2012;307:47–5517. Snyder WSCM, Nasset ES, Karhausen LR,Howells GP, Tipton IH. Report of the Task Groupon Reference Man. International Commissionon Radiological Protection publication no. 23.Oxford, United Kingdom, Elsevier, 197518. Larson-Meyer DE, Newcomer BR, HunterGR. Influence of endurance running and recov-ery diet on intramyocellular lipid content inwomen: a 1H NMR study. Am J Physiol Endocri-nol Metab 2002;282:E95–E10619. Naressi A, Couturier C, Devos JM, et al.Java-based graphical user interface for theMRUI quantitation package. MAGMA 2001;12:141–152

2796 Adipocyte Size and Overfeeding Diabetes Care Volume 37, October 2014

20. Rico-Sanz J, Thomas EL, Jenkinson G,Mierisova S, Iles R, Bell JD. Diversity in levelsof intracellular total creatine and triglyceridesin human skeletal muscles observed by (1)H-MRS. J Appl Physiol (1985) 1999;87:2068–207221. Vanhamme L, van denBoogaart A, VanHuffelS. Improved method for accurate and efficientquantification of MRS data with use of priorknowledge. J Magn Reson 1997;129:35–4322. Perseghin G, Scifo P, De Cobelli F, et al. In-tramyocellular triglyceride content is a determi-nant of in vivo insulin resistance in humans:a 1H-13C nuclear magnetic resonance spectros-copy assessment in offspring of type 2 diabeticparents. Diabetes 1999;48:1600–160623. DeFronzo RA, Tobin JD, Andres R. Glucoseclamp technique: a method for quantifying in-sulin secretion and resistance. Am J Physiol1979;237:E214–E22324. Lillioja S, Bogardus C. Obesity and insulinresistance: lessons learned from the Pima Indi-ans. Diabetes Metab Rev 1988;4:517–54025. Bergstrom J. Percutaneous needle biopsy ofskeletal muscle in physiological and clinical re-search. Scand J Clin Lab Invest 1975;35:609–61626. Kissebah AH, VydelingumN,Murray R, et al.Relation of body fat distribution to metaboliccomplications of obesity. J Clin EndocrinolMetab 1982;54:254–260

27. Hirsch J, Gallian E. Methods for the deter-mination of adipose cell size in man and ani-mals. J Lipid Res 1968;9:110–11928. Pasarica M, Xie H, Hymel D, et al. Lower totaladipocyte number but no evidence for small adi-pocyte depletion in patients with type 2 diabetes.Diabetes Care 2009;32:900–90229. Tam CS, Covington JD, Bajpeyi S, et al.Weight gain reveals dramatic increases in skel-etal muscle extracellular matrix remodeling. JClin Endocrinol Metab 2014;99:1749–175730. Weyer C, Foley JE, Bogardus C, TataranniPA, Pratley RE. Enlarged subcutaneous abdom-inal adipocyte size, but not obesity itself, predictstype II diabetes independent of insulin resistance.Diabetologia 2000;43:1498–150631. Yang J, Eliasson B, Smith U, Cushman SW,Sherman AS. The size of large adipose cells is apredictor of insulin resistance in first-degree rel-atives of type 2 diabetic patients. Obesity (SilverSpring) 2012;20:932–93832. Kursawe R, Eszlinger M, Narayan D, et al. Cel-lularity and adipogenic profile of the abdominalsubcutaneous adipose tissue from obese adoles-cents: association with insulin resistance and he-patic steatosis. Diabetes 2010;59:2288–229633. Ravussin E, Swinburn BA. Metabolicpredictors of obesity: cross-sectional versuslongitudinal data. Int J Obes Relat Metab

Disord 1993;17(Suppl. 3):S28–S31; discussionS41–S4234. Boden G, Shulman GI. Free fatty acids inobesity and type 2 diabetes: defining their rolein the development of insulin resistance andbeta-cell dysfunction. Eur J Clin Invest 2002;32(Suppl. 3):14–2335. Kraegen EW, Cooney GJ, Turner N. Muscleinsulin resistance: a case of fat overconsump-tion, not mitochondrial dysfunction. Proc NatlAcad Sci U S A 2008;105:7627–762836. Yu C, Chen Y, Cline GW, et al. Mechanism bywhich fatty acids inhibit insulin activation of in-sulin receptor substrate-1 (IRS-1)-associatedphosphatidylinositol 3-kinase activity in muscle.J Biol Chem 2002;277:50230–5023637. Varma V, Yao-Borengasser A, Rasouli N,et al. Muscle inflammatory response and insulinresistance: synergistic interaction betweenmacrophages and fatty acids leads to impairedinsulin action. Am J Physiol Endocrinol Metab2009;296:E1300–E131038. Marques BG, Hausman DB, Martin RJ. Asso-ciation of fat cell size and paracrine growth fac-tors in development of hyperplastic obesity. AmJ Physiol 1998;275:R1898–R190839. Sun K, Scherer PE. Adipose Tissue Dysfunc-tion: A Multistep Process. Berlin, Germany,Springer-Verlag Berlin Heidelberg, 2010

care.diabetesjournals.org Johannsen and Associates 2797